diff options
Diffstat (limited to 'parse.y')
| -rw-r--r-- | parse.y | 10748 |
1 files changed, 6155 insertions, 4593 deletions
@@ -19,16 +19,28 @@ #define YYDEBUG 1 #define YYERROR_VERBOSE 1 #define YYSTACK_USE_ALLOCA 0 -#define YYLTYPE rb_code_location_t -#define YYLTYPE_IS_DECLARED 1 + +/* For Ripper */ +#ifdef RUBY_EXTCONF_H +# include RUBY_EXTCONF_H +#endif #include "ruby/internal/config.h" -#include <ctype.h> #include <errno.h> -#include <stdio.h> -struct lex_context; +#ifdef UNIVERSAL_PARSER + +#include "internal/ruby_parser.h" +#include "parser_node.h" +#include "universal_parser.c" + +#ifdef RIPPER +#define STATIC_ID2SYM p->config->static_id2sym +#define rb_str_coderange_scan_restartable p->config->str_coderange_scan_restartable +#endif + +#else #include "internal.h" #include "internal/compile.h" @@ -37,16 +49,17 @@ struct lex_context; #include "internal/encoding.h" #include "internal/error.h" #include "internal/hash.h" -#include "internal/imemo.h" #include "internal/io.h" #include "internal/numeric.h" #include "internal/parse.h" #include "internal/rational.h" #include "internal/re.h" +#include "internal/ruby_parser.h" #include "internal/symbol.h" #include "internal/thread.h" #include "internal/variable.h" #include "node.h" +#include "parser_node.h" #include "probes.h" #include "regenc.h" #include "ruby/encoding.h" @@ -57,11 +70,241 @@ struct lex_context; #include "ruby/ractor.h" #include "symbol.h" -enum shareability { - shareable_none, - shareable_literal, - shareable_copy, - shareable_everything, +#ifndef RIPPER +static VALUE +syntax_error_new(void) +{ + return rb_class_new_instance(0, 0, rb_eSyntaxError); +} +#endif + +static NODE *reg_named_capture_assign(struct parser_params* p, VALUE regexp, const YYLTYPE *loc); + +#define compile_callback rb_suppress_tracing +#endif /* !UNIVERSAL_PARSER */ + +#define NODE_SPECIAL_EMPTY_ARGS ((NODE *)-1) +#define NODE_EMPTY_ARGS_P(node) ((node) == NODE_SPECIAL_EMPTY_ARGS) + +static int rb_parser_string_hash_cmp(rb_parser_string_t *str1, rb_parser_string_t *str2); +static rb_parser_string_t *rb_parser_string_deep_copy(struct parser_params *p, const rb_parser_string_t *original); + +static int +node_integer_cmp(rb_node_integer_t *n1, rb_node_integer_t *n2) +{ + return (n1->minus != n2->minus || + n1->base != n2->base || + strcmp(n1->val, n2->val)); +} + +static int +node_float_cmp(rb_node_float_t *n1, rb_node_float_t *n2) +{ + return (n1->minus != n2->minus || + strcmp(n1->val, n2->val)); +} + +static int +node_rational_cmp(rb_node_rational_t *n1, rb_node_rational_t *n2) +{ + return (n1->minus != n2->minus || + n1->base != n2->base || + n1->seen_point != n2->seen_point || + strcmp(n1->val, n2->val)); +} + +static int +node_imaginary_cmp(rb_node_imaginary_t *n1, rb_node_imaginary_t *n2) +{ + return (n1->minus != n2->minus || + n1->base != n2->base || + n1->seen_point != n2->seen_point || + n1->type != n2->type || + strcmp(n1->val, n2->val)); +} + +static int +rb_parser_regx_hash_cmp(rb_node_regx_t *n1, rb_node_regx_t *n2) +{ + return (n1->options != n2->options || + rb_parser_string_hash_cmp(n1->string, n2->string)); +} + +static st_index_t rb_parser_str_hash(rb_parser_string_t *str); +static st_index_t rb_char_p_hash(const char *c); + +static int +literal_cmp(st_data_t val, st_data_t lit) +{ + if (val == lit) return 0; + + NODE *node_val = RNODE(val); + NODE *node_lit = RNODE(lit); + enum node_type type_val = nd_type(node_val); + enum node_type type_lit = nd_type(node_lit); + + if (type_val != type_lit) { + return -1; + } + + switch (type_lit) { + case NODE_INTEGER: + return node_integer_cmp(RNODE_INTEGER(node_val), RNODE_INTEGER(node_lit)); + case NODE_FLOAT: + return node_float_cmp(RNODE_FLOAT(node_val), RNODE_FLOAT(node_lit)); + case NODE_RATIONAL: + return node_rational_cmp(RNODE_RATIONAL(node_val), RNODE_RATIONAL(node_lit)); + case NODE_IMAGINARY: + return node_imaginary_cmp(RNODE_IMAGINARY(node_val), RNODE_IMAGINARY(node_lit)); + case NODE_STR: + return rb_parser_string_hash_cmp(RNODE_STR(node_val)->string, RNODE_STR(node_lit)->string); + case NODE_SYM: + return rb_parser_string_hash_cmp(RNODE_SYM(node_val)->string, RNODE_SYM(node_lit)->string); + case NODE_REGX: + return rb_parser_regx_hash_cmp(RNODE_REGX(node_val), RNODE_REGX(node_lit)); + case NODE_LINE: + return node_val->nd_loc.beg_pos.lineno != node_lit->nd_loc.beg_pos.lineno; + case NODE_FILE: + return rb_parser_string_hash_cmp(RNODE_FILE(node_val)->path, RNODE_FILE(node_lit)->path); + case NODE_ENCODING: + return RNODE_ENCODING(node_val)->enc != RNODE_ENCODING(node_lit)->enc; + default: +#ifdef UNIVERSAL_PARSER + abort(); +#else + rb_bug("unexpected node: %s, %s", ruby_node_name(type_val), ruby_node_name(type_lit)); +#endif + } +} + +static st_index_t +literal_hash(st_data_t a) +{ + NODE *node = (NODE *)a; + enum node_type type = nd_type(node); + + switch (type) { + case NODE_INTEGER: + return rb_char_p_hash(RNODE_INTEGER(node)->val); + case NODE_FLOAT: + return rb_char_p_hash(RNODE_FLOAT(node)->val); + case NODE_RATIONAL: + return rb_char_p_hash(RNODE_RATIONAL(node)->val); + case NODE_IMAGINARY: + return rb_char_p_hash(RNODE_IMAGINARY(node)->val); + case NODE_STR: + return rb_parser_str_hash(RNODE_STR(node)->string); + case NODE_SYM: + return rb_parser_str_hash(RNODE_SYM(node)->string); + case NODE_REGX: + return rb_parser_str_hash(RNODE_REGX(node)->string); + case NODE_LINE: + return (st_index_t)node->nd_loc.beg_pos.lineno; + case NODE_FILE: + return rb_parser_str_hash(RNODE_FILE(node)->path); + case NODE_ENCODING: + return (st_index_t)RNODE_ENCODING(node)->enc; + default: +#ifdef UNIVERSAL_PARSER + abort(); +#else + rb_bug("unexpected node: %s", ruby_node_name(type)); +#endif + } +} + +static inline int +parse_isascii(int c) +{ + return '\0' <= c && c <= '\x7f'; +} + +#undef ISASCII +#define ISASCII parse_isascii + +static inline int +parse_isspace(int c) +{ + return c == ' ' || ('\t' <= c && c <= '\r'); +} + +#undef ISSPACE +#define ISSPACE parse_isspace + +static inline int +parse_iscntrl(int c) +{ + return ('\0' <= c && c < ' ') || c == '\x7f'; +} + +#undef ISCNTRL +#define ISCNTRL(c) parse_iscntrl(c) + +static inline int +parse_isupper(int c) +{ + return 'A' <= c && c <= 'Z'; +} + +static inline int +parse_islower(int c) +{ + return 'a' <= c && c <= 'z'; +} + +static inline int +parse_isalpha(int c) +{ + return parse_isupper(c) || parse_islower(c); +} + +#undef ISALPHA +#define ISALPHA(c) parse_isalpha(c) + +static inline int +parse_isdigit(int c) +{ + return '0' <= c && c <= '9'; +} + +#undef ISDIGIT +#define ISDIGIT(c) parse_isdigit(c) + +static inline int +parse_isalnum(int c) +{ + return parse_isalpha(c) || parse_isdigit(c); +} + +#undef ISALNUM +#define ISALNUM(c) parse_isalnum(c) + +static inline int +parse_isxdigit(int c) +{ + return parse_isdigit(c) || ('A' <= c && c <= 'F') || ('a' <= c && c <= 'f'); +} + +#undef ISXDIGIT +#define ISXDIGIT(c) parse_isxdigit(c) + +#include "parser_st.h" + +#undef STRCASECMP +#define STRCASECMP rb_parser_st_locale_insensitive_strcasecmp + +#undef STRNCASECMP +#define STRNCASECMP rb_parser_st_locale_insensitive_strncasecmp + +#ifdef RIPPER +#include "ripper_init.h" +#endif + +enum rescue_context { + before_rescue, + after_rescue, + after_else, + after_ensure, }; struct lex_context { @@ -70,9 +313,13 @@ struct lex_context { unsigned int in_argdef: 1; unsigned int in_def: 1; unsigned int in_class: 1; - BITFIELD(enum shareability, shareable_constant_value, 2); + BITFIELD(enum rb_parser_shareability, shareable_constant_value, 2); + BITFIELD(enum rescue_context, in_rescue, 2); }; +typedef struct RNode_DEF_TEMP rb_node_def_temp_t; +typedef struct RNode_EXITS rb_node_exits_t; + #if defined(__GNUC__) && !defined(__clang__) // Suppress "parameter passing for argument of type 'struct // lex_context' changed" notes. `struct lex_context` is file scope, @@ -87,8 +334,6 @@ RBIMPL_WARNING_POP() #define NO_LEX_CTXT (struct lex_context){0} -#define AREF(ary, i) RARRAY_AREF(ary, i) - #ifndef WARN_PAST_SCOPE # define WARN_PAST_SCOPE 0 #endif @@ -97,10 +342,6 @@ RBIMPL_WARNING_POP() #define yydebug (p->debug) /* disable the global variable definition */ -#define YYMALLOC(size) rb_parser_malloc(p, (size)) -#define YYREALLOC(ptr, size) rb_parser_realloc(p, (ptr), (size)) -#define YYCALLOC(nelem, size) rb_parser_calloc(p, (nelem), (size)) -#define YYFREE(ptr) rb_parser_free(p, (ptr)) #define YYFPRINTF(out, ...) rb_parser_printf(p, __VA_ARGS__) #define YY_LOCATION_PRINT(File, loc, p) \ rb_parser_printf(p, "%d.%d-%d.%d", \ @@ -141,44 +382,6 @@ RBIMPL_WARNING_POP() {p->ruby_sourceline, (int)(p->lex.pcur - p->lex.pbeg)}, \ } -enum lex_state_bits { - EXPR_BEG_bit, /* ignore newline, +/- is a sign. */ - EXPR_END_bit, /* newline significant, +/- is an operator. */ - EXPR_ENDARG_bit, /* ditto, and unbound braces. */ - EXPR_ENDFN_bit, /* ditto, and unbound braces. */ - EXPR_ARG_bit, /* newline significant, +/- is an operator. */ - EXPR_CMDARG_bit, /* newline significant, +/- is an operator. */ - EXPR_MID_bit, /* newline significant, +/- is an operator. */ - EXPR_FNAME_bit, /* ignore newline, no reserved words. */ - EXPR_DOT_bit, /* right after `.', `&.' or `::', no reserved words. */ - EXPR_CLASS_bit, /* immediate after `class', no here document. */ - EXPR_LABEL_bit, /* flag bit, label is allowed. */ - EXPR_LABELED_bit, /* flag bit, just after a label. */ - EXPR_FITEM_bit, /* symbol literal as FNAME. */ - EXPR_MAX_STATE -}; -/* examine combinations */ -enum lex_state_e { -#define DEF_EXPR(n) EXPR_##n = (1 << EXPR_##n##_bit) - DEF_EXPR(BEG), - DEF_EXPR(END), - DEF_EXPR(ENDARG), - DEF_EXPR(ENDFN), - DEF_EXPR(ARG), - DEF_EXPR(CMDARG), - DEF_EXPR(MID), - DEF_EXPR(FNAME), - DEF_EXPR(DOT), - DEF_EXPR(CLASS), - DEF_EXPR(LABEL), - DEF_EXPR(LABELED), - DEF_EXPR(FITEM), - EXPR_VALUE = EXPR_BEG, - EXPR_BEG_ANY = (EXPR_BEG | EXPR_MID | EXPR_CLASS), - EXPR_ARG_ANY = (EXPR_ARG | EXPR_CMDARG), - EXPR_END_ANY = (EXPR_END | EXPR_ENDARG | EXPR_ENDFN), - EXPR_NONE = 0 -}; #define IS_lex_state_for(x, ls) ((x) & (ls)) #define IS_lex_state_all_for(x, ls) (((x) & (ls)) == (ls)) #define IS_lex_state(ls) IS_lex_state_for(p->lex.state, (ls)) @@ -227,11 +430,10 @@ struct local_vars { struct vtable *past; # endif struct local_vars *prev; -# ifndef RIPPER struct { NODE *outer, *inner, *current; } numparam; -# endif + NODE *it; }; enum { @@ -240,18 +442,6 @@ enum { NUMPARAM_MAX = 9, }; -#define NUMPARAM_ID_P(id) numparam_id_p(id) -#define NUMPARAM_ID_TO_IDX(id) (unsigned int)(((id) >> ID_SCOPE_SHIFT) - (tNUMPARAM_1 - 1)) -#define NUMPARAM_IDX_TO_ID(idx) TOKEN2LOCALID((tNUMPARAM_1 - 1 + (idx))) -static int -numparam_id_p(ID id) -{ - if (!is_local_id(id) || id < (tNUMPARAM_1 << ID_SCOPE_SHIFT)) return 0; - unsigned int idx = NUMPARAM_ID_TO_IDX(id); - return idx > 0 && idx <= NUMPARAM_MAX; -} -static void numparam_name(struct parser_params *p, ID id); - #define DVARS_INHERIT ((void*)1) #define DVARS_TOPSCOPE NULL #define DVARS_TERMINAL_P(tbl) ((tbl) == DVARS_INHERIT || (tbl) == DVARS_TOPSCOPE) @@ -264,7 +454,24 @@ typedef struct token_info { struct token_info *next; } token_info; -typedef struct rb_strterm_struct rb_strterm_t; +typedef struct end_expect_token_locations { + const rb_code_position_t *pos; + struct end_expect_token_locations *prev; +} end_expect_token_locations_t; + +typedef struct parser_string_buffer_elem { + struct parser_string_buffer_elem *next; + long len; /* Total length of allocated buf */ + long used; /* Current usage of buf */ + rb_parser_string_t *buf[FLEX_ARY_LEN]; +} parser_string_buffer_elem_t; + +typedef struct parser_string_buffer { + parser_string_buffer_elem_t *head; + parser_string_buffer_elem_t *last; +} parser_string_buffer_t; + +#define AFTER_HEREDOC_WITHOUT_TERMINTOR ((rb_parser_string_t *)1) /* Structure of Lexer Buffer: @@ -276,25 +483,20 @@ typedef struct rb_strterm_struct rb_strterm_t; token */ struct parser_params { - rb_imemo_tmpbuf_t *heap; - YYSTYPE *lval; YYLTYPE *yylloc; struct { rb_strterm_t *strterm; - VALUE (*gets)(struct parser_params*,VALUE); - VALUE input; - VALUE lastline; - VALUE nextline; + rb_parser_lex_gets_func *gets; + rb_parser_input_data input; + parser_string_buffer_t string_buffer; + rb_parser_string_t *lastline; + rb_parser_string_t *nextline; const char *pbeg; const char *pcur; const char *pend; const char *ptok; - union { - long ptr; - VALUE (*call)(VALUE, int); - } gets_; enum lex_state_e state; /* track the nest level of any parens "()[]{}" */ int paren_nest; @@ -307,7 +509,6 @@ struct parser_params { stack_type cmdarg_stack; int tokidx; int toksiz; - int tokline; int heredoc_end; int heredoc_indent; int heredoc_line_indent; @@ -318,17 +519,17 @@ struct parser_params { int line_count; int ruby_sourceline; /* current line no. */ const char *ruby_sourcefile; /* current source file */ - VALUE ruby_sourcefile_string; + rb_parser_string_t *ruby_sourcefile_string; rb_encoding *enc; token_info *token_info; - VALUE case_labels; - VALUE compile_option; + st_table *case_labels; + rb_node_exits_t *exits; VALUE debug_buffer; VALUE debug_output; struct { - VALUE token; + rb_parser_string_t *token; int beg_line; int beg_col; int end_line; @@ -341,9 +542,20 @@ struct parser_params { int node_id; int max_numparam; + ID it_id; struct lex_context ctxt; + NODE *eval_tree_begin; + NODE *eval_tree; + const struct rb_iseq_struct *parent_iseq; + +#ifdef UNIVERSAL_PARSER + const rb_parser_config_t *config; +#endif + /* compile_option */ + signed int frozen_string_literal:2; /* -1: not specified, 0: false, 1: true */ + unsigned int command_start:1; unsigned int eofp: 1; unsigned int ruby__end__seen: 1; @@ -364,45 +576,142 @@ struct parser_params { unsigned int do_loop: 1; unsigned int do_chomp: 1; unsigned int do_split: 1; - unsigned int keep_script_lines: 1; unsigned int error_tolerant: 1; unsigned int keep_tokens: 1; - NODE *eval_tree_begin; - NODE *eval_tree; VALUE error_buffer; - VALUE debug_lines; - const struct rb_iseq_struct *parent_iseq; - /* store specific keyword locations to generate dummy end token */ - VALUE end_expect_token_locations; + rb_parser_ary_t *debug_lines; + /* + * Store specific keyword locations to generate dummy end token. + * Refer to the tail of list element. + */ + end_expect_token_locations_t *end_expect_token_locations; /* id for terms */ int token_id; /* Array for term tokens */ - VALUE tokens; + rb_parser_ary_t *tokens; #else /* Ripper only */ VALUE value; VALUE result; VALUE parsing_thread; + VALUE s_value; /* Token VALUE */ + VALUE s_lvalue; /* VALUE generated by rule action (reduce) */ + VALUE s_value_stack; #endif }; +#define NUMPARAM_ID_P(id) numparam_id_p(p, id) +#define NUMPARAM_ID_TO_IDX(id) (unsigned int)(((id) >> ID_SCOPE_SHIFT) - (tNUMPARAM_1 - 1)) +#define NUMPARAM_IDX_TO_ID(idx) TOKEN2LOCALID((tNUMPARAM_1 - 1 + (idx))) +static int +numparam_id_p(struct parser_params *p, ID id) +{ + if (!is_local_id(id) || id < (tNUMPARAM_1 << ID_SCOPE_SHIFT)) return 0; + unsigned int idx = NUMPARAM_ID_TO_IDX(id); + return idx > 0 && idx <= NUMPARAM_MAX; +} +static void numparam_name(struct parser_params *p, ID id); + +#ifdef RIPPER +static void +after_shift(struct parser_params *p) +{ + if (p->debug) { + rb_parser_printf(p, "after-shift: %+"PRIsVALUE"\n", p->s_value); + } + rb_ary_push(p->s_value_stack, p->s_value); + p->s_value = Qnil; +} + +static void +before_reduce(int len, struct parser_params *p) +{ + // Initialize $$ with $1. + if (len) p->s_lvalue = rb_ary_entry(p->s_value_stack, -len); +} + +static void +after_reduce(int len, struct parser_params *p) +{ + for (int i = 0; i < len; i++) { + VALUE tos = rb_ary_pop(p->s_value_stack); + if (p->debug) { + rb_parser_printf(p, "after-reduce pop: %+"PRIsVALUE"\n", tos); + } + } + if (p->debug) { + rb_parser_printf(p, "after-reduce push: %+"PRIsVALUE"\n", p->s_lvalue); + } + rb_ary_push(p->s_value_stack, p->s_lvalue); + p->s_lvalue = Qnil; +} + +static void +after_shift_error_token(struct parser_params *p) +{ + if (p->debug) { + rb_parser_printf(p, "after-shift-error-token:\n"); + } + rb_ary_push(p->s_value_stack, Qnil); +} + +static void +after_pop_stack(int len, struct parser_params *p) +{ + for (int i = 0; i < len; i++) { + VALUE tos = rb_ary_pop(p->s_value_stack); + if (p->debug) { + rb_parser_printf(p, "after-pop-stack pop: %+"PRIsVALUE"\n", tos); + } + } +} +#else +static void +after_shift(struct parser_params *p) +{ +} + +static void +before_reduce(int len, struct parser_params *p) +{ +} + +static void +after_reduce(int len, struct parser_params *p) +{ +} + +static void +after_shift_error_token(struct parser_params *p) +{ +} + +static void +after_pop_stack(int len, struct parser_params *p) +{ +} +#endif + #define intern_cstr(n,l,en) rb_intern3(n,l,en) +#define STRING_NEW0() rb_parser_encoding_string_new(p,0,0,p->enc) + #define STR_NEW(ptr,len) rb_enc_str_new((ptr),(len),p->enc) #define STR_NEW0() rb_enc_str_new(0,0,p->enc) #define STR_NEW2(ptr) rb_enc_str_new((ptr),strlen(ptr),p->enc) -#define STR_NEW3(ptr,len,e,func) parser_str_new((ptr),(len),(e),(func),p->enc) +#define STR_NEW3(ptr,len,e,func) parser_str_new(p, (ptr),(len),(e),(func),p->enc) #define TOK_INTERN() intern_cstr(tok(p), toklen(p), p->enc) +#define VALID_SYMNAME_P(s, l, enc, type) (rb_enc_symname_type(s, l, enc, (1U<<(type))) == (int)(type)) -static st_table * -push_pvtbl(struct parser_params *p) +#ifndef RIPPER +static inline bool +end_with_newline_p(struct parser_params *p, VALUE str) { - st_table *tbl = p->pvtbl; - p->pvtbl = st_init_numtable(); - return tbl; + return RSTRING_LEN(str) > 0 && RSTRING_END(str)[-1] == '\n'; } +#endif static void pop_pvtbl(struct parser_params *p, st_table *tbl) @@ -411,14 +720,6 @@ pop_pvtbl(struct parser_params *p, st_table *tbl) p->pvtbl = tbl; } -static st_table * -push_pktbl(struct parser_params *p) -{ - st_table *tbl = p->pktbl; - p->pktbl = 0; - return tbl; -} - static void pop_pktbl(struct parser_params *p, st_table *tbl) { @@ -426,6 +727,59 @@ pop_pktbl(struct parser_params *p, st_table *tbl) p->pktbl = tbl; } +#define STRING_BUF_DEFAULT_LEN 16 + +static void +string_buffer_init(struct parser_params *p) +{ + parser_string_buffer_t *buf = &p->lex.string_buffer; + const size_t size = offsetof(parser_string_buffer_elem_t, buf) + sizeof(rb_parser_string_t *) * STRING_BUF_DEFAULT_LEN; + + buf->head = buf->last = xmalloc(size); + buf->head->len = STRING_BUF_DEFAULT_LEN; + buf->head->used = 0; + buf->head->next = NULL; +} + +static void +string_buffer_append(struct parser_params *p, rb_parser_string_t *str) +{ + parser_string_buffer_t *buf = &p->lex.string_buffer; + + if (buf->head->used >= buf->head->len) { + parser_string_buffer_elem_t *elem; + long n = buf->head->len * 2; + const size_t size = offsetof(parser_string_buffer_elem_t, buf) + sizeof(rb_parser_string_t *) * n; + + elem = xmalloc(size); + elem->len = n; + elem->used = 0; + elem->next = NULL; + buf->last->next = elem; + buf->last = elem; + } + buf->last->buf[buf->last->used++] = str; +} + +static void rb_parser_string_free(rb_parser_t *p, rb_parser_string_t *str); + +static void +string_buffer_free(struct parser_params *p) +{ + parser_string_buffer_elem_t *elem = p->lex.string_buffer.head; + + while (elem) { + parser_string_buffer_elem_t *next_elem = elem->next; + + for (long i = 0; i < elem->used; i++) { + rb_parser_string_free(p, elem->buf[i]); + } + + xfree(elem); + elem = next_elem; + } +} + #ifndef RIPPER static void flush_debug_buffer(struct parser_params *p, VALUE out, VALUE str); @@ -433,8 +787,15 @@ static void debug_end_expect_token_locations(struct parser_params *p, const char *name) { if(p->debug) { - VALUE mesg = rb_sprintf("%s: ", name); - rb_str_catf(mesg, " %"PRIsVALUE"\n", p->end_expect_token_locations); + VALUE mesg = rb_sprintf("%s: [", name); + int i = 0; + for (end_expect_token_locations_t *loc = p->end_expect_token_locations; loc; loc = loc->prev) { + if (i > 0) + rb_str_cat_cstr(mesg, ", "); + rb_str_catf(mesg, "[%d, %d]", loc->pos->lineno, loc->pos->column); + i++; + } + rb_str_cat_cstr(mesg, "]\n"); flush_debug_buffer(p, p->debug_output, mesg); } } @@ -442,197 +803,215 @@ debug_end_expect_token_locations(struct parser_params *p, const char *name) static void push_end_expect_token_locations(struct parser_params *p, const rb_code_position_t *pos) { - if(NIL_P(p->end_expect_token_locations)) return; - rb_ary_push(p->end_expect_token_locations, rb_ary_new_from_args(2, INT2NUM(pos->lineno), INT2NUM(pos->column))); + if(!p->error_tolerant) return; + + end_expect_token_locations_t *locations; + locations = ALLOC(end_expect_token_locations_t); + locations->pos = pos; + locations->prev = p->end_expect_token_locations; + p->end_expect_token_locations = locations; + debug_end_expect_token_locations(p, "push_end_expect_token_locations"); } static void pop_end_expect_token_locations(struct parser_params *p) { - if(NIL_P(p->end_expect_token_locations)) return; - rb_ary_pop(p->end_expect_token_locations); + if(!p->end_expect_token_locations) return; + + end_expect_token_locations_t *locations = p->end_expect_token_locations->prev; + ruby_sized_xfree(p->end_expect_token_locations, sizeof(end_expect_token_locations_t)); + p->end_expect_token_locations = locations; + debug_end_expect_token_locations(p, "pop_end_expect_token_locations"); } -static VALUE +static end_expect_token_locations_t * peek_end_expect_token_locations(struct parser_params *p) { - if(NIL_P(p->end_expect_token_locations)) return Qnil; - return rb_ary_last(0, 0, p->end_expect_token_locations); + return p->end_expect_token_locations; } -static ID -parser_token2id(enum yytokentype tok) +static const char * +parser_token2char(struct parser_params *p, enum yytokentype tok) { switch ((int) tok) { -#define TOKEN2ID(tok) case tok: return rb_intern(#tok); -#define TOKEN2ID2(tok, name) case tok: return rb_intern(name); - TOKEN2ID2(' ', "words_sep") - TOKEN2ID2('!', "!") - TOKEN2ID2('%', "%"); - TOKEN2ID2('&', "&"); - TOKEN2ID2('*', "*"); - TOKEN2ID2('+', "+"); - TOKEN2ID2('-', "-"); - TOKEN2ID2('/', "/"); - TOKEN2ID2('<', "<"); - TOKEN2ID2('=', "="); - TOKEN2ID2('>', ">"); - TOKEN2ID2('?', "?"); - TOKEN2ID2('^', "^"); - TOKEN2ID2('|', "|"); - TOKEN2ID2('~', "~"); - TOKEN2ID2(':', ":"); - TOKEN2ID2(',', ","); - TOKEN2ID2('.', "."); - TOKEN2ID2(';', ";"); - TOKEN2ID2('`', "`"); - TOKEN2ID2('\n', "nl"); - TOKEN2ID2('{', "{"); - TOKEN2ID2('}', "}"); - TOKEN2ID2('[', "["); - TOKEN2ID2(']', "]"); - TOKEN2ID2('(', "("); - TOKEN2ID2(')', ")"); - TOKEN2ID2('\\', "backslash"); - TOKEN2ID(keyword_class); - TOKEN2ID(keyword_module); - TOKEN2ID(keyword_def); - TOKEN2ID(keyword_undef); - TOKEN2ID(keyword_begin); - TOKEN2ID(keyword_rescue); - TOKEN2ID(keyword_ensure); - TOKEN2ID(keyword_end); - TOKEN2ID(keyword_if); - TOKEN2ID(keyword_unless); - TOKEN2ID(keyword_then); - TOKEN2ID(keyword_elsif); - TOKEN2ID(keyword_else); - TOKEN2ID(keyword_case); - TOKEN2ID(keyword_when); - TOKEN2ID(keyword_while); - TOKEN2ID(keyword_until); - TOKEN2ID(keyword_for); - TOKEN2ID(keyword_break); - TOKEN2ID(keyword_next); - TOKEN2ID(keyword_redo); - TOKEN2ID(keyword_retry); - TOKEN2ID(keyword_in); - TOKEN2ID(keyword_do); - TOKEN2ID(keyword_do_cond); - TOKEN2ID(keyword_do_block); - TOKEN2ID(keyword_do_LAMBDA); - TOKEN2ID(keyword_return); - TOKEN2ID(keyword_yield); - TOKEN2ID(keyword_super); - TOKEN2ID(keyword_self); - TOKEN2ID(keyword_nil); - TOKEN2ID(keyword_true); - TOKEN2ID(keyword_false); - TOKEN2ID(keyword_and); - TOKEN2ID(keyword_or); - TOKEN2ID(keyword_not); - TOKEN2ID(modifier_if); - TOKEN2ID(modifier_unless); - TOKEN2ID(modifier_while); - TOKEN2ID(modifier_until); - TOKEN2ID(modifier_rescue); - TOKEN2ID(keyword_alias); - TOKEN2ID(keyword_defined); - TOKEN2ID(keyword_BEGIN); - TOKEN2ID(keyword_END); - TOKEN2ID(keyword__LINE__); - TOKEN2ID(keyword__FILE__); - TOKEN2ID(keyword__ENCODING__); - TOKEN2ID(tIDENTIFIER); - TOKEN2ID(tFID); - TOKEN2ID(tGVAR); - TOKEN2ID(tIVAR); - TOKEN2ID(tCONSTANT); - TOKEN2ID(tCVAR); - TOKEN2ID(tLABEL); - TOKEN2ID(tINTEGER); - TOKEN2ID(tFLOAT); - TOKEN2ID(tRATIONAL); - TOKEN2ID(tIMAGINARY); - TOKEN2ID(tCHAR); - TOKEN2ID(tNTH_REF); - TOKEN2ID(tBACK_REF); - TOKEN2ID(tSTRING_CONTENT); - TOKEN2ID(tREGEXP_END); - TOKEN2ID(tDUMNY_END); - TOKEN2ID(tSP); - TOKEN2ID(tUPLUS); - TOKEN2ID(tUMINUS); - TOKEN2ID(tPOW); - TOKEN2ID(tCMP); - TOKEN2ID(tEQ); - TOKEN2ID(tEQQ); - TOKEN2ID(tNEQ); - TOKEN2ID(tGEQ); - TOKEN2ID(tLEQ); - TOKEN2ID(tANDOP); - TOKEN2ID(tOROP); - TOKEN2ID(tMATCH); - TOKEN2ID(tNMATCH); - TOKEN2ID(tDOT2); - TOKEN2ID(tDOT3); - TOKEN2ID(tBDOT2); - TOKEN2ID(tBDOT3); - TOKEN2ID(tAREF); - TOKEN2ID(tASET); - TOKEN2ID(tLSHFT); - TOKEN2ID(tRSHFT); - TOKEN2ID(tANDDOT); - TOKEN2ID(tCOLON2); - TOKEN2ID(tCOLON3); - TOKEN2ID(tOP_ASGN); - TOKEN2ID(tASSOC); - TOKEN2ID(tLPAREN); - TOKEN2ID(tLPAREN_ARG); - TOKEN2ID(tRPAREN); - TOKEN2ID(tLBRACK); - TOKEN2ID(tLBRACE); - TOKEN2ID(tLBRACE_ARG); - TOKEN2ID(tSTAR); - TOKEN2ID(tDSTAR); - TOKEN2ID(tAMPER); - TOKEN2ID(tLAMBDA); - TOKEN2ID(tSYMBEG); - TOKEN2ID(tSTRING_BEG); - TOKEN2ID(tXSTRING_BEG); - TOKEN2ID(tREGEXP_BEG); - TOKEN2ID(tWORDS_BEG); - TOKEN2ID(tQWORDS_BEG); - TOKEN2ID(tSYMBOLS_BEG); - TOKEN2ID(tQSYMBOLS_BEG); - TOKEN2ID(tSTRING_END); - TOKEN2ID(tSTRING_DEND); - TOKEN2ID(tSTRING_DBEG); - TOKEN2ID(tSTRING_DVAR); - TOKEN2ID(tLAMBEG); - TOKEN2ID(tLABEL_END); - TOKEN2ID(tIGNORED_NL); - TOKEN2ID(tCOMMENT); - TOKEN2ID(tEMBDOC_BEG); - TOKEN2ID(tEMBDOC); - TOKEN2ID(tEMBDOC_END); - TOKEN2ID(tHEREDOC_BEG); - TOKEN2ID(tHEREDOC_END); - TOKEN2ID(k__END__); - TOKEN2ID(tLOWEST); - TOKEN2ID(tUMINUS_NUM); - TOKEN2ID(tLAST_TOKEN); -#undef TOKEN2ID -#undef TOKEN2ID2 +#define TOKEN2CHAR(tok) case tok: return (#tok); +#define TOKEN2CHAR2(tok, name) case tok: return (name); + TOKEN2CHAR2(' ', "word_sep"); + TOKEN2CHAR2('!', "!") + TOKEN2CHAR2('%', "%"); + TOKEN2CHAR2('&', "&"); + TOKEN2CHAR2('*', "*"); + TOKEN2CHAR2('+', "+"); + TOKEN2CHAR2('-', "-"); + TOKEN2CHAR2('/', "/"); + TOKEN2CHAR2('<', "<"); + TOKEN2CHAR2('=', "="); + TOKEN2CHAR2('>', ">"); + TOKEN2CHAR2('?', "?"); + TOKEN2CHAR2('^', "^"); + TOKEN2CHAR2('|', "|"); + TOKEN2CHAR2('~', "~"); + TOKEN2CHAR2(':', ":"); + TOKEN2CHAR2(',', ","); + TOKEN2CHAR2('.', "."); + TOKEN2CHAR2(';', ";"); + TOKEN2CHAR2('`', "`"); + TOKEN2CHAR2('\n', "nl"); + TOKEN2CHAR2('{', "\"{\""); + TOKEN2CHAR2('}', "\"}\""); + TOKEN2CHAR2('[', "\"[\""); + TOKEN2CHAR2(']', "\"]\""); + TOKEN2CHAR2('(', "\"(\""); + TOKEN2CHAR2(')', "\")\""); + TOKEN2CHAR2('\\', "backslash"); + TOKEN2CHAR(keyword_class); + TOKEN2CHAR(keyword_module); + TOKEN2CHAR(keyword_def); + TOKEN2CHAR(keyword_undef); + TOKEN2CHAR(keyword_begin); + TOKEN2CHAR(keyword_rescue); + TOKEN2CHAR(keyword_ensure); + TOKEN2CHAR(keyword_end); + TOKEN2CHAR(keyword_if); + TOKEN2CHAR(keyword_unless); + TOKEN2CHAR(keyword_then); + TOKEN2CHAR(keyword_elsif); + TOKEN2CHAR(keyword_else); + TOKEN2CHAR(keyword_case); + TOKEN2CHAR(keyword_when); + TOKEN2CHAR(keyword_while); + TOKEN2CHAR(keyword_until); + TOKEN2CHAR(keyword_for); + TOKEN2CHAR(keyword_break); + TOKEN2CHAR(keyword_next); + TOKEN2CHAR(keyword_redo); + TOKEN2CHAR(keyword_retry); + TOKEN2CHAR(keyword_in); + TOKEN2CHAR(keyword_do); + TOKEN2CHAR(keyword_do_cond); + TOKEN2CHAR(keyword_do_block); + TOKEN2CHAR(keyword_do_LAMBDA); + TOKEN2CHAR(keyword_return); + TOKEN2CHAR(keyword_yield); + TOKEN2CHAR(keyword_super); + TOKEN2CHAR(keyword_self); + TOKEN2CHAR(keyword_nil); + TOKEN2CHAR(keyword_true); + TOKEN2CHAR(keyword_false); + TOKEN2CHAR(keyword_and); + TOKEN2CHAR(keyword_or); + TOKEN2CHAR(keyword_not); + TOKEN2CHAR(modifier_if); + TOKEN2CHAR(modifier_unless); + TOKEN2CHAR(modifier_while); + TOKEN2CHAR(modifier_until); + TOKEN2CHAR(modifier_rescue); + TOKEN2CHAR(keyword_alias); + TOKEN2CHAR(keyword_defined); + TOKEN2CHAR(keyword_BEGIN); + TOKEN2CHAR(keyword_END); + TOKEN2CHAR(keyword__LINE__); + TOKEN2CHAR(keyword__FILE__); + TOKEN2CHAR(keyword__ENCODING__); + TOKEN2CHAR(tIDENTIFIER); + TOKEN2CHAR(tFID); + TOKEN2CHAR(tGVAR); + TOKEN2CHAR(tIVAR); + TOKEN2CHAR(tCONSTANT); + TOKEN2CHAR(tCVAR); + TOKEN2CHAR(tLABEL); + TOKEN2CHAR(tINTEGER); + TOKEN2CHAR(tFLOAT); + TOKEN2CHAR(tRATIONAL); + TOKEN2CHAR(tIMAGINARY); + TOKEN2CHAR(tCHAR); + TOKEN2CHAR(tNTH_REF); + TOKEN2CHAR(tBACK_REF); + TOKEN2CHAR(tSTRING_CONTENT); + TOKEN2CHAR(tREGEXP_END); + TOKEN2CHAR(tDUMNY_END); + TOKEN2CHAR(tSP); + TOKEN2CHAR(tUPLUS); + TOKEN2CHAR(tUMINUS); + TOKEN2CHAR(tPOW); + TOKEN2CHAR(tCMP); + TOKEN2CHAR(tEQ); + TOKEN2CHAR(tEQQ); + TOKEN2CHAR(tNEQ); + TOKEN2CHAR(tGEQ); + TOKEN2CHAR(tLEQ); + TOKEN2CHAR(tANDOP); + TOKEN2CHAR(tOROP); + TOKEN2CHAR(tMATCH); + TOKEN2CHAR(tNMATCH); + TOKEN2CHAR(tDOT2); + TOKEN2CHAR(tDOT3); + TOKEN2CHAR(tBDOT2); + TOKEN2CHAR(tBDOT3); + TOKEN2CHAR(tAREF); + TOKEN2CHAR(tASET); + TOKEN2CHAR(tLSHFT); + TOKEN2CHAR(tRSHFT); + TOKEN2CHAR(tANDDOT); + TOKEN2CHAR(tCOLON2); + TOKEN2CHAR(tCOLON3); + TOKEN2CHAR(tOP_ASGN); + TOKEN2CHAR(tASSOC); + TOKEN2CHAR(tLPAREN); + TOKEN2CHAR(tLPAREN_ARG); + TOKEN2CHAR(tRPAREN); + TOKEN2CHAR(tLBRACK); + TOKEN2CHAR(tLBRACE); + TOKEN2CHAR(tLBRACE_ARG); + TOKEN2CHAR(tSTAR); + TOKEN2CHAR(tDSTAR); + TOKEN2CHAR(tAMPER); + TOKEN2CHAR(tLAMBDA); + TOKEN2CHAR(tSYMBEG); + TOKEN2CHAR(tSTRING_BEG); + TOKEN2CHAR(tXSTRING_BEG); + TOKEN2CHAR(tREGEXP_BEG); + TOKEN2CHAR(tWORDS_BEG); + TOKEN2CHAR(tQWORDS_BEG); + TOKEN2CHAR(tSYMBOLS_BEG); + TOKEN2CHAR(tQSYMBOLS_BEG); + TOKEN2CHAR(tSTRING_END); + TOKEN2CHAR(tSTRING_DEND); + TOKEN2CHAR(tSTRING_DBEG); + TOKEN2CHAR(tSTRING_DVAR); + TOKEN2CHAR(tLAMBEG); + TOKEN2CHAR(tLABEL_END); + TOKEN2CHAR(tIGNORED_NL); + TOKEN2CHAR(tCOMMENT); + TOKEN2CHAR(tEMBDOC_BEG); + TOKEN2CHAR(tEMBDOC); + TOKEN2CHAR(tEMBDOC_END); + TOKEN2CHAR(tHEREDOC_BEG); + TOKEN2CHAR(tHEREDOC_END); + TOKEN2CHAR(k__END__); + TOKEN2CHAR(tLOWEST); + TOKEN2CHAR(tUMINUS_NUM); + TOKEN2CHAR(tLAST_TOKEN); +#undef TOKEN2CHAR +#undef TOKEN2CHAR2 } rb_bug("parser_token2id: unknown token %d", tok); UNREACHABLE_RETURN(0); } +#else +static void +push_end_expect_token_locations(struct parser_params *p, const rb_code_position_t *pos) +{ +} +static void +pop_end_expect_token_locations(struct parser_params *p) +{ +} #endif RBIMPL_ATTR_NONNULL((1, 2, 3)) @@ -644,8 +1023,10 @@ static int parser_yyerror0(struct parser_params*, const char*); #define yyerror(yylloc, p, msg) parser_yyerror(p, yylloc, msg) #define token_flush(ptr) ((ptr)->lex.ptok = (ptr)->lex.pcur) #define lex_goto_eol(p) ((p)->lex.pcur = (p)->lex.pend) -#define lex_eol_p(p) ((p)->lex.pcur >= (p)->lex.pend) -#define lex_eol_n_p(p,n) ((p)->lex.pcur+(n) >= (p)->lex.pend) +#define lex_eol_p(p) lex_eol_n_p(p, 0) +#define lex_eol_n_p(p,n) lex_eol_ptr_n_p(p, (p)->lex.pcur, n) +#define lex_eol_ptr_p(p,ptr) lex_eol_ptr_n_p(p,ptr,0) +#define lex_eol_ptr_n_p(p,ptr,n) ((ptr)+(n) >= (p)->lex.pend) static void token_info_setup(token_info *ptinfo, const char *ptr, const rb_code_location_t *loc); static void token_info_push(struct parser_params*, const char *token, const rb_code_location_t *loc); @@ -661,44 +1042,290 @@ static void token_info_drop(struct parser_params *p, const char *token, rb_code_ #define token_column ((int)(p->lex.ptok - p->lex.pbeg)) -#define CALL_Q_P(q) ((q) == TOKEN2VAL(tANDDOT)) -#define NODE_CALL_Q(q) (CALL_Q_P(q) ? NODE_QCALL : NODE_CALL) -#define NEW_QCALL(q,r,m,a,loc) NEW_NODE(NODE_CALL_Q(q),r,m,a,loc) +#define CALL_Q_P(q) ((q) == tANDDOT) +#define NEW_QCALL(q,r,m,a,loc) (CALL_Q_P(q) ? NEW_QCALL0(r,m,a,loc) : NEW_CALL(r,m,a,loc)) #define lambda_beginning_p() (p->lex.lpar_beg == p->lex.paren_nest) static enum yytokentype yylex(YYSTYPE*, YYLTYPE*, struct parser_params*); -#ifndef RIPPER static inline void rb_discard_node(struct parser_params *p, NODE *n) { rb_ast_delete_node(p->ast, n); } -#endif -#ifdef RIPPER -static inline VALUE -add_mark_object(struct parser_params *p, VALUE obj) +static rb_node_scope_t *rb_node_scope_new(struct parser_params *p, rb_node_args_t *nd_args, NODE *nd_body, const YYLTYPE *loc); +static rb_node_scope_t *rb_node_scope_new2(struct parser_params *p, rb_ast_id_table_t *nd_tbl, rb_node_args_t *nd_args, NODE *nd_body, const YYLTYPE *loc); +static rb_node_block_t *rb_node_block_new(struct parser_params *p, NODE *nd_head, const YYLTYPE *loc); +static rb_node_if_t *rb_node_if_new(struct parser_params *p, NODE *nd_cond, NODE *nd_body, NODE *nd_else, const YYLTYPE *loc); +static rb_node_unless_t *rb_node_unless_new(struct parser_params *p, NODE *nd_cond, NODE *nd_body, NODE *nd_else, const YYLTYPE *loc); +static rb_node_case_t *rb_node_case_new(struct parser_params *p, NODE *nd_head, NODE *nd_body, const YYLTYPE *loc); +static rb_node_case2_t *rb_node_case2_new(struct parser_params *p, NODE *nd_body, const YYLTYPE *loc); +static rb_node_case3_t *rb_node_case3_new(struct parser_params *p, NODE *nd_head, NODE *nd_body, const YYLTYPE *loc); +static rb_node_when_t *rb_node_when_new(struct parser_params *p, NODE *nd_head, NODE *nd_body, NODE *nd_next, const YYLTYPE *loc); +static rb_node_in_t *rb_node_in_new(struct parser_params *p, NODE *nd_head, NODE *nd_body, NODE *nd_next, const YYLTYPE *loc); +static rb_node_while_t *rb_node_while_new(struct parser_params *p, NODE *nd_cond, NODE *nd_body, long nd_state, const YYLTYPE *loc); +static rb_node_until_t *rb_node_until_new(struct parser_params *p, NODE *nd_cond, NODE *nd_body, long nd_state, const YYLTYPE *loc); +static rb_node_iter_t *rb_node_iter_new(struct parser_params *p, rb_node_args_t *nd_args, NODE *nd_body, const YYLTYPE *loc); +static rb_node_for_t *rb_node_for_new(struct parser_params *p, NODE *nd_iter, NODE *nd_body, const YYLTYPE *loc); +static rb_node_for_masgn_t *rb_node_for_masgn_new(struct parser_params *p, NODE *nd_var, const YYLTYPE *loc); +static rb_node_retry_t *rb_node_retry_new(struct parser_params *p, const YYLTYPE *loc); +static rb_node_begin_t *rb_node_begin_new(struct parser_params *p, NODE *nd_body, const YYLTYPE *loc); +static rb_node_rescue_t *rb_node_rescue_new(struct parser_params *p, NODE *nd_head, NODE *nd_resq, NODE *nd_else, const YYLTYPE *loc); +static rb_node_resbody_t *rb_node_resbody_new(struct parser_params *p, NODE *nd_args, NODE *nd_body, NODE *nd_next, const YYLTYPE *loc); +static rb_node_ensure_t *rb_node_ensure_new(struct parser_params *p, NODE *nd_head, NODE *nd_ensr, const YYLTYPE *loc); +static rb_node_and_t *rb_node_and_new(struct parser_params *p, NODE *nd_1st, NODE *nd_2nd, const YYLTYPE *loc); +static rb_node_or_t *rb_node_or_new(struct parser_params *p, NODE *nd_1st, NODE *nd_2nd, const YYLTYPE *loc); +static rb_node_masgn_t *rb_node_masgn_new(struct parser_params *p, NODE *nd_head, NODE *nd_args, const YYLTYPE *loc); +static rb_node_lasgn_t *rb_node_lasgn_new(struct parser_params *p, ID nd_vid, NODE *nd_value, const YYLTYPE *loc); +static rb_node_dasgn_t *rb_node_dasgn_new(struct parser_params *p, ID nd_vid, NODE *nd_value, const YYLTYPE *loc); +static rb_node_gasgn_t *rb_node_gasgn_new(struct parser_params *p, ID nd_vid, NODE *nd_value, const YYLTYPE *loc); +static rb_node_iasgn_t *rb_node_iasgn_new(struct parser_params *p, ID nd_vid, NODE *nd_value, const YYLTYPE *loc); +static rb_node_cdecl_t *rb_node_cdecl_new(struct parser_params *p, ID nd_vid, NODE *nd_value, NODE *nd_else, enum rb_parser_shareability shareability, const YYLTYPE *loc); +static rb_node_cvasgn_t *rb_node_cvasgn_new(struct parser_params *p, ID nd_vid, NODE *nd_value, const YYLTYPE *loc); +static rb_node_op_asgn1_t *rb_node_op_asgn1_new(struct parser_params *p, NODE *nd_recv, ID nd_mid, NODE *index, NODE *rvalue, const YYLTYPE *loc); +static rb_node_op_asgn2_t *rb_node_op_asgn2_new(struct parser_params *p, NODE *nd_recv, NODE *nd_value, ID nd_vid, ID nd_mid, bool nd_aid, const YYLTYPE *loc); +static rb_node_op_asgn_or_t *rb_node_op_asgn_or_new(struct parser_params *p, NODE *nd_head, NODE *nd_value, const YYLTYPE *loc); +static rb_node_op_asgn_and_t *rb_node_op_asgn_and_new(struct parser_params *p, NODE *nd_head, NODE *nd_value, const YYLTYPE *loc); +static rb_node_op_cdecl_t *rb_node_op_cdecl_new(struct parser_params *p, NODE *nd_head, NODE *nd_value, ID nd_aid, enum rb_parser_shareability shareability, const YYLTYPE *loc); +static rb_node_call_t *rb_node_call_new(struct parser_params *p, NODE *nd_recv, ID nd_mid, NODE *nd_args, const YYLTYPE *loc); +static rb_node_opcall_t *rb_node_opcall_new(struct parser_params *p, NODE *nd_recv, ID nd_mid, NODE *nd_args, const YYLTYPE *loc); +static rb_node_fcall_t *rb_node_fcall_new(struct parser_params *p, ID nd_mid, NODE *nd_args, const YYLTYPE *loc); +static rb_node_vcall_t *rb_node_vcall_new(struct parser_params *p, ID nd_mid, const YYLTYPE *loc); +static rb_node_qcall_t *rb_node_qcall_new(struct parser_params *p, NODE *nd_recv, ID nd_mid, NODE *nd_args, const YYLTYPE *loc); +static rb_node_super_t *rb_node_super_new(struct parser_params *p, NODE *nd_args, const YYLTYPE *loc); +static rb_node_zsuper_t * rb_node_zsuper_new(struct parser_params *p, const YYLTYPE *loc); +static rb_node_list_t *rb_node_list_new(struct parser_params *p, NODE *nd_head, const YYLTYPE *loc); +static rb_node_list_t *rb_node_list_new2(struct parser_params *p, NODE *nd_head, long nd_alen, NODE *nd_next, const YYLTYPE *loc); +static rb_node_zlist_t *rb_node_zlist_new(struct parser_params *p, const YYLTYPE *loc); +static rb_node_hash_t *rb_node_hash_new(struct parser_params *p, NODE *nd_head, const YYLTYPE *loc); +static rb_node_return_t *rb_node_return_new(struct parser_params *p, NODE *nd_stts, const YYLTYPE *loc); +static rb_node_yield_t *rb_node_yield_new(struct parser_params *p, NODE *nd_head, const YYLTYPE *loc); +static rb_node_lvar_t *rb_node_lvar_new(struct parser_params *p, ID nd_vid, const YYLTYPE *loc); +static rb_node_dvar_t *rb_node_dvar_new(struct parser_params *p, ID nd_vid, const YYLTYPE *loc); +static rb_node_gvar_t *rb_node_gvar_new(struct parser_params *p, ID nd_vid, const YYLTYPE *loc); +static rb_node_ivar_t *rb_node_ivar_new(struct parser_params *p, ID nd_vid, const YYLTYPE *loc); +static rb_node_const_t *rb_node_const_new(struct parser_params *p, ID nd_vid, const YYLTYPE *loc); +static rb_node_cvar_t *rb_node_cvar_new(struct parser_params *p, ID nd_vid, const YYLTYPE *loc); +static rb_node_nth_ref_t *rb_node_nth_ref_new(struct parser_params *p, long nd_nth, const YYLTYPE *loc); +static rb_node_back_ref_t *rb_node_back_ref_new(struct parser_params *p, long nd_nth, const YYLTYPE *loc); +static rb_node_match2_t *rb_node_match2_new(struct parser_params *p, NODE *nd_recv, NODE *nd_value, const YYLTYPE *loc); +static rb_node_match3_t *rb_node_match3_new(struct parser_params *p, NODE *nd_recv, NODE *nd_value, const YYLTYPE *loc); +static rb_node_integer_t * rb_node_integer_new(struct parser_params *p, char* val, int base, const YYLTYPE *loc); +static rb_node_float_t * rb_node_float_new(struct parser_params *p, char* val, const YYLTYPE *loc); +static rb_node_rational_t * rb_node_rational_new(struct parser_params *p, char* val, int base, int seen_point, const YYLTYPE *loc); +static rb_node_imaginary_t * rb_node_imaginary_new(struct parser_params *p, char* val, int base, int seen_point, enum rb_numeric_type, const YYLTYPE *loc); +static rb_node_str_t *rb_node_str_new(struct parser_params *p, rb_parser_string_t *string, const YYLTYPE *loc); +static rb_node_dstr_t *rb_node_dstr_new0(struct parser_params *p, rb_parser_string_t *string, long nd_alen, NODE *nd_next, const YYLTYPE *loc); +static rb_node_dstr_t *rb_node_dstr_new(struct parser_params *p, rb_parser_string_t *string, const YYLTYPE *loc); +static rb_node_xstr_t *rb_node_xstr_new(struct parser_params *p, rb_parser_string_t *string, const YYLTYPE *loc); +static rb_node_dxstr_t *rb_node_dxstr_new(struct parser_params *p, rb_parser_string_t *string, long nd_alen, NODE *nd_next, const YYLTYPE *loc); +static rb_node_evstr_t *rb_node_evstr_new(struct parser_params *p, NODE *nd_body, const YYLTYPE *loc); +static rb_node_regx_t *rb_node_regx_new(struct parser_params *p, rb_parser_string_t *string, int options, const YYLTYPE *loc); +static rb_node_once_t *rb_node_once_new(struct parser_params *p, NODE *nd_body, const YYLTYPE *loc); +static rb_node_args_t *rb_node_args_new(struct parser_params *p, const YYLTYPE *loc); +static rb_node_args_aux_t *rb_node_args_aux_new(struct parser_params *p, ID nd_pid, int nd_plen, const YYLTYPE *loc); +static rb_node_opt_arg_t *rb_node_opt_arg_new(struct parser_params *p, NODE *nd_body, const YYLTYPE *loc); +static rb_node_kw_arg_t *rb_node_kw_arg_new(struct parser_params *p, NODE *nd_body, const YYLTYPE *loc); +static rb_node_postarg_t *rb_node_postarg_new(struct parser_params *p, NODE *nd_1st, NODE *nd_2nd, const YYLTYPE *loc); +static rb_node_argscat_t *rb_node_argscat_new(struct parser_params *p, NODE *nd_head, NODE *nd_body, const YYLTYPE *loc); +static rb_node_argspush_t *rb_node_argspush_new(struct parser_params *p, NODE *nd_head, NODE *nd_body, const YYLTYPE *loc); +static rb_node_splat_t *rb_node_splat_new(struct parser_params *p, NODE *nd_head, const YYLTYPE *loc); +static rb_node_block_pass_t *rb_node_block_pass_new(struct parser_params *p, NODE *nd_body, const YYLTYPE *loc); +static rb_node_defn_t *rb_node_defn_new(struct parser_params *p, ID nd_mid, NODE *nd_defn, const YYLTYPE *loc); +static rb_node_defs_t *rb_node_defs_new(struct parser_params *p, NODE *nd_recv, ID nd_mid, NODE *nd_defn, const YYLTYPE *loc); +static rb_node_alias_t *rb_node_alias_new(struct parser_params *p, NODE *nd_1st, NODE *nd_2nd, const YYLTYPE *loc); +static rb_node_valias_t *rb_node_valias_new(struct parser_params *p, ID nd_alias, ID nd_orig, const YYLTYPE *loc); +static rb_node_undef_t *rb_node_undef_new(struct parser_params *p, NODE *nd_undef, const YYLTYPE *loc); +static rb_node_class_t *rb_node_class_new(struct parser_params *p, NODE *nd_cpath, NODE *nd_body, NODE *nd_super, const YYLTYPE *loc); +static rb_node_module_t *rb_node_module_new(struct parser_params *p, NODE *nd_cpath, NODE *nd_body, const YYLTYPE *loc); +static rb_node_sclass_t *rb_node_sclass_new(struct parser_params *p, NODE *nd_recv, NODE *nd_body, const YYLTYPE *loc); +static rb_node_colon2_t *rb_node_colon2_new(struct parser_params *p, NODE *nd_head, ID nd_mid, const YYLTYPE *loc); +static rb_node_colon3_t *rb_node_colon3_new(struct parser_params *p, ID nd_mid, const YYLTYPE *loc); +static rb_node_dot2_t *rb_node_dot2_new(struct parser_params *p, NODE *nd_beg, NODE *nd_end, const YYLTYPE *loc); +static rb_node_dot3_t *rb_node_dot3_new(struct parser_params *p, NODE *nd_beg, NODE *nd_end, const YYLTYPE *loc); +static rb_node_self_t *rb_node_self_new(struct parser_params *p, const YYLTYPE *loc); +static rb_node_nil_t *rb_node_nil_new(struct parser_params *p, const YYLTYPE *loc); +static rb_node_true_t *rb_node_true_new(struct parser_params *p, const YYLTYPE *loc); +static rb_node_false_t *rb_node_false_new(struct parser_params *p, const YYLTYPE *loc); +static rb_node_errinfo_t *rb_node_errinfo_new(struct parser_params *p, const YYLTYPE *loc); +static rb_node_defined_t *rb_node_defined_new(struct parser_params *p, NODE *nd_head, const YYLTYPE *loc); +static rb_node_postexe_t *rb_node_postexe_new(struct parser_params *p, NODE *nd_body, const YYLTYPE *loc); +static rb_node_sym_t *rb_node_sym_new(struct parser_params *p, VALUE str, const YYLTYPE *loc); +static rb_node_dsym_t *rb_node_dsym_new(struct parser_params *p, rb_parser_string_t *string, long nd_alen, NODE *nd_next, const YYLTYPE *loc); +static rb_node_attrasgn_t *rb_node_attrasgn_new(struct parser_params *p, NODE *nd_recv, ID nd_mid, NODE *nd_args, const YYLTYPE *loc); +static rb_node_lambda_t *rb_node_lambda_new(struct parser_params *p, rb_node_args_t *nd_args, NODE *nd_body, const YYLTYPE *loc); +static rb_node_aryptn_t *rb_node_aryptn_new(struct parser_params *p, NODE *pre_args, NODE *rest_arg, NODE *post_args, const YYLTYPE *loc); +static rb_node_hshptn_t *rb_node_hshptn_new(struct parser_params *p, NODE *nd_pconst, NODE *nd_pkwargs, NODE *nd_pkwrestarg, const YYLTYPE *loc); +static rb_node_fndptn_t *rb_node_fndptn_new(struct parser_params *p, NODE *pre_rest_arg, NODE *args, NODE *post_rest_arg, const YYLTYPE *loc); +static rb_node_line_t *rb_node_line_new(struct parser_params *p, const YYLTYPE *loc); +static rb_node_file_t *rb_node_file_new(struct parser_params *p, rb_parser_string_t *str, const YYLTYPE *loc); +static rb_node_error_t *rb_node_error_new(struct parser_params *p, const YYLTYPE *loc); + +#define NEW_SCOPE(a,b,loc) (NODE *)rb_node_scope_new(p,a,b,loc) +#define NEW_SCOPE2(t,a,b,loc) (NODE *)rb_node_scope_new2(p,t,a,b,loc) +#define NEW_BLOCK(a,loc) (NODE *)rb_node_block_new(p,a,loc) +#define NEW_IF(c,t,e,loc) (NODE *)rb_node_if_new(p,c,t,e,loc) +#define NEW_UNLESS(c,t,e,loc) (NODE *)rb_node_unless_new(p,c,t,e,loc) +#define NEW_CASE(h,b,loc) (NODE *)rb_node_case_new(p,h,b,loc) +#define NEW_CASE2(b,loc) (NODE *)rb_node_case2_new(p,b,loc) +#define NEW_CASE3(h,b,loc) (NODE *)rb_node_case3_new(p,h,b,loc) +#define NEW_WHEN(c,t,e,loc) (NODE *)rb_node_when_new(p,c,t,e,loc) +#define NEW_IN(c,t,e,loc) (NODE *)rb_node_in_new(p,c,t,e,loc) +#define NEW_WHILE(c,b,n,loc) (NODE *)rb_node_while_new(p,c,b,n,loc) +#define NEW_UNTIL(c,b,n,loc) (NODE *)rb_node_until_new(p,c,b,n,loc) +#define NEW_ITER(a,b,loc) (NODE *)rb_node_iter_new(p,a,b,loc) +#define NEW_FOR(i,b,loc) (NODE *)rb_node_for_new(p,i,b,loc) +#define NEW_FOR_MASGN(v,loc) (NODE *)rb_node_for_masgn_new(p,v,loc) +#define NEW_RETRY(loc) (NODE *)rb_node_retry_new(p,loc) +#define NEW_BEGIN(b,loc) (NODE *)rb_node_begin_new(p,b,loc) +#define NEW_RESCUE(b,res,e,loc) (NODE *)rb_node_rescue_new(p,b,res,e,loc) +#define NEW_RESBODY(a,ex,n,loc) (NODE *)rb_node_resbody_new(p,a,ex,n,loc) +#define NEW_ENSURE(b,en,loc) (NODE *)rb_node_ensure_new(p,b,en,loc) +#define NEW_AND(f,s,loc) (NODE *)rb_node_and_new(p,f,s,loc) +#define NEW_OR(f,s,loc) (NODE *)rb_node_or_new(p,f,s,loc) +#define NEW_MASGN(l,r,loc) rb_node_masgn_new(p,l,r,loc) +#define NEW_LASGN(v,val,loc) (NODE *)rb_node_lasgn_new(p,v,val,loc) +#define NEW_DASGN(v,val,loc) (NODE *)rb_node_dasgn_new(p,v,val,loc) +#define NEW_GASGN(v,val,loc) (NODE *)rb_node_gasgn_new(p,v,val,loc) +#define NEW_IASGN(v,val,loc) (NODE *)rb_node_iasgn_new(p,v,val,loc) +#define NEW_CDECL(v,val,path,share,loc) (NODE *)rb_node_cdecl_new(p,v,val,path,share,loc) +#define NEW_CVASGN(v,val,loc) (NODE *)rb_node_cvasgn_new(p,v,val,loc) +#define NEW_OP_ASGN1(r,id,idx,rval,loc) (NODE *)rb_node_op_asgn1_new(p,r,id,idx,rval,loc) +#define NEW_OP_ASGN2(r,t,i,o,val,loc) (NODE *)rb_node_op_asgn2_new(p,r,val,i,o,t,loc) +#define NEW_OP_ASGN_OR(i,val,loc) (NODE *)rb_node_op_asgn_or_new(p,i,val,loc) +#define NEW_OP_ASGN_AND(i,val,loc) (NODE *)rb_node_op_asgn_and_new(p,i,val,loc) +#define NEW_OP_CDECL(v,op,val,share,loc) (NODE *)rb_node_op_cdecl_new(p,v,val,op,share,loc) +#define NEW_CALL(r,m,a,loc) (NODE *)rb_node_call_new(p,r,m,a,loc) +#define NEW_OPCALL(r,m,a,loc) (NODE *)rb_node_opcall_new(p,r,m,a,loc) +#define NEW_FCALL(m,a,loc) rb_node_fcall_new(p,m,a,loc) +#define NEW_VCALL(m,loc) (NODE *)rb_node_vcall_new(p,m,loc) +#define NEW_QCALL0(r,m,a,loc) (NODE *)rb_node_qcall_new(p,r,m,a,loc) +#define NEW_SUPER(a,loc) (NODE *)rb_node_super_new(p,a,loc) +#define NEW_ZSUPER(loc) (NODE *)rb_node_zsuper_new(p,loc) +#define NEW_LIST(a,loc) (NODE *)rb_node_list_new(p,a,loc) +#define NEW_LIST2(h,l,n,loc) (NODE *)rb_node_list_new2(p,h,l,n,loc) +#define NEW_ZLIST(loc) (NODE *)rb_node_zlist_new(p,loc) +#define NEW_HASH(a,loc) (NODE *)rb_node_hash_new(p,a,loc) +#define NEW_RETURN(s,loc) (NODE *)rb_node_return_new(p,s,loc) +#define NEW_YIELD(a,loc) (NODE *)rb_node_yield_new(p,a,loc) +#define NEW_LVAR(v,loc) (NODE *)rb_node_lvar_new(p,v,loc) +#define NEW_DVAR(v,loc) (NODE *)rb_node_dvar_new(p,v,loc) +#define NEW_GVAR(v,loc) (NODE *)rb_node_gvar_new(p,v,loc) +#define NEW_IVAR(v,loc) (NODE *)rb_node_ivar_new(p,v,loc) +#define NEW_CONST(v,loc) (NODE *)rb_node_const_new(p,v,loc) +#define NEW_CVAR(v,loc) (NODE *)rb_node_cvar_new(p,v,loc) +#define NEW_NTH_REF(n,loc) (NODE *)rb_node_nth_ref_new(p,n,loc) +#define NEW_BACK_REF(n,loc) (NODE *)rb_node_back_ref_new(p,n,loc) +#define NEW_MATCH2(n1,n2,loc) (NODE *)rb_node_match2_new(p,n1,n2,loc) +#define NEW_MATCH3(r,n2,loc) (NODE *)rb_node_match3_new(p,r,n2,loc) +#define NEW_INTEGER(val, base,loc) (NODE *)rb_node_integer_new(p,val,base,loc) +#define NEW_FLOAT(val,loc) (NODE *)rb_node_float_new(p,val,loc) +#define NEW_RATIONAL(val,base,seen_point,loc) (NODE *)rb_node_rational_new(p,val,base,seen_point,loc) +#define NEW_IMAGINARY(val,base,seen_point,numeric_type,loc) (NODE *)rb_node_imaginary_new(p,val,base,seen_point,numeric_type,loc) +#define NEW_STR(s,loc) (NODE *)rb_node_str_new(p,s,loc) +#define NEW_DSTR0(s,l,n,loc) (NODE *)rb_node_dstr_new0(p,s,l,n,loc) +#define NEW_DSTR(s,loc) (NODE *)rb_node_dstr_new(p,s,loc) +#define NEW_XSTR(s,loc) (NODE *)rb_node_xstr_new(p,s,loc) +#define NEW_DXSTR(s,l,n,loc) (NODE *)rb_node_dxstr_new(p,s,l,n,loc) +#define NEW_EVSTR(n,loc) (NODE *)rb_node_evstr_new(p,n,loc) +#define NEW_REGX(str,opts,loc) (NODE *)rb_node_regx_new(p,str,opts,loc) +#define NEW_ONCE(b,loc) (NODE *)rb_node_once_new(p,b,loc) +#define NEW_ARGS(loc) rb_node_args_new(p,loc) +#define NEW_ARGS_AUX(r,b,loc) rb_node_args_aux_new(p,r,b,loc) +#define NEW_OPT_ARG(v,loc) rb_node_opt_arg_new(p,v,loc) +#define NEW_KW_ARG(v,loc) rb_node_kw_arg_new(p,v,loc) +#define NEW_POSTARG(i,v,loc) (NODE *)rb_node_postarg_new(p,i,v,loc) +#define NEW_ARGSCAT(a,b,loc) (NODE *)rb_node_argscat_new(p,a,b,loc) +#define NEW_ARGSPUSH(a,b,loc) (NODE *)rb_node_argspush_new(p,a,b,loc) +#define NEW_SPLAT(a,loc) (NODE *)rb_node_splat_new(p,a,loc) +#define NEW_BLOCK_PASS(b,loc) rb_node_block_pass_new(p,b,loc) +#define NEW_DEFN(i,s,loc) (NODE *)rb_node_defn_new(p,i,s,loc) +#define NEW_DEFS(r,i,s,loc) (NODE *)rb_node_defs_new(p,r,i,s,loc) +#define NEW_ALIAS(n,o,loc) (NODE *)rb_node_alias_new(p,n,o,loc) +#define NEW_VALIAS(n,o,loc) (NODE *)rb_node_valias_new(p,n,o,loc) +#define NEW_UNDEF(i,loc) (NODE *)rb_node_undef_new(p,i,loc) +#define NEW_CLASS(n,b,s,loc) (NODE *)rb_node_class_new(p,n,b,s,loc) +#define NEW_MODULE(n,b,loc) (NODE *)rb_node_module_new(p,n,b,loc) +#define NEW_SCLASS(r,b,loc) (NODE *)rb_node_sclass_new(p,r,b,loc) +#define NEW_COLON2(c,i,loc) (NODE *)rb_node_colon2_new(p,c,i,loc) +#define NEW_COLON3(i,loc) (NODE *)rb_node_colon3_new(p,i,loc) +#define NEW_DOT2(b,e,loc) (NODE *)rb_node_dot2_new(p,b,e,loc) +#define NEW_DOT3(b,e,loc) (NODE *)rb_node_dot3_new(p,b,e,loc) +#define NEW_SELF(loc) (NODE *)rb_node_self_new(p,loc) +#define NEW_NIL(loc) (NODE *)rb_node_nil_new(p,loc) +#define NEW_TRUE(loc) (NODE *)rb_node_true_new(p,loc) +#define NEW_FALSE(loc) (NODE *)rb_node_false_new(p,loc) +#define NEW_ERRINFO(loc) (NODE *)rb_node_errinfo_new(p,loc) +#define NEW_DEFINED(e,loc) (NODE *)rb_node_defined_new(p,e,loc) +#define NEW_POSTEXE(b,loc) (NODE *)rb_node_postexe_new(p,b,loc) +#define NEW_SYM(str,loc) (NODE *)rb_node_sym_new(p,str,loc) +#define NEW_DSYM(s,l,n,loc) (NODE *)rb_node_dsym_new(p,s,l,n,loc) +#define NEW_ATTRASGN(r,m,a,loc) (NODE *)rb_node_attrasgn_new(p,r,m,a,loc) +#define NEW_LAMBDA(a,b,loc) (NODE *)rb_node_lambda_new(p,a,b,loc) +#define NEW_ARYPTN(pre,r,post,loc) (NODE *)rb_node_aryptn_new(p,pre,r,post,loc) +#define NEW_HSHPTN(c,kw,kwrest,loc) (NODE *)rb_node_hshptn_new(p,c,kw,kwrest,loc) +#define NEW_FNDPTN(pre,a,post,loc) (NODE *)rb_node_fndptn_new(p,pre,a,post,loc) +#define NEW_LINE(loc) (NODE *)rb_node_line_new(p,loc) +#define NEW_FILE(str,loc) (NODE *)rb_node_file_new(p,str,loc) +#define NEW_ENCODING(loc) (NODE *)rb_node_encoding_new(p,loc) +#define NEW_ERROR(loc) (NODE *)rb_node_error_new(p,loc) + +enum internal_node_type { + NODE_INTERNAL_ONLY = NODE_LAST, + NODE_DEF_TEMP, + NODE_EXITS, + NODE_INTERNAL_LAST +}; + +static const char * +parser_node_name(int node) { - if (!SPECIAL_CONST_P(obj) - && !RB_TYPE_P(obj, T_NODE) /* Ripper jumbles NODE objects and other objects... */ - ) { - rb_ast_add_mark_object(p->ast, obj); + switch (node) { + case NODE_DEF_TEMP: + return "NODE_DEF_TEMP"; + case NODE_EXITS: + return "NODE_EXITS"; + default: + return ruby_node_name(node); } - return obj; } -#else -static NODE* node_newnode_with_locals(struct parser_params *, enum node_type, VALUE, VALUE, const rb_code_location_t*); -#endif -static NODE* node_newnode(struct parser_params *, enum node_type, VALUE, VALUE, VALUE, const rb_code_location_t*); -#define rb_node_newnode(type, a1, a2, a3, loc) node_newnode(p, (type), (a1), (a2), (a3), (loc)) +/* This node is parse.y internal */ +struct RNode_DEF_TEMP { + NODE node; + + /* for NODE_DEFN/NODE_DEFS */ + + struct RNode *nd_def; + ID nd_mid; + + struct { + ID cur_arg; + int max_numparam; + NODE *numparam_save; + struct lex_context ctxt; + } save; +}; + +#define RNODE_DEF_TEMP(node) ((struct RNode_DEF_TEMP *)(node)) + +static rb_node_break_t *rb_node_break_new(struct parser_params *p, NODE *nd_stts, const YYLTYPE *loc); +static rb_node_next_t *rb_node_next_new(struct parser_params *p, NODE *nd_stts, const YYLTYPE *loc); +static rb_node_redo_t *rb_node_redo_new(struct parser_params *p, const YYLTYPE *loc); +static rb_node_def_temp_t *rb_node_def_temp_new(struct parser_params *p, const YYLTYPE *loc); +static rb_node_def_temp_t *def_head_save(struct parser_params *p, rb_node_def_temp_t *n); + +#define NEW_BREAK(s,loc) (NODE *)rb_node_break_new(p,s,loc) +#define NEW_NEXT(s,loc) (NODE *)rb_node_next_new(p,s,loc) +#define NEW_REDO(loc) (NODE *)rb_node_redo_new(p,loc) +#define NEW_DEF_TEMP(loc) rb_node_def_temp_new(p,loc) -/* Make a new temporal node, which should not be appeared in the +/* Make a new internal node, which should not be appeared in the * result AST and does not have node_id and location. */ -static NODE* node_new_temporal(struct parser_params *p, enum node_type type, VALUE a0, VALUE a1, VALUE a2); -#define NODE_NEW_TEMPORAL(t,a0,a1,a2) node_new_temporal(p, (t),(VALUE)(a0),(VALUE)(a1),(VALUE)(a2)) +static NODE* node_new_internal(struct parser_params *p, enum node_type type, size_t size, size_t alignment); +#define NODE_NEW_INTERNAL(ndtype, type) (type *)node_new_internal(p, (enum node_type)(ndtype), sizeof(type), RUBY_ALIGNOF(type)) static NODE *nd_set_loc(NODE *nd, const YYLTYPE *loc); @@ -714,11 +1341,10 @@ static void anddot_multiple_assignment_check(struct parser_params* p, const YYLTYPE *loc, ID id) { if (id == tANDDOT) { - yyerror1(loc, "&. inside multiple assignment destination"); + yyerror1(loc, "&. inside multiple assignment destination"); } } -#ifndef RIPPER static inline void set_line_body(NODE *body, int line) { @@ -730,7 +1356,33 @@ set_line_body(NODE *body, int line) } } +static void +set_embraced_location(NODE *node, const rb_code_location_t *beg, const rb_code_location_t *end) +{ + RNODE_ITER(node)->nd_body->nd_loc = code_loc_gen(beg, end); + nd_set_line(node, beg->end_pos.lineno); +} + +static NODE * +last_expr_node(NODE *expr) +{ + while (expr) { + if (nd_type_p(expr, NODE_BLOCK)) { + expr = RNODE_BLOCK(RNODE_BLOCK(expr)->nd_end)->nd_head; + } + else if (nd_type_p(expr, NODE_BEGIN) && RNODE_BEGIN(expr)->nd_body) { + expr = RNODE_BEGIN(expr)->nd_body; + } + else { + break; + } + } + return expr; +} + +#ifndef RIPPER #define yyparse ruby_yyparse +#endif static NODE* cond(struct parser_params *p, NODE *node, const YYLTYPE *loc); static NODE* method_cond(struct parser_params *p, NODE *node, const YYLTYPE *loc); @@ -746,7 +1398,6 @@ static void fixpos(NODE*,NODE*); static int value_expr_gen(struct parser_params*,NODE*); static void void_expr(struct parser_params*,NODE*); static NODE *remove_begin(NODE*); -static NODE *remove_begin_all(NODE*); #define value_expr(node) value_expr_gen(p, (node)) static NODE *void_stmts(struct parser_params*,NODE*); static void reduce_nodes(struct parser_params*,NODE**); @@ -761,6 +1412,7 @@ static NODE *rest_arg_append(struct parser_params *p, NODE *args, NODE *rest_arg static NODE *literal_concat(struct parser_params*,NODE*,NODE*,const YYLTYPE*); static NODE *new_evstr(struct parser_params*,NODE*,const YYLTYPE*); static NODE *new_dstr(struct parser_params*,NODE*,const YYLTYPE*); +static NODE *str2dstr(struct parser_params*,NODE*); static NODE *evstr2dstr(struct parser_params*,NODE*); static NODE *splat_array(NODE*); static void mark_lvar_used(struct parser_params *p, NODE *rhs); @@ -769,11 +1421,11 @@ static NODE *call_bin_op(struct parser_params*,NODE*,ID,NODE*,const YYLTYPE*,con static NODE *call_uni_op(struct parser_params*,NODE*,ID,const YYLTYPE*,const YYLTYPE*); static NODE *new_qcall(struct parser_params* p, ID atype, NODE *recv, ID mid, NODE *args, const YYLTYPE *op_loc, const YYLTYPE *loc); static NODE *new_command_qcall(struct parser_params* p, ID atype, NODE *recv, ID mid, NODE *args, NODE *block, const YYLTYPE *op_loc, const YYLTYPE *loc); -static NODE *method_add_block(struct parser_params*p, NODE *m, NODE *b, const YYLTYPE *loc) {b->nd_iter = m; b->nd_loc = *loc; return b;} +static NODE *method_add_block(struct parser_params*p, NODE *m, NODE *b, const YYLTYPE *loc) {RNODE_ITER(b)->nd_iter = m; b->nd_loc = *loc; return b;} static bool args_info_empty_p(struct rb_args_info *args); -static NODE *new_args(struct parser_params*,NODE*,NODE*,ID,NODE*,NODE*,const YYLTYPE*); -static NODE *new_args_tail(struct parser_params*,NODE*,ID,ID,const YYLTYPE*); +static rb_node_args_t *new_args(struct parser_params*,rb_node_args_aux_t*,rb_node_opt_arg_t*,ID,rb_node_args_aux_t*,rb_node_args_t*,const YYLTYPE*); +static rb_node_args_t *new_args_tail(struct parser_params*,rb_node_kw_arg_t*,ID,ID,const YYLTYPE*); static NODE *new_array_pattern(struct parser_params *p, NODE *constant, NODE *pre_arg, NODE *aryptn, const YYLTYPE *loc); static NODE *new_array_pattern_tail(struct parser_params *p, NODE *pre_args, int has_rest, NODE *rest_arg, NODE *post_args, const YYLTYPE *loc); static NODE *new_find_pattern(struct parser_params *p, NODE *constant, NODE *fndptn, const YYLTYPE *loc); @@ -781,12 +1433,12 @@ static NODE *new_find_pattern_tail(struct parser_params *p, NODE *pre_rest_arg, static NODE *new_hash_pattern(struct parser_params *p, NODE *constant, NODE *hshptn, const YYLTYPE *loc); static NODE *new_hash_pattern_tail(struct parser_params *p, NODE *kw_args, ID kw_rest_arg, const YYLTYPE *loc); -static NODE *new_kw_arg(struct parser_params *p, NODE *k, const YYLTYPE *loc); -static NODE *args_with_numbered(struct parser_params*,NODE*,int); +static rb_node_kw_arg_t *new_kw_arg(struct parser_params *p, NODE *k, const YYLTYPE *loc); +static rb_node_args_t *args_with_numbered(struct parser_params*,rb_node_args_t*,int,ID); -static VALUE negate_lit(struct parser_params*, VALUE); +static NODE* negate_lit(struct parser_params*, NODE*); static NODE *ret_args(struct parser_params*,NODE*); -static NODE *arg_blk_pass(NODE*,NODE*); +static NODE *arg_blk_pass(NODE*,rb_node_block_pass_t*); static NODE *new_yield(struct parser_params*,NODE*,const YYLTYPE*); static NODE *dsym_node(struct parser_params*,NODE*,const YYLTYPE*); @@ -807,8 +1459,8 @@ static NODE *new_bodystmt(struct parser_params *p, NODE *head, NODE *rescue, NOD static NODE *const_decl(struct parser_params *p, NODE* path, const YYLTYPE *loc); -static NODE *opt_arg_append(NODE*, NODE*); -static NODE *kwd_append(NODE*, NODE*); +static rb_node_opt_arg_t *opt_arg_append(rb_node_opt_arg_t*, rb_node_opt_arg_t*); +static rb_node_kw_arg_t *kwd_append(rb_node_kw_arg_t*, rb_node_kw_arg_t*); static NODE *new_hash(struct parser_params *p, NODE *hash, const YYLTYPE *loc); static NODE *new_unique_key_hash(struct parser_params *p, NODE *hash, const YYLTYPE *loc); @@ -827,92 +1479,29 @@ static NODE *match_op(struct parser_params*,NODE*,NODE*,const YYLTYPE*,const YYL static rb_ast_id_table_t *local_tbl(struct parser_params*); -static VALUE reg_compile(struct parser_params*, VALUE, int); -static void reg_fragment_setenc(struct parser_params*, VALUE, int); -static int reg_fragment_check(struct parser_params*, VALUE, int); -static NODE *reg_named_capture_assign(struct parser_params* p, VALUE regexp, const YYLTYPE *loc); +static VALUE reg_compile(struct parser_params*, rb_parser_string_t*, int); +static void reg_fragment_setenc(struct parser_params*, rb_parser_string_t*, int); +#define reg_fragment_check rb_parser_reg_fragment_check +int reg_fragment_check(struct parser_params*, rb_parser_string_t*, int); -static int literal_concat0(struct parser_params *p, VALUE head, VALUE tail); +static int literal_concat0(struct parser_params *p, rb_parser_string_t *head, rb_parser_string_t *tail); static NODE *heredoc_dedent(struct parser_params*,NODE*); static void check_literal_when(struct parser_params *p, NODE *args, const YYLTYPE *loc); -#define get_id(id) (id) -#define get_value(val) (val) -#define get_num(num) (num) -#else /* RIPPER */ -#define NODE_RIPPER NODE_CDECL -#define NEW_RIPPER(a,b,c,loc) (VALUE)NEW_CDECL(a,b,c,loc) -#define NODE_RIPPER2 NODE_OP_CDECL -#define NEW_RIPPER2(a,b,c,loc) (VALUE)NEW_OP_CDECL(a,c,b,loc) - -static inline int ripper_is_node_yylval(VALUE n); - -static inline VALUE -ripper_new_yylval(struct parser_params *p, ID a, VALUE b, VALUE c) -{ - if (ripper_is_node_yylval(c)) c = RNODE(c)->nd_cval; - add_mark_object(p, b); - add_mark_object(p, c); - return NEW_RIPPER(a, b, c, &NULL_LOC); -} - -static inline VALUE -ripper_new_yylval2(struct parser_params *p, VALUE a, VALUE b, VALUE c) -{ - add_mark_object(p, a); - add_mark_object(p, b); - add_mark_object(p, c); - return NEW_RIPPER2(a, b, c, &NULL_LOC); -} - -static inline int -ripper_is_node_yylval(VALUE n) -{ - return RB_TYPE_P(n, T_NODE) && nd_type_p(RNODE(n), NODE_RIPPER); -} - -#define value_expr(node) ((void)(node)) -#define remove_begin(node) (node) -#define void_stmts(p,x) (x) -#define rb_dvar_defined(id, base) 0 -#define rb_local_defined(id, base) 0 -static ID ripper_get_id(VALUE); -#define get_id(id) ripper_get_id(id) -static VALUE ripper_get_value(VALUE); -#define get_value(val) ripper_get_value(val) -#define get_num(num) (int)get_id(num) -static VALUE assignable(struct parser_params*,VALUE); -static int id_is_var(struct parser_params *p, ID id); - -#define method_cond(p,node,loc) (node) -#define call_bin_op(p, recv,id,arg1,op_loc,loc) dispatch3(binary, (recv), STATIC_ID2SYM(id), (arg1)) -#define match_op(p,node1,node2,op_loc,loc) call_bin_op(0, (node1), idEqTilde, (node2), op_loc, loc) -#define call_uni_op(p, recv,id,op_loc,loc) dispatch2(unary, STATIC_ID2SYM(id), (recv)) -#define logop(p,id,node1,node2,op_loc,loc) call_bin_op(0, (node1), (id), (node2), op_loc, loc) - -#define new_nil(loc) Qnil - -static VALUE new_regexp(struct parser_params *, VALUE, VALUE, const YYLTYPE *); - -static VALUE const_decl(struct parser_params *p, VALUE path); - -static VALUE var_field(struct parser_params *p, VALUE a); -static VALUE assign_error(struct parser_params *p, const char *mesg, VALUE a); - -static VALUE parser_reg_compile(struct parser_params*, VALUE, int, VALUE *); - +#ifdef RIPPER +#define get_value(idx) (rb_ary_entry(p->s_value_stack, idx)) +#define set_value(val) (p->s_lvalue = val) static VALUE backref_error(struct parser_params*, NODE *, VALUE); -#endif /* !RIPPER */ - -/* forward declaration */ -typedef struct rb_strterm_heredoc_struct rb_strterm_heredoc_t; +static VALUE assign_error(struct parser_params *p, const char *mesg, VALUE a); +static int id_is_var(struct parser_params *p, ID id); +#endif RUBY_SYMBOL_EXPORT_BEGIN VALUE rb_parser_reg_compile(struct parser_params* p, VALUE str, int options); -int rb_reg_fragment_setenc(struct parser_params*, VALUE, int); +int rb_reg_fragment_setenc(struct parser_params*, rb_parser_string_t *, int); enum lex_state_e rb_parser_trace_lex_state(struct parser_params *, enum lex_state_e, enum lex_state_e, int); -VALUE rb_parser_lex_state_name(enum lex_state_e state); +VALUE rb_parser_lex_state_name(struct parser_params *p, enum lex_state_e state); void rb_parser_show_bitstack(struct parser_params *, stack_type, const char *, int); PRINTF_ARGS(void rb_parser_fatal(struct parser_params *p, const char *fmt, ...), 2, 3); YYLTYPE *rb_parser_set_location_from_strterm_heredoc(struct parser_params *p, rb_strterm_heredoc_t *here, YYLTYPE *yylloc); @@ -921,15 +1510,12 @@ YYLTYPE *rb_parser_set_location_of_heredoc_end(struct parser_params *p, YYLTYPE YYLTYPE *rb_parser_set_location_of_dummy_end(struct parser_params *p, YYLTYPE *yylloc); YYLTYPE *rb_parser_set_location_of_none(struct parser_params *p, YYLTYPE *yylloc); YYLTYPE *rb_parser_set_location(struct parser_params *p, YYLTYPE *yylloc); +void ruby_show_error_line(struct parser_params *p, VALUE errbuf, const YYLTYPE *yylloc, int lineno, rb_parser_string_t *str); RUBY_SYMBOL_EXPORT_END static void error_duplicate_pattern_variable(struct parser_params *p, ID id, const YYLTYPE *loc); static void error_duplicate_pattern_key(struct parser_params *p, ID id, const YYLTYPE *loc); -#ifndef RIPPER static ID formal_argument(struct parser_params*, ID); -#else -static ID formal_argument(struct parser_params*, VALUE); -#endif static ID shadowing_lvar(struct parser_params*,ID); static void new_bv(struct parser_params*,ID); @@ -939,19 +1525,20 @@ static void local_var(struct parser_params*, ID); static void arg_var(struct parser_params*, ID); static int local_id(struct parser_params *p, ID id); static int local_id_ref(struct parser_params*, ID, ID **); -#ifndef RIPPER -static ID internal_id(struct parser_params*); +#define internal_id rb_parser_internal_id +ID internal_id(struct parser_params*); static NODE *new_args_forward_call(struct parser_params*, NODE*, const YYLTYPE*, const YYLTYPE*); -#endif static int check_forwarding_args(struct parser_params*); static void add_forwarding_args(struct parser_params *p); +static void forwarding_arg_check(struct parser_params *p, ID arg, ID all, const char *var); static const struct vtable *dyna_push(struct parser_params *); static void dyna_pop(struct parser_params*, const struct vtable *); static int dyna_in_block(struct parser_params*); #define dyna_var(p, id) local_var(p, id) static int dvar_defined(struct parser_params*, ID); -static int dvar_defined_ref(struct parser_params*, ID, ID**); +#define dvar_defined_ref rb_parser_dvar_defined_ref +int dvar_defined_ref(struct parser_params*, ID, ID**); static int dvar_curr(struct parser_params*,ID); static int lvar_defined(struct parser_params*, ID); @@ -959,18 +1546,17 @@ static int lvar_defined(struct parser_params*, ID); static NODE *numparam_push(struct parser_params *p); static void numparam_pop(struct parser_params *p, NODE *prev_inner); -#ifdef RIPPER -# define METHOD_NOT idNOT -#else -# define METHOD_NOT '!' -#endif +#define METHOD_NOT '!' #define idFWD_REST '*' #define idFWD_KWREST idPow /* Use simple "**", as tDSTAR is "**arg" */ #define idFWD_BLOCK '&' #define idFWD_ALL idDot3 -#define FORWARD_ARGS_WITH_RUBY2_KEYWORDS +#define arg_FWD_BLOCK idFWD_BLOCK +#define RE_ONIG_OPTION_IGNORECASE 1 +#define RE_ONIG_OPTION_EXTEND (RE_ONIG_OPTION_IGNORECASE<<1) +#define RE_ONIG_OPTION_MULTILINE (RE_ONIG_OPTION_EXTEND<<1) #define RE_OPTION_ONCE (1<<16) #define RE_OPTION_ENCODING_SHIFT 8 #define RE_OPTION_ENCODING(e) (((e)&0xff)<<RE_OPTION_ENCODING_SHIFT) @@ -979,71 +1565,8 @@ static void numparam_pop(struct parser_params *p, NODE *prev_inner); #define RE_OPTION_MASK 0xff #define RE_OPTION_ARG_ENCODING_NONE 32 -/* structs for managing terminator of string literal and heredocment */ -typedef struct rb_strterm_literal_struct { - union { - VALUE dummy; - long nest; - } u0; - union { - VALUE dummy; - long func; /* STR_FUNC_* (e.g., STR_FUNC_ESCAPE and STR_FUNC_EXPAND) */ - } u1; - union { - VALUE dummy; - long paren; /* '(' of `%q(...)` */ - } u2; - union { - VALUE dummy; - long term; /* ')' of `%q(...)` */ - } u3; -} rb_strterm_literal_t; - -#define HERETERM_LENGTH_BITS ((SIZEOF_VALUE - 1) * CHAR_BIT - 1) - -struct rb_strterm_heredoc_struct { - VALUE lastline; /* the string of line that contains `<<"END"` */ - long offset; /* the column of END in `<<"END"` */ - int sourceline; /* lineno of the line that contains `<<"END"` */ - unsigned length /* the length of END in `<<"END"` */ -#if HERETERM_LENGTH_BITS < SIZEOF_INT * CHAR_BIT - : HERETERM_LENGTH_BITS -# define HERETERM_LENGTH_MAX ((1U << HERETERM_LENGTH_BITS) - 1) -#else -# define HERETERM_LENGTH_MAX UINT_MAX -#endif - ; -#if HERETERM_LENGTH_BITS < SIZEOF_INT * CHAR_BIT - unsigned quote: 1; - unsigned func: 8; -#else - uint8_t quote; - uint8_t func; -#endif -}; -STATIC_ASSERT(rb_strterm_heredoc_t, sizeof(rb_strterm_heredoc_t) <= 4 * SIZEOF_VALUE); - -#define STRTERM_HEREDOC IMEMO_FL_USER0 - -struct rb_strterm_struct { - VALUE flags; - union { - rb_strterm_literal_t literal; - rb_strterm_heredoc_t heredoc; - } u; -}; - -#ifndef RIPPER -void -rb_strterm_mark(VALUE obj) -{ - rb_strterm_t *strterm = (rb_strterm_t*)obj; - if (RBASIC(obj)->flags & STRTERM_HEREDOC) { - rb_strterm_heredoc_t *heredoc = &strterm->u.heredoc; - rb_gc_mark(heredoc->lastline); - } -} -#endif +#define CHECK_LITERAL_WHEN (st_table *)1 +#define CASE_LABELS_ENABLED_P(case_labels) (case_labels && case_labels != CHECK_LITERAL_WHEN) #define yytnamerr(yyres, yystr) (YYSIZE_T)rb_yytnamerr(p, yyres, yystr) size_t rb_yytnamerr(struct parser_params *p, char *yyres, const char *yystr); @@ -1060,12 +1583,11 @@ size_t rb_yytnamerr(struct parser_params *p, char *yyres, const char *yystr); /****** Ripper *******/ #ifdef RIPPER -#define RIPPER_VERSION "0.1.0" -static inline VALUE intern_sym(const char *name); +#include "eventids1.h" +#include "eventids2.h" -#include "eventids1.c" -#include "eventids2.c" +extern const struct ripper_parser_ids ripper_parser_ids; static VALUE ripper_dispatch0(struct parser_params*,ID); static VALUE ripper_dispatch1(struct parser_params*,ID,VALUE); @@ -1074,7 +1596,7 @@ static VALUE ripper_dispatch3(struct parser_params*,ID,VALUE,VALUE,VALUE); static VALUE ripper_dispatch4(struct parser_params*,ID,VALUE,VALUE,VALUE,VALUE); static VALUE ripper_dispatch5(struct parser_params*,ID,VALUE,VALUE,VALUE,VALUE,VALUE); static VALUE ripper_dispatch7(struct parser_params*,ID,VALUE,VALUE,VALUE,VALUE,VALUE,VALUE,VALUE); -static void ripper_error(struct parser_params *p); +void ripper_error(struct parser_params *p); #define dispatch0(n) ripper_dispatch0(p, TOKEN_PASTE(ripper_id_, n)) #define dispatch1(n,a) ripper_dispatch1(p, TOKEN_PASTE(ripper_id_, n), (a)) @@ -1086,45 +1608,11 @@ static void ripper_error(struct parser_params *p); #define yyparse ripper_yyparse -#define ID2VAL(id) STATIC_ID2SYM(id) -#define TOKEN2VAL(t) ID2VAL(TOKEN2ID(t)) -#define KWD2EID(t, v) ripper_new_yylval(p, keyword_##t, get_value(v), 0) - -#define params_new(pars, opts, rest, pars2, kws, kwrest, blk) \ - dispatch7(params, (pars), (opts), (rest), (pars2), (kws), (kwrest), (blk)) - -#define escape_Qundef(x) ((x)==Qundef ? Qnil : (x)) - -static inline VALUE -new_args(struct parser_params *p, VALUE pre_args, VALUE opt_args, VALUE rest_arg, VALUE post_args, VALUE tail, YYLTYPE *loc) -{ - NODE *t = (NODE *)tail; - VALUE kw_args = t->u1.value, kw_rest_arg = t->u2.value, block = t->u3.value; - return params_new(pre_args, opt_args, rest_arg, post_args, kw_args, kw_rest_arg, escape_Qundef(block)); -} - -static inline VALUE -new_args_tail(struct parser_params *p, VALUE kw_args, VALUE kw_rest_arg, VALUE block, YYLTYPE *loc) -{ - NODE *t = rb_node_newnode(NODE_ARGS_AUX, kw_args, kw_rest_arg, block, &NULL_LOC); - add_mark_object(p, kw_args); - add_mark_object(p, kw_rest_arg); - add_mark_object(p, block); - return (VALUE)t; -} - -static inline VALUE -args_with_numbered(struct parser_params *p, VALUE args, int max_numparam) -{ - return args; -} +static VALUE ripper_formal_argument(struct parser_params *p, ID id, VALUE lhs); static VALUE -new_array_pattern(struct parser_params *p, VALUE constant, VALUE pre_arg, VALUE aryptn, const YYLTYPE *loc) +aryptn_pre_args(struct parser_params *p, VALUE pre_arg, VALUE pre_args) { - NODE *t = (NODE *)aryptn; - VALUE pre_args = t->u1.value, rest_arg = t->u2.value, post_args = t->u3.value; - if (!NIL_P(pre_arg)) { if (!NIL_P(pre_args)) { rb_ary_unshift(pre_args, pre_arg); @@ -1133,75 +1621,22 @@ new_array_pattern(struct parser_params *p, VALUE constant, VALUE pre_arg, VALUE pre_args = rb_ary_new_from_args(1, pre_arg); } } - return dispatch4(aryptn, constant, pre_args, rest_arg, post_args); -} - -static VALUE -new_array_pattern_tail(struct parser_params *p, VALUE pre_args, VALUE has_rest, VALUE rest_arg, VALUE post_args, const YYLTYPE *loc) -{ - return ripper_new_yylval2(p, pre_args, rest_arg, post_args); -} - -static VALUE -new_find_pattern(struct parser_params *p, VALUE constant, VALUE fndptn, const YYLTYPE *loc) -{ - NODE *t = (NODE *)fndptn; - VALUE pre_rest_arg = t->u1.value, args = t->u2.value, post_rest_arg = t->u3.value; - - return dispatch4(fndptn, constant, pre_rest_arg, args, post_rest_arg); -} - -static VALUE -new_find_pattern_tail(struct parser_params *p, VALUE pre_rest_arg, VALUE args, VALUE post_rest_arg, const YYLTYPE *loc) -{ - return ripper_new_yylval2(p, pre_rest_arg, args, post_rest_arg); -} - -#define new_hash(p,h,l) rb_ary_new_from_args(0) - -static VALUE -new_unique_key_hash(struct parser_params *p, VALUE ary, const YYLTYPE *loc) -{ - return ary; -} - -static VALUE -new_hash_pattern(struct parser_params *p, VALUE constant, VALUE hshptn, const YYLTYPE *loc) -{ - NODE *t = (NODE *)hshptn; - VALUE kw_args = t->u1.value, kw_rest_arg = t->u2.value; - return dispatch3(hshptn, constant, kw_args, kw_rest_arg); -} - -static VALUE -new_hash_pattern_tail(struct parser_params *p, VALUE kw_args, VALUE kw_rest_arg, const YYLTYPE *loc) -{ - if (kw_rest_arg) { - kw_rest_arg = dispatch1(var_field, kw_rest_arg); - } - else { - kw_rest_arg = Qnil; - } - return ripper_new_yylval2(p, kw_args, kw_rest_arg, Qnil); + return pre_args; } -#define new_defined(p,expr,loc) dispatch1(defined, (expr)) - -static VALUE heredoc_dedent(struct parser_params*,VALUE); +#define ID2VAL(id) STATIC_ID2SYM(id) +#define TOKEN2VAL(t) ID2VAL(TOKEN2ID(t)) +#endif /* RIPPER */ -#else -#define ID2VAL(id) (id) -#define TOKEN2VAL(t) ID2VAL(t) #define KWD2EID(t, v) keyword_##t static NODE * -set_defun_body(struct parser_params *p, NODE *n, NODE *args, NODE *body, const YYLTYPE *loc) +new_scope_body(struct parser_params *p, rb_node_args_t *args, NODE *body, const YYLTYPE *loc) { body = remove_begin(body); reduce_nodes(p, &body); - n->nd_defn = NEW_SCOPE(args, body, loc); - n->nd_loc = *loc; - nd_set_line(n->nd_defn, loc->end_pos.lineno); + NODE *n = NEW_SCOPE(args, body, loc); + nd_set_line(n, loc->end_pos.lineno); set_line_body(body, loc->beg_pos.lineno); return n; } @@ -1216,27 +1651,35 @@ rescued_expr(struct parser_params *p, NODE *arg, NODE *rescue, return NEW_RESCUE(arg, rescue, 0, &loc); } -#endif /* RIPPER */ +static NODE *add_block_exit(struct parser_params *p, NODE *node); +static rb_node_exits_t *init_block_exit(struct parser_params *p); +static rb_node_exits_t *allow_block_exit(struct parser_params *p); +static void restore_block_exit(struct parser_params *p, rb_node_exits_t *exits); +static void clear_block_exit(struct parser_params *p, bool error); static void -restore_defun(struct parser_params *p, NODE *name) +next_rescue_context(struct lex_context *next, const struct lex_context *outer, enum rescue_context def) { - NODE *save = name->nd_next; - YYSTYPE c = {.val = save->nd_cval}; - p->cur_arg = name->nd_vid; - p->ctxt.in_def = c.ctxt.in_def; - p->ctxt.shareable_constant_value = c.ctxt.shareable_constant_value; - p->max_numparam = (int)save->nd_nth; - numparam_pop(p, save->nd_head); + next->in_rescue = outer->in_rescue == after_rescue ? after_rescue : def; } static void -endless_method_name(struct parser_params *p, NODE *defn, const YYLTYPE *loc) +restore_defun(struct parser_params *p, rb_node_def_temp_t *temp) +{ + /* See: def_name action */ + struct lex_context ctxt = temp->save.ctxt; + p->cur_arg = temp->save.cur_arg; + p->ctxt.in_def = ctxt.in_def; + p->ctxt.shareable_constant_value = ctxt.shareable_constant_value; + p->ctxt.in_rescue = ctxt.in_rescue; + p->max_numparam = temp->save.max_numparam; + numparam_pop(p, temp->save.numparam_save); + clear_block_exit(p, true); +} + +static void +endless_method_name(struct parser_params *p, ID mid, const YYLTYPE *loc) { -#ifdef RIPPER - defn = defn->nd_defn; -#endif - ID mid = defn->nd_mid; if (is_attrset_id(mid)) { yyerror1(loc, "setter method cannot be defined in an endless method definition"); } @@ -1253,14 +1696,24 @@ endless_method_name(struct parser_params *p, NODE *defn, const YYLTYPE *loc) } \ } while (0) +#define begin_definition(k, loc_beg, loc_end) \ + do { \ + if (!(p->ctxt.in_class = (k)[0] != 0)) { \ + p->ctxt.in_def = 0; \ + } \ + else if (p->ctxt.in_def) { \ + YYLTYPE loc = code_loc_gen(loc_beg, loc_end); \ + yyerror1(&loc, k " definition in method body"); \ + } \ + local_push(p, 0); \ + } while (0) + #ifndef RIPPER -# define Qnone 0 -# define Qnull 0 # define ifndef_ripper(x) (x) +# define ifdef_ripper(r,x) (x) #else -# define Qnone Qnil -# define Qnull Qundef # define ifndef_ripper(x) +# define ifdef_ripper(r,x) (r) #endif # define rb_warn0(fmt) WARN_CALL(WARN_ARGS(fmt, 1)) @@ -1284,14 +1737,13 @@ endless_method_name(struct parser_params *p, NODE *defn, const YYLTYPE *loc) # define rb_warning3L(l,fmt,a,b,c) WARNING_CALL(WARNING_ARGS_L(l, fmt, 4), (a), (b), (c)) # define rb_warning4L(l,fmt,a,b,c,d) WARNING_CALL(WARNING_ARGS_L(l, fmt, 5), (a), (b), (c), (d)) #ifdef RIPPER -static ID id_warn, id_warning, id_gets, id_assoc; +extern const ID id_warn, id_warning, id_gets, id_assoc; # define ERR_MESG() STR_NEW2(mesg) /* to bypass Ripper DSL */ # define WARN_S_L(s,l) STR_NEW(s,l) # define WARN_S(s) STR_NEW2(s) # define WARN_I(i) INT2NUM(i) # define WARN_ID(i) rb_id2str(i) -# define WARN_IVAL(i) i -# define PRIsWARN "s" +# define PRIsWARN PRIsVALUE # define rb_warn0L_experimental(l,fmt) WARN_CALL(WARN_ARGS_L(l, fmt, 1)) # define WARN_ARGS(fmt,n) p->value, id_warn, n, rb_usascii_str_new_lit(fmt) # define WARN_ARGS_L(l,fmt,n) WARN_ARGS(fmt,n) @@ -1307,14 +1759,12 @@ static ID id_warn, id_warning, id_gets, id_assoc; # else # define WARNING_CALL rb_funcall # endif -PRINTF_ARGS(static void ripper_compile_error(struct parser_params*, const char *fmt, ...), 2, 3); # define compile_error ripper_compile_error #else # define WARN_S_L(s,l) s # define WARN_S(s) s # define WARN_I(i) i # define WARN_ID(i) rb_id2name(i) -# define WARN_IVAL(i) NUM2INT(i) # define PRIsWARN PRIsVALUE # define WARN_ARGS(fmt,n) WARN_ARGS_L(p->ruby_sourceline,fmt,n) # define WARN_ARGS_L(l,fmt,n) p->ruby_sourcefile, (l), (fmt) @@ -1323,125 +1773,999 @@ PRINTF_ARGS(static void ripper_compile_error(struct parser_params*, const char * # define WARNING_ARGS(fmt,n) WARN_ARGS(fmt,n) # define WARNING_ARGS_L(l,fmt,n) WARN_ARGS_L(l,fmt,n) # define WARNING_CALL rb_compile_warning -PRINTF_ARGS(static void parser_compile_error(struct parser_params*, const char *fmt, ...), 2, 3); -# define compile_error parser_compile_error +PRINTF_ARGS(static void parser_compile_error(struct parser_params*, const rb_code_location_t *loc, const char *fmt, ...), 3, 4); +# define compile_error(p, ...) parser_compile_error(p, NULL, __VA_ARGS__) #endif +struct RNode_EXITS { + NODE node; + + NODE *nd_chain; /* Assume NODE_BREAK, NODE_NEXT, NODE_REDO have nd_chain here */ + NODE *nd_end; +}; + +#define RNODE_EXITS(node) ((rb_node_exits_t*)(node)) + +static NODE * +add_block_exit(struct parser_params *p, NODE *node) +{ + if (!node) { + compile_error(p, "unexpected null node"); + return 0; + } + switch (nd_type(node)) { + case NODE_BREAK: case NODE_NEXT: case NODE_REDO: break; + default: + compile_error(p, "unexpected node: %s", parser_node_name(nd_type(node))); + return node; + } + if (!p->ctxt.in_defined) { + rb_node_exits_t *exits = p->exits; + if (exits) { + RNODE_EXITS(exits->nd_end)->nd_chain = node; + exits->nd_end = node; + } + } + return node; +} + +static rb_node_exits_t * +init_block_exit(struct parser_params *p) +{ + rb_node_exits_t *old = p->exits; + rb_node_exits_t *exits = NODE_NEW_INTERNAL(NODE_EXITS, rb_node_exits_t); + exits->nd_chain = 0; + exits->nd_end = RNODE(exits); + p->exits = exits; + return old; +} + +static rb_node_exits_t * +allow_block_exit(struct parser_params *p) +{ + rb_node_exits_t *exits = p->exits; + p->exits = 0; + return exits; +} + +static void +restore_block_exit(struct parser_params *p, rb_node_exits_t *exits) +{ + p->exits = exits; +} + +static void +clear_block_exit(struct parser_params *p, bool error) +{ + rb_node_exits_t *exits = p->exits; + if (!exits) return; + if (error && !compile_for_eval) { + for (NODE *e = RNODE(exits); (e = RNODE_EXITS(e)->nd_chain) != 0; ) { + switch (nd_type(e)) { + case NODE_BREAK: + yyerror1(&e->nd_loc, "Invalid break"); + break; + case NODE_NEXT: + yyerror1(&e->nd_loc, "Invalid next"); + break; + case NODE_REDO: + yyerror1(&e->nd_loc, "Invalid redo"); + break; + default: + yyerror1(&e->nd_loc, "unexpected node"); + goto end_checks; /* no nd_chain */ + } + } + end_checks:; + } + exits->nd_end = RNODE(exits); + exits->nd_chain = 0; +} + #define WARN_EOL(tok) \ (looking_at_eol_p(p) ? \ - (void)rb_warning0("`" tok "' at the end of line without an expression") : \ + (void)rb_warning0("'" tok "' at the end of line without an expression") : \ (void)0) static int looking_at_eol_p(struct parser_params *p); + +static NODE * +get_nd_value(struct parser_params *p, NODE *node) +{ + switch (nd_type(node)) { + case NODE_GASGN: + return RNODE_GASGN(node)->nd_value; + case NODE_IASGN: + return RNODE_IASGN(node)->nd_value; + case NODE_LASGN: + return RNODE_LASGN(node)->nd_value; + case NODE_DASGN: + return RNODE_DASGN(node)->nd_value; + case NODE_MASGN: + return RNODE_MASGN(node)->nd_value; + case NODE_CVASGN: + return RNODE_CVASGN(node)->nd_value; + case NODE_CDECL: + return RNODE_CDECL(node)->nd_value; + default: + compile_error(p, "unexpected node: %s", parser_node_name(nd_type(node))); + return 0; + } +} + +static void +set_nd_value(struct parser_params *p, NODE *node, NODE *rhs) +{ + switch (nd_type(node)) { + case NODE_CDECL: + RNODE_CDECL(node)->nd_value = rhs; + break; + case NODE_GASGN: + RNODE_GASGN(node)->nd_value = rhs; + break; + case NODE_IASGN: + RNODE_IASGN(node)->nd_value = rhs; + break; + case NODE_LASGN: + RNODE_LASGN(node)->nd_value = rhs; + break; + case NODE_DASGN: + RNODE_DASGN(node)->nd_value = rhs; + break; + case NODE_MASGN: + RNODE_MASGN(node)->nd_value = rhs; + break; + case NODE_CVASGN: + RNODE_CVASGN(node)->nd_value = rhs; + break; + default: + compile_error(p, "unexpected node: %s", parser_node_name(nd_type(node))); + break; + } +} + +static ID +get_nd_vid(struct parser_params *p, NODE *node) +{ + switch (nd_type(node)) { + case NODE_CDECL: + return RNODE_CDECL(node)->nd_vid; + case NODE_GASGN: + return RNODE_GASGN(node)->nd_vid; + case NODE_IASGN: + return RNODE_IASGN(node)->nd_vid; + case NODE_LASGN: + return RNODE_LASGN(node)->nd_vid; + case NODE_DASGN: + return RNODE_DASGN(node)->nd_vid; + case NODE_CVASGN: + return RNODE_CVASGN(node)->nd_vid; + default: + compile_error(p, "unexpected node: %s", parser_node_name(nd_type(node))); + return 0; + } +} + +static NODE * +get_nd_args(struct parser_params *p, NODE *node) +{ + switch (nd_type(node)) { + case NODE_CALL: + return RNODE_CALL(node)->nd_args; + case NODE_OPCALL: + return RNODE_OPCALL(node)->nd_args; + case NODE_FCALL: + return RNODE_FCALL(node)->nd_args; + case NODE_QCALL: + return RNODE_QCALL(node)->nd_args; + case NODE_SUPER: + return RNODE_SUPER(node)->nd_args; + case NODE_VCALL: + case NODE_ZSUPER: + case NODE_YIELD: + case NODE_RETURN: + case NODE_BREAK: + case NODE_NEXT: + return 0; + default: + compile_error(p, "unexpected node: %s", parser_node_name(nd_type(node))); + return 0; + } +} + +static st_index_t +djb2(const uint8_t *str, size_t len) +{ + st_index_t hash = 5381; + + for (size_t i = 0; i < len; i++) { + hash = ((hash << 5) + hash) + str[i]; + } + + return hash; +} + +static st_index_t +parser_memhash(const void *ptr, long len) +{ + return djb2(ptr, len); +} + +#define PARSER_STRING_PTR(str) (str->ptr) +#define PARSER_STRING_LEN(str) (str->len) +#define PARSER_STRING_END(str) (&str->ptr[str->len]) +#define STRING_SIZE(str) ((size_t)str->len + 1) +#define STRING_TERM_LEN(str) (1) +#define STRING_TERM_FILL(str) (str->ptr[str->len] = '\0') +#define PARSER_STRING_RESIZE_CAPA_TERM(p,str,capacity,termlen) do {\ + SIZED_REALLOC_N(str->ptr, char, (size_t)total + termlen, STRING_SIZE(str)); \ + str->len = total; \ +} while (0) +#define STRING_SET_LEN(str, n) do { \ + (str)->len = (n); \ +} while (0) +#define PARSER_STRING_GETMEM(str, ptrvar, lenvar) \ + ((ptrvar) = str->ptr, \ + (lenvar) = str->len) + +static inline bool +parser_string_end_with_newline_p(struct parser_params *p, rb_parser_string_t *str) +{ + return PARSER_STRING_LEN(str) > 0 && PARSER_STRING_END(str)[-1] == '\n'; +} + +static rb_parser_string_t * +rb_parser_string_new(rb_parser_t *p, const char *ptr, long len) +{ + rb_parser_string_t *str; + + if (len < 0) { + rb_bug("negative string size (or size too big): %ld", len); + } + + str = xcalloc(1, sizeof(rb_parser_string_t)); + str->ptr = xcalloc(len + 1, sizeof(char)); + + if (ptr) { + memcpy(PARSER_STRING_PTR(str), ptr, len); + } + STRING_SET_LEN(str, len); + STRING_TERM_FILL(str); + return str; +} + +static rb_parser_string_t * +rb_parser_encoding_string_new(rb_parser_t *p, const char *ptr, long len, rb_encoding *enc) +{ + rb_parser_string_t *str = rb_parser_string_new(p, ptr, len); + str->coderange = RB_PARSER_ENC_CODERANGE_UNKNOWN; + str->enc = enc; + return str; +} + +#ifndef RIPPER +static bool +zero_filled(const char *s, int n) +{ + for (; n > 0; --n) { + if (*s++) return false; + } + return true; +} + +static bool +str_null_char(rb_parser_t *p, const char *s, long len, const int minlen, rb_encoding *enc) +{ + const char *e = s + len; + + for (; s + minlen <= e; s += rb_enc_mbclen(s, e, enc)) { + if (zero_filled(s, minlen)) return true; + } + return false; +} + +static bool +cstr_null_check(rb_parser_t *p, const char *s, long len, rb_encoding *enc, int *w) +{ + const int minlen = rb_enc_mbminlen(enc); + + if (minlen > 1) { + *w = 1; + return str_null_char(p, s, len, minlen, enc); + } + else { + *w = 0; + return (!s || memchr(s, 0, len)); + } +} + +rb_parser_string_t * +rb_str_to_parser_string(rb_parser_t *p, VALUE str) +{ + /* Type check */ + return rb_parser_encoding_string_new(p, RSTRING_PTR(str), RSTRING_LEN(str), rb_enc_get(str)); +} +#endif + +static void +rb_parser_string_free(rb_parser_t *p, rb_parser_string_t *str) +{ + if (!str) return; + xfree(PARSER_STRING_PTR(str)); + xfree(str); +} + +static st_index_t +rb_parser_str_hash(rb_parser_string_t *str) +{ + return parser_memhash((const void *)PARSER_STRING_PTR(str), PARSER_STRING_LEN(str)); +} + +static st_index_t +rb_char_p_hash(const char *c) +{ + return parser_memhash((const void *)c, strlen(c)); +} + +static size_t +rb_parser_str_capacity(rb_parser_string_t *str, const int termlen) +{ + return PARSER_STRING_LEN(str); +} + +#ifndef RIPPER +static char * +rb_parser_string_end(rb_parser_string_t *str) +{ + return &str->ptr[str->len]; +} +#endif + +static void +rb_parser_string_set_encoding(rb_parser_string_t *str, rb_encoding *enc) +{ + str->enc = enc; +} + +static rb_encoding * +rb_parser_str_get_encoding(rb_parser_string_t *str) +{ + return str->enc; +} + +#ifndef RIPPER +static bool +PARSER_ENCODING_IS_ASCII8BIT(struct parser_params *p, rb_parser_string_t *str) +{ + return rb_parser_str_get_encoding(str) == rb_ascii8bit_encoding(); +} +#endif + +static int +PARSER_ENC_CODERANGE(rb_parser_string_t *str) +{ + return str->coderange; +} + +static void +PARSER_ENC_CODERANGE_SET(rb_parser_string_t *str, int coderange) +{ + str->coderange = coderange; +} + +static void +PARSER_ENCODING_CODERANGE_SET(rb_parser_string_t *str, rb_encoding *enc, enum rb_parser_string_coderange_type cr) +{ + rb_parser_string_set_encoding(str, enc); + PARSER_ENC_CODERANGE_SET(str, cr); +} + +static void +PARSER_ENC_CODERANGE_CLEAR(rb_parser_string_t *str) +{ + str->coderange = RB_PARSER_ENC_CODERANGE_UNKNOWN; +} + +static bool +PARSER_ENC_CODERANGE_ASCIIONLY(rb_parser_string_t *str) +{ + return PARSER_ENC_CODERANGE(str) == RB_PARSER_ENC_CODERANGE_7BIT; +} + +static bool +PARSER_ENC_CODERANGE_CLEAN_P(int cr) +{ + return cr == RB_PARSER_ENC_CODERANGE_7BIT || cr == RB_PARSER_ENC_CODERANGE_VALID; +} + +static const char * +rb_parser_search_nonascii(const char *p, const char *e) +{ + const char *s = p; + + for (; s < e; s++) { + if (*s & 0x80) return s; + } + + return NULL; +} + +static int +rb_parser_coderange_scan(struct parser_params *p, const char *ptr, long len, rb_encoding *enc) +{ + const char *e = ptr + len; + + if (enc == rb_ascii8bit_encoding()) { + /* enc is ASCII-8BIT. ASCII-8BIT string never be broken. */ + ptr = rb_parser_search_nonascii(ptr, e); + return ptr ? RB_PARSER_ENC_CODERANGE_VALID : RB_PARSER_ENC_CODERANGE_7BIT; + } + + /* parser string encoding is always asciicompat */ + ptr = rb_parser_search_nonascii(ptr, e); + if (!ptr) return RB_PARSER_ENC_CODERANGE_7BIT; + for (;;) { + int ret = rb_enc_precise_mbclen(ptr, e, enc); + if (!MBCLEN_CHARFOUND_P(ret)) return RB_PARSER_ENC_CODERANGE_BROKEN; + ptr += MBCLEN_CHARFOUND_LEN(ret); + if (ptr == e) break; + ptr = rb_parser_search_nonascii(ptr, e); + if (!ptr) break; + } + + return RB_PARSER_ENC_CODERANGE_VALID; +} + +static int +rb_parser_enc_coderange_scan(struct parser_params *p, rb_parser_string_t *str, rb_encoding *enc) +{ + return rb_parser_coderange_scan(p, PARSER_STRING_PTR(str), PARSER_STRING_LEN(str), enc); +} + +static int +rb_parser_enc_str_coderange(struct parser_params *p, rb_parser_string_t *str) +{ + int cr = PARSER_ENC_CODERANGE(str); + + if (cr == RB_PARSER_ENC_CODERANGE_UNKNOWN) { + cr = rb_parser_enc_coderange_scan(p, str, rb_parser_str_get_encoding(str)); + PARSER_ENC_CODERANGE_SET(str, cr); + } + + return cr; +} + +static rb_parser_string_t * +rb_parser_enc_associate(struct parser_params *p, rb_parser_string_t *str, rb_encoding *enc) +{ + if (rb_parser_str_get_encoding(str) == enc) + return str; + if (!PARSER_ENC_CODERANGE_ASCIIONLY(str) || + !rb_enc_asciicompat(enc)) { + PARSER_ENC_CODERANGE_CLEAR(str); + } + rb_parser_string_set_encoding(str, enc); + return str; +} + +static bool +rb_parser_is_ascii_string(struct parser_params *p, rb_parser_string_t *str) +{ + return rb_parser_enc_str_coderange(p, str) == RB_PARSER_ENC_CODERANGE_7BIT; +} + +static int +rb_parser_enc_str_asciionly_p(struct parser_params *p, rb_parser_string_t *str) +{ + rb_encoding *enc = rb_parser_str_get_encoding(str); + + if (!rb_enc_asciicompat(enc)) + return FALSE; + else if (rb_parser_is_ascii_string(p, str)) + return TRUE; + return FALSE; +} + +static rb_encoding * +rb_parser_enc_compatible_latter(struct parser_params *p, rb_parser_string_t *str1, rb_parser_string_t *str2, rb_encoding *enc1, rb_encoding *enc2) +{ + int cr1, cr2; + + if (PARSER_STRING_LEN(str2) == 0) + return enc1; + if (PARSER_STRING_LEN(str1) == 0) + return (rb_enc_asciicompat(enc1) && rb_parser_enc_str_asciionly_p(p, str2)) ? enc1 : enc2; + if (!rb_enc_asciicompat(enc1) || !rb_enc_asciicompat(enc2)) { + return 0; + } + + cr1 = rb_parser_enc_str_coderange(p, str1); + cr2 = rb_parser_enc_str_coderange(p, str2); + + if (cr1 != cr2) { + if (cr1 == RB_PARSER_ENC_CODERANGE_7BIT) return enc2; + if (cr2 == RB_PARSER_ENC_CODERANGE_7BIT) return enc1; + } + + if (cr2 == RB_PARSER_ENC_CODERANGE_7BIT) { + return enc1; + } + + if (cr1 == RB_PARSER_ENC_CODERANGE_7BIT) { + return enc2; + } + + return 0; +} + +static rb_encoding * +rb_parser_enc_compatible(struct parser_params *p, rb_parser_string_t *str1, rb_parser_string_t *str2) +{ + rb_encoding *enc1 = rb_parser_str_get_encoding(str1); + rb_encoding *enc2 = rb_parser_str_get_encoding(str2); + + if (enc1 == NULL || enc2 == NULL) + return 0; + + if (enc1 == enc2) { + return enc1; + } + + return rb_parser_enc_compatible_latter(p, str1, str2, enc1, enc2); +} + +static void +rb_parser_str_modify(rb_parser_string_t *str) +{ + PARSER_ENC_CODERANGE_CLEAR(str); +} + +static void +rb_parser_str_set_len(struct parser_params *p, rb_parser_string_t *str, long len) +{ + long capa; + const int termlen = STRING_TERM_LEN(str); + + if (len > (capa = (long)(rb_parser_str_capacity(str, termlen))) || len < 0) { + rb_bug("probable buffer overflow: %ld for %ld", len, capa); + } + + int cr = PARSER_ENC_CODERANGE(str); + if (cr == RB_PARSER_ENC_CODERANGE_UNKNOWN) { + /* Leave unknown. */ + } + else if (len > PARSER_STRING_LEN(str)) { + PARSER_ENC_CODERANGE_SET(str, RB_PARSER_ENC_CODERANGE_UNKNOWN); + } + else if (len < PARSER_STRING_LEN(str)) { + if (cr != RB_PARSER_ENC_CODERANGE_7BIT) { + /* ASCII-only string is keeping after truncated. Valid + * and broken may be invalid or valid, leave unknown. */ + PARSER_ENC_CODERANGE_SET(str, RB_PARSER_ENC_CODERANGE_UNKNOWN); + } + } + + STRING_SET_LEN(str, len); + STRING_TERM_FILL(str); +} + +static rb_parser_string_t * +rb_parser_str_buf_cat(struct parser_params *p, rb_parser_string_t *str, const char *ptr, long len) +{ + rb_parser_str_modify(str); + if (len == 0) return 0; + + long total, olen, off = -1; + char *sptr; + const int termlen = STRING_TERM_LEN(str); + + PARSER_STRING_GETMEM(str, sptr, olen); + if (ptr >= sptr && ptr <= sptr + olen) { + off = ptr - sptr; + } + + if (olen > LONG_MAX - len) { + compile_error(p, "string sizes too big"); + return 0; + } + total = olen + len; + PARSER_STRING_RESIZE_CAPA_TERM(p, str, total, termlen); + sptr = PARSER_STRING_PTR(str); + if (off != -1) { + ptr = sptr + off; + } + memcpy(sptr + olen, ptr, len); + STRING_SET_LEN(str, total); + STRING_TERM_FILL(str); + + return str; +} + +static rb_parser_string_t * +rb_parser_enc_cr_str_buf_cat(struct parser_params *p, rb_parser_string_t *str, const char *ptr, long len, + rb_encoding *ptr_enc, int ptr_cr, int *ptr_cr_ret) +{ + int str_cr, res_cr; + rb_encoding *str_enc, *res_enc; + + str_enc = rb_parser_str_get_encoding(str); + str_cr = PARSER_STRING_LEN(str) ? PARSER_ENC_CODERANGE(str) : RB_PARSER_ENC_CODERANGE_7BIT; + + if (str_enc == ptr_enc) { + if (str_cr != RB_PARSER_ENC_CODERANGE_UNKNOWN && ptr_cr == RB_PARSER_ENC_CODERANGE_UNKNOWN) { + ptr_cr = rb_parser_coderange_scan(p, ptr, len, ptr_enc); + } + } + else { + /* parser string encoding is always asciicompat */ + if (ptr_cr == RB_PARSER_ENC_CODERANGE_UNKNOWN) { + ptr_cr = rb_parser_coderange_scan(p, ptr, len, ptr_enc); + } + if (str_cr == RB_PARSER_ENC_CODERANGE_UNKNOWN) { + if (str_enc == rb_ascii8bit_encoding() || ptr_cr != RB_PARSER_ENC_CODERANGE_7BIT) { + str_cr = rb_parser_enc_str_coderange(p, str); + } + } + } + if (ptr_cr_ret) + *ptr_cr_ret = ptr_cr; + + if (str_enc != ptr_enc && + str_cr != RB_PARSER_ENC_CODERANGE_7BIT && + ptr_cr != RB_PARSER_ENC_CODERANGE_7BIT) { + goto incompatible; + } + + if (str_cr == RB_PARSER_ENC_CODERANGE_UNKNOWN) { + res_enc = str_enc; + res_cr = RB_PARSER_ENC_CODERANGE_UNKNOWN; + } + else if (str_cr == RB_PARSER_ENC_CODERANGE_7BIT) { + if (ptr_cr == RB_PARSER_ENC_CODERANGE_7BIT) { + res_enc = str_enc; + res_cr = RB_PARSER_ENC_CODERANGE_7BIT; + } + else { + res_enc = ptr_enc; + res_cr = ptr_cr; + } + } + else if (str_cr == RB_PARSER_ENC_CODERANGE_VALID) { + res_enc = str_enc; + if (PARSER_ENC_CODERANGE_CLEAN_P(ptr_cr)) + res_cr = str_cr; + else + res_cr = ptr_cr; + } + else { /* str_cr == RB_PARSER_ENC_CODERANGE_BROKEN */ + res_enc = str_enc; + res_cr = str_cr; + if (0 < len) res_cr = RB_PARSER_ENC_CODERANGE_UNKNOWN; + } + + if (len < 0) { + compile_error(p, "negative string size (or size too big)"); + } + rb_parser_str_buf_cat(p, str, ptr, len); + PARSER_ENCODING_CODERANGE_SET(str, res_enc, res_cr); + return str; + + incompatible: + compile_error(p, "incompatible character encodings: %s and %s", + rb_enc_name(str_enc), rb_enc_name(ptr_enc)); + UNREACHABLE_RETURN(0); + +} + +static rb_parser_string_t * +rb_parser_enc_str_buf_cat(struct parser_params *p, rb_parser_string_t *str, const char *ptr, long len, + rb_encoding *ptr_enc) +{ + return rb_parser_enc_cr_str_buf_cat(p, str, ptr, len, ptr_enc, RB_PARSER_ENC_CODERANGE_UNKNOWN, NULL); +} + +static rb_parser_string_t * +rb_parser_str_buf_append(struct parser_params *p, rb_parser_string_t *str, rb_parser_string_t *str2) +{ + int str2_cr = rb_parser_enc_str_coderange(p, str2); + + rb_parser_enc_cr_str_buf_cat(p, str, PARSER_STRING_PTR(str2), PARSER_STRING_LEN(str2), + rb_parser_str_get_encoding(str2), str2_cr, &str2_cr); + + PARSER_ENC_CODERANGE_SET(str2, str2_cr); + + return str; +} + +static rb_parser_string_t * +rb_parser_str_resize(struct parser_params *p, rb_parser_string_t *str, long len) +{ + if (len < 0) { + rb_bug("negative string size (or size too big)"); + } + + long slen = PARSER_STRING_LEN(str); + + if (slen > len && PARSER_ENC_CODERANGE(str) != RB_PARSER_ENC_CODERANGE_7BIT) { + PARSER_ENC_CODERANGE_CLEAR(str); + } + + { + long capa; + const int termlen = STRING_TERM_LEN(str); + + if ((capa = slen) < len) { + SIZED_REALLOC_N(str->ptr, char, (size_t)len + termlen, STRING_SIZE(str)); + } + else if (len == slen) return str; + STRING_SET_LEN(str, len); + STRING_TERM_FILL(str); + } + return str; +} + +# define PARSER_ENC_STRING_GETMEM(str, ptrvar, lenvar, encvar) \ + ((ptrvar) = str->ptr, \ + (lenvar) = str->len, \ + (encvar) = str->enc) + +static int +rb_parser_string_hash_cmp(rb_parser_string_t *str1, rb_parser_string_t *str2) +{ + long len1, len2; + const char *ptr1, *ptr2; + rb_encoding *enc1, *enc2; + + PARSER_ENC_STRING_GETMEM(str1, ptr1, len1, enc1); + PARSER_ENC_STRING_GETMEM(str2, ptr2, len2, enc2); + + return (len1 != len2 || + enc1 != enc2 || + memcmp(ptr1, ptr2, len1) != 0); +} + +#ifndef RIPPER +static void +rb_parser_ary_extend(rb_parser_t *p, rb_parser_ary_t *ary, long len) +{ + long i; + if (ary->capa < len) { + ary->capa = len; + ary->data = (rb_parser_ary_data *)xrealloc(ary->data, sizeof(rb_parser_ary_data) * len); + for (i = ary->len; i < len; i++) { + ary->data[i] = 0; + } + } +} + +/* + * Do not call this directly. + * Use rb_parser_ary_new_capa_for_script_line() or rb_parser_ary_new_capa_for_ast_token() instead. + */ +static rb_parser_ary_t * +parser_ary_new_capa(rb_parser_t *p, long len) +{ + if (len < 0) { + rb_bug("negative array size (or size too big): %ld", len); + } + rb_parser_ary_t *ary = xcalloc(1, sizeof(rb_parser_ary_t)); + ary->len = 0; + ary->capa = len; + if (0 < len) { + ary->data = (rb_parser_ary_data *)xcalloc(len, sizeof(rb_parser_ary_data)); + } + else { + ary->data = NULL; + } + return ary; +} + +static rb_parser_ary_t * +rb_parser_ary_new_capa_for_script_line(rb_parser_t *p, long len) +{ + rb_parser_ary_t *ary = parser_ary_new_capa(p, len); + ary->data_type = PARSER_ARY_DATA_SCRIPT_LINE; + return ary; +} + +static rb_parser_ary_t * +rb_parser_ary_new_capa_for_ast_token(rb_parser_t *p, long len) +{ + rb_parser_ary_t *ary = parser_ary_new_capa(p, len); + ary->data_type = PARSER_ARY_DATA_AST_TOKEN; + return ary; +} + +/* + * Do not call this directly. + * Use rb_parser_ary_push_script_line() or rb_parser_ary_push_ast_token() instead. + */ +static rb_parser_ary_t * +parser_ary_push(rb_parser_t *p, rb_parser_ary_t *ary, rb_parser_ary_data val) +{ + if (ary->len == ary->capa) { + rb_parser_ary_extend(p, ary, ary->len == 0 ? 1 : ary->len * 2); + } + ary->data[ary->len++] = val; + return ary; +} + +static rb_parser_ary_t * +rb_parser_ary_push_ast_token(rb_parser_t *p, rb_parser_ary_t *ary, rb_parser_ast_token_t *val) +{ + if (ary->data_type != PARSER_ARY_DATA_AST_TOKEN) { + rb_bug("unexpected rb_parser_ary_data_type: %d", ary->data_type); + } + return parser_ary_push(p, ary, val); +} + +static rb_parser_ary_t * +rb_parser_ary_push_script_line(rb_parser_t *p, rb_parser_ary_t *ary, rb_parser_string_t *val) +{ + if (ary->data_type != PARSER_ARY_DATA_SCRIPT_LINE) { + rb_bug("unexpected rb_parser_ary_data_type: %d", ary->data_type); + } + return parser_ary_push(p, ary, val); +} + +static void +rb_parser_ast_token_free(rb_parser_t *p, rb_parser_ast_token_t *token) +{ + if (!token) return; + rb_parser_string_free(p, token->str); + xfree(token); +} + +static void +rb_parser_ary_free(rb_parser_t *p, rb_parser_ary_t *ary) +{ + void (*free_func)(rb_parser_t *, rb_parser_ary_data) = NULL; + switch (ary->data_type) { + case PARSER_ARY_DATA_AST_TOKEN: + free_func = (void (*)(rb_parser_t *, rb_parser_ary_data))rb_parser_ast_token_free; + break; + case PARSER_ARY_DATA_SCRIPT_LINE: + free_func = (void (*)(rb_parser_t *, rb_parser_ary_data))rb_parser_string_free; + break; + default: + rb_bug("unexpected rb_parser_ary_data_type: %d", ary->data_type); + break; + } + for (long i = 0; i < ary->len; i++) { + free_func(p, ary->data[i]); + } + xfree(ary); +} + +#endif /* !RIPPER */ %} %expect 0 %define api.pure %define parse.error verbose %printer { -#ifndef RIPPER - if ($$) { - rb_parser_printf(p, "%s", ruby_node_name(nd_type($$))); + if ((NODE *)$$ == (NODE *)-1) { + rb_parser_printf(p, "NODE_SPECIAL"); } -#else -#endif -} <node> + else if ($$) { + rb_parser_printf(p, "%s", parser_node_name(nd_type(RNODE($$)))); + } +} <node> <node_fcall> <node_args> <node_args_aux> <node_opt_arg> + <node_kw_arg> <node_block_pass> <node_masgn> <node_def_temp> <node_exits> %printer { -#ifndef RIPPER rb_parser_printf(p, "%"PRIsVALUE, rb_id2str($$)); -#else - rb_parser_printf(p, "%"PRIsVALUE, RNODE($$)->nd_rval); -#endif -} tIDENTIFIER tFID tGVAR tIVAR tCONSTANT tCVAR tLABEL tOP_ASGN +} <id> %printer { -#ifndef RIPPER - rb_parser_printf(p, "%+"PRIsVALUE, $$->nd_lit); -#else - rb_parser_printf(p, "%+"PRIsVALUE, get_value($$)); -#endif + switch (nd_type(RNODE($$))) { + case NODE_INTEGER: + rb_parser_printf(p, "%+"PRIsVALUE, rb_node_integer_literal_val($$)); + break; + case NODE_FLOAT: + rb_parser_printf(p, "%+"PRIsVALUE, rb_node_float_literal_val($$)); + break; + case NODE_RATIONAL: + rb_parser_printf(p, "%+"PRIsVALUE, rb_node_rational_literal_val($$)); + break; + case NODE_IMAGINARY: + rb_parser_printf(p, "%+"PRIsVALUE, rb_node_imaginary_literal_val($$)); + break; + default: + break; + } } tINTEGER tFLOAT tRATIONAL tIMAGINARY tSTRING_CONTENT tCHAR %printer { -#ifndef RIPPER - rb_parser_printf(p, "$%ld", $$->nd_nth); -#else - rb_parser_printf(p, "%"PRIsVALUE, $$); -#endif + rb_parser_printf(p, "$%ld", RNODE_NTH_REF($$)->nd_nth); } tNTH_REF %printer { -#ifndef RIPPER - rb_parser_printf(p, "$%c", (int)$$->nd_nth); -#else - rb_parser_printf(p, "%"PRIsVALUE, $$); -#endif + rb_parser_printf(p, "$%c", (int)RNODE_BACK_REF($$)->nd_nth); } tBACK_REF +%destructor { + if (CASE_LABELS_ENABLED_P($$)) st_free_table($$); +} <labels> + %lex-param {struct parser_params *p} %parse-param {struct parser_params *p} %initial-action { RUBY_SET_YYLLOC_OF_NONE(@$); }; +%after-shift after_shift +%before-reduce before_reduce +%after-reduce after_reduce +%after-shift-error-token after_shift_error_token +%after-pop-stack after_pop_stack %union { - VALUE val; NODE *node; + rb_node_fcall_t *node_fcall; + rb_node_args_t *node_args; + rb_node_args_aux_t *node_args_aux; + rb_node_opt_arg_t *node_opt_arg; + rb_node_kw_arg_t *node_kw_arg; + rb_node_block_pass_t *node_block_pass; + rb_node_masgn_t *node_masgn; + rb_node_def_temp_t *node_def_temp; + rb_node_exits_t *node_exits; ID id; int num; st_table *tbl; + st_table *labels; const struct vtable *vars; struct rb_strterm_struct *strterm; struct lex_context ctxt; + enum lex_state_e state; } %token <id> - keyword_class "`class'" - keyword_module "`module'" - keyword_def "`def'" - keyword_undef "`undef'" - keyword_begin "`begin'" - keyword_rescue "`rescue'" - keyword_ensure "`ensure'" - keyword_end "`end'" - keyword_if "`if'" - keyword_unless "`unless'" - keyword_then "`then'" - keyword_elsif "`elsif'" - keyword_else "`else'" - keyword_case "`case'" - keyword_when "`when'" - keyword_while "`while'" - keyword_until "`until'" - keyword_for "`for'" - keyword_break "`break'" - keyword_next "`next'" - keyword_redo "`redo'" - keyword_retry "`retry'" - keyword_in "`in'" - keyword_do "`do'" - keyword_do_cond "`do' for condition" - keyword_do_block "`do' for block" - keyword_do_LAMBDA "`do' for lambda" - keyword_return "`return'" - keyword_yield "`yield'" - keyword_super "`super'" - keyword_self "`self'" - keyword_nil "`nil'" - keyword_true "`true'" - keyword_false "`false'" - keyword_and "`and'" - keyword_or "`or'" - keyword_not "`not'" - modifier_if "`if' modifier" - modifier_unless "`unless' modifier" - modifier_while "`while' modifier" - modifier_until "`until' modifier" - modifier_rescue "`rescue' modifier" - keyword_alias "`alias'" - keyword_defined "`defined?'" - keyword_BEGIN "`BEGIN'" - keyword_END "`END'" - keyword__LINE__ "`__LINE__'" - keyword__FILE__ "`__FILE__'" - keyword__ENCODING__ "`__ENCODING__'" + keyword_class "'class'" + keyword_module "'module'" + keyword_def "'def'" + keyword_undef "'undef'" + keyword_begin "'begin'" + keyword_rescue "'rescue'" + keyword_ensure "'ensure'" + keyword_end "'end'" + keyword_if "'if'" + keyword_unless "'unless'" + keyword_then "'then'" + keyword_elsif "'elsif'" + keyword_else "'else'" + keyword_case "'case'" + keyword_when "'when'" + keyword_while "'while'" + keyword_until "'until'" + keyword_for "'for'" + keyword_break "'break'" + keyword_next "'next'" + keyword_redo "'redo'" + keyword_retry "'retry'" + keyword_in "'in'" + keyword_do "'do'" + keyword_do_cond "'do' for condition" + keyword_do_block "'do' for block" + keyword_do_LAMBDA "'do' for lambda" + keyword_return "'return'" + keyword_yield "'yield'" + keyword_super "'super'" + keyword_self "'self'" + keyword_nil "'nil'" + keyword_true "'true'" + keyword_false "'false'" + keyword_and "'and'" + keyword_or "'or'" + keyword_not "'not'" + modifier_if "'if' modifier" + modifier_unless "'unless' modifier" + modifier_while "'while' modifier" + modifier_until "'until' modifier" + modifier_rescue "'rescue' modifier" + keyword_alias "'alias'" + keyword_defined "'defined?'" + keyword_BEGIN "'BEGIN'" + keyword_END "'END'" + keyword__LINE__ "'__LINE__'" + keyword__FILE__ "'__FILE__'" + keyword__ENCODING__ "'__ENCODING__'" %token <id> tIDENTIFIER "local variable or method" %token <id> tFID "method" @@ -1464,26 +2788,35 @@ static int looking_at_eol_p(struct parser_params *p); %type <node> singleton strings string string1 xstring regexp %type <node> string_contents xstring_contents regexp_contents string_content %type <node> words symbols symbol_list qwords qsymbols word_list qword_list qsym_list word -%type <node> literal numeric simple_numeric ssym dsym symbol cpath def_name defn_head defs_head +%type <node> literal numeric simple_numeric ssym dsym symbol cpath +%type <node_def_temp> defn_head defs_head k_def +%type <node_exits> block_open k_while k_until k_for allow_exits %type <node> top_compstmt top_stmts top_stmt begin_block endless_arg endless_command %type <node> bodystmt compstmt stmts stmt_or_begin stmt expr arg primary command command_call method_call -%type <node> expr_value expr_value_do arg_value primary_value fcall rel_expr +%type <node> expr_value expr_value_do arg_value primary_value rel_expr +%type <node_fcall> fcall %type <node> if_tail opt_else case_body case_args cases opt_rescue exc_list exc_var opt_ensure -%type <node> args call_args opt_call_args -%type <node> paren_args opt_paren_args args_tail opt_args_tail block_args_tail opt_block_args_tail -%type <node> command_args aref_args opt_block_arg block_arg var_ref var_lhs +%type <node> args arg_splat call_args opt_call_args +%type <node> paren_args opt_paren_args +%type <node_args> args_tail block_args_tail +%type <node> command_args aref_args +%type <node_block_pass> opt_block_arg block_arg +%type <node> var_ref var_lhs %type <node> command_rhs arg_rhs %type <node> command_asgn mrhs mrhs_arg superclass block_call block_command -%type <node> f_block_optarg f_block_opt -%type <node> f_arglist f_opt_paren_args f_paren_args f_args f_arg f_arg_item -%type <node> f_optarg f_marg f_marg_list f_margs f_rest_marg +%type <node_args> f_arglist f_opt_paren_args f_paren_args f_args +%type <node_args_aux> f_arg f_arg_item +%type <node> f_marg f_marg_list f_rest_marg +%type <node_masgn> f_margs %type <node> assoc_list assocs assoc undef_list backref string_dvar for_var -%type <node> block_param opt_block_param block_param_def f_opt -%type <node> f_kwarg f_kw f_block_kwarg f_block_kw -%type <node> bv_decls opt_bv_decl bvar -%type <node> lambda f_larglist lambda_body brace_body do_body +%type <node_args> block_param opt_block_param block_param_def +%type <node_kw_arg> f_kw f_block_kw +%type <id> bv_decls opt_bv_decl bvar +%type <node> lambda lambda_body brace_body do_body +%type <node_args> f_larglist %type <node> brace_block cmd_brace_block do_block lhs none fitem -%type <node> mlhs mlhs_head mlhs_basic mlhs_item mlhs_node mlhs_post mlhs_inner +%type <node> mlhs_head mlhs_item mlhs_node mlhs_post +%type <node_masgn> mlhs mlhs_basic mlhs_inner %type <node> p_case_body p_cases p_top_expr p_top_expr_body %type <node> p_expr p_as p_alt p_expr_basic p_find %type <node> p_args p_args_head p_args_tail p_args_post p_arg p_rest @@ -1493,10 +2826,16 @@ static int looking_at_eol_p(struct parser_params *p); %type <id> cname fname op f_rest_arg f_block_arg opt_f_block_arg f_norm_arg f_bad_arg %type <id> f_kwrest f_label f_arg_asgn call_op call_op2 reswords relop dot_or_colon %type <id> p_kwrest p_kwnorest p_any_kwrest p_kw_label -%type <id> f_no_kwarg f_any_kwrest args_forward excessed_comma nonlocal_var - %type <ctxt> lex_ctxt /* keep <ctxt> in ripper */ +%type <id> f_no_kwarg f_any_kwrest args_forward excessed_comma nonlocal_var def_name +%type <ctxt> lex_ctxt begin_defined k_class k_module k_END k_rescue k_ensure after_rescue +%type <ctxt> p_in_kwarg +%type <tbl> p_lparen p_lbracket p_pktbl p_pvtbl +%type <num> max_numparam +%type <node> numparam +%type <id> it_id %token END_OF_INPUT 0 "end-of-input" %token <id> '.' + /* escaped chars, should be ignored otherwise */ %token <id> '\\' "backslash" %token tSP "escaped space" @@ -1539,7 +2878,7 @@ static int looking_at_eol_p(struct parser_params *p); %token tSTAR "*" %token tDSTAR "**arg" %token tAMPER "&" -%token tLAMBDA "->" +%token <num> tLAMBDA "->" %token tSYMBEG "symbol literal" %token tSTRING_BEG "string literal" %token tXSTRING_BEG "backtick literal" @@ -1550,7 +2889,8 @@ static int looking_at_eol_p(struct parser_params *p); %token tQSYMBOLS_BEG "verbatim symbol list" %token tSTRING_END "terminator" %token tSTRING_DEND "'}'" -%token tSTRING_DBEG tSTRING_DVAR tLAMBEG tLABEL_END +%token <state> tSTRING_DBEG "'#{'" +%token tSTRING_DVAR tLAMBEG tLABEL_END %token tIGNORED_NL tCOMMENT tEMBDOC_BEG tEMBDOC tEMBDOC_END %token tHEREDOC_BEG tHEREDOC_END k__END__ @@ -1585,34 +2925,89 @@ static int looking_at_eol_p(struct parser_params *p); %token tLAST_TOKEN +/* + * parameterizing rules + */ +%rule f_opt(value) <node_opt_arg>: f_arg_asgn f_eq value + { + p->cur_arg = 0; + p->ctxt.in_argdef = 1; + $$ = NEW_OPT_ARG(assignable(p, $1, $3, &@$), &@$); + /*% ripper: [$:$, $:3] %*/ + } + ; + +%rule f_optarg(value) <node_opt_arg>: f_opt(value) + { + $$ = $1; + /*% ripper: rb_ary_new3(1, $:1) %*/ + } + | f_optarg(value) ',' f_opt(value) + { + $$ = opt_arg_append($1, $3); + /*% ripper: rb_ary_push($:1, $:3) %*/ + } + ; + +%rule f_kwarg(kw) <node_kw_arg>: kw + { + $$ = $1; + /*% ripper: rb_ary_new3(1, $:1) %*/ + } + | f_kwarg(kw) ',' kw + { + $$ = kwd_append($1, $3); + /*% ripper: rb_ary_push($:1, $:3) %*/ + } + ; + +%rule opt_args_tail(tail) <node_args>: ',' tail + { + $$ = $2; + /*% ripper: $:2 %*/ + } + | /* none */ + { + $$ = new_args_tail(p, 0, 0, 0, &@0); + /*% ripper: [Qnil, Qnil, Qnil] %*/ + } + ; + +%rule words(begin, word_list): begin ' '+ word_list tSTRING_END + { + $$ = make_list($3, &@$); + /*% ripper: array!($:3) %*/ + } + ; + %% program : { SET_LEX_STATE(EXPR_BEG); local_push(p, ifndef_ripper(1)+0); + /* jumps are possible in the top-level loop. */ + if (!ifndef_ripper(p->do_loop) + 0) init_block_exit(p); } top_compstmt { - /*%%%*/ if ($2 && !compile_for_eval) { NODE *node = $2; /* last expression should not be void */ if (nd_type_p(node, NODE_BLOCK)) { - while (node->nd_next) { - node = node->nd_next; + while (RNODE_BLOCK(node)->nd_next) { + node = RNODE_BLOCK(node)->nd_next; } - node = node->nd_head; + node = RNODE_BLOCK(node)->nd_head; } node = remove_begin(node); void_expr(p, node); } p->eval_tree = NEW_SCOPE(0, block_append(p, p->eval_tree, $2), &@$); - /*% %*/ - /*% ripper[final]: program!($2) %*/ + /*% ripper[final]: program!($:2) %*/ local_pop(p); } ; -top_compstmt : top_stmts opt_terms +top_compstmt : top_stmts terms? { $$ = void_stmts(p, $1); } @@ -1620,68 +3015,76 @@ top_compstmt : top_stmts opt_terms top_stmts : none { - /*%%%*/ $$ = NEW_BEGIN(0, &@$); - /*% %*/ /*% ripper: stmts_add!(stmts_new!, void_stmt!) %*/ } | top_stmt { - /*%%%*/ $$ = newline_node($1); - /*% %*/ - /*% ripper: stmts_add!(stmts_new!, $1) %*/ + /*% ripper: stmts_add!(stmts_new!, $:1) %*/ } | top_stmts terms top_stmt { - /*%%%*/ $$ = block_append(p, $1, newline_node($3)); - /*% %*/ - /*% ripper: stmts_add!($1, $3) %*/ + /*% ripper: stmts_add!($:1, $:3) %*/ } ; top_stmt : stmt + { + clear_block_exit(p, true); + $$ = $1; + } | keyword_BEGIN begin_block { $$ = $2; + /*% ripper: $:2 %*/ } ; -begin_block : '{' top_compstmt '}' +block_open : '{' {$$ = init_block_exit(p);}; + +begin_block : block_open top_compstmt '}' { - /*%%%*/ + restore_block_exit(p, $block_open); p->eval_tree_begin = block_append(p, p->eval_tree_begin, NEW_BEGIN($2, &@$)); $$ = NEW_BEGIN(0, &@$); - /*% %*/ - /*% ripper: BEGIN!($2) %*/ + /*% ripper: BEGIN!($:2) %*/ } ; -bodystmt : compstmt +bodystmt : compstmt[body] + lex_ctxt[ctxt] opt_rescue - k_else {if (!$2) {yyerror1(&@3, "else without rescue is useless");}} - compstmt + k_else + { + if (!$opt_rescue) yyerror1(&@k_else, "else without rescue is useless"); + next_rescue_context(&p->ctxt, &$ctxt, after_else); + } + compstmt[elsebody] + { + next_rescue_context(&p->ctxt, &$ctxt, after_ensure); + } opt_ensure { - /*%%%*/ - $$ = new_bodystmt(p, $1, $2, $5, $6, &@$); - /*% %*/ - /*% ripper: bodystmt!(escape_Qundef($1), escape_Qundef($2), escape_Qundef($5), escape_Qundef($6)) %*/ + $$ = new_bodystmt(p, $body, $opt_rescue, $elsebody, $opt_ensure, &@$); + /*% ripper: bodystmt!($:body, $:opt_rescue, $:elsebody, $:opt_ensure) %*/ } - | compstmt + | compstmt[body] + lex_ctxt[ctxt] opt_rescue + { + next_rescue_context(&p->ctxt, &$ctxt, after_ensure); + } opt_ensure { - /*%%%*/ - $$ = new_bodystmt(p, $1, $2, 0, $3, &@$); - /*% %*/ - /*% ripper: bodystmt!(escape_Qundef($1), escape_Qundef($2), Qnil, escape_Qundef($3)) %*/ + $$ = new_bodystmt(p, $body, $opt_rescue, 0, $opt_ensure, &@$); + /*% ripper: bodystmt!($:body, $:opt_rescue, Qnil, $:opt_ensure) %*/ } ; -compstmt : stmts opt_terms +compstmt : stmts terms? { $$ = void_stmts(p, $1); } @@ -1689,24 +3092,18 @@ compstmt : stmts opt_terms stmts : none { - /*%%%*/ $$ = NEW_BEGIN(0, &@$); - /*% %*/ /*% ripper: stmts_add!(stmts_new!, void_stmt!) %*/ } | stmt_or_begin { - /*%%%*/ $$ = newline_node($1); - /*% %*/ - /*% ripper: stmts_add!(stmts_new!, $1) %*/ + /*% ripper: stmts_add!(stmts_new!, $:1) %*/ } | stmts terms stmt_or_begin { - /*%%%*/ $$ = block_append(p, $1, newline_node($3)); - /*% %*/ - /*% ripper: stmts_add!($1, $3) %*/ + /*% ripper: stmts_add!($:1, $:3) %*/ } ; @@ -1724,245 +3121,219 @@ stmt_or_begin : stmt } ; +allow_exits : {$$ = allow_block_exit(p);}; + +k_END : keyword_END lex_ctxt + { + $$ = $2; + p->ctxt.in_rescue = before_rescue; + /*% ripper: $:2 %*/ + }; + stmt : keyword_alias fitem {SET_LEX_STATE(EXPR_FNAME|EXPR_FITEM);} fitem { - /*%%%*/ $$ = NEW_ALIAS($2, $4, &@$); - /*% %*/ - /*% ripper: alias!($2, $4) %*/ + /*% ripper: alias!($:2, $:4) %*/ } | keyword_alias tGVAR tGVAR { - /*%%%*/ $$ = NEW_VALIAS($2, $3, &@$); - /*% %*/ - /*% ripper: var_alias!($2, $3) %*/ + /*% ripper: var_alias!($:2, $:3) %*/ } | keyword_alias tGVAR tBACK_REF { - /*%%%*/ char buf[2]; buf[0] = '$'; - buf[1] = (char)$3->nd_nth; + buf[1] = (char)RNODE_BACK_REF($3)->nd_nth; $$ = NEW_VALIAS($2, rb_intern2(buf, 2), &@$); - /*% %*/ - /*% ripper: var_alias!($2, $3) %*/ + /*% ripper: var_alias!($:2, $:3) %*/ } | keyword_alias tGVAR tNTH_REF { static const char mesg[] = "can't make alias for the number variables"; /*%%%*/ yyerror1(&@3, mesg); - $$ = NEW_BEGIN(0, &@$); /*% %*/ - /*% ripper[error]: alias_error!(ERR_MESG(), $3) %*/ + $$ = NEW_ERROR(&@$); + /*% ripper[error]: alias_error!(ERR_MESG(), $:3) %*/ } | keyword_undef undef_list { - /*%%%*/ $$ = $2; - /*% %*/ - /*% ripper: undef!($2) %*/ + /*% ripper: undef!($:2) %*/ } | stmt modifier_if expr_value { - /*%%%*/ $$ = new_if(p, $3, remove_begin($1), 0, &@$); fixpos($$, $3); - /*% %*/ - /*% ripper: if_mod!($3, $1) %*/ + /*% ripper: if_mod!($:3, $:1) %*/ } | stmt modifier_unless expr_value { - /*%%%*/ $$ = new_unless(p, $3, remove_begin($1), 0, &@$); fixpos($$, $3); - /*% %*/ - /*% ripper: unless_mod!($3, $1) %*/ + /*% ripper: unless_mod!($:3, $:1) %*/ } | stmt modifier_while expr_value { - /*%%%*/ + clear_block_exit(p, false); if ($1 && nd_type_p($1, NODE_BEGIN)) { - $$ = NEW_WHILE(cond(p, $3, &@3), $1->nd_body, 0, &@$); + $$ = NEW_WHILE(cond(p, $3, &@3), RNODE_BEGIN($1)->nd_body, 0, &@$); } else { $$ = NEW_WHILE(cond(p, $3, &@3), $1, 1, &@$); } - /*% %*/ - /*% ripper: while_mod!($3, $1) %*/ + /*% ripper: while_mod!($:3, $:1) %*/ } | stmt modifier_until expr_value { - /*%%%*/ + clear_block_exit(p, false); if ($1 && nd_type_p($1, NODE_BEGIN)) { - $$ = NEW_UNTIL(cond(p, $3, &@3), $1->nd_body, 0, &@$); + $$ = NEW_UNTIL(cond(p, $3, &@3), RNODE_BEGIN($1)->nd_body, 0, &@$); } else { $$ = NEW_UNTIL(cond(p, $3, &@3), $1, 1, &@$); } - /*% %*/ - /*% ripper: until_mod!($3, $1) %*/ + /*% ripper: until_mod!($:3, $:1) %*/ } - | stmt modifier_rescue stmt + | stmt modifier_rescue after_rescue stmt { - /*%%%*/ + p->ctxt.in_rescue = $3.in_rescue; NODE *resq; - YYLTYPE loc = code_loc_gen(&@2, &@3); - resq = NEW_RESBODY(0, remove_begin($3), 0, &loc); + YYLTYPE loc = code_loc_gen(&@2, &@4); + resq = NEW_RESBODY(0, remove_begin($4), 0, &loc); $$ = NEW_RESCUE(remove_begin($1), resq, 0, &@$); - /*% %*/ - /*% ripper: rescue_mod!($1, $3) %*/ + /*% ripper: rescue_mod!($:1, $:4) %*/ } - | keyword_END '{' compstmt '}' + | k_END allow_exits '{' compstmt '}' { if (p->ctxt.in_def) { rb_warn0("END in method; use at_exit"); } - /*%%%*/ + restore_block_exit(p, $allow_exits); + p->ctxt = $k_END; { - NODE *scope = NEW_NODE( - NODE_SCOPE, 0 /* tbl */, $3 /* body */, 0 /* args */, &@$); + NODE *scope = NEW_SCOPE2(0 /* tbl */, 0 /* args */, $compstmt /* body */, &@$); $$ = NEW_POSTEXE(scope, &@$); } - /*% %*/ - /*% ripper: END!($3) %*/ + /*% ripper: END!($:compstmt) %*/ } | command_asgn | mlhs '=' lex_ctxt command_call { - /*%%%*/ value_expr($4); - $$ = node_assign(p, $1, $4, $3, &@$); - /*% %*/ - /*% ripper: massign!($1, $4) %*/ + $$ = node_assign(p, (NODE *)$1, $4, $3, &@$); + /*% ripper: massign!($:1, $:4) %*/ } | lhs '=' lex_ctxt mrhs { - /*%%%*/ $$ = node_assign(p, $1, $4, $3, &@$); - /*% %*/ - /*% ripper: assign!($1, $4) %*/ + /*% ripper: assign!($:1, $:4) %*/ } - | mlhs '=' lex_ctxt mrhs_arg modifier_rescue stmt + | mlhs '=' lex_ctxt mrhs_arg modifier_rescue + after_rescue stmt[resbody] { - /*%%%*/ - YYLTYPE loc = code_loc_gen(&@5, &@6); - $$ = node_assign(p, $1, NEW_RESCUE($4, NEW_RESBODY(0, remove_begin($6), 0, &loc), 0, &@$), $3, &@$); - /*% %*/ - /*% ripper: massign!($1, rescue_mod!($4, $6)) %*/ + p->ctxt.in_rescue = $3.in_rescue; + YYLTYPE loc = code_loc_gen(&@modifier_rescue, &@resbody); + $resbody = NEW_RESBODY(0, remove_begin($resbody), 0, &loc); + loc.beg_pos = @mrhs_arg.beg_pos; + $mrhs_arg = NEW_RESCUE($mrhs_arg, $resbody, 0, &loc); + $$ = node_assign(p, (NODE *)$mlhs, $mrhs_arg, $lex_ctxt, &@$); + /*% ripper: massign!($:1, rescue_mod!($:4, $:7)) %*/ } | mlhs '=' lex_ctxt mrhs_arg { - /*%%%*/ - $$ = node_assign(p, $1, $4, $3, &@$); - /*% %*/ - /*% ripper: massign!($1, $4) %*/ + $$ = node_assign(p, (NODE *)$1, $4, $3, &@$); + /*% ripper: massign!($:1, $:4) %*/ } | expr | error { (void)yynerrs; - /*%%%*/ $$ = NEW_ERROR(&@$); - /*% %*/ } ; command_asgn : lhs '=' lex_ctxt command_rhs { - /*%%%*/ $$ = node_assign(p, $1, $4, $3, &@$); - /*% %*/ - /*% ripper: assign!($1, $4) %*/ + /*% ripper: assign!($:1, $:4) %*/ } | var_lhs tOP_ASGN lex_ctxt command_rhs { - /*%%%*/ $$ = new_op_assign(p, $1, $2, $4, $3, &@$); - /*% %*/ - /*% ripper: opassign!($1, $2, $4) %*/ + /*% ripper: opassign!($:1, $:2, $:4) %*/ } | primary_value '[' opt_call_args rbracket tOP_ASGN lex_ctxt command_rhs { - /*%%%*/ $$ = new_ary_op_assign(p, $1, $3, $5, $7, &@3, &@$); - /*% %*/ - /*% ripper: opassign!(aref_field!($1, escape_Qundef($3)), $5, $7) %*/ + /*% ripper: opassign!(aref_field!($:1, $:3), $:5, $:7) %*/ } | primary_value call_op tIDENTIFIER tOP_ASGN lex_ctxt command_rhs { - /*%%%*/ $$ = new_attr_op_assign(p, $1, $2, $3, $4, $6, &@$); - /*% %*/ - /*% ripper: opassign!(field!($1, $2, $3), $4, $6) %*/ + /*% ripper: opassign!(field!($:1, $:2, $:3), $:4, $:6) %*/ } | primary_value call_op tCONSTANT tOP_ASGN lex_ctxt command_rhs { - /*%%%*/ $$ = new_attr_op_assign(p, $1, $2, $3, $4, $6, &@$); - /*% %*/ - /*% ripper: opassign!(field!($1, $2, $3), $4, $6) %*/ + /*% ripper: opassign!(field!($:1, $:2, $:3), $:4, $:6) %*/ } | primary_value tCOLON2 tCONSTANT tOP_ASGN lex_ctxt command_rhs { - /*%%%*/ YYLTYPE loc = code_loc_gen(&@1, &@3); $$ = new_const_op_assign(p, NEW_COLON2($1, $3, &loc), $4, $6, $5, &@$); - /*% %*/ - /*% ripper: opassign!(const_path_field!($1, $3), $4, $6) %*/ + /*% ripper: opassign!(const_path_field!($:1, $:3), $:4, $:6) %*/ } | primary_value tCOLON2 tIDENTIFIER tOP_ASGN lex_ctxt command_rhs { - /*%%%*/ - $$ = new_attr_op_assign(p, $1, ID2VAL(idCOLON2), $3, $4, $6, &@$); - /*% %*/ - /*% ripper: opassign!(field!($1, ID2VAL(idCOLON2), $3), $4, $6) %*/ + $$ = new_attr_op_assign(p, $1, idCOLON2, $3, $4, $6, &@$); + /*% ripper: opassign!(field!($:1, $:2, $:3), $:4, $:6) %*/ } - | defn_head f_opt_paren_args '=' endless_command + | defn_head[head] f_opt_paren_args[args] '=' endless_command[bodystmt] { - endless_method_name(p, $<node>1, &@1); - restore_defun(p, $<node>1->nd_defn); - /*%%%*/ - $$ = set_defun_body(p, $1, $2, $4, &@$); - /*% %*/ - /*% ripper: def!(get_value($1), $2, bodystmt!($4, Qnil, Qnil, Qnil)) %*/ + endless_method_name(p, $head->nd_mid, &@head); + restore_defun(p, $head); + $bodystmt = new_scope_body(p, $args, $bodystmt, &@$); + ($$ = $head->nd_def)->nd_loc = @$; + RNODE_DEFN($$)->nd_defn = $bodystmt; + /*% ripper: bodystmt!($:bodystmt, Qnil, Qnil, Qnil) %*/ + /*% ripper: def!($:head, $:args, $:$) %*/ local_pop(p); } - | defs_head f_opt_paren_args '=' endless_command + | defs_head[head] f_opt_paren_args[args] '=' endless_command[bodystmt] { - endless_method_name(p, $<node>1, &@1); - restore_defun(p, $<node>1->nd_defn); - /*%%%*/ - $$ = set_defun_body(p, $1, $2, $4, &@$); - /*% - $1 = get_value($1); - %*/ - /*% ripper: defs!(AREF($1, 0), AREF($1, 1), AREF($1, 2), $2, bodystmt!($4, Qnil, Qnil, Qnil)) %*/ + endless_method_name(p, $head->nd_mid, &@head); + restore_defun(p, $head); + $bodystmt = new_scope_body(p, $args, $bodystmt, &@$); + ($$ = $head->nd_def)->nd_loc = @$; + RNODE_DEFS($$)->nd_defn = $bodystmt; + /*% ripper: bodystmt!($:bodystmt, Qnil, Qnil, Qnil) %*/ + /*% ripper: defs!(*$:head[0..2], $:args, $:$) %*/ local_pop(p); } | backref tOP_ASGN lex_ctxt command_rhs { /*%%%*/ rb_backref_error(p, $1); - $$ = NEW_BEGIN(0, &@$); /*% %*/ - /*% ripper[error]: backref_error(p, RNODE($1), assign!(var_field(p, $1), $4)) %*/ + $$ = NEW_ERROR(&@$); + /*% ripper[error]: backref_error(p, $1, opassign!(var_field!($:1), $:2, $:4)) %*/ } ; endless_command : command - | endless_command modifier_rescue arg + | endless_command modifier_rescue after_rescue arg { - /*%%%*/ - $$ = rescued_expr(p, $1, $3, &@1, &@2, &@3); - /*% %*/ - /*% ripper: rescue_mod!($1, $3) %*/ + p->ctxt.in_rescue = $3.in_rescue; + $$ = rescued_expr(p, $1, $4, &@1, &@2, &@4); + /*% ripper: rescue_mod!($:1, $:4) %*/ } - | keyword_not opt_nl endless_command + | keyword_not '\n'? endless_command { $$ = call_uni_op(p, method_cond(p, $3, &@3), METHOD_NOT, &@1, &@$); + /*% ripper: unary!(ID2VAL(idNOT), $:3) %*/ } ; @@ -1971,14 +3342,13 @@ command_rhs : command_call %prec tOP_ASGN value_expr($1); $$ = $1; } - | command_call modifier_rescue stmt + | command_call modifier_rescue after_rescue stmt { - /*%%%*/ - YYLTYPE loc = code_loc_gen(&@2, &@3); + p->ctxt.in_rescue = $3.in_rescue; + YYLTYPE loc = code_loc_gen(&@2, &@4); value_expr($1); - $$ = NEW_RESCUE($1, NEW_RESBODY(0, remove_begin($3), 0, &loc), 0, &@$); - /*% %*/ - /*% ripper: rescue_mod!($1, $3) %*/ + $$ = NEW_RESCUE($1, NEW_RESBODY(0, remove_begin($4), 0, &loc), 0, &@$); + /*% ripper: rescue_mod!($:1, $:4) %*/ } | command_asgn ; @@ -1987,94 +3357,70 @@ expr : command_call | expr keyword_and expr { $$ = logop(p, idAND, $1, $3, &@2, &@$); + /*% ripper: binary!($:1, ID2VAL(idAND), $:3) %*/ } | expr keyword_or expr { $$ = logop(p, idOR, $1, $3, &@2, &@$); + /*% ripper: binary!($:1, ID2VAL(idOR), $:3) %*/ } - | keyword_not opt_nl expr + | keyword_not '\n'? expr { $$ = call_uni_op(p, method_cond(p, $3, &@3), METHOD_NOT, &@1, &@$); + /*% ripper: unary!(ID2VAL(idNOT), $:3) %*/ } | '!' command_call { $$ = call_uni_op(p, method_cond(p, $2, &@2), '!', &@1, &@$); + /*% ripper: unary!(ID2VAL('\'!\''), $:2) %*/ } | arg tASSOC { - value_expr($1); - SET_LEX_STATE(EXPR_BEG|EXPR_LABEL); - p->command_start = FALSE; - $<ctxt>2 = p->ctxt; - p->ctxt.in_kwarg = 1; - $<tbl>$ = push_pvtbl(p); + value_expr($arg); } + p_in_kwarg[ctxt] p_pvtbl p_pktbl + p_top_expr_body[body] { - $<tbl>$ = push_pktbl(p); - } - p_top_expr_body - { - pop_pktbl(p, $<tbl>4); - pop_pvtbl(p, $<tbl>3); - p->ctxt.in_kwarg = $<ctxt>2.in_kwarg; - /*%%%*/ - $$ = NEW_CASE3($1, NEW_IN($5, 0, 0, &@5), &@$); - /*% %*/ - /*% ripper: case!($1, in!($5, Qnil, Qnil)) %*/ + pop_pktbl(p, $p_pktbl); + pop_pvtbl(p, $p_pvtbl); + p->ctxt.in_kwarg = $ctxt.in_kwarg; + $$ = NEW_CASE3($arg, NEW_IN($body, 0, 0, &@body), &@$); + /*% ripper: case!($:arg, in!($:body, Qnil, Qnil)) %*/ } | arg keyword_in { - value_expr($1); - SET_LEX_STATE(EXPR_BEG|EXPR_LABEL); - p->command_start = FALSE; - $<ctxt>2 = p->ctxt; - p->ctxt.in_kwarg = 1; - $<tbl>$ = push_pvtbl(p); + value_expr($arg); } + p_in_kwarg[ctxt] p_pvtbl p_pktbl + p_top_expr_body[body] { - $<tbl>$ = push_pktbl(p); - } - p_top_expr_body - { - pop_pktbl(p, $<tbl>4); - pop_pvtbl(p, $<tbl>3); - p->ctxt.in_kwarg = $<ctxt>2.in_kwarg; - /*%%%*/ - $$ = NEW_CASE3($1, NEW_IN($5, NEW_TRUE(&@5), NEW_FALSE(&@5), &@5), &@$); - /*% %*/ - /*% ripper: case!($1, in!($5, Qnil, Qnil)) %*/ + pop_pktbl(p, $p_pktbl); + pop_pvtbl(p, $p_pvtbl); + p->ctxt.in_kwarg = $ctxt.in_kwarg; + $$ = NEW_CASE3($arg, NEW_IN($body, NEW_TRUE(&@body), NEW_FALSE(&@body), &@body), &@$); + /*% ripper: case!($:arg, in!($:body, Qnil, Qnil)) %*/ } | arg %prec tLBRACE_ARG ; def_name : fname { - ID fname = get_id($1); - ID cur_arg = p->cur_arg; - YYSTYPE c = {.ctxt = p->ctxt}; + ID fname = $1; numparam_name(p, fname); - NODE *save = - NODE_NEW_TEMPORAL(NODE_SELF, - /*head*/numparam_push(p), - /*nth*/p->max_numparam, - /*cval*/c.val); local_push(p, 0); p->cur_arg = 0; p->ctxt.in_def = 1; - $<node>$ = NEW_NODE(NODE_SELF, /*vid*/cur_arg, /*mid*/fname, /*args*/save, &@$); - /*%%%*/ - /*% - $$ = NEW_RIPPER(fname, get_value($1), $$, &NULL_LOC); - %*/ + p->ctxt.in_rescue = before_rescue; + $$ = $1; } ; defn_head : k_def def_name { - $$ = $2; - /*%%%*/ - $$ = NEW_NODE(NODE_DEFN, 0, $$->nd_mid, $$, &@$); - /*% %*/ + $$ = def_head_save(p, $k_def); + $$->nd_mid = $def_name; + $$->nd_def = NEW_DEFN($def_name, 0, &@$); + /*% ripper: $:def_name %*/ } ; @@ -2086,14 +3432,10 @@ defs_head : k_def singleton dot_or_colon def_name { SET_LEX_STATE(EXPR_ENDFN|EXPR_LABEL); /* force for args */ - $$ = $5; - /*%%%*/ - $$ = NEW_NODE(NODE_DEFS, $2, $$->nd_mid, $$, &@$); - /*% - VALUE ary = rb_ary_new_from_args(3, $2, $3, get_value($$)); - add_mark_object(p, ary); - $<node>$->nd_rval = ary; - %*/ + $$ = def_head_save(p, $k_def); + $$->nd_mid = $def_name; + $$->nd_def = NEW_DEFS($singleton, $def_name, 0, &@$); + /*% ripper: [$:singleton, $:dot_or_colon, $:def_name] %*/ } ; @@ -2104,15 +3446,14 @@ expr_value : expr } | error { - /*%%%*/ $$ = NEW_ERROR(&@$); - /*% %*/ } ; expr_value_do : {COND_PUSH(1);} expr_value do {COND_POP();} { $$ = $2; + /*% ripper: $:2 %*/ } ; @@ -2123,385 +3464,300 @@ command_call : command block_command : block_call | block_call call_op2 operation2 command_args { - /*%%%*/ $$ = new_qcall(p, $2, $1, $3, $4, &@3, &@$); - /*% %*/ - /*% ripper: method_add_arg!(call!($1, $2, $3), $4) %*/ + /*% ripper: method_add_arg!(call!($:1, $:2, $:3), $:4) %*/ } ; cmd_brace_block : tLBRACE_ARG brace_body '}' { $$ = $2; - /*%%%*/ - $$->nd_body->nd_loc = code_loc_gen(&@1, &@3); - nd_set_line($$, @1.end_pos.lineno); - /*% %*/ + set_embraced_location($$, &@1, &@3); + /*% ripper: $:2 %*/ } ; fcall : operation { - /*%%%*/ $$ = NEW_FCALL($1, 0, &@$); - nd_set_line($$, p->tokline); - /*% %*/ - /*% ripper: $1 %*/ + /*% ripper: $:1 %*/ } ; command : fcall command_args %prec tLOWEST { - /*%%%*/ $1->nd_args = $2; nd_set_last_loc($1, @2.end_pos); - $$ = $1; - /*% %*/ - /*% ripper: command!($1, $2) %*/ + $$ = (NODE *)$1; + /*% ripper: command!($:1, $:2) %*/ } | fcall command_args cmd_brace_block { - /*%%%*/ block_dup_check(p, $2, $3); $1->nd_args = $2; - $$ = method_add_block(p, $1, $3, &@$); - fixpos($$, $1); + $$ = method_add_block(p, (NODE *)$1, $3, &@$); + fixpos($$, RNODE($1)); nd_set_last_loc($1, @2.end_pos); - /*% %*/ - /*% ripper: method_add_block!(command!($1, $2), $3) %*/ + /*% ripper: method_add_block!(command!($:1, $:2), $:3) %*/ } | primary_value call_op operation2 command_args %prec tLOWEST { - /*%%%*/ - $$ = new_command_qcall(p, $2, $1, $3, $4, Qnull, &@3, &@$); - /*% %*/ - /*% ripper: command_call!($1, $2, $3, $4) %*/ + $$ = new_command_qcall(p, $2, $1, $3, $4, 0, &@3, &@$); + /*% ripper: command_call!($:1, $:2, $:3, $:4) %*/ } | primary_value call_op operation2 command_args cmd_brace_block { - /*%%%*/ $$ = new_command_qcall(p, $2, $1, $3, $4, $5, &@3, &@$); - /*% %*/ - /*% ripper: method_add_block!(command_call!($1, $2, $3, $4), $5) %*/ + /*% ripper: method_add_block!(command_call!($:1, $:2, $:3, $:4), $:5) %*/ } | primary_value tCOLON2 operation2 command_args %prec tLOWEST { - /*%%%*/ - $$ = new_command_qcall(p, ID2VAL(idCOLON2), $1, $3, $4, Qnull, &@3, &@$); - /*% %*/ - /*% ripper: command_call!($1, ID2VAL(idCOLON2), $3, $4) %*/ + $$ = new_command_qcall(p, idCOLON2, $1, $3, $4, 0, &@3, &@$); + /*% ripper: command_call!($:1, $:2, $:3, $:4) %*/ } | primary_value tCOLON2 operation2 command_args cmd_brace_block { - /*%%%*/ - $$ = new_command_qcall(p, ID2VAL(idCOLON2), $1, $3, $4, $5, &@3, &@$); - /*% %*/ - /*% ripper: method_add_block!(command_call!($1, ID2VAL(idCOLON2), $3, $4), $5) %*/ + $$ = new_command_qcall(p, idCOLON2, $1, $3, $4, $5, &@3, &@$); + /*% ripper: method_add_block!(command_call!($:1, $:2, $:3, $:4), $:5) %*/ + } + | primary_value tCOLON2 tCONSTANT '{' brace_body '}' + { + set_embraced_location($5, &@4, &@6); + $$ = new_command_qcall(p, idCOLON2, $1, $3, 0, $5, &@3, &@$); + /*% ripper: method_add_block!(command_call!($:1, $:2, $:3, Qnil), $:5) %*/ } | keyword_super command_args { - /*%%%*/ $$ = NEW_SUPER($2, &@$); fixpos($$, $2); - /*% %*/ - /*% ripper: super!($2) %*/ + /*% ripper: super!($:2) %*/ } - | keyword_yield command_args + | k_yield command_args { - /*%%%*/ $$ = new_yield(p, $2, &@$); fixpos($$, $2); - /*% %*/ - /*% ripper: yield!($2) %*/ + /*% ripper: yield!($:2) %*/ } | k_return call_args { - /*%%%*/ $$ = NEW_RETURN(ret_args(p, $2), &@$); - /*% %*/ - /*% ripper: return!($2) %*/ + /*% ripper: return!($:2) %*/ } | keyword_break call_args { - /*%%%*/ - $$ = NEW_BREAK(ret_args(p, $2), &@$); - /*% %*/ - /*% ripper: break!($2) %*/ + NODE *args = 0; + args = ret_args(p, $2); + $$ = add_block_exit(p, NEW_BREAK(args, &@$)); + /*% ripper: break!($:2) %*/ } | keyword_next call_args { - /*%%%*/ - $$ = NEW_NEXT(ret_args(p, $2), &@$); - /*% %*/ - /*% ripper: next!($2) %*/ + NODE *args = 0; + args = ret_args(p, $2); + $$ = add_block_exit(p, NEW_NEXT(args, &@$)); + /*% ripper: next!($:2) %*/ } ; mlhs : mlhs_basic | tLPAREN mlhs_inner rparen { - /*%%%*/ $$ = $2; - /*% %*/ - /*% ripper: mlhs_paren!($2) %*/ + /*% ripper: mlhs_paren!($:2) %*/ } ; mlhs_inner : mlhs_basic | tLPAREN mlhs_inner rparen { - /*%%%*/ - $$ = NEW_MASGN(NEW_LIST($2, &@$), 0, &@$); - /*% %*/ - /*% ripper: mlhs_paren!($2) %*/ + $$ = NEW_MASGN(NEW_LIST((NODE *)$2, &@$), 0, &@$); + /*% ripper: mlhs_paren!($:2) %*/ } ; mlhs_basic : mlhs_head { - /*%%%*/ $$ = NEW_MASGN($1, 0, &@$); - /*% %*/ - /*% ripper: $1 %*/ + /*% ripper: $:1 %*/ } | mlhs_head mlhs_item { - /*%%%*/ - $$ = NEW_MASGN(list_append(p, $1,$2), 0, &@$); - /*% %*/ - /*% ripper: mlhs_add!($1, $2) %*/ + $$ = NEW_MASGN(list_append(p, $1, $2), 0, &@$); + /*% ripper: mlhs_add!($:1, $:2) %*/ } | mlhs_head tSTAR mlhs_node { - /*%%%*/ $$ = NEW_MASGN($1, $3, &@$); - /*% %*/ - /*% ripper: mlhs_add_star!($1, $3) %*/ + /*% ripper: mlhs_add_star!($:1, $:3) %*/ } | mlhs_head tSTAR mlhs_node ',' mlhs_post { - /*%%%*/ $$ = NEW_MASGN($1, NEW_POSTARG($3,$5,&@$), &@$); - /*% %*/ - /*% ripper: mlhs_add_post!(mlhs_add_star!($1, $3), $5) %*/ + /*% ripper: mlhs_add_post!(mlhs_add_star!($:1, $:3), $:5) %*/ } | mlhs_head tSTAR { - /*%%%*/ $$ = NEW_MASGN($1, NODE_SPECIAL_NO_NAME_REST, &@$); - /*% %*/ - /*% ripper: mlhs_add_star!($1, Qnil) %*/ + /*% ripper: mlhs_add_star!($:1, Qnil) %*/ } | mlhs_head tSTAR ',' mlhs_post { - /*%%%*/ $$ = NEW_MASGN($1, NEW_POSTARG(NODE_SPECIAL_NO_NAME_REST, $4, &@$), &@$); - /*% %*/ - /*% ripper: mlhs_add_post!(mlhs_add_star!($1, Qnil), $4) %*/ + /*% ripper: mlhs_add_post!(mlhs_add_star!($:1, Qnil), $:4) %*/ } | tSTAR mlhs_node { - /*%%%*/ $$ = NEW_MASGN(0, $2, &@$); - /*% %*/ - /*% ripper: mlhs_add_star!(mlhs_new!, $2) %*/ + /*% ripper: mlhs_add_star!(mlhs_new!, $:2) %*/ } | tSTAR mlhs_node ',' mlhs_post { - /*%%%*/ $$ = NEW_MASGN(0, NEW_POSTARG($2,$4,&@$), &@$); - /*% %*/ - /*% ripper: mlhs_add_post!(mlhs_add_star!(mlhs_new!, $2), $4) %*/ + /*% ripper: mlhs_add_post!(mlhs_add_star!(mlhs_new!, $:2), $:4) %*/ } | tSTAR { - /*%%%*/ $$ = NEW_MASGN(0, NODE_SPECIAL_NO_NAME_REST, &@$); - /*% %*/ /*% ripper: mlhs_add_star!(mlhs_new!, Qnil) %*/ } | tSTAR ',' mlhs_post { - /*%%%*/ $$ = NEW_MASGN(0, NEW_POSTARG(NODE_SPECIAL_NO_NAME_REST, $3, &@$), &@$); - /*% %*/ - /*% ripper: mlhs_add_post!(mlhs_add_star!(mlhs_new!, Qnil), $3) %*/ + /*% ripper: mlhs_add_post!(mlhs_add_star!(mlhs_new!, Qnil), $:3) %*/ } ; mlhs_item : mlhs_node | tLPAREN mlhs_inner rparen { - /*%%%*/ - $$ = $2; - /*% %*/ - /*% ripper: mlhs_paren!($2) %*/ + $$ = (NODE *)$2; + /*% ripper: mlhs_paren!($:2) %*/ } ; mlhs_head : mlhs_item ',' { - /*%%%*/ $$ = NEW_LIST($1, &@1); - /*% %*/ - /*% ripper: mlhs_add!(mlhs_new!, $1) %*/ + /*% ripper: mlhs_add!(mlhs_new!, $:1) %*/ } | mlhs_head mlhs_item ',' { - /*%%%*/ $$ = list_append(p, $1, $2); - /*% %*/ - /*% ripper: mlhs_add!($1, $2) %*/ + /*% ripper: mlhs_add!($:1, $:2) %*/ } ; mlhs_post : mlhs_item { - /*%%%*/ $$ = NEW_LIST($1, &@$); - /*% %*/ - /*% ripper: mlhs_add!(mlhs_new!, $1) %*/ + /*% ripper: mlhs_add!(mlhs_new!, $:1) %*/ } | mlhs_post ',' mlhs_item { - /*%%%*/ $$ = list_append(p, $1, $3); - /*% %*/ - /*% ripper: mlhs_add!($1, $3) %*/ + /*% ripper: mlhs_add!($:1, $:3) %*/ } ; mlhs_node : user_variable { - /*%%%*/ + /*% ripper: var_field!($:1) %*/ $$ = assignable(p, $1, 0, &@$); - /*% %*/ - /*% ripper: assignable(p, var_field(p, $1)) %*/ } | keyword_variable { - /*%%%*/ + /*% ripper: var_field!($:1) %*/ $$ = assignable(p, $1, 0, &@$); - /*% %*/ - /*% ripper: assignable(p, var_field(p, $1)) %*/ } | primary_value '[' opt_call_args rbracket { - /*%%%*/ $$ = aryset(p, $1, $3, &@$); - /*% %*/ - /*% ripper: aref_field!($1, escape_Qundef($3)) %*/ + /*% ripper: aref_field!($:1, $:3) %*/ } | primary_value call_op tIDENTIFIER { anddot_multiple_assignment_check(p, &@2, $2); - /*%%%*/ $$ = attrset(p, $1, $2, $3, &@$); - /*% %*/ - /*% ripper: field!($1, $2, $3) %*/ + /*% ripper: field!($:1, $:2, $:3) %*/ } | primary_value tCOLON2 tIDENTIFIER { - /*%%%*/ $$ = attrset(p, $1, idCOLON2, $3, &@$); - /*% %*/ - /*% ripper: const_path_field!($1, $3) %*/ + /*% ripper: const_path_field!($:1, $:3) %*/ } | primary_value call_op tCONSTANT { anddot_multiple_assignment_check(p, &@2, $2); - /*%%%*/ $$ = attrset(p, $1, $2, $3, &@$); - /*% %*/ - /*% ripper: field!($1, $2, $3) %*/ + /*% ripper: field!($:1, $:2, $:3) %*/ } | primary_value tCOLON2 tCONSTANT { - /*%%%*/ + /*% ripper: const_path_field!($:1, $:3) %*/ $$ = const_decl(p, NEW_COLON2($1, $3, &@$), &@$); - /*% %*/ - /*% ripper: const_decl(p, const_path_field!($1, $3)) %*/ } | tCOLON3 tCONSTANT { - /*%%%*/ + /*% ripper: top_const_field!($:2) %*/ $$ = const_decl(p, NEW_COLON3($2, &@$), &@$); - /*% %*/ - /*% ripper: const_decl(p, top_const_field!($2)) %*/ } | backref { /*%%%*/ rb_backref_error(p, $1); - $$ = NEW_BEGIN(0, &@$); /*% %*/ - /*% ripper[error]: backref_error(p, RNODE($1), var_field(p, $1)) %*/ + $$ = NEW_ERROR(&@$); + /*% ripper[error]: backref_error(p, $1, var_field!($:1)) %*/ } ; lhs : user_variable { - /*%%%*/ + /*% ripper: var_field!($:1) %*/ $$ = assignable(p, $1, 0, &@$); - /*% %*/ - /*% ripper: assignable(p, var_field(p, $1)) %*/ } | keyword_variable { - /*%%%*/ + /*% ripper: var_field!($:1) %*/ $$ = assignable(p, $1, 0, &@$); - /*% %*/ - /*% ripper: assignable(p, var_field(p, $1)) %*/ } | primary_value '[' opt_call_args rbracket { - /*%%%*/ $$ = aryset(p, $1, $3, &@$); - /*% %*/ - /*% ripper: aref_field!($1, escape_Qundef($3)) %*/ + /*% ripper: aref_field!($:1, $:3) %*/ } | primary_value call_op tIDENTIFIER { - /*%%%*/ $$ = attrset(p, $1, $2, $3, &@$); - /*% %*/ - /*% ripper: field!($1, $2, $3) %*/ + /*% ripper: field!($:1, $:2, $:3) %*/ } | primary_value tCOLON2 tIDENTIFIER { - /*%%%*/ $$ = attrset(p, $1, idCOLON2, $3, &@$); - /*% %*/ - /*% ripper: field!($1, ID2VAL(idCOLON2), $3) %*/ + /*% ripper: field!($:1, $:2, $:3) %*/ } | primary_value call_op tCONSTANT { - /*%%%*/ $$ = attrset(p, $1, $2, $3, &@$); - /*% %*/ - /*% ripper: field!($1, $2, $3) %*/ + /*% ripper: field!($:1, $:2, $:3) %*/ } | primary_value tCOLON2 tCONSTANT { - /*%%%*/ + /*% ripper: const_path_field!($:1, $:3) %*/ $$ = const_decl(p, NEW_COLON2($1, $3, &@$), &@$); - /*% %*/ - /*% ripper: const_decl(p, const_path_field!($1, $3)) %*/ } | tCOLON3 tCONSTANT { - /*%%%*/ + /*% ripper: top_const_field!($:2) %*/ $$ = const_decl(p, NEW_COLON3($2, &@$), &@$); - /*% %*/ - /*% ripper: const_decl(p, top_const_field!($2)) %*/ } | backref { /*%%%*/ rb_backref_error(p, $1); - $$ = NEW_BEGIN(0, &@$); /*% %*/ - /*% ripper[error]: backref_error(p, RNODE($1), var_field(p, $1)) %*/ + $$ = NEW_ERROR(&@$); + /*% ripper[error]: backref_error(p, $1, var_field!($:1)) %*/ } ; @@ -2511,31 +3767,25 @@ cname : tIDENTIFIER /*%%%*/ yyerror1(&@1, mesg); /*% %*/ - /*% ripper[error]: class_name_error!(ERR_MESG(), $1) %*/ + /*% ripper[error]: class_name_error!(ERR_MESG(), $:1) %*/ } | tCONSTANT ; cpath : tCOLON3 cname { - /*%%%*/ $$ = NEW_COLON3($2, &@$); - /*% %*/ - /*% ripper: top_const_ref!($2) %*/ + /*% ripper: top_const_ref!($:2) %*/ } | cname { - /*%%%*/ - $$ = NEW_COLON2(0, $$, &@$); - /*% %*/ - /*% ripper: const_ref!($1) %*/ + $$ = NEW_COLON2(0, $1, &@$); + /*% ripper: const_ref!($:1) %*/ } | primary_value tCOLON2 cname { - /*%%%*/ $$ = NEW_COLON2($1, $3, &@$); - /*% %*/ - /*% ripper: const_path_ref!($1, $3) %*/ + /*% ripper: const_path_ref!($:1, $:3) %*/ } ; @@ -2552,61 +3802,55 @@ fname : tIDENTIFIER fitem : fname { - /*%%%*/ - $$ = NEW_LIT(ID2SYM($1), &@$); - /*% %*/ - /*% ripper: symbol_literal!($1) %*/ + $$ = NEW_SYM(rb_id2str($1), &@$); + /*% ripper: symbol_literal!($:1) %*/ } | symbol ; undef_list : fitem { - /*%%%*/ $$ = NEW_UNDEF($1, &@$); - /*% %*/ - /*% ripper: rb_ary_new3(1, get_value($1)) %*/ + /*% ripper: rb_ary_new3(1, $:1) %*/ } | undef_list ',' {SET_LEX_STATE(EXPR_FNAME|EXPR_FITEM);} fitem { - /*%%%*/ NODE *undef = NEW_UNDEF($4, &@4); $$ = block_append(p, $1, undef); - /*% %*/ - /*% ripper: rb_ary_push($1, get_value($4)) %*/ - } - ; - -op : '|' { ifndef_ripper($$ = '|'); } - | '^' { ifndef_ripper($$ = '^'); } - | '&' { ifndef_ripper($$ = '&'); } - | tCMP { ifndef_ripper($$ = tCMP); } - | tEQ { ifndef_ripper($$ = tEQ); } - | tEQQ { ifndef_ripper($$ = tEQQ); } - | tMATCH { ifndef_ripper($$ = tMATCH); } - | tNMATCH { ifndef_ripper($$ = tNMATCH); } - | '>' { ifndef_ripper($$ = '>'); } - | tGEQ { ifndef_ripper($$ = tGEQ); } - | '<' { ifndef_ripper($$ = '<'); } - | tLEQ { ifndef_ripper($$ = tLEQ); } - | tNEQ { ifndef_ripper($$ = tNEQ); } - | tLSHFT { ifndef_ripper($$ = tLSHFT); } - | tRSHFT { ifndef_ripper($$ = tRSHFT); } - | '+' { ifndef_ripper($$ = '+'); } - | '-' { ifndef_ripper($$ = '-'); } - | '*' { ifndef_ripper($$ = '*'); } - | tSTAR { ifndef_ripper($$ = '*'); } - | '/' { ifndef_ripper($$ = '/'); } - | '%' { ifndef_ripper($$ = '%'); } - | tPOW { ifndef_ripper($$ = tPOW); } - | tDSTAR { ifndef_ripper($$ = tDSTAR); } - | '!' { ifndef_ripper($$ = '!'); } - | '~' { ifndef_ripper($$ = '~'); } - | tUPLUS { ifndef_ripper($$ = tUPLUS); } - | tUMINUS { ifndef_ripper($$ = tUMINUS); } - | tAREF { ifndef_ripper($$ = tAREF); } - | tASET { ifndef_ripper($$ = tASET); } - | '`' { ifndef_ripper($$ = '`'); } + /*% ripper: rb_ary_push($:1, $:4) %*/ + } + ; + +op : '|' { $$ = '|'; } + | '^' { $$ = '^'; } + | '&' { $$ = '&'; } + | tCMP { $$ = tCMP; } + | tEQ { $$ = tEQ; } + | tEQQ { $$ = tEQQ; } + | tMATCH { $$ = tMATCH; } + | tNMATCH { $$ = tNMATCH; } + | '>' { $$ = '>'; } + | tGEQ { $$ = tGEQ; } + | '<' { $$ = '<'; } + | tLEQ { $$ = tLEQ; } + | tNEQ { $$ = tNEQ; } + | tLSHFT { $$ = tLSHFT; } + | tRSHFT { $$ = tRSHFT; } + | '+' { $$ = '+'; } + | '-' { $$ = '-'; } + | '*' { $$ = '*'; } + | tSTAR { $$ = '*'; } + | '/' { $$ = '/'; } + | '%' { $$ = '%'; } + | tPOW { $$ = tPOW; } + | tDSTAR { $$ = tDSTAR; } + | '!' { $$ = '!'; } + | '~' { $$ = '~'; } + | tUPLUS { $$ = tUPLUS; } + | tUMINUS { $$ = tUMINUS; } + | tAREF { $$ = tAREF; } + | tASET { $$ = tASET; } + | '`' { $$ = '`'; } ; reswords : keyword__LINE__ | keyword__FILE__ | keyword__ENCODING__ @@ -2625,251 +3869,246 @@ reswords : keyword__LINE__ | keyword__FILE__ | keyword__ENCODING__ arg : lhs '=' lex_ctxt arg_rhs { - /*%%%*/ $$ = node_assign(p, $1, $4, $3, &@$); - /*% %*/ - /*% ripper: assign!($1, $4) %*/ + /*% ripper: assign!($:1, $:4) %*/ } | var_lhs tOP_ASGN lex_ctxt arg_rhs { - /*%%%*/ $$ = new_op_assign(p, $1, $2, $4, $3, &@$); - /*% %*/ - /*% ripper: opassign!($1, $2, $4) %*/ + /*% ripper: opassign!($:1, $:2, $:4) %*/ } | primary_value '[' opt_call_args rbracket tOP_ASGN lex_ctxt arg_rhs { - /*%%%*/ $$ = new_ary_op_assign(p, $1, $3, $5, $7, &@3, &@$); - /*% %*/ - /*% ripper: opassign!(aref_field!($1, escape_Qundef($3)), $5, $7) %*/ + /*% ripper: opassign!(aref_field!($:1, $:3), $:5, $:7) %*/ } | primary_value call_op tIDENTIFIER tOP_ASGN lex_ctxt arg_rhs { - /*%%%*/ $$ = new_attr_op_assign(p, $1, $2, $3, $4, $6, &@$); - /*% %*/ - /*% ripper: opassign!(field!($1, $2, $3), $4, $6) %*/ + /*% ripper: opassign!(field!($:1, $:2, $:3), $:4, $:6) %*/ } | primary_value call_op tCONSTANT tOP_ASGN lex_ctxt arg_rhs { - /*%%%*/ $$ = new_attr_op_assign(p, $1, $2, $3, $4, $6, &@$); - /*% %*/ - /*% ripper: opassign!(field!($1, $2, $3), $4, $6) %*/ + /*% ripper: opassign!(field!($:1, $:2, $:3), $:4, $:6) %*/ } | primary_value tCOLON2 tIDENTIFIER tOP_ASGN lex_ctxt arg_rhs { - /*%%%*/ - $$ = new_attr_op_assign(p, $1, ID2VAL(idCOLON2), $3, $4, $6, &@$); - /*% %*/ - /*% ripper: opassign!(field!($1, ID2VAL(idCOLON2), $3), $4, $6) %*/ + $$ = new_attr_op_assign(p, $1, idCOLON2, $3, $4, $6, &@$); + /*% ripper: opassign!(field!($:1, $:2, $:3), $:4, $:6) %*/ } | primary_value tCOLON2 tCONSTANT tOP_ASGN lex_ctxt arg_rhs { - /*%%%*/ YYLTYPE loc = code_loc_gen(&@1, &@3); $$ = new_const_op_assign(p, NEW_COLON2($1, $3, &loc), $4, $6, $5, &@$); - /*% %*/ - /*% ripper: opassign!(const_path_field!($1, $3), $4, $6) %*/ + /*% ripper: opassign!(const_path_field!($:1, $:3), $:4, $:6) %*/ } | tCOLON3 tCONSTANT tOP_ASGN lex_ctxt arg_rhs { - /*%%%*/ YYLTYPE loc = code_loc_gen(&@1, &@2); $$ = new_const_op_assign(p, NEW_COLON3($2, &loc), $3, $5, $4, &@$); - /*% %*/ - /*% ripper: opassign!(top_const_field!($2), $3, $5) %*/ + /*% ripper: opassign!(top_const_field!($:2), $:3, $:5) %*/ } | backref tOP_ASGN lex_ctxt arg_rhs { - /*%%%*/ rb_backref_error(p, $1); - $$ = NEW_BEGIN(0, &@$); + /*%%%*/ + $$ = NEW_ERROR(&@$); /*% %*/ - /*% ripper[error]: backref_error(p, RNODE($1), opassign!(var_field(p, $1), $2, $4)) %*/ + /*% ripper[error]: backref_error(p, $1, opassign!(var_field!($:1), $:2, $:4)) %*/ } | arg tDOT2 arg { - /*%%%*/ value_expr($1); value_expr($3); $$ = NEW_DOT2($1, $3, &@$); - /*% %*/ - /*% ripper: dot2!($1, $3) %*/ + /*% ripper: dot2!($:1, $:3) %*/ } | arg tDOT3 arg { - /*%%%*/ value_expr($1); value_expr($3); $$ = NEW_DOT3($1, $3, &@$); - /*% %*/ - /*% ripper: dot3!($1, $3) %*/ + /*% ripper: dot3!($:1, $:3) %*/ } | arg tDOT2 { - /*%%%*/ value_expr($1); $$ = NEW_DOT2($1, new_nil_at(p, &@2.end_pos), &@$); - /*% %*/ - /*% ripper: dot2!($1, Qnil) %*/ + /*% ripper: dot2!($:1, Qnil) %*/ } | arg tDOT3 { - /*%%%*/ value_expr($1); $$ = NEW_DOT3($1, new_nil_at(p, &@2.end_pos), &@$); - /*% %*/ - /*% ripper: dot3!($1, Qnil) %*/ + /*% ripper: dot3!($:1, Qnil) %*/ } | tBDOT2 arg { - /*%%%*/ value_expr($2); $$ = NEW_DOT2(new_nil_at(p, &@1.beg_pos), $2, &@$); - /*% %*/ - /*% ripper: dot2!(Qnil, $2) %*/ + /*% ripper: dot2!(Qnil, $:2) %*/ } | tBDOT3 arg { - /*%%%*/ value_expr($2); $$ = NEW_DOT3(new_nil_at(p, &@1.beg_pos), $2, &@$); - /*% %*/ - /*% ripper: dot3!(Qnil, $2) %*/ + /*% ripper: dot3!(Qnil, $:2) %*/ } | arg '+' arg { $$ = call_bin_op(p, $1, '+', $3, &@2, &@$); + /*% ripper: binary!($:1, ID2VAL('\'+\''), $:3) %*/ } | arg '-' arg { $$ = call_bin_op(p, $1, '-', $3, &@2, &@$); + /*% ripper: binary!($:1, ID2VAL('\'-\''), $:3) %*/ } | arg '*' arg { $$ = call_bin_op(p, $1, '*', $3, &@2, &@$); + /*% ripper: binary!($:1, ID2VAL('\'*\''), $:3) %*/ } | arg '/' arg { $$ = call_bin_op(p, $1, '/', $3, &@2, &@$); + /*% ripper: binary!($:1, ID2VAL('\'/\''), $:3) %*/ } | arg '%' arg { $$ = call_bin_op(p, $1, '%', $3, &@2, &@$); + /*% ripper: binary!($:1, ID2VAL('\'%\''), $:3) %*/ } | arg tPOW arg { $$ = call_bin_op(p, $1, idPow, $3, &@2, &@$); + /*% ripper: binary!($:1, ID2VAL(idPow), $:3) %*/ } | tUMINUS_NUM simple_numeric tPOW arg { $$ = call_uni_op(p, call_bin_op(p, $2, idPow, $4, &@2, &@$), idUMinus, &@1, &@$); + /*% ripper: unary!(ID2VAL(idUMinus), binary!($:2, ID2VAL(idPow), $:4)) %*/ } | tUPLUS arg { $$ = call_uni_op(p, $2, idUPlus, &@1, &@$); + /*% ripper: unary!(ID2VAL(idUPlus), $:2) %*/ } | tUMINUS arg { $$ = call_uni_op(p, $2, idUMinus, &@1, &@$); + /*% ripper: unary!(ID2VAL(idUMinus), $:2) %*/ } | arg '|' arg { $$ = call_bin_op(p, $1, '|', $3, &@2, &@$); + /*% ripper: binary!($:1, ID2VAL('\'|\''), $:3) %*/ } | arg '^' arg { $$ = call_bin_op(p, $1, '^', $3, &@2, &@$); + /*% ripper: binary!($:1, ID2VAL('\'^\''), $:3) %*/ } | arg '&' arg { $$ = call_bin_op(p, $1, '&', $3, &@2, &@$); + /*% ripper: binary!($:1, ID2VAL('\'&\''), $:3) %*/ } | arg tCMP arg { $$ = call_bin_op(p, $1, idCmp, $3, &@2, &@$); + /*% ripper: binary!($:1, ID2VAL(idCmp), $:3) %*/ } | rel_expr %prec tCMP | arg tEQ arg { $$ = call_bin_op(p, $1, idEq, $3, &@2, &@$); + /*% ripper: binary!($:1, ID2VAL(idEq), $:3) %*/ } | arg tEQQ arg { $$ = call_bin_op(p, $1, idEqq, $3, &@2, &@$); + /*% ripper: binary!($:1, ID2VAL(idEqq), $:3) %*/ } | arg tNEQ arg { $$ = call_bin_op(p, $1, idNeq, $3, &@2, &@$); + /*% ripper: binary!($:1, ID2VAL(idNeq), $:3) %*/ } | arg tMATCH arg { $$ = match_op(p, $1, $3, &@2, &@$); + /*% ripper: binary!($:1, ID2VAL(idEqTilde), $:3) %*/ } | arg tNMATCH arg { $$ = call_bin_op(p, $1, idNeqTilde, $3, &@2, &@$); + /*% ripper: binary!($:1, ID2VAL(idNeqTilde), $:3) %*/ } | '!' arg { $$ = call_uni_op(p, method_cond(p, $2, &@2), '!', &@1, &@$); + /*% ripper: unary!(ID2VAL('\'!\''), $:2) %*/ } | '~' arg { $$ = call_uni_op(p, $2, '~', &@1, &@$); + /*% ripper: unary!(ID2VAL('\'~\''), $:2) %*/ } | arg tLSHFT arg { $$ = call_bin_op(p, $1, idLTLT, $3, &@2, &@$); + /*% ripper: binary!($:1, ID2VAL(idLTLT), $:3) %*/ } | arg tRSHFT arg { $$ = call_bin_op(p, $1, idGTGT, $3, &@2, &@$); + /*% ripper: binary!($:1, ID2VAL(idGTGT), $:3) %*/ } | arg tANDOP arg { $$ = logop(p, idANDOP, $1, $3, &@2, &@$); + /*% ripper: binary!($:1, ID2VAL(idANDOP), $:3) %*/ } | arg tOROP arg { $$ = logop(p, idOROP, $1, $3, &@2, &@$); + /*% ripper: binary!($:1, ID2VAL(idOROP), $:3) %*/ } - | keyword_defined opt_nl {p->ctxt.in_defined = 1;} arg + | keyword_defined '\n'? begin_defined arg { - p->ctxt.in_defined = 0; + p->ctxt.in_defined = $3.in_defined; $$ = new_defined(p, $4, &@$); + /*% ripper: defined!($:4) %*/ } - | arg '?' arg opt_nl ':' arg + | arg '?' arg '\n'? ':' arg { - /*%%%*/ value_expr($1); $$ = new_if(p, $1, $3, $6, &@$); fixpos($$, $1); - /*% %*/ - /*% ripper: ifop!($1, $3, $6) %*/ + /*% ripper: ifop!($:1, $:3, $:6) %*/ } - | defn_head f_opt_paren_args '=' endless_arg + | defn_head[head] f_opt_paren_args[args] '=' endless_arg[bodystmt] { - endless_method_name(p, $<node>1, &@1); - restore_defun(p, $<node>1->nd_defn); - /*%%%*/ - $$ = set_defun_body(p, $1, $2, $4, &@$); - /*% %*/ - /*% ripper: def!(get_value($1), $2, bodystmt!($4, Qnil, Qnil, Qnil)) %*/ + endless_method_name(p, $head->nd_mid, &@head); + restore_defun(p, $head); + $bodystmt = new_scope_body(p, $args, $bodystmt, &@$); + ($$ = $head->nd_def)->nd_loc = @$; + RNODE_DEFN($$)->nd_defn = $bodystmt; + /*% ripper: bodystmt!($:bodystmt, Qnil, Qnil, Qnil) %*/ + /*% ripper: def!($:head, $:args, $:$) %*/ local_pop(p); } - | defs_head f_opt_paren_args '=' endless_arg + | defs_head[head] f_opt_paren_args[args] '=' endless_arg[bodystmt] { - endless_method_name(p, $<node>1, &@1); - restore_defun(p, $<node>1->nd_defn); - /*%%%*/ - $$ = set_defun_body(p, $1, $2, $4, &@$); - /*% - $1 = get_value($1); - %*/ - /*% ripper: defs!(AREF($1, 0), AREF($1, 1), AREF($1, 2), $2, bodystmt!($4, Qnil, Qnil, Qnil)) %*/ + endless_method_name(p, $head->nd_mid, &@head); + restore_defun(p, $head); + $bodystmt = new_scope_body(p, $args, $bodystmt, &@$); + ($$ = $head->nd_def)->nd_loc = @$; + RNODE_DEFS($$)->nd_defn = $bodystmt; + /*% ripper: bodystmt!($:bodystmt, Qnil, Qnil, Qnil) %*/ + /*% ripper: defs!(*$:head[0..2], $:args, $:$) %*/ local_pop(p); } | primary @@ -2879,16 +4118,16 @@ arg : lhs '=' lex_ctxt arg_rhs ; endless_arg : arg %prec modifier_rescue - | endless_arg modifier_rescue arg + | endless_arg modifier_rescue after_rescue arg { - /*%%%*/ - $$ = rescued_expr(p, $1, $3, &@1, &@2, &@3); - /*% %*/ - /*% ripper: rescue_mod!($1, $3) %*/ + p->ctxt.in_rescue = $3.in_rescue; + $$ = rescued_expr(p, $1, $4, &@1, &@2, &@4); + /*% ripper: rescue_mod!($:1, $:4) %*/ } - | keyword_not opt_nl endless_arg + | keyword_not '\n'? endless_arg { $$ = call_uni_op(p, method_cond(p, $3, &@3), METHOD_NOT, &@1, &@$); + /*% ripper: unary!(ID2VAL(idNOT), $:3) %*/ } ; @@ -2901,11 +4140,13 @@ relop : '>' {$$ = '>';} rel_expr : arg relop arg %prec '>' { $$ = call_bin_op(p, $1, $2, $3, &@2, &@$); + /*% ripper: binary!($:1, ID2VAL($2), $:3) %*/ } | rel_expr relop arg %prec '>' { rb_warning1("comparison '%s' after comparison", WARN_ID($2)); $$ = call_bin_op(p, $1, $2, $3, &@2, &@$); + /*% ripper: binary!($:1, ID2VAL($2), $:3) %*/ } ; @@ -2915,6 +4156,20 @@ lex_ctxt : none } ; +begin_defined : lex_ctxt + { + p->ctxt.in_defined = 1; + $$ = $1; + } + ; + +after_rescue : lex_ctxt + { + p->ctxt.in_rescue = after_rescue; + $$ = $1; + } + ; + arg_value : arg { value_expr($1); @@ -2929,17 +4184,13 @@ aref_args : none } | args ',' assocs trailer { - /*%%%*/ $$ = $3 ? arg_append(p, $1, new_hash(p, $3, &@3), &@$) : $1; - /*% %*/ - /*% ripper: args_add!($1, bare_assoc_hash!($3)) %*/ + /*% ripper: args_add!($:1, bare_assoc_hash!($:3)) %*/ } | assocs trailer { - /*%%%*/ $$ = $1 ? NEW_LIST(new_hash(p, $1, &@1), &@$) : 0; - /*% %*/ - /*% ripper: args_add!(args_new!, bare_assoc_hash!($1)) %*/ + /*% ripper: args_add!(args_new!, bare_assoc_hash!($:1)) %*/ } ; @@ -2948,108 +4199,92 @@ arg_rhs : arg %prec tOP_ASGN value_expr($1); $$ = $1; } - | arg modifier_rescue arg + | arg modifier_rescue after_rescue arg { - /*%%%*/ + p->ctxt.in_rescue = $3.in_rescue; value_expr($1); - $$ = rescued_expr(p, $1, $3, &@1, &@2, &@3); - /*% %*/ - /*% ripper: rescue_mod!($1, $3) %*/ + $$ = rescued_expr(p, $1, $4, &@1, &@2, &@4); + /*% ripper: rescue_mod!($:1, $:4) %*/ } ; paren_args : '(' opt_call_args rparen { - /*%%%*/ $$ = $2; - /*% %*/ - /*% ripper: arg_paren!(escape_Qundef($2)) %*/ + /*% ripper: arg_paren!($:2) %*/ } | '(' args ',' args_forward rparen { if (!check_forwarding_args(p)) { - $$ = Qnone; + $$ = 0; } else { - /*%%%*/ $$ = new_args_forward_call(p, $2, &@4, &@$); - /*% %*/ - /*% ripper: arg_paren!(args_add!($2, $4)) %*/ + /*% ripper: arg_paren!(args_add!($:2, $:4)) %*/ } } | '(' args_forward rparen { if (!check_forwarding_args(p)) { - $$ = Qnone; + $$ = 0; } else { - /*%%%*/ $$ = new_args_forward_call(p, 0, &@2, &@$); - /*% %*/ - /*% ripper: arg_paren!($2) %*/ + /*% ripper: arg_paren!($:2) %*/ } } ; opt_paren_args : none | paren_args + { + $$ = $1 ? $1 : NODE_SPECIAL_EMPTY_ARGS; + } ; opt_call_args : none | call_args | args ',' { - $$ = $1; + $$ = $1; } | args ',' assocs ',' { - /*%%%*/ $$ = $3 ? arg_append(p, $1, new_hash(p, $3, &@3), &@$) : $1; - /*% %*/ - /*% ripper: args_add!($1, bare_assoc_hash!($3)) %*/ + /*% ripper: args_add!($:1, bare_assoc_hash!($:3)) %*/ } | assocs ',' { - /*%%%*/ $$ = $1 ? NEW_LIST(new_hash(p, $1, &@1), &@1) : 0; - /*% %*/ - /*% ripper: args_add!(args_new!, bare_assoc_hash!($1)) %*/ + /*% ripper: args_add!(args_new!, bare_assoc_hash!($:1)) %*/ } ; call_args : command { - /*%%%*/ value_expr($1); $$ = NEW_LIST($1, &@$); - /*% %*/ - /*% ripper: args_add!(args_new!, $1) %*/ + /*% ripper: args_add!(args_new!, $:1) %*/ } | args opt_block_arg { - /*%%%*/ $$ = arg_blk_pass($1, $2); - /*% %*/ - /*% ripper: args_add_block!($1, $2) %*/ + /*% ripper: args_add_block!($:1, $:2) %*/ } | assocs opt_block_arg { - /*%%%*/ $$ = $1 ? NEW_LIST(new_hash(p, $1, &@1), &@1) : 0; $$ = arg_blk_pass($$, $2); - /*% %*/ - /*% ripper: args_add_block!(args_add!(args_new!, bare_assoc_hash!($1)), $2) %*/ + /*% ripper: args_add_block!(args_add!(args_new!, bare_assoc_hash!($:1)), $:2) %*/ } | args ',' assocs opt_block_arg { - /*%%%*/ $$ = $3 ? arg_append(p, $1, new_hash(p, $3, &@3), &@$) : $1; $$ = arg_blk_pass($$, $4); - /*% %*/ - /*% ripper: args_add_block!(args_add!($1, bare_assoc_hash!($3)), $4) %*/ + /*% ripper: args_add_block!(args_add!($:1, bare_assoc_hash!($:3)), $:4) %*/ } | block_arg - /*% ripper[brace]: args_add_block!(args_new!, $1) %*/ + /*% ripper: args_add_block!(args_new!, $:1) %*/ ; command_args : { @@ -3086,24 +4321,19 @@ command_args : { CMDARG_POP(); if (lookahead) CMDARG_PUSH(0); $$ = $2; + /*% ripper: $:2 %*/ } ; block_arg : tAMPER arg_value { - /*%%%*/ $$ = NEW_BLOCK_PASS($2, &@$); - /*% %*/ - /*% ripper: $2 %*/ + /*% ripper: $:2 %*/ } | tAMPER { - if (!local_id(p, idFWD_BLOCK)) { - compile_error(p, "no anonymous block parameter"); - } - /*%%%*/ + forwarding_arg_check(p, idFWD_BLOCK, idFWD_ALL, "block"); $$ = NEW_BLOCK_PASS(NEW_LVAR(idFWD_BLOCK, &@1), &@$); - /*% %*/ /*% ripper: Qnil %*/ } ; @@ -3111,63 +4341,49 @@ block_arg : tAMPER arg_value opt_block_arg : ',' block_arg { $$ = $2; + /*% ripper: $:2 %*/ } | none { $$ = 0; + /*% ripper: Qfalse %*/ } ; /* value */ args : arg_value { - /*%%%*/ $$ = NEW_LIST($1, &@$); - /*% %*/ - /*% ripper: args_add!(args_new!, $1) %*/ - } - | tSTAR arg_value - { - /*%%%*/ - $$ = NEW_SPLAT($2, &@$); - /*% %*/ - /*% ripper: args_add_star!(args_new!, $2) %*/ + /*% ripper: args_add!(args_new!, $:1) %*/ } - | tSTAR + | arg_splat { - if (!local_id(p, idFWD_REST) || - local_id(p, idFWD_ALL)) { - compile_error(p, "no anonymous rest parameter"); - } - /*%%%*/ - $$ = NEW_SPLAT(NEW_LVAR(idFWD_REST, &@1), &@$); - /*% %*/ - /*% ripper: args_add_star!(args_new!, Qnil) %*/ + $$ = NEW_SPLAT($arg_splat, &@$); + /*% ripper: args_add_star!(args_new!, $:arg_splat) %*/ } | args ',' arg_value { - /*%%%*/ $$ = last_arg_append(p, $1, $3, &@$); - /*% %*/ - /*% ripper: args_add!($1, $3) %*/ + /*% ripper: args_add!($:1, $:3) %*/ } - | args ',' tSTAR arg_value + | args ',' arg_splat { - /*%%%*/ - $$ = rest_arg_append(p, $1, $4, &@$); - /*% %*/ - /*% ripper: args_add_star!($1, $4) %*/ + $$ = rest_arg_append(p, $1, $3, &@$); + /*% ripper: args_add_star!($:1, $:3) %*/ } - | args ',' tSTAR + ; + +/* value */ +arg_splat : tSTAR arg_value { - if (!local_id(p, idFWD_REST) || - local_id(p, idFWD_ALL)) { - compile_error(p, "no anonymous rest parameter"); - } - /*%%%*/ - $$ = rest_arg_append(p, $1, NEW_LVAR(idFWD_REST, &@3), &@$); - /*% %*/ - /*% ripper: args_add_star!($1, Qnil) %*/ + $$ = $2; + /*% ripper: $:2 %*/ + } + | tSTAR /* none */ + { + forwarding_arg_check(p, idFWD_REST, idFWD_ALL, "rest"); + $$ = NEW_LVAR(idFWD_REST, &@1); + /*% ripper: Qnil %*/ } ; @@ -3179,24 +4395,18 @@ mrhs_arg : mrhs /* value */ mrhs : args ',' arg_value { - /*%%%*/ $$ = last_arg_append(p, $1, $3, &@$); - /*% %*/ - /*% ripper: mrhs_add!(mrhs_new_from_args!($1), $3) %*/ + /*% ripper: mrhs_add!(mrhs_new_from_args!($:1), $:3) %*/ } | args ',' tSTAR arg_value { - /*%%%*/ $$ = rest_arg_append(p, $1, $4, &@$); - /*% %*/ - /*% ripper: mrhs_add_star!(mrhs_new_from_args!($1), $4) %*/ + /*% ripper: mrhs_add_star!(mrhs_new_from_args!($:1), $:4) %*/ } | tSTAR arg_value { - /*%%%*/ $$ = NEW_SPLAT($2, &@$); - /*% %*/ - /*% ripper: mrhs_add_star!(mrhs_new!, $2) %*/ + /*% ripper: mrhs_add_star!(mrhs_new!, $:2) %*/ } ; @@ -3212,10 +4422,8 @@ primary : literal | backref | tFID { - /*%%%*/ - $$ = NEW_FCALL($1, 0, &@$); - /*% %*/ - /*% ripper: method_add_arg!(fcall!($1), args_new!) %*/ + $$ = (NODE *)NEW_FCALL($1, 0, &@$); + /*% ripper: method_add_arg!(fcall!($:1), args_new!) %*/ } | k_begin { @@ -3225,121 +4433,91 @@ primary : literal k_end { CMDARG_POP(); - /*%%%*/ set_line_body($3, @1.end_pos.lineno); $$ = NEW_BEGIN($3, &@$); nd_set_line($$, @1.end_pos.lineno); - /*% %*/ - /*% ripper: begin!($3) %*/ - } - | tLPAREN_ARG {SET_LEX_STATE(EXPR_ENDARG);} rparen - { - /*%%%*/ - $$ = NEW_BEGIN(0, &@$); - /*% %*/ - /*% ripper: paren!(0) %*/ + /*% ripper: begin!($:3) %*/ } - | tLPAREN_ARG stmt {SET_LEX_STATE(EXPR_ENDARG);} rparen + | tLPAREN_ARG compstmt {SET_LEX_STATE(EXPR_ENDARG);} ')' { - /*%%%*/ - if (nd_type_p($2, NODE_SELF)) $2->nd_state = 0; + if (nd_type_p($2, NODE_SELF)) RNODE_SELF($2)->nd_state = 0; $$ = $2; - /*% %*/ - /*% ripper: paren!($2) %*/ + /*% ripper: paren!($:2) %*/ } | tLPAREN compstmt ')' { - /*%%%*/ - if (nd_type_p($2, NODE_SELF)) $2->nd_state = 0; - $$ = $2; - /*% %*/ - /*% ripper: paren!($2) %*/ + if (nd_type_p($2, NODE_SELF)) RNODE_SELF($2)->nd_state = 0; + $$ = NEW_BLOCK($2, &@$); + /*% ripper: paren!($:2) %*/ } | primary_value tCOLON2 tCONSTANT { - /*%%%*/ $$ = NEW_COLON2($1, $3, &@$); - /*% %*/ - /*% ripper: const_path_ref!($1, $3) %*/ + /*% ripper: const_path_ref!($:1, $:3) %*/ } | tCOLON3 tCONSTANT { - /*%%%*/ $$ = NEW_COLON3($2, &@$); - /*% %*/ - /*% ripper: top_const_ref!($2) %*/ + /*% ripper: top_const_ref!($:2) %*/ } | tLBRACK aref_args ']' { - /*%%%*/ $$ = make_list($2, &@$); - /*% %*/ - /*% ripper: array!(escape_Qundef($2)) %*/ + /*% ripper: array!($:2) %*/ } | tLBRACE assoc_list '}' { - /*%%%*/ $$ = new_hash(p, $2, &@$); - $$->nd_brace = TRUE; - /*% %*/ - /*% ripper: hash!(escape_Qundef($2)) %*/ + RNODE_HASH($$)->nd_brace = TRUE; + /*% ripper: hash!($:2) %*/ } | k_return { - /*%%%*/ $$ = NEW_RETURN(0, &@$); - /*% %*/ /*% ripper: return0! %*/ } - | keyword_yield '(' call_args rparen + | k_yield '(' call_args rparen { - /*%%%*/ $$ = new_yield(p, $3, &@$); - /*% %*/ - /*% ripper: yield!(paren!($3)) %*/ + /*% ripper: yield!(paren!($:3)) %*/ } - | keyword_yield '(' rparen + | k_yield '(' rparen { - /*%%%*/ $$ = NEW_YIELD(0, &@$); - /*% %*/ /*% ripper: yield!(paren!(args_new!)) %*/ } - | keyword_yield + | k_yield { - /*%%%*/ $$ = NEW_YIELD(0, &@$); - /*% %*/ /*% ripper: yield0! %*/ } - | keyword_defined opt_nl '(' {p->ctxt.in_defined = 1;} expr rparen + | keyword_defined '\n'? '(' begin_defined expr rparen { - p->ctxt.in_defined = 0; + p->ctxt.in_defined = $4.in_defined; $$ = new_defined(p, $5, &@$); + /*% ripper: defined!($:5) %*/ } | keyword_not '(' expr rparen { $$ = call_uni_op(p, method_cond(p, $3, &@3), METHOD_NOT, &@1, &@$); + /*% ripper: unary!(ID2VAL(idNOT), $:3) %*/ } | keyword_not '(' rparen { $$ = call_uni_op(p, method_cond(p, new_nil(&@2), &@2), METHOD_NOT, &@1, &@$); + /*% ripper: unary!(ID2VAL(idNOT), Qnil) %*/ } | fcall brace_block { - /*%%%*/ - $$ = method_add_block(p, $1, $2, &@$); - /*% %*/ - /*% ripper: method_add_block!(method_add_arg!(fcall!($1), args_new!), $2) %*/ + $$ = method_add_block(p, (NODE *)$1, $2, &@$); + /*% ripper: method_add_block!(method_add_arg!(fcall!($:1), args_new!), $:2) %*/ } | method_call | method_call brace_block { - /*%%%*/ - block_dup_check(p, $1->nd_args, $2); + block_dup_check(p, get_nd_args(p, $1), $2); $$ = method_add_block(p, $1, $2, &@$); - /*% %*/ - /*% ripper: method_add_block!($1, $2) %*/ + /*% ripper: method_add_block!($:1, $:2) %*/ } | lambda | k_if expr_value then @@ -3347,88 +4525,76 @@ primary : literal if_tail k_end { - /*%%%*/ $$ = new_if(p, $2, $4, $5, &@$); fixpos($$, $2); - /*% %*/ - /*% ripper: if!($2, $4, escape_Qundef($5)) %*/ + /*% ripper: if!($:2, $:4, $:5) %*/ } | k_unless expr_value then compstmt opt_else k_end { - /*%%%*/ $$ = new_unless(p, $2, $4, $5, &@$); fixpos($$, $2); - /*% %*/ - /*% ripper: unless!($2, $4, escape_Qundef($5)) %*/ + /*% ripper: unless!($:2, $:4, $:5) %*/ } | k_while expr_value_do compstmt k_end { - /*%%%*/ + restore_block_exit(p, $1); $$ = NEW_WHILE(cond(p, $2, &@2), $3, 1, &@$); fixpos($$, $2); - /*% %*/ - /*% ripper: while!($2, $3) %*/ + /*% ripper: while!($:2, $:3) %*/ } | k_until expr_value_do compstmt k_end { - /*%%%*/ + restore_block_exit(p, $1); $$ = NEW_UNTIL(cond(p, $2, &@2), $3, 1, &@$); fixpos($$, $2); - /*% %*/ - /*% ripper: until!($2, $3) %*/ + /*% ripper: until!($:2, $:3) %*/ } - | k_case expr_value opt_terms + | k_case expr_value terms? { - $<val>$ = p->case_labels; - p->case_labels = Qnil; - } + $$ = p->case_labels; + p->case_labels = CHECK_LITERAL_WHEN; + }<labels> case_body k_end { - if (RTEST(p->case_labels)) rb_hash_clear(p->case_labels); - p->case_labels = $<val>4; - /*%%%*/ + if (CASE_LABELS_ENABLED_P(p->case_labels)) st_free_table(p->case_labels); + p->case_labels = $4; $$ = NEW_CASE($2, $5, &@$); fixpos($$, $2); - /*% %*/ - /*% ripper: case!($2, $5) %*/ + /*% ripper: case!($:2, $:5) %*/ } - | k_case opt_terms + | k_case terms? { - $<val>$ = p->case_labels; + $$ = p->case_labels; p->case_labels = 0; - } + }<labels> case_body k_end { - if (RTEST(p->case_labels)) rb_hash_clear(p->case_labels); - p->case_labels = $<val>3; - /*%%%*/ + if (p->case_labels) st_free_table(p->case_labels); + p->case_labels = $3; $$ = NEW_CASE2($4, &@$); - /*% %*/ - /*% ripper: case!(Qnil, $4) %*/ + /*% ripper: case!(Qnil, $:4) %*/ } - | k_case expr_value opt_terms + | k_case expr_value terms? p_case_body k_end { - /*%%%*/ $$ = NEW_CASE3($2, $4, &@$); - /*% %*/ - /*% ripper: case!($2, $4) %*/ + /*% ripper: case!($:2, $:4) %*/ } | k_for for_var keyword_in expr_value_do compstmt k_end { - /*%%%*/ + restore_block_exit(p, $1); /* * for a, b, c in e * #=> @@ -3439,15 +4605,16 @@ primary : literal * e.each{|x| a, = x} */ ID id = internal_id(p); - NODE *m = NEW_ARGS_AUX(0, 0, &NULL_LOC); - NODE *args, *scope, *internal_var = NEW_DVAR(id, &@2); + rb_node_args_aux_t *m = NEW_ARGS_AUX(0, 0, &NULL_LOC); + rb_node_args_t *args; + NODE *scope, *internal_var = NEW_DVAR(id, &@2); rb_ast_id_table_t *tbl = rb_ast_new_local_table(p->ast, 1); tbl->ids[0] = id; /* internal id */ switch (nd_type($2)) { case NODE_LASGN: case NODE_DASGN: /* e.each {|internal_var| a = internal_var; ... } */ - $2->nd_value = internal_var; + set_nd_value(p, $2, internal_var); id = 0; m->nd_plen = 1; m->nd_next = $2; @@ -3456,146 +4623,121 @@ primary : literal m->nd_next = node_assign(p, $2, NEW_FOR_MASGN(internal_var, &@2), NO_LEX_CTXT, &@2); break; default: /* e.each {|*internal_var| @a, B, c[1], d.attr = internal_val; ... } */ - m->nd_next = node_assign(p, NEW_MASGN(NEW_LIST($2, &@2), 0, &@2), internal_var, NO_LEX_CTXT, &@2); + m->nd_next = node_assign(p, (NODE *)NEW_MASGN(NEW_LIST($2, &@2), 0, &@2), internal_var, NO_LEX_CTXT, &@2); } /* {|*internal_id| <m> = internal_id; ... } */ args = new_args(p, m, 0, id, 0, new_args_tail(p, 0, 0, 0, &@2), &@2); - scope = NEW_NODE(NODE_SCOPE, tbl, $5, args, &@$); + scope = NEW_SCOPE2(tbl, args, $5, &@$); $$ = NEW_FOR($4, scope, &@$); fixpos($$, $2); - /*% %*/ - /*% ripper: for!($2, $4, $5) %*/ + /*% ripper: for!($:2, $:4, $:5) %*/ } | k_class cpath superclass { - if (p->ctxt.in_def) { - YYLTYPE loc = code_loc_gen(&@1, &@2); - yyerror1(&loc, "class definition in method body"); - } - p->ctxt.in_class = 1; - local_push(p, 0); + begin_definition("class", &@k_class, &@cpath); } bodystmt k_end { - /*%%%*/ - $$ = NEW_CLASS($2, $5, $3, &@$); - nd_set_line($$->nd_body, @6.end_pos.lineno); - set_line_body($5, @3.end_pos.lineno); - nd_set_line($$, @3.end_pos.lineno); - /*% %*/ - /*% ripper: class!($2, $3, $5) %*/ + $$ = NEW_CLASS($cpath, $bodystmt, $superclass, &@$); + nd_set_line(RNODE_CLASS($$)->nd_body, @k_end.end_pos.lineno); + set_line_body($bodystmt, @superclass.end_pos.lineno); + nd_set_line($$, @superclass.end_pos.lineno); + /*% ripper: class!($:cpath, $:superclass, $:bodystmt) %*/ local_pop(p); - p->ctxt.in_class = $<ctxt>1.in_class; - p->ctxt.shareable_constant_value = $<ctxt>1.shareable_constant_value; + p->ctxt.in_class = $k_class.in_class; + p->ctxt.shareable_constant_value = $k_class.shareable_constant_value; } - | k_class tLSHFT expr + | k_class tLSHFT expr_value { - p->ctxt.in_def = 0; - p->ctxt.in_class = 0; - local_push(p, 0); + begin_definition("", &@k_class, &@tLSHFT); } term bodystmt k_end { - /*%%%*/ - $$ = NEW_SCLASS($3, $6, &@$); - nd_set_line($$->nd_body, @7.end_pos.lineno); - set_line_body($6, nd_line($3)); - fixpos($$, $3); - /*% %*/ - /*% ripper: sclass!($3, $6) %*/ + $$ = NEW_SCLASS($expr_value, $bodystmt, &@$); + nd_set_line(RNODE_SCLASS($$)->nd_body, @k_end.end_pos.lineno); + set_line_body($bodystmt, nd_line($expr_value)); + fixpos($$, $expr_value); + /*% ripper: sclass!($:expr_value, $:bodystmt) %*/ local_pop(p); - p->ctxt.in_def = $<ctxt>1.in_def; - p->ctxt.in_class = $<ctxt>1.in_class; - p->ctxt.shareable_constant_value = $<ctxt>1.shareable_constant_value; + p->ctxt.in_def = $k_class.in_def; + p->ctxt.in_class = $k_class.in_class; + p->ctxt.shareable_constant_value = $k_class.shareable_constant_value; } | k_module cpath { - if (p->ctxt.in_def) { - YYLTYPE loc = code_loc_gen(&@1, &@2); - yyerror1(&loc, "module definition in method body"); - } - p->ctxt.in_class = 1; - local_push(p, 0); + begin_definition("module", &@k_module, &@cpath); } bodystmt k_end { - /*%%%*/ - $$ = NEW_MODULE($2, $4, &@$); - nd_set_line($$->nd_body, @5.end_pos.lineno); - set_line_body($4, @2.end_pos.lineno); - nd_set_line($$, @2.end_pos.lineno); - /*% %*/ - /*% ripper: module!($2, $4) %*/ + $$ = NEW_MODULE($cpath, $bodystmt, &@$); + nd_set_line(RNODE_MODULE($$)->nd_body, @k_end.end_pos.lineno); + set_line_body($bodystmt, @cpath.end_pos.lineno); + nd_set_line($$, @cpath.end_pos.lineno); + /*% ripper: module!($:cpath, $:bodystmt) %*/ local_pop(p); - p->ctxt.in_class = $<ctxt>1.in_class; - p->ctxt.shareable_constant_value = $<ctxt>1.shareable_constant_value; + p->ctxt.in_class = $k_module.in_class; + p->ctxt.shareable_constant_value = $k_module.shareable_constant_value; } - | defn_head - f_arglist + | defn_head[head] + f_arglist[args] { - /*%%%*/ - push_end_expect_token_locations(p, &@1.beg_pos); - /*% %*/ + push_end_expect_token_locations(p, &@head.beg_pos); } bodystmt k_end { - restore_defun(p, $<node>1->nd_defn); - /*%%%*/ - $$ = set_defun_body(p, $1, $2, $4, &@$); - /*% %*/ - /*% ripper: def!(get_value($1), $2, $4) %*/ + restore_defun(p, $head); + $bodystmt = new_scope_body(p, $args, $bodystmt, &@$); + ($$ = $head->nd_def)->nd_loc = @$; + RNODE_DEFN($$)->nd_defn = $bodystmt; + /*% ripper: def!($:head, $:args, $:bodystmt) %*/ local_pop(p); } - | defs_head - f_arglist + | defs_head[head] + f_arglist[args] { - /*%%%*/ - push_end_expect_token_locations(p, &@1.beg_pos); - /*% %*/ + push_end_expect_token_locations(p, &@head.beg_pos); } bodystmt k_end { - restore_defun(p, $<node>1->nd_defn); - /*%%%*/ - $$ = set_defun_body(p, $1, $2, $4, &@$); - /*% - $1 = get_value($1); - %*/ - /*% ripper: defs!(AREF($1, 0), AREF($1, 1), AREF($1, 2), $2, $4) %*/ + restore_defun(p, $head); + $bodystmt = new_scope_body(p, $args, $bodystmt, &@$); + ($$ = $head->nd_def)->nd_loc = @$; + RNODE_DEFS($$)->nd_defn = $bodystmt; + /*% ripper: defs!(*$:head[0..2], $:args, $:bodystmt) %*/ local_pop(p); } | keyword_break { - /*%%%*/ - $$ = NEW_BREAK(0, &@$); - /*% %*/ + $$ = add_block_exit(p, NEW_BREAK(0, &@$)); /*% ripper: break!(args_new!) %*/ } | keyword_next { - /*%%%*/ - $$ = NEW_NEXT(0, &@$); - /*% %*/ + $$ = add_block_exit(p, NEW_NEXT(0, &@$)); /*% ripper: next!(args_new!) %*/ } | keyword_redo { - /*%%%*/ - $$ = NEW_REDO(&@$); - /*% %*/ + $$ = add_block_exit(p, NEW_REDO(&@$)); /*% ripper: redo! %*/ } | keyword_retry { - /*%%%*/ + if (!p->ctxt.in_defined) { + switch (p->ctxt.in_rescue) { + case before_rescue: yyerror1(&@1, "Invalid retry without rescue"); break; + case after_rescue: /* ok */ break; + case after_else: yyerror1(&@1, "Invalid retry after else"); break; + case after_ensure: yyerror1(&@1, "Invalid retry after ensure"); break; + } + } $$ = NEW_RETRY(&@$); - /*% %*/ /*% ripper: retry! %*/ } ; @@ -3610,9 +4752,7 @@ primary_value : primary k_begin : keyword_begin { token_info_push(p, "begin", &@$); - /*%%%*/ push_end_expect_token_locations(p, &@1.beg_pos); - /*% %*/ } ; @@ -3630,80 +4770,70 @@ k_if : keyword_if p->token_info->nonspc = 0; } } - /*%%%*/ push_end_expect_token_locations(p, &@1.beg_pos); - /*% %*/ } ; k_unless : keyword_unless { token_info_push(p, "unless", &@$); - /*%%%*/ push_end_expect_token_locations(p, &@1.beg_pos); - /*% %*/ } ; -k_while : keyword_while +k_while : keyword_while allow_exits { + $$ = $allow_exits; token_info_push(p, "while", &@$); - /*%%%*/ push_end_expect_token_locations(p, &@1.beg_pos); - /*% %*/ } ; -k_until : keyword_until +k_until : keyword_until allow_exits { + $$ = $allow_exits; token_info_push(p, "until", &@$); - /*%%%*/ push_end_expect_token_locations(p, &@1.beg_pos); - /*% %*/ } ; k_case : keyword_case { token_info_push(p, "case", &@$); - /*%%%*/ push_end_expect_token_locations(p, &@1.beg_pos); - /*% %*/ } ; -k_for : keyword_for +k_for : keyword_for allow_exits { + $$ = $allow_exits; token_info_push(p, "for", &@$); - /*%%%*/ push_end_expect_token_locations(p, &@1.beg_pos); - /*% %*/ } ; k_class : keyword_class { token_info_push(p, "class", &@$); - $<ctxt>$ = p->ctxt; - /*%%%*/ + $$ = p->ctxt; + p->ctxt.in_rescue = before_rescue; push_end_expect_token_locations(p, &@1.beg_pos); - /*% %*/ } ; k_module : keyword_module { token_info_push(p, "module", &@$); - $<ctxt>$ = p->ctxt; - /*%%%*/ + $$ = p->ctxt; + p->ctxt.in_rescue = before_rescue; push_end_expect_token_locations(p, &@1.beg_pos); - /*% %*/ } ; k_def : keyword_def { token_info_push(p, "def", &@$); + $$ = NEW_DEF_TEMP(&@$); p->ctxt.in_argdef = 1; } ; @@ -3711,31 +4841,29 @@ k_def : keyword_def k_do : keyword_do { token_info_push(p, "do", &@$); - /*%%%*/ push_end_expect_token_locations(p, &@1.beg_pos); - /*% %*/ - } ; k_do_block : keyword_do_block { token_info_push(p, "do", &@$); - /*%%%*/ push_end_expect_token_locations(p, &@1.beg_pos); - /*% %*/ } ; k_rescue : keyword_rescue { token_info_warn(p, "rescue", p->token_info, 1, &@$); + $$ = p->ctxt; + p->ctxt.in_rescue = after_rescue; } ; k_ensure : keyword_ensure { token_info_warn(p, "ensure", p->token_info, 1, &@$); + $$ = p->ctxt; } ; @@ -3770,9 +4898,7 @@ k_elsif : keyword_elsif k_end : keyword_end { token_info_pop(p, "end", &@$); - /*%%%*/ pop_end_expect_token_locations(p); - /*% %*/ } | tDUMNY_END { @@ -3787,6 +4913,13 @@ k_return : keyword_return } ; +k_yield : keyword_yield + { + if (!p->ctxt.in_defined && !p->ctxt.in_def && !compile_for_eval) + yyerror1(&@1, "Invalid yield"); + } + ; + then : term | keyword_then | term keyword_then @@ -3801,21 +4934,17 @@ if_tail : opt_else compstmt if_tail { - /*%%%*/ $$ = new_if(p, $2, $4, $5, &@$); fixpos($$, $2); - /*% %*/ - /*% ripper: elsif!($2, $4, escape_Qundef($5)) %*/ + /*% ripper: elsif!($:2, $:4, $:5) %*/ } ; opt_else : none | k_else compstmt { - /*%%%*/ $$ = $2; - /*% %*/ - /*% ripper: else!($2) %*/ + /*% ripper: else!($:2) %*/ } ; @@ -3825,195 +4954,183 @@ for_var : lhs f_marg : f_norm_arg { - /*%%%*/ $$ = assignable(p, $1, 0, &@$); mark_lvar_used(p, $$); - /*% %*/ - /*% ripper: assignable(p, $1) %*/ } | tLPAREN f_margs rparen { - /*%%%*/ - $$ = $2; - /*% %*/ - /*% ripper: mlhs_paren!($2) %*/ + $$ = (NODE *)$2; + /*% ripper: mlhs_paren!($:2) %*/ } ; f_marg_list : f_marg { - /*%%%*/ $$ = NEW_LIST($1, &@$); - /*% %*/ - /*% ripper: mlhs_add!(mlhs_new!, $1) %*/ + /*% ripper: mlhs_add!(mlhs_new!, $:1) %*/ } | f_marg_list ',' f_marg { - /*%%%*/ $$ = list_append(p, $1, $3); - /*% %*/ - /*% ripper: mlhs_add!($1, $3) %*/ + /*% ripper: mlhs_add!($:1, $:3) %*/ } ; f_margs : f_marg_list { - /*%%%*/ $$ = NEW_MASGN($1, 0, &@$); - /*% %*/ - /*% ripper: $1 %*/ + /*% ripper: $:1 %*/ } | f_marg_list ',' f_rest_marg { - /*%%%*/ $$ = NEW_MASGN($1, $3, &@$); - /*% %*/ - /*% ripper: mlhs_add_star!($1, $3) %*/ + /*% ripper: mlhs_add_star!($:1, $:3) %*/ } | f_marg_list ',' f_rest_marg ',' f_marg_list { - /*%%%*/ $$ = NEW_MASGN($1, NEW_POSTARG($3, $5, &@$), &@$); - /*% %*/ - /*% ripper: mlhs_add_post!(mlhs_add_star!($1, $3), $5) %*/ + /*% ripper: mlhs_add_post!(mlhs_add_star!($:1, $:3), $:5) %*/ } | f_rest_marg { - /*%%%*/ $$ = NEW_MASGN(0, $1, &@$); - /*% %*/ - /*% ripper: mlhs_add_star!(mlhs_new!, $1) %*/ + /*% ripper: mlhs_add_star!(mlhs_new!, $:1) %*/ } | f_rest_marg ',' f_marg_list { - /*%%%*/ $$ = NEW_MASGN(0, NEW_POSTARG($1, $3, &@$), &@$); - /*% %*/ - /*% ripper: mlhs_add_post!(mlhs_add_star!(mlhs_new!, $1), $3) %*/ + /*% ripper: mlhs_add_post!(mlhs_add_star!(mlhs_new!, $:1), $:3) %*/ } ; f_rest_marg : tSTAR f_norm_arg { - /*%%%*/ + /*% ripper: $:2 %*/ $$ = assignable(p, $2, 0, &@$); mark_lvar_used(p, $$); - /*% %*/ - /*% ripper: assignable(p, $2) %*/ } | tSTAR { - /*%%%*/ $$ = NODE_SPECIAL_NO_NAME_REST; - /*% %*/ /*% ripper: Qnil %*/ } ; f_any_kwrest : f_kwrest - | f_no_kwarg {$$ = ID2VAL(idNil);} + | f_no_kwarg + { + $$ = idNil; + /*% ripper: ID2VAL(idNil) %*/ + } ; f_eq : {p->ctxt.in_argdef = 0;} '='; -block_args_tail : f_block_kwarg ',' f_kwrest opt_f_block_arg +block_args_tail : f_kwarg(f_block_kw) ',' f_kwrest opt_f_block_arg { $$ = new_args_tail(p, $1, $3, $4, &@3); + /*% ripper: [$:1, $:3, $:4] %*/ } - | f_block_kwarg opt_f_block_arg + | f_kwarg(f_block_kw) opt_f_block_arg { - $$ = new_args_tail(p, $1, Qnone, $2, &@1); + $$ = new_args_tail(p, $1, 0, $2, &@1); + /*% ripper: [$:1, Qnil, $:2] %*/ } | f_any_kwrest opt_f_block_arg { - $$ = new_args_tail(p, Qnone, $1, $2, &@1); + $$ = new_args_tail(p, 0, $1, $2, &@1); + /*% ripper: [Qnil, $:1, $:2] %*/ } | f_block_arg { - $$ = new_args_tail(p, Qnone, Qnone, $1, &@1); - } - ; - -opt_block_args_tail : ',' block_args_tail - { - $$ = $2; - } - | /* none */ - { - $$ = new_args_tail(p, Qnone, Qnone, Qnone, &@0); + $$ = new_args_tail(p, 0, 0, $1, &@1); + /*% ripper: [Qnil, Qnil, $:1] %*/ } ; excessed_comma : ',' { /* magic number for rest_id in iseq_set_arguments() */ - /*%%%*/ $$ = NODE_SPECIAL_EXCESSIVE_COMMA; - /*% %*/ /*% ripper: excessed_comma! %*/ } ; -block_param : f_arg ',' f_block_optarg ',' f_rest_arg opt_block_args_tail +block_param : f_arg ',' f_optarg(primary_value) ',' f_rest_arg opt_args_tail(block_args_tail) { - $$ = new_args(p, $1, $3, $5, Qnone, $6, &@$); + $$ = new_args(p, $1, $3, $5, 0, $6, &@$); + /*% ripper: params!($:1, $:3, $:5, Qnil, *$:6[0..2]) %*/ } - | f_arg ',' f_block_optarg ',' f_rest_arg ',' f_arg opt_block_args_tail + | f_arg ',' f_optarg(primary_value) ',' f_rest_arg ',' f_arg opt_args_tail(block_args_tail) { $$ = new_args(p, $1, $3, $5, $7, $8, &@$); + /*% ripper: params!($:1, $:3, $:5, $:7, *$:8[0..2]) %*/ } - | f_arg ',' f_block_optarg opt_block_args_tail + | f_arg ',' f_optarg(primary_value) opt_args_tail(block_args_tail) { - $$ = new_args(p, $1, $3, Qnone, Qnone, $4, &@$); + $$ = new_args(p, $1, $3, 0, 0, $4, &@$); + /*% ripper: params!($:1, $:3, Qnil, Qnil, *$:4[0..2]) %*/ } - | f_arg ',' f_block_optarg ',' f_arg opt_block_args_tail + | f_arg ',' f_optarg(primary_value) ',' f_arg opt_args_tail(block_args_tail) { - $$ = new_args(p, $1, $3, Qnone, $5, $6, &@$); + $$ = new_args(p, $1, $3, 0, $5, $6, &@$); + /*% ripper: params!($:1, $:3, Qnil, $:5, *$:6[0..2]) %*/ } - | f_arg ',' f_rest_arg opt_block_args_tail + | f_arg ',' f_rest_arg opt_args_tail(block_args_tail) { - $$ = new_args(p, $1, Qnone, $3, Qnone, $4, &@$); + $$ = new_args(p, $1, 0, $3, 0, $4, &@$); + /*% ripper: params!($:1, Qnil, $:3, Qnil, *$:4[0..2]) %*/ } | f_arg excessed_comma { - $$ = new_args_tail(p, Qnone, Qnone, Qnone, &@2); - $$ = new_args(p, $1, Qnone, $2, Qnone, $$, &@$); + $$ = new_args_tail(p, 0, 0, 0, &@2); + $$ = new_args(p, $1, 0, $2, 0, $$, &@$); + /*% ripper: params!($:1, Qnil, $:2, Qnil, Qnil, Qnil, Qnil) %*/ } - | f_arg ',' f_rest_arg ',' f_arg opt_block_args_tail + | f_arg ',' f_rest_arg ',' f_arg opt_args_tail(block_args_tail) { - $$ = new_args(p, $1, Qnone, $3, $5, $6, &@$); + $$ = new_args(p, $1, 0, $3, $5, $6, &@$); + /*% ripper: params!($:1, Qnil, $:3, $:5, *$:6[0..2]) %*/ } - | f_arg opt_block_args_tail + | f_arg opt_args_tail(block_args_tail) { - $$ = new_args(p, $1, Qnone, Qnone, Qnone, $2, &@$); + $$ = new_args(p, $1, 0, 0, 0, $2, &@$); + /*% ripper: params!($:1, Qnil, Qnil, Qnil, *$:2[0..2]) %*/ } - | f_block_optarg ',' f_rest_arg opt_block_args_tail + | f_optarg(primary_value) ',' f_rest_arg opt_args_tail(block_args_tail) { - $$ = new_args(p, Qnone, $1, $3, Qnone, $4, &@$); + $$ = new_args(p, 0, $1, $3, 0, $4, &@$); + /*% ripper: params!(Qnil, $:1, $:3, Qnil, *$:4[0..2]) %*/ } - | f_block_optarg ',' f_rest_arg ',' f_arg opt_block_args_tail + | f_optarg(primary_value) ',' f_rest_arg ',' f_arg opt_args_tail(block_args_tail) { - $$ = new_args(p, Qnone, $1, $3, $5, $6, &@$); + $$ = new_args(p, 0, $1, $3, $5, $6, &@$); + /*% ripper: params!(Qnil, $:1, $:3, $:5, *$:6[0..2]) %*/ } - | f_block_optarg opt_block_args_tail + | f_optarg(primary_value) opt_args_tail(block_args_tail) { - $$ = new_args(p, Qnone, $1, Qnone, Qnone, $2, &@$); + $$ = new_args(p, 0, $1, 0, 0, $2, &@$); + /*% ripper: params!(Qnil, $:1, Qnil, Qnil, *$:2[0..2]) %*/ } - | f_block_optarg ',' f_arg opt_block_args_tail + | f_optarg(primary_value) ',' f_arg opt_args_tail(block_args_tail) { - $$ = new_args(p, Qnone, $1, Qnone, $3, $4, &@$); + $$ = new_args(p, 0, $1, 0, $3, $4, &@$); + /*% ripper: params!(Qnil, $:1, Qnil, $:3, *$:4[0..2]) %*/ } - | f_rest_arg opt_block_args_tail + | f_rest_arg opt_args_tail(block_args_tail) { - $$ = new_args(p, Qnone, Qnone, $1, Qnone, $2, &@$); + $$ = new_args(p, 0, 0, $1, 0, $2, &@$); + /*% ripper: params!(Qnil, Qnil, $:1, Qnil, *$:2[0..2]) %*/ } - | f_rest_arg ',' f_arg opt_block_args_tail + | f_rest_arg ',' f_arg opt_args_tail(block_args_tail) { - $$ = new_args(p, Qnone, Qnone, $1, $3, $4, &@$); + $$ = new_args(p, 0, 0, $1, $3, $4, &@$); + /*% ripper: params!(Qnil, Qnil, $:1, $:3, *$:4[0..2]) %*/ } | block_args_tail { - $$ = new_args(p, Qnone, Qnone, Qnone, Qnone, $1, &@$); + $$ = new_args(p, 0, 0, 0, 0, $1, &@$); + /*% ripper: params!(Qnil, Qnil, Qnil, Qnil, *$:1[0..2]) %*/ } ; @@ -4029,47 +5146,42 @@ block_param_def : '|' opt_bv_decl '|' p->cur_arg = 0; p->max_numparam = ORDINAL_PARAM; p->ctxt.in_argdef = 0; - /*%%%*/ $$ = 0; - /*% %*/ - /*% ripper: block_var!(params!(Qnil,Qnil,Qnil,Qnil,Qnil,Qnil,Qnil), escape_Qundef($2)) %*/ + /*% ripper: block_var!(params!(Qnil,Qnil,Qnil,Qnil,Qnil,Qnil,Qnil), $:2) %*/ } | '|' block_param opt_bv_decl '|' { p->cur_arg = 0; p->max_numparam = ORDINAL_PARAM; p->ctxt.in_argdef = 0; - /*%%%*/ $$ = $2; - /*% %*/ - /*% ripper: block_var!(escape_Qundef($2), escape_Qundef($3)) %*/ + /*% ripper: block_var!($:2, $:3) %*/ } ; -opt_bv_decl : opt_nl +opt_bv_decl : '\n'? { $$ = 0; + /*% ripper: Qfalse %*/ } - | opt_nl ';' bv_decls opt_nl + | '\n'? ';' bv_decls '\n'? { - /*%%%*/ $$ = 0; - /*% %*/ - /*% ripper: $3 %*/ + /*% ripper: $:3 %*/ } ; bv_decls : bvar - /*% ripper[brace]: rb_ary_new3(1, get_value($1)) %*/ + /*% ripper[brace]: rb_ary_new3(1, $:1) %*/ | bv_decls ',' bvar - /*% ripper[brace]: rb_ary_push($1, get_value($3)) %*/ + /*% ripper[brace]: rb_ary_push($:1, $:3) %*/ ; bvar : tIDENTIFIER { - new_bv(p, get_id($1)); - /*% ripper: get_value($1) %*/ + new_bv(p, $1); + /*% ripper: $:1 %*/ } | f_bad_arg { @@ -4077,62 +5189,69 @@ bvar : tIDENTIFIER } ; -lambda : tLAMBDA - { - token_info_push(p, "->", &@1); - $<vars>1 = dyna_push(p); - $<num>$ = p->lex.lpar_beg; - p->lex.lpar_beg = p->lex.paren_nest; - } - { - $<num>$ = p->max_numparam; +max_numparam : { + $$ = p->max_numparam; p->max_numparam = 0; } - { - $<node>$ = numparam_push(p); + ; + +numparam : { + $$ = numparam_push(p); } - f_larglist + ; + +it_id : { + $$ = p->it_id; + p->it_id = 0; + } + ; + +lambda : tLAMBDA[lpar] + { + token_info_push(p, "->", &@1); + $$ = dyna_push(p); + p->lex.lpar_beg = p->lex.paren_nest; + }[dyna]<vars> + max_numparam numparam it_id allow_exits + f_larglist[args] { CMDARG_PUSH(0); } - lambda_body + lambda_body[body] { int max_numparam = p->max_numparam; - p->lex.lpar_beg = $<num>2; - p->max_numparam = $<num>3; + ID it_id = p->it_id; + p->lex.lpar_beg = $lpar; + p->max_numparam = $max_numparam; + p->it_id = $it_id; + restore_block_exit(p, $allow_exits); CMDARG_POP(); - $5 = args_with_numbered(p, $5, max_numparam); - /*%%%*/ + $args = args_with_numbered(p, $args, max_numparam, it_id); { - YYLTYPE loc = code_loc_gen(&@5, &@7); - $$ = NEW_LAMBDA($5, $7, &loc); - nd_set_line($$->nd_body, @7.end_pos.lineno); - nd_set_line($$, @5.end_pos.lineno); + YYLTYPE loc = code_loc_gen(&@args, &@body); + $$ = NEW_LAMBDA($args, $body, &loc); + nd_set_line(RNODE_LAMBDA($$)->nd_body, @body.end_pos.lineno); + nd_set_line($$, @args.end_pos.lineno); nd_set_first_loc($$, @1.beg_pos); } - /*% %*/ - /*% ripper: lambda!($5, $7) %*/ - numparam_pop(p, $<node>4); - dyna_pop(p, $<vars>1); + /*% ripper: lambda!($:args, $:body) %*/ + numparam_pop(p, $numparam); + dyna_pop(p, $dyna); } ; f_larglist : '(' f_args opt_bv_decl ')' { p->ctxt.in_argdef = 0; - /*%%%*/ $$ = $2; p->max_numparam = ORDINAL_PARAM; - /*% %*/ - /*% ripper: paren!($2) %*/ + /*% ripper: paren!($:2) %*/ } | f_args { p->ctxt.in_argdef = 0; - /*%%%*/ - if (!args_info_empty_p($1->nd_ainfo)) + if (!args_info_empty_p(&$1->nd_ainfo)) p->max_numparam = ORDINAL_PARAM; - /*% %*/ $$ = $1; } ; @@ -4141,231 +5260,198 @@ lambda_body : tLAMBEG compstmt '}' { token_info_pop(p, "}", &@3); $$ = $2; + /*% ripper: $:2 %*/ } | keyword_do_LAMBDA { - /*%%%*/ push_end_expect_token_locations(p, &@1.beg_pos); - /*% %*/ } bodystmt k_end { $$ = $3; + /*% ripper: $:3 %*/ } ; do_block : k_do_block do_body k_end { $$ = $2; - /*%%%*/ - $$->nd_body->nd_loc = code_loc_gen(&@1, &@3); - nd_set_line($$, @1.end_pos.lineno); - /*% %*/ + set_embraced_location($$, &@1, &@3); + /*% ripper: $:2 %*/ } ; block_call : command do_block { - /*%%%*/ if (nd_type_p($1, NODE_YIELD)) { compile_error(p, "block given to yield"); } else { - block_dup_check(p, $1->nd_args, $2); + block_dup_check(p, get_nd_args(p, $1), $2); } $$ = method_add_block(p, $1, $2, &@$); fixpos($$, $1); - /*% %*/ - /*% ripper: method_add_block!($1, $2) %*/ + /*% ripper: method_add_block!($:1, $:2) %*/ } | block_call call_op2 operation2 opt_paren_args { - /*%%%*/ + bool has_args = $4 != 0; + if (NODE_EMPTY_ARGS_P($4)) $4 = 0; $$ = new_qcall(p, $2, $1, $3, $4, &@3, &@$); - /*% %*/ - /*% ripper: opt_event(:method_add_arg!, call!($1, $2, $3), $4) %*/ + /*% ripper: call!($:1, $:2, $:3) %*/ + if (has_args) { + /*% ripper: method_add_arg!($:$, $:4) %*/ + } } | block_call call_op2 operation2 opt_paren_args brace_block { - /*%%%*/ + bool has_args = $5 != 0; + if (NODE_EMPTY_ARGS_P($5)) $5 = 0; $$ = new_command_qcall(p, $2, $1, $3, $4, $5, &@3, &@$); - /*% %*/ - /*% ripper: opt_event(:method_add_block!, command_call!($1, $2, $3, $4), $5) %*/ + /*% ripper: command_call!($:1, $:2, $:3, $:4) %*/ + if (has_args) { + /*% ripper: method_add_block!($:$, $:5) %*/ + } } | block_call call_op2 operation2 command_args do_block { - /*%%%*/ $$ = new_command_qcall(p, $2, $1, $3, $4, $5, &@3, &@$); - /*% %*/ - /*% ripper: method_add_block!(command_call!($1, $2, $3, $4), $5) %*/ + /*% ripper: method_add_block!(command_call!($:1, $:2, $:3, $:4), $:5) %*/ } ; method_call : fcall paren_args { - /*%%%*/ - $$ = $1; - $$->nd_args = $2; + $1->nd_args = $2; + $$ = (NODE *)$1; nd_set_last_loc($1, @2.end_pos); - /*% %*/ - /*% ripper: method_add_arg!(fcall!($1), $2) %*/ + /*% ripper: method_add_arg!(fcall!($:1), $:2) %*/ } | primary_value call_op operation2 opt_paren_args { - /*%%%*/ + bool has_args = $4 != 0; + if (NODE_EMPTY_ARGS_P($4)) $4 = 0; $$ = new_qcall(p, $2, $1, $3, $4, &@3, &@$); nd_set_line($$, @3.end_pos.lineno); - /*% %*/ - /*% ripper: opt_event(:method_add_arg!, call!($1, $2, $3), $4) %*/ + /*% ripper: call!($:1, $:2, $:3) %*/ + if (has_args) { + /*% ripper: method_add_arg!($:$, $:4) %*/ + } } | primary_value tCOLON2 operation2 paren_args { - /*%%%*/ - $$ = new_qcall(p, ID2VAL(idCOLON2), $1, $3, $4, &@3, &@$); + $$ = new_qcall(p, idCOLON2, $1, $3, $4, &@3, &@$); nd_set_line($$, @3.end_pos.lineno); - /*% %*/ - /*% ripper: method_add_arg!(call!($1, ID2VAL(idCOLON2), $3), $4) %*/ + /*% ripper: method_add_arg!(call!($:1, $:2, $:3), $:4) %*/ } | primary_value tCOLON2 operation3 { - /*%%%*/ - $$ = new_qcall(p, ID2VAL(idCOLON2), $1, $3, Qnull, &@3, &@$); - /*% %*/ - /*% ripper: call!($1, ID2VAL(idCOLON2), $3) %*/ + $$ = new_qcall(p, idCOLON2, $1, $3, 0, &@3, &@$); + /*% ripper: call!($:1, $:2, $:3) %*/ } | primary_value call_op paren_args { - /*%%%*/ - $$ = new_qcall(p, $2, $1, ID2VAL(idCall), $3, &@2, &@$); + $$ = new_qcall(p, $2, $1, idCall, $3, &@2, &@$); nd_set_line($$, @2.end_pos.lineno); - /*% %*/ - /*% ripper: method_add_arg!(call!($1, $2, ID2VAL(idCall)), $3) %*/ + /*% ripper: method_add_arg!(call!($:1, $:2, ID2VAL(idCall)), $:3) %*/ } | primary_value tCOLON2 paren_args { - /*%%%*/ - $$ = new_qcall(p, ID2VAL(idCOLON2), $1, ID2VAL(idCall), $3, &@2, &@$); + $$ = new_qcall(p, idCOLON2, $1, idCall, $3, &@2, &@$); nd_set_line($$, @2.end_pos.lineno); - /*% %*/ - /*% ripper: method_add_arg!(call!($1, ID2VAL(idCOLON2), ID2VAL(idCall)), $3) %*/ + /*% ripper: method_add_arg!(call!($:1, $:2, ID2VAL(idCall)), $:3) %*/ } | keyword_super paren_args { - /*%%%*/ $$ = NEW_SUPER($2, &@$); - /*% %*/ - /*% ripper: super!($2) %*/ + /*% ripper: super!($:2) %*/ } | keyword_super { - /*%%%*/ $$ = NEW_ZSUPER(&@$); - /*% %*/ /*% ripper: zsuper! %*/ } | primary_value '[' opt_call_args rbracket { - /*%%%*/ $$ = NEW_CALL($1, tAREF, $3, &@$); fixpos($$, $1); - /*% %*/ - /*% ripper: aref!($1, escape_Qundef($3)) %*/ + /*% ripper: aref!($:1, $:3) %*/ } ; brace_block : '{' brace_body '}' { $$ = $2; - /*%%%*/ - $$->nd_body->nd_loc = code_loc_gen(&@1, &@3); - nd_set_line($$, @1.end_pos.lineno); - /*% %*/ + set_embraced_location($$, &@1, &@3); + /*% ripper: $:2 %*/ } | k_do do_body k_end { $$ = $2; - /*%%%*/ - $$->nd_body->nd_loc = code_loc_gen(&@1, &@3); - nd_set_line($$, @1.end_pos.lineno); - /*% %*/ + set_embraced_location($$, &@1, &@3); + /*% ripper: $:2 %*/ } ; -brace_body : {$<vars>$ = dyna_push(p);} - { - $<num>$ = p->max_numparam; - p->max_numparam = 0; - } - { - $<node>$ = numparam_push(p); - } - opt_block_param compstmt +brace_body : {$$ = dyna_push(p);}[dyna]<vars> + max_numparam numparam it_id allow_exits + opt_block_param[args] compstmt { int max_numparam = p->max_numparam; - p->max_numparam = $<num>2; - $4 = args_with_numbered(p, $4, max_numparam); - /*%%%*/ - $$ = NEW_ITER($4, $5, &@$); - /*% %*/ - /*% ripper: brace_block!(escape_Qundef($4), $5) %*/ - numparam_pop(p, $<node>3); - dyna_pop(p, $<vars>1); + ID it_id = p->it_id; + p->max_numparam = $max_numparam; + p->it_id = $it_id; + $args = args_with_numbered(p, $args, max_numparam, it_id); + $$ = NEW_ITER($args, $compstmt, &@$); + /*% ripper: brace_block!($:args, $:compstmt) %*/ + restore_block_exit(p, $allow_exits); + numparam_pop(p, $numparam); + dyna_pop(p, $dyna); } ; -do_body : {$<vars>$ = dyna_push(p);} - { - $<num>$ = p->max_numparam; - p->max_numparam = 0; - } - { - $<node>$ = numparam_push(p); +do_body : { + $$ = dyna_push(p); CMDARG_PUSH(0); - } - opt_block_param bodystmt + }[dyna]<vars> + max_numparam numparam it_id allow_exits + opt_block_param[args] bodystmt { int max_numparam = p->max_numparam; - p->max_numparam = $<num>2; - $4 = args_with_numbered(p, $4, max_numparam); - /*%%%*/ - $$ = NEW_ITER($4, $5, &@$); - /*% %*/ - /*% ripper: do_block!(escape_Qundef($4), $5) %*/ + ID it_id = p->it_id; + p->max_numparam = $max_numparam; + p->it_id = $it_id; + $args = args_with_numbered(p, $args, max_numparam, it_id); + $$ = NEW_ITER($args, $bodystmt, &@$); + /*% ripper: do_block!($:args, $:bodystmt) %*/ CMDARG_POP(); - numparam_pop(p, $<node>3); - dyna_pop(p, $<vars>1); + restore_block_exit(p, $allow_exits); + numparam_pop(p, $numparam); + dyna_pop(p, $dyna); } ; case_args : arg_value { - /*%%%*/ check_literal_when(p, $1, &@1); $$ = NEW_LIST($1, &@$); - /*% %*/ - /*% ripper: args_add!(args_new!, $1) %*/ + /*% ripper: args_add!(args_new!, $:1) %*/ } | tSTAR arg_value { - /*%%%*/ $$ = NEW_SPLAT($2, &@$); - /*% %*/ - /*% ripper: args_add_star!(args_new!, $2) %*/ + /*% ripper: args_add_star!(args_new!, $:2) %*/ } | case_args ',' arg_value { - /*%%%*/ check_literal_when(p, $3, &@3); $$ = last_arg_append(p, $1, $3, &@$); - /*% %*/ - /*% ripper: args_add!($1, $3) %*/ + /*% ripper: args_add!($:1, $:3) %*/ } | case_args ',' tSTAR arg_value { - /*%%%*/ $$ = rest_arg_append(p, $1, $4, &@$); - /*% %*/ - /*% ripper: args_add_star!($1, $4) %*/ + /*% ripper: args_add_star!($:1, $:4) %*/ } ; @@ -4373,11 +5459,9 @@ case_body : k_when case_args then compstmt cases { - /*%%%*/ $$ = NEW_WHEN($2, $4, $5, &@$); fixpos($$, $2); - /*% %*/ - /*% ripper: when!($2, $4, escape_Qundef($5)) %*/ + /*% ripper: when!($:2, $:4, $:5) %*/ } ; @@ -4385,30 +5469,30 @@ cases : opt_else | case_body ; -p_case_body : keyword_in - { +p_pvtbl : {$$ = p->pvtbl; p->pvtbl = st_init_numtable();}; +p_pktbl : {$$ = p->pktbl; p->pktbl = 0;}; + +p_in_kwarg : { + $$ = p->ctxt; SET_LEX_STATE(EXPR_BEG|EXPR_LABEL); p->command_start = FALSE; - $<ctxt>1 = p->ctxt; p->ctxt.in_kwarg = 1; - $<tbl>$ = push_pvtbl(p); } + ; + +p_case_body : keyword_in + p_in_kwarg[ctxt] p_pvtbl p_pktbl + p_top_expr[expr] then { - $<tbl>$ = push_pktbl(p); - } - p_top_expr then - { - pop_pktbl(p, $<tbl>3); - pop_pvtbl(p, $<tbl>2); - p->ctxt.in_kwarg = $<ctxt>1.in_kwarg; + pop_pktbl(p, $p_pktbl); + pop_pvtbl(p, $p_pvtbl); + p->ctxt.in_kwarg = $ctxt.in_kwarg; } compstmt - p_cases + p_cases[cases] { - /*%%%*/ - $$ = NEW_IN($4, $7, $8, &@$); - /*% %*/ - /*% ripper: in!($4, $7, escape_Qundef($8)) %*/ + $$ = NEW_IN($expr, $compstmt, $cases, &@$); + /*% ripper: in!($:expr, $:compstmt, $:cases) %*/ } ; @@ -4419,47 +5503,45 @@ p_cases : opt_else p_top_expr : p_top_expr_body | p_top_expr_body modifier_if expr_value { - /*%%%*/ $$ = new_if(p, $3, $1, 0, &@$); fixpos($$, $3); - /*% %*/ - /*% ripper: if_mod!($3, $1) %*/ + /*% ripper: if_mod!($:3, $:1) %*/ } | p_top_expr_body modifier_unless expr_value { - /*%%%*/ $$ = new_unless(p, $3, $1, 0, &@$); fixpos($$, $3); - /*% %*/ - /*% ripper: unless_mod!($3, $1) %*/ + /*% ripper: unless_mod!($:3, $:1) %*/ } ; p_top_expr_body : p_expr | p_expr ',' { - $$ = new_array_pattern_tail(p, Qnone, 1, Qnone, Qnone, &@$); - $$ = new_array_pattern(p, Qnone, get_value($1), $$, &@$); + $$ = new_array_pattern_tail(p, 0, 1, 0, 0, &@$); + $$ = new_array_pattern(p, 0, $1, $$, &@$); + /*% ripper: aryptn!(Qnil, [$:1], Qnil, Qnil) %*/ } | p_expr ',' p_args { - $$ = new_array_pattern(p, Qnone, get_value($1), $3, &@$); - /*%%%*/ + $$ = new_array_pattern(p, 0, $1, $3, &@$); nd_set_first_loc($$, @1.beg_pos); - /*% - %*/ + /*% ripper: aryptn!(Qnil, aryptn_pre_args(p, $:1, $:3[0]), *$:3[1..2]) %*/ } | p_find { - $$ = new_find_pattern(p, Qnone, $1, &@$); + $$ = new_find_pattern(p, 0, $1, &@$); + /*% ripper: fndptn!(Qnil, *$:1[0..2]) %*/ } | p_args_tail { - $$ = new_array_pattern(p, Qnone, Qnone, $1, &@$); + $$ = new_array_pattern(p, 0, 0, $1, &@$); + /*% ripper: aryptn!(Qnil, *$:1[0..2]) %*/ } | p_kwargs { - $$ = new_hash_pattern(p, Qnone, $1, &@$); + $$ = new_hash_pattern(p, 0, $1, &@$); + /*% ripper: hshptn!(Qnil, *$:1[0..1]) %*/ } ; @@ -4468,161 +5550,158 @@ p_expr : p_as p_as : p_expr tASSOC p_variable { - /*%%%*/ NODE *n = NEW_LIST($1, &@$); n = list_append(p, n, $3); $$ = new_hash(p, n, &@$); - /*% %*/ - /*% ripper: binary!($1, STATIC_ID2SYM((id_assoc)), $3) %*/ + /*% ripper: binary!($:1, ID2VAL((id_assoc)), $:3) %*/ } | p_alt ; p_alt : p_alt '|' p_expr_basic { - /*%%%*/ - $$ = NEW_NODE(NODE_OR, $1, $3, 0, &@$); - /*% %*/ - /*% ripper: binary!($1, STATIC_ID2SYM(idOr), $3) %*/ + $$ = NEW_OR($1, $3, &@$); + /*% ripper: binary!($:1, ID2VAL(idOr), $:3) %*/ } | p_expr_basic ; -p_lparen : '(' {$<tbl>$ = push_pktbl(p);}; -p_lbracket : '[' {$<tbl>$ = push_pktbl(p);}; +p_lparen : '(' p_pktbl + { + $$ = $2; + /*% ripper: $:2 %*/ + } + ; + +p_lbracket : '[' p_pktbl + { + $$ = $2; + /*% ripper: $:2 %*/ + } + ; p_expr_basic : p_value | p_variable - | p_const p_lparen p_args rparen + | p_const p_lparen[p_pktbl] p_args rparen { - pop_pktbl(p, $<tbl>2); - $$ = new_array_pattern(p, $1, Qnone, $3, &@$); - /*%%%*/ - nd_set_first_loc($$, @1.beg_pos); - /*% - %*/ + pop_pktbl(p, $p_pktbl); + $$ = new_array_pattern(p, $p_const, 0, $p_args, &@$); + nd_set_first_loc($$, @p_const.beg_pos); + /*% ripper: aryptn!($:p_const, *$:p_args[0..2]) %*/ } - | p_const p_lparen p_find rparen + | p_const p_lparen[p_pktbl] p_find rparen { - pop_pktbl(p, $<tbl>2); - $$ = new_find_pattern(p, $1, $3, &@$); - /*%%%*/ - nd_set_first_loc($$, @1.beg_pos); - /*% - %*/ + pop_pktbl(p, $p_pktbl); + $$ = new_find_pattern(p, $p_const, $p_find, &@$); + nd_set_first_loc($$, @p_const.beg_pos); + /*% ripper: fndptn!($:p_const, *$:p_find[0..2]) %*/ } - | p_const p_lparen p_kwargs rparen + | p_const p_lparen[p_pktbl] p_kwargs rparen { - pop_pktbl(p, $<tbl>2); - $$ = new_hash_pattern(p, $1, $3, &@$); - /*%%%*/ - nd_set_first_loc($$, @1.beg_pos); - /*% - %*/ + pop_pktbl(p, $p_pktbl); + $$ = new_hash_pattern(p, $p_const, $p_kwargs, &@$); + nd_set_first_loc($$, @p_const.beg_pos); + /*% ripper: hshptn!($:p_const, *$:p_kwargs[0..1]) %*/ } | p_const '(' rparen { - $$ = new_array_pattern_tail(p, Qnone, 0, Qnone, Qnone, &@$); - $$ = new_array_pattern(p, $1, Qnone, $$, &@$); + $$ = new_array_pattern_tail(p, 0, 0, 0, 0, &@$); + $$ = new_array_pattern(p, $p_const, 0, $$, &@$); + /*% ripper: aryptn!($:p_const, Qnil, Qnil, Qnil) %*/ } - | p_const p_lbracket p_args rbracket + | p_const p_lbracket[p_pktbl] p_args rbracket { - pop_pktbl(p, $<tbl>2); - $$ = new_array_pattern(p, $1, Qnone, $3, &@$); - /*%%%*/ - nd_set_first_loc($$, @1.beg_pos); - /*% - %*/ + pop_pktbl(p, $p_pktbl); + $$ = new_array_pattern(p, $p_const, 0, $p_args, &@$); + nd_set_first_loc($$, @p_const.beg_pos); + /*% ripper: aryptn!($:p_const, *$:p_args[0..2]) %*/ } - | p_const p_lbracket p_find rbracket + | p_const p_lbracket[p_pktbl] p_find rbracket { - pop_pktbl(p, $<tbl>2); - $$ = new_find_pattern(p, $1, $3, &@$); - /*%%%*/ - nd_set_first_loc($$, @1.beg_pos); - /*% - %*/ + pop_pktbl(p, $p_pktbl); + $$ = new_find_pattern(p, $p_const, $p_find, &@$); + nd_set_first_loc($$, @p_const.beg_pos); + /*% ripper: fndptn!($:p_const, *$:p_find[0..2]) %*/ } - | p_const p_lbracket p_kwargs rbracket + | p_const p_lbracket[p_pktbl] p_kwargs rbracket { - pop_pktbl(p, $<tbl>2); - $$ = new_hash_pattern(p, $1, $3, &@$); - /*%%%*/ - nd_set_first_loc($$, @1.beg_pos); - /*% - %*/ + pop_pktbl(p, $p_pktbl); + $$ = new_hash_pattern(p, $p_const, $p_kwargs, &@$); + nd_set_first_loc($$, @p_const.beg_pos); + /*% ripper: hshptn!($:p_const, *$:p_kwargs[0..1]) %*/ } | p_const '[' rbracket { - $$ = new_array_pattern_tail(p, Qnone, 0, Qnone, Qnone, &@$); - $$ = new_array_pattern(p, $1, Qnone, $$, &@$); + $$ = new_array_pattern_tail(p, 0, 0, 0, 0, &@$); + $$ = new_array_pattern(p, $1, 0, $$, &@$); + /*% ripper: aryptn!($:1, Qnil, Qnil, Qnil) %*/ } | tLBRACK p_args rbracket { - $$ = new_array_pattern(p, Qnone, Qnone, $2, &@$); + $$ = new_array_pattern(p, 0, 0, $p_args, &@$); + /*% ripper: aryptn!(Qnil, *$:p_args[0..2]) %*/ } | tLBRACK p_find rbracket { - $$ = new_find_pattern(p, Qnone, $2, &@$); + $$ = new_find_pattern(p, 0, $p_find, &@$); + /*% ripper: fndptn!(Qnil, *$:p_find[0..2]) %*/ } | tLBRACK rbracket { - $$ = new_array_pattern_tail(p, Qnone, 0, Qnone, Qnone, &@$); - $$ = new_array_pattern(p, Qnone, Qnone, $$, &@$); + $$ = new_array_pattern_tail(p, 0, 0, 0, 0, &@$); + $$ = new_array_pattern(p, 0, 0, $$, &@$); + /*% ripper: aryptn!(Qnil, Qnil, Qnil, Qnil) %*/ } - | tLBRACE + | tLBRACE p_pktbl lex_ctxt[ctxt] { - $<tbl>$ = push_pktbl(p); - $<ctxt>1 = p->ctxt; p->ctxt.in_kwarg = 0; } p_kwargs rbrace { - pop_pktbl(p, $<tbl>2); - p->ctxt.in_kwarg = $<ctxt>1.in_kwarg; - $$ = new_hash_pattern(p, Qnone, $3, &@$); + pop_pktbl(p, $p_pktbl); + p->ctxt.in_kwarg = $ctxt.in_kwarg; + $$ = new_hash_pattern(p, 0, $p_kwargs, &@$); + /*% ripper: hshptn!(Qnil, *$:p_kwargs[0..1]) %*/ } | tLBRACE rbrace { - $$ = new_hash_pattern_tail(p, Qnone, 0, &@$); - $$ = new_hash_pattern(p, Qnone, $$, &@$); + $$ = new_hash_pattern_tail(p, 0, 0, &@$); + $$ = new_hash_pattern(p, 0, $$, &@$); + /*% ripper: hshptn!(Qnil, Qnil, Qnil) %*/ } - | tLPAREN {$<tbl>$ = push_pktbl(p);} p_expr rparen + | tLPAREN p_pktbl p_expr rparen { - pop_pktbl(p, $<tbl>2); - $$ = $3; + pop_pktbl(p, $p_pktbl); + $$ = $p_expr; + /*% ripper: $:p_expr %*/ } ; p_args : p_expr { - /*%%%*/ NODE *pre_args = NEW_LIST($1, &@$); - $$ = new_array_pattern_tail(p, pre_args, 0, Qnone, Qnone, &@$); - /*% - $$ = new_array_pattern_tail(p, rb_ary_new_from_args(1, get_value($1)), 0, Qnone, Qnone, &@$); - %*/ + $$ = new_array_pattern_tail(p, pre_args, 0, 0, 0, &@$); + /*% ripper: [[$:1], Qnil, Qnil] %*/ } | p_args_head { - $$ = new_array_pattern_tail(p, $1, 1, Qnone, Qnone, &@$); + $$ = new_array_pattern_tail(p, $1, 1, 0, 0, &@$); + /*% ripper: [$:1, Qnil, Qnil] %*/ } | p_args_head p_arg { - /*%%%*/ - $$ = new_array_pattern_tail(p, list_concat($1, $2), 0, Qnone, Qnone, &@$); - /*% - VALUE pre_args = rb_ary_concat($1, get_value($2)); - $$ = new_array_pattern_tail(p, pre_args, 0, Qnone, Qnone, &@$); - %*/ + $$ = new_array_pattern_tail(p, list_concat($1, $2), 0, 0, 0, &@$); + /*% ripper: [rb_ary_concat($:1, $:2), Qnil, Qnil] %*/ } | p_args_head p_rest { - $$ = new_array_pattern_tail(p, $1, 1, $2, Qnone, &@$); + $$ = new_array_pattern_tail(p, $1, 1, $2, 0, &@$); + /*% ripper: [$:1, $:2, Qnil] %*/ } | p_args_head p_rest ',' p_args_post { $$ = new_array_pattern_tail(p, $1, 1, $2, $4, &@$); + /*% ripper: [$:1, $:2, $:4] %*/ } | p_args_tail ; @@ -4633,114 +5712,105 @@ p_args_head : p_arg ',' } | p_args_head p_arg ',' { - /*%%%*/ $$ = list_concat($1, $2); - /*% %*/ - /*% ripper: rb_ary_concat($1, get_value($2)) %*/ + /*% ripper: rb_ary_concat($:1, $:2) %*/ } ; p_args_tail : p_rest { - $$ = new_array_pattern_tail(p, Qnone, 1, $1, Qnone, &@$); + $$ = new_array_pattern_tail(p, 0, 1, $1, 0, &@$); + /*% ripper: [Qnil, $:1, Qnil] %*/ } | p_rest ',' p_args_post { - $$ = new_array_pattern_tail(p, Qnone, 1, $1, $3, &@$); + $$ = new_array_pattern_tail(p, 0, 1, $1, $3, &@$); + /*% ripper: [Qnil, $:1, $:3] %*/ } ; p_find : p_rest ',' p_args_post ',' p_rest { $$ = new_find_pattern_tail(p, $1, $3, $5, &@$); + /*% ripper: [$:1, $:3, $:5] %*/ } ; p_rest : tSTAR tIDENTIFIER { - /*%%%*/ error_duplicate_pattern_variable(p, $2, &@2); + /*% ripper: var_field!($:2) %*/ $$ = assignable(p, $2, 0, &@$); - /*% %*/ - /*% ripper: assignable(p, var_field(p, $2)) %*/ } | tSTAR { - /*%%%*/ $$ = 0; - /*% %*/ - /*% ripper: var_field(p, Qnil) %*/ + /*% ripper: var_field!(Qnil) %*/ } ; p_args_post : p_arg | p_args_post ',' p_arg { - /*%%%*/ $$ = list_concat($1, $3); - /*% %*/ - /*% ripper: rb_ary_concat($1, get_value($3)) %*/ + /*% ripper: rb_ary_concat($:1, $:3) %*/ } ; p_arg : p_expr { - /*%%%*/ $$ = NEW_LIST($1, &@$); - /*% %*/ - /*% ripper: rb_ary_new_from_args(1, get_value($1)) %*/ + /*% ripper: [$:1] %*/ } ; p_kwargs : p_kwarg ',' p_any_kwrest { $$ = new_hash_pattern_tail(p, new_unique_key_hash(p, $1, &@$), $3, &@$); + /*% ripper: [$:1, $:3] %*/ } | p_kwarg { $$ = new_hash_pattern_tail(p, new_unique_key_hash(p, $1, &@$), 0, &@$); + /*% ripper: [$:1, Qnil] %*/ } | p_kwarg ',' { $$ = new_hash_pattern_tail(p, new_unique_key_hash(p, $1, &@$), 0, &@$); + /*% ripper: [$:1, Qnil] %*/ } | p_any_kwrest { - $$ = new_hash_pattern_tail(p, new_hash(p, Qnone, &@$), $1, &@$); + $$ = new_hash_pattern_tail(p, new_hash(p, 0, &@$), $1, &@$); + /*% ripper: [[], $:1] %*/ } ; p_kwarg : p_kw - /*% ripper[brace]: rb_ary_new_from_args(1, $1) %*/ + /*% ripper[brace]: [$:1] %*/ | p_kwarg ',' p_kw { - /*%%%*/ $$ = list_concat($1, $3); - /*% %*/ - /*% ripper: rb_ary_push($1, $3) %*/ + /*% ripper: rb_ary_push($:1, $:3) %*/ } ; p_kw : p_kw_label p_expr { - error_duplicate_pattern_key(p, get_id($1), &@1); - /*%%%*/ - $$ = list_append(p, NEW_LIST(NEW_LIT(ID2SYM($1), &@1), &@$), $2); - /*% %*/ - /*% ripper: rb_ary_new_from_args(2, get_value($1), get_value($2)) %*/ + error_duplicate_pattern_key(p, $1, &@1); + $$ = list_append(p, NEW_LIST(NEW_SYM(rb_id2str($1), &@1), &@$), $2); + /*% ripper: [$:1, $:2] %*/ } | p_kw_label { - error_duplicate_pattern_key(p, get_id($1), &@1); - if ($1 && !is_local_id(get_id($1))) { + error_duplicate_pattern_key(p, $1, &@1); + if ($1 && !is_local_id($1)) { yyerror1(&@1, "key must be valid as local variables"); } - error_duplicate_pattern_variable(p, get_id($1), &@1); - /*%%%*/ - $$ = list_append(p, NEW_LIST(NEW_LIT(ID2SYM($1), &@$), &@$), assignable(p, $1, 0, &@$)); - /*% %*/ - /*% ripper: rb_ary_new_from_args(2, get_value(assignable(p, $1)), Qnil) %*/ + error_duplicate_pattern_variable(p, $1, &@1); + $$ = list_append(p, NEW_LIST(NEW_SYM(rb_id2str($1), &@$), &@$), assignable(p, $1, 0, &@$)); + /*% ripper: [$:1, Qnil] %*/ } ; @@ -4748,33 +5818,27 @@ p_kw_label : tLABEL | tSTRING_BEG string_contents tLABEL_END { YYLTYPE loc = code_loc_gen(&@1, &@3); - /*%%%*/ if (!$2 || nd_type_p($2, NODE_STR)) { NODE *node = dsym_node(p, $2, &loc); - $$ = SYM2ID(node->nd_lit); - } - /*% - if (ripper_is_node_yylval($2) && RNODE($2)->nd_cval) { - VALUE label = RNODE($2)->nd_cval; - VALUE rval = RNODE($2)->nd_rval; - $$ = ripper_new_yylval(p, rb_intern_str(label), rval, label); - RNODE($$)->nd_loc = loc; + $$ = rb_sym2id(rb_node_sym_string_val(node)); } - %*/ else { yyerror1(&loc, "symbol literal with interpolation is not allowed"); - $$ = 0; + $$ = rb_intern_str(STR_NEW0()); } + /*% ripper: $:2 %*/ } ; p_kwrest : kwrest_mark tIDENTIFIER { $$ = $2; + /*% ripper: var_field!($:2) %*/ } | kwrest_mark { $$ = 0; + /*% ripper: Qnil %*/ } ; @@ -4785,62 +5849,54 @@ p_kwnorest : kwrest_mark keyword_nil ; p_any_kwrest : p_kwrest - | p_kwnorest {$$ = ID2VAL(idNil);} + | p_kwnorest + { + $$ = idNil; + /*% ripper: var_field!(ID2VAL(idNil)) %*/ + } ; p_value : p_primitive | p_primitive tDOT2 p_primitive { - /*%%%*/ value_expr($1); value_expr($3); $$ = NEW_DOT2($1, $3, &@$); - /*% %*/ - /*% ripper: dot2!($1, $3) %*/ + /*% ripper: dot2!($:1, $:3) %*/ } | p_primitive tDOT3 p_primitive { - /*%%%*/ value_expr($1); value_expr($3); $$ = NEW_DOT3($1, $3, &@$); - /*% %*/ - /*% ripper: dot3!($1, $3) %*/ + /*% ripper: dot3!($:1, $:3) %*/ } | p_primitive tDOT2 { - /*%%%*/ value_expr($1); $$ = NEW_DOT2($1, new_nil_at(p, &@2.end_pos), &@$); - /*% %*/ - /*% ripper: dot2!($1, Qnil) %*/ + /*% ripper: dot2!($:1, Qnil) %*/ } | p_primitive tDOT3 { - /*%%%*/ value_expr($1); $$ = NEW_DOT3($1, new_nil_at(p, &@2.end_pos), &@$); - /*% %*/ - /*% ripper: dot3!($1, Qnil) %*/ + /*% ripper: dot3!($:1, Qnil) %*/ } | p_var_ref | p_expr_ref | p_const | tBDOT2 p_primitive { - /*%%%*/ value_expr($2); $$ = NEW_DOT2(new_nil_at(p, &@1.beg_pos), $2, &@$); - /*% %*/ - /*% ripper: dot2!(Qnil, $2) %*/ + /*% ripper: dot2!(Qnil, $:2) %*/ } | tBDOT3 p_primitive { - /*%%%*/ value_expr($2); $$ = NEW_DOT3(new_nil_at(p, &@1.beg_pos), $2, &@$); - /*% %*/ - /*% ripper: dot3!(Qnil, $2) %*/ + /*% ripper: dot3!(Qnil, $:2) %*/ } ; @@ -4854,73 +5910,60 @@ p_primitive : literal | qsymbols | keyword_variable { - /*%%%*/ - if (!($$ = gettable(p, $1, &@$))) $$ = NEW_BEGIN(0, &@$); - /*% %*/ - /*% ripper: var_ref!($1) %*/ + if (!($$ = gettable(p, $1, &@$))) $$ = NEW_ERROR(&@$); + /*% ripper: var_ref!($:1) %*/ } | lambda ; p_variable : tIDENTIFIER { - /*%%%*/ error_duplicate_pattern_variable(p, $1, &@1); + /*% ripper: var_field!($:1) %*/ $$ = assignable(p, $1, 0, &@$); - /*% %*/ - /*% ripper: assignable(p, var_field(p, $1)) %*/ } ; p_var_ref : '^' tIDENTIFIER { - /*%%%*/ NODE *n = gettable(p, $2, &@$); - if (!(nd_type_p(n, NODE_LVAR) || nd_type_p(n, NODE_DVAR))) { + if (!n) { + n = NEW_ERROR(&@$); + } + else if (!(nd_type_p(n, NODE_LVAR) || nd_type_p(n, NODE_DVAR))) { compile_error(p, "%"PRIsVALUE": no such local variable", rb_id2str($2)); } $$ = n; - /*% %*/ - /*% ripper: var_ref!($2) %*/ + /*% ripper: var_ref!($:2) %*/ } | '^' nonlocal_var { - /*%%%*/ - if (!($$ = gettable(p, $2, &@$))) $$ = NEW_BEGIN(0, &@$); - /*% %*/ - /*% ripper: var_ref!($2) %*/ + if (!($$ = gettable(p, $2, &@$))) $$ = NEW_ERROR(&@$); + /*% ripper: var_ref!($:2) %*/ } ; p_expr_ref : '^' tLPAREN expr_value rparen { - /*%%%*/ - $$ = NEW_BEGIN($3, &@$); - /*% %*/ - /*% ripper: begin!($3) %*/ + $$ = NEW_BLOCK($3, &@$); + /*% ripper: begin!($:3) %*/ } ; p_const : tCOLON3 cname { - /*%%%*/ $$ = NEW_COLON3($2, &@$); - /*% %*/ - /*% ripper: top_const_ref!($2) %*/ + /*% ripper: top_const_ref!($:2) %*/ } | p_const tCOLON2 cname { - /*%%%*/ $$ = NEW_COLON2($1, $3, &@$); - /*% %*/ - /*% ripper: const_path_ref!($1, $3) %*/ + /*% ripper: const_path_ref!($:1, $:3) %*/ } | tCONSTANT { - /*%%%*/ $$ = gettable(p, $1, &@$); - /*% %*/ - /*% ripper: var_ref!($1) %*/ + /*% ripper: var_ref!($:1) %*/ } ; @@ -4928,11 +5971,13 @@ opt_rescue : k_rescue exc_list exc_var then compstmt opt_rescue { - /*%%%*/ - $$ = NEW_RESBODY($2, - $3 ? block_append(p, node_assign(p, $3, NEW_ERRINFO(&@3), NO_LEX_CTXT, &@3), $5) : $5, - $6, &@$); - + NODE *body = $5; + if ($3) { + NODE *err = NEW_ERRINFO(&@3); + err = node_assign(p, $3, err, NO_LEX_CTXT, &@3); + body = block_append(p, err, body); + } + $$ = NEW_RESBODY($2, body, $6, &@$); if ($2) { fixpos($$, $2); } @@ -4942,25 +5987,19 @@ opt_rescue : k_rescue exc_list exc_var then else { fixpos($$, $5); } - /*% %*/ - /*% ripper: rescue!(escape_Qundef($2), escape_Qundef($3), escape_Qundef($5), escape_Qundef($6)) %*/ + /*% ripper: rescue!($:2, $:3, $:5, $:6) %*/ } | none ; exc_list : arg_value { - /*%%%*/ $$ = NEW_LIST($1, &@$); - /*% %*/ - /*% ripper: rb_ary_new3(1, get_value($1)) %*/ + /*% ripper: rb_ary_new3(1, $:1) %*/ } | mrhs { - /*%%%*/ if (!($$ = splat_array($1))) $$ = $1; - /*% %*/ - /*% ripper: $1 %*/ } | none ; @@ -4968,16 +6007,16 @@ exc_list : arg_value exc_var : tASSOC lhs { $$ = $2; + /*% ripper: $:2 %*/ } | none ; opt_ensure : k_ensure compstmt { - /*%%%*/ + p->ctxt.in_rescue = $1.in_rescue; $$ = $2; - /*% %*/ - /*% ripper: ensure!($2) %*/ + /*% ripper: ensure!($:2) %*/ } | none ; @@ -4988,18 +6027,15 @@ literal : numeric strings : string { - /*%%%*/ NODE *node = $1; if (!node) { - node = NEW_STR(STR_NEW0(), &@$); - RB_OBJ_WRITTEN(p->ast, Qnil, node->nd_lit); + node = NEW_STR(STRING_NEW0(), &@$); } else { node = evstr2dstr(p, node); } $$ = node; - /*% %*/ - /*% ripper: $1 %*/ + /*% ripper: $:1 %*/ } ; @@ -5007,212 +6043,145 @@ string : tCHAR | string1 | string string1 { - /*%%%*/ $$ = literal_concat(p, $1, $2, &@$); - /*% %*/ - /*% ripper: string_concat!($1, $2) %*/ + /*% ripper: string_concat!($:1, $:2) %*/ } ; string1 : tSTRING_BEG string_contents tSTRING_END { - /*%%%*/ $$ = heredoc_dedent(p, $2); if ($$) nd_set_loc($$, &@$); - /*% %*/ - /*% ripper: string_literal!(heredoc_dedent(p, $2)) %*/ + /*% ripper: $:2 %*/ + if (p->heredoc_indent > 0) { + /*% ripper: heredoc_dedent!($:$, INT2NUM(%{p->heredoc_indent})) %*/ + p->heredoc_indent = 0; + } + /*% ripper: string_literal!($:$) %*/ } ; xstring : tXSTRING_BEG xstring_contents tSTRING_END { - /*%%%*/ $$ = new_xstring(p, heredoc_dedent(p, $2), &@$); - /*% %*/ - /*% ripper: xstring_literal!(heredoc_dedent(p, $2)) %*/ + /*% ripper: $:2 %*/ + if (p->heredoc_indent > 0) { + /*% ripper: heredoc_dedent!($:$, INT2NUM(%{p->heredoc_indent})) %*/ + p->heredoc_indent = 0; + } + /*% ripper: xstring_literal!($:$) %*/ } ; regexp : tREGEXP_BEG regexp_contents tREGEXP_END { $$ = new_regexp(p, $2, $3, &@$); + /*% ripper: regexp_literal!($:2, $:3) %*/ } ; -words_sep : ' ' {} - | words_sep ' ' - ; - -words : tWORDS_BEG words_sep word_list tSTRING_END - { - /*%%%*/ - $$ = make_list($3, &@$); - /*% %*/ - /*% ripper: array!($3) %*/ - } +words : words(tWORDS_BEG, word_list) <node> ; word_list : /* none */ { - /*%%%*/ $$ = 0; - /*% %*/ /*% ripper: words_new! %*/ } - | word_list word words_sep + | word_list word ' '+ { - /*%%%*/ $$ = list_append(p, $1, evstr2dstr(p, $2)); - /*% %*/ - /*% ripper: words_add!($1, $2) %*/ + /*% ripper: words_add!($:1, $:2) %*/ } ; word : string_content - /*% ripper[brace]: word_add!(word_new!, $1) %*/ + /*% ripper[brace]: word_add!(word_new!, $:1) %*/ | word string_content { - /*%%%*/ $$ = literal_concat(p, $1, $2, &@$); - /*% %*/ - /*% ripper: word_add!($1, $2) %*/ + /*% ripper: word_add!($:1, $:2) %*/ } ; -symbols : tSYMBOLS_BEG words_sep symbol_list tSTRING_END - { - /*%%%*/ - $$ = make_list($3, &@$); - /*% %*/ - /*% ripper: array!($3) %*/ - } +symbols : words(tSYMBOLS_BEG, symbol_list) <node> ; symbol_list : /* none */ { - /*%%%*/ $$ = 0; - /*% %*/ /*% ripper: symbols_new! %*/ } - | symbol_list word words_sep + | symbol_list word ' '+ { - /*%%%*/ $$ = symbol_append(p, $1, evstr2dstr(p, $2)); - /*% %*/ - /*% ripper: symbols_add!($1, $2) %*/ + /*% ripper: symbols_add!($:1, $:2) %*/ } ; -qwords : tQWORDS_BEG words_sep qword_list tSTRING_END - { - /*%%%*/ - $$ = make_list($3, &@$); - /*% %*/ - /*% ripper: array!($3) %*/ - } +qwords : words(tQWORDS_BEG, qword_list) <node> ; -qsymbols : tQSYMBOLS_BEG words_sep qsym_list tSTRING_END - { - /*%%%*/ - $$ = make_list($3, &@$); - /*% %*/ - /*% ripper: array!($3) %*/ - } +qsymbols : words(tQSYMBOLS_BEG, qsym_list) <node> ; qword_list : /* none */ { - /*%%%*/ $$ = 0; - /*% %*/ /*% ripper: qwords_new! %*/ } - | qword_list tSTRING_CONTENT words_sep + | qword_list tSTRING_CONTENT ' '+ { - /*%%%*/ $$ = list_append(p, $1, $2); - /*% %*/ - /*% ripper: qwords_add!($1, $2) %*/ + /*% ripper: qwords_add!($:1, $:2) %*/ } ; qsym_list : /* none */ { - /*%%%*/ $$ = 0; - /*% %*/ /*% ripper: qsymbols_new! %*/ } - | qsym_list tSTRING_CONTENT words_sep + | qsym_list tSTRING_CONTENT ' '+ { - /*%%%*/ $$ = symbol_append(p, $1, $2); - /*% %*/ - /*% ripper: qsymbols_add!($1, $2) %*/ + /*% ripper: qsymbols_add!($:1, $:2) %*/ } ; string_contents : /* none */ { - /*%%%*/ $$ = 0; + /*%%%*/ /*% %*/ /*% ripper: string_content! %*/ - /*%%%*/ - /*% - $$ = ripper_new_yylval(p, 0, $$, 0); - %*/ } | string_contents string_content { - /*%%%*/ $$ = literal_concat(p, $1, $2, &@$); - /*% %*/ - /*% ripper: string_add!($1, $2) %*/ - /*%%%*/ - /*% - if (ripper_is_node_yylval($1) && ripper_is_node_yylval($2) && - !RNODE($1)->nd_cval) { - RNODE($1)->nd_cval = RNODE($2)->nd_cval; - RNODE($1)->nd_rval = add_mark_object(p, $$); - $$ = $1; - } - %*/ + /*% ripper: string_add!($:1, $:2) %*/ } ; xstring_contents: /* none */ { - /*%%%*/ $$ = 0; - /*% %*/ /*% ripper: xstring_new! %*/ } | xstring_contents string_content { - /*%%%*/ $$ = literal_concat(p, $1, $2, &@$); - /*% %*/ - /*% ripper: xstring_add!($1, $2) %*/ + /*% ripper: xstring_add!($:1, $:2) %*/ } ; regexp_contents: /* none */ { - /*%%%*/ $$ = 0; - /*% %*/ /*% ripper: regexp_new! %*/ - /*%%%*/ - /*% - $$ = ripper_new_yylval(p, 0, $$, 0); - %*/ } | regexp_contents string_content { - /*%%%*/ NODE *head = $1, *tail = $2; if (!head) { $$ = tail; @@ -5223,111 +6192,76 @@ regexp_contents: /* none */ else { switch (nd_type(head)) { case NODE_STR: - nd_set_type(head, NODE_DSTR); + head = str2dstr(p, head); break; case NODE_DSTR: break; default: - head = list_append(p, NEW_DSTR(Qnil, &@$), head); + head = list_append(p, NEW_DSTR(0, &@$), head); break; } $$ = list_append(p, head, tail); } - /*% - VALUE s1 = 1, s2 = 0, n1 = $1, n2 = $2; - if (ripper_is_node_yylval(n1)) { - s1 = RNODE(n1)->nd_cval; - n1 = RNODE(n1)->nd_rval; - } - if (ripper_is_node_yylval(n2)) { - s2 = RNODE(n2)->nd_cval; - n2 = RNODE(n2)->nd_rval; - } - $$ = dispatch2(regexp_add, n1, n2); - if (!s1 && s2) { - $$ = ripper_new_yylval(p, 0, $$, s2); - } - %*/ + /*% ripper: regexp_add!($:1, $:2) %*/ } ; string_content : tSTRING_CONTENT - /*% ripper[brace]: ripper_new_yylval(p, 0, get_value($1), $1) %*/ + /*% ripper[brace]: $:1 %*/ | tSTRING_DVAR { /* need to backup p->lex.strterm so that a string literal `%&foo,#$&,bar&` can be parsed */ - $<strterm>$ = p->lex.strterm; + $$ = p->lex.strterm; p->lex.strterm = 0; SET_LEX_STATE(EXPR_BEG); - } + }<strterm> string_dvar { - p->lex.strterm = $<strterm>2; - /*%%%*/ + p->lex.strterm = $2; $$ = NEW_EVSTR($3, &@$); nd_set_line($$, @3.end_pos.lineno); - /*% %*/ - /*% ripper: string_dvar!($3) %*/ + /*% ripper: string_dvar!($:3) %*/ } - | tSTRING_DBEG + | tSTRING_DBEG[state] { CMDARG_PUSH(0); COND_PUSH(0); - } - { /* need to backup p->lex.strterm so that a string literal `%!foo,#{ !0 },bar!` can be parsed */ - $<strterm>$ = p->lex.strterm; + $$ = p->lex.strterm; p->lex.strterm = 0; - } - { - $<num>$ = p->lex.state; SET_LEX_STATE(EXPR_BEG); - } + }[term]<strterm> { - $<num>$ = p->lex.brace_nest; + $$ = p->lex.brace_nest; p->lex.brace_nest = 0; - } + }[brace]<num> { - $<num>$ = p->heredoc_indent; + $$ = p->heredoc_indent; p->heredoc_indent = 0; - } - compstmt tSTRING_DEND + }[indent]<num> + compstmt string_dend { COND_POP(); CMDARG_POP(); - p->lex.strterm = $<strterm>3; - SET_LEX_STATE($<num>4); - p->lex.brace_nest = $<num>5; - p->heredoc_indent = $<num>6; + p->lex.strterm = $term; + SET_LEX_STATE($state); + p->lex.brace_nest = $brace; + p->heredoc_indent = $indent; p->heredoc_line_indent = -1; - /*%%%*/ - if ($7) $7->flags &= ~NODE_FL_NEWLINE; - $$ = new_evstr(p, $7, &@$); - /*% %*/ - /*% ripper: string_embexpr!($7) %*/ + if ($compstmt) nd_unset_fl_newline($compstmt); + $$ = new_evstr(p, $compstmt, &@$); + /*% ripper: string_embexpr!($:compstmt) %*/ } ; -string_dvar : tGVAR - { - /*%%%*/ - $$ = NEW_GVAR($1, &@$); - /*% %*/ - /*% ripper: var_ref!($1) %*/ - } - | tIVAR - { - /*%%%*/ - $$ = NEW_IVAR($1, &@$); - /*% %*/ - /*% ripper: var_ref!($1) %*/ - } - | tCVAR +string_dend : tSTRING_DEND + | END_OF_INPUT + ; + +string_dvar : nonlocal_var { - /*%%%*/ - $$ = NEW_CVAR($1, &@$); - /*% %*/ - /*% ripper: var_ref!($1) %*/ + if (!($$ = gettable(p, $1, &@$))) $$ = NEW_ERROR(&@$); + /*% ripper: var_ref!($:1) %*/ } | backref ; @@ -5339,10 +6273,16 @@ symbol : ssym ssym : tSYMBEG sym { SET_LEX_STATE(EXPR_END); - /*%%%*/ - $$ = NEW_LIT(ID2SYM($2), &@$); - /*% %*/ - /*% ripper: symbol_literal!(symbol!($2)) %*/ + VALUE str = rb_id2str($2); + /* + * TODO: + * set_yylval_noname sets invalid id to yylval. + * This branch can be removed once yylval is changed to + * hold lexed string. + */ + if (!str) str = STR_NEW0(); + $$ = NEW_SYM(str, &@$); + /*% ripper: symbol_literal!(symbol!($:2)) %*/ } ; @@ -5353,21 +6293,17 @@ sym : fname dsym : tSYMBEG string_contents tSTRING_END { SET_LEX_STATE(EXPR_END); - /*%%%*/ $$ = dsym_node(p, $2, &@$); - /*% %*/ - /*% ripper: dyna_symbol!($2) %*/ + /*% ripper: dyna_symbol!($:2) %*/ } ; numeric : simple_numeric | tUMINUS_NUM simple_numeric %prec tLOWEST { - /*%%%*/ $$ = $2; - RB_OBJ_WRITE(p->ast, &$$->nd_lit, negate_lit(p, $$->nd_lit)); - /*% %*/ - /*% ripper: unary!(ID2VAL(idUMinus), $2) %*/ + negate_lit(p, $$); + /*% ripper: unary!(ID2VAL(idUMinus), $:2) %*/ } ; @@ -5398,39 +6334,30 @@ keyword_variable: keyword_nil {$$ = KWD2EID(nil, $1);} var_ref : user_variable { - /*%%%*/ - if (!($$ = gettable(p, $1, &@$))) $$ = NEW_BEGIN(0, &@$); - /*% - if (id_is_var(p, get_id($1))) { - $$ = dispatch1(var_ref, $1); + if (!($$ = gettable(p, $1, &@$))) $$ = NEW_ERROR(&@$); + if (ifdef_ripper(id_is_var(p, $1), false)) { + /*% ripper: var_ref!($:1) %*/ } else { - $$ = dispatch1(vcall, $1); + /*% ripper: vcall!($:1) %*/ } - %*/ } | keyword_variable { - /*%%%*/ - if (!($$ = gettable(p, $1, &@$))) $$ = NEW_BEGIN(0, &@$); - /*% %*/ - /*% ripper: var_ref!($1) %*/ + if (!($$ = gettable(p, $1, &@$))) $$ = NEW_ERROR(&@$); + /*% ripper: var_ref!($:1) %*/ } ; var_lhs : user_variable { - /*%%%*/ + /*% ripper: var_field!($:1) %*/ $$ = assignable(p, $1, 0, &@$); - /*% %*/ - /*% ripper: assignable(p, var_field(p, $1)) %*/ } | keyword_variable { - /*%%%*/ + /*% ripper: var_field!($:1) %*/ $$ = assignable(p, $1, 0, &@$); - /*% %*/ - /*% ripper: assignable(p, var_field(p, $1)) %*/ } ; @@ -5446,12 +6373,11 @@ superclass : '<' expr_value term { $$ = $3; + /*% ripper: $:3 %*/ } | /* none */ { - /*%%%*/ $$ = 0; - /*% %*/ /*% ripper: Qnil %*/ } ; @@ -5460,17 +6386,16 @@ f_opt_paren_args: f_paren_args | none { p->ctxt.in_argdef = 0; - $$ = new_args_tail(p, Qnone, Qnone, Qnone, &@0); - $$ = new_args(p, Qnone, Qnone, Qnone, Qnone, $$, &@0); + $$ = new_args_tail(p, 0, 0, 0, &@0); + $$ = new_args(p, 0, 0, 0, 0, $$, &@0); + /*% ripper: params!(Qnil, Qnil, Qnil, Qnil, Qnil, Qnil, Qnil) %*/ } ; f_paren_args : '(' f_args rparen { - /*%%%*/ $$ = $2; - /*% %*/ - /*% ripper: paren!($2) %*/ + /*% ripper: paren!($:2) %*/ SET_LEX_STATE(EXPR_BEG); p->command_start = TRUE; p->ctxt.in_argdef = 0; @@ -5479,129 +6404,136 @@ f_paren_args : '(' f_args rparen f_arglist : f_paren_args | { - $<ctxt>$ = p->ctxt; + $$ = p->ctxt; p->ctxt.in_kwarg = 1; p->ctxt.in_argdef = 1; SET_LEX_STATE(p->lex.state|EXPR_LABEL); /* force for args */ - } + }<ctxt> f_args term { - p->ctxt.in_kwarg = $<ctxt>1.in_kwarg; + p->ctxt.in_kwarg = $1.in_kwarg; p->ctxt.in_argdef = 0; $$ = $2; SET_LEX_STATE(EXPR_BEG); p->command_start = TRUE; + /*% ripper: $:2 %*/ } ; -args_tail : f_kwarg ',' f_kwrest opt_f_block_arg +args_tail : f_kwarg(f_kw) ',' f_kwrest opt_f_block_arg { $$ = new_args_tail(p, $1, $3, $4, &@3); + /*% ripper: [$:1, $:3, $:4] %*/ } - | f_kwarg opt_f_block_arg + | f_kwarg(f_kw) opt_f_block_arg { - $$ = new_args_tail(p, $1, Qnone, $2, &@1); + $$ = new_args_tail(p, $1, 0, $2, &@1); + /*% ripper: [$:1, Qnil, $:2] %*/ } | f_any_kwrest opt_f_block_arg { - $$ = new_args_tail(p, Qnone, $1, $2, &@1); + $$ = new_args_tail(p, 0, $1, $2, &@1); + /*% ripper: [Qnil, $:1, $:2] %*/ } | f_block_arg { - $$ = new_args_tail(p, Qnone, Qnone, $1, &@1); + $$ = new_args_tail(p, 0, 0, $1, &@1); + /*% ripper: [Qnil, Qnil, $:1] %*/ } | args_forward { add_forwarding_args(p); - $$ = new_args_tail(p, Qnone, $1, ID2VAL(idFWD_BLOCK), &@1); - /*%%%*/ - ($$->nd_ainfo)->forwarding = 1; - /*% %*/ + $$ = new_args_tail(p, 0, $1, arg_FWD_BLOCK, &@1); + $$->nd_ainfo.forwarding = 1; + /*% ripper: [Qnil, $:1, Qnil] %*/ } ; -opt_args_tail : ',' args_tail +f_args : f_arg ',' f_optarg(arg_value) ',' f_rest_arg opt_args_tail(args_tail) { - $$ = $2; + $$ = new_args(p, $1, $3, $5, 0, $6, &@$); + /*% ripper: params!($:1, $:3, $:5, Qnil, *$:6[0..2]) %*/ } - | /* none */ - { - $$ = new_args_tail(p, Qnone, Qnone, Qnone, &@0); - } - ; - -f_args : f_arg ',' f_optarg ',' f_rest_arg opt_args_tail - { - $$ = new_args(p, $1, $3, $5, Qnone, $6, &@$); - } - | f_arg ',' f_optarg ',' f_rest_arg ',' f_arg opt_args_tail + | f_arg ',' f_optarg(arg_value) ',' f_rest_arg ',' f_arg opt_args_tail(args_tail) { $$ = new_args(p, $1, $3, $5, $7, $8, &@$); + /*% ripper: params!($:1, $:3, $:5, $:7, *$:8[0..2]) %*/ } - | f_arg ',' f_optarg opt_args_tail + | f_arg ',' f_optarg(arg_value) opt_args_tail(args_tail) { - $$ = new_args(p, $1, $3, Qnone, Qnone, $4, &@$); + $$ = new_args(p, $1, $3, 0, 0, $4, &@$); + /*% ripper: params!($:1, $:3, Qnil, Qnil, *$:4[0..2]) %*/ } - | f_arg ',' f_optarg ',' f_arg opt_args_tail + | f_arg ',' f_optarg(arg_value) ',' f_arg opt_args_tail(args_tail) { - $$ = new_args(p, $1, $3, Qnone, $5, $6, &@$); + $$ = new_args(p, $1, $3, 0, $5, $6, &@$); + /*% ripper: params!($:1, $:3, Qnil, $:5, *$:6[0..2]) %*/ } - | f_arg ',' f_rest_arg opt_args_tail + | f_arg ',' f_rest_arg opt_args_tail(args_tail) { - $$ = new_args(p, $1, Qnone, $3, Qnone, $4, &@$); + $$ = new_args(p, $1, 0, $3, 0, $4, &@$); + /*% ripper: params!($:1, Qnil, $:3, Qnil, *$:4[0..2]) %*/ } - | f_arg ',' f_rest_arg ',' f_arg opt_args_tail + | f_arg ',' f_rest_arg ',' f_arg opt_args_tail(args_tail) { - $$ = new_args(p, $1, Qnone, $3, $5, $6, &@$); + $$ = new_args(p, $1, 0, $3, $5, $6, &@$); + /*% ripper: params!($:1, Qnil, $:3, $:5, *$:6[0..2]) %*/ } - | f_arg opt_args_tail + | f_arg opt_args_tail(args_tail) { - $$ = new_args(p, $1, Qnone, Qnone, Qnone, $2, &@$); + $$ = new_args(p, $1, 0, 0, 0, $2, &@$); + /*% ripper: params!($:1, Qnil, Qnil, Qnil, *$:2[0..2]) %*/ } - | f_optarg ',' f_rest_arg opt_args_tail + | f_optarg(arg_value) ',' f_rest_arg opt_args_tail(args_tail) { - $$ = new_args(p, Qnone, $1, $3, Qnone, $4, &@$); + $$ = new_args(p, 0, $1, $3, 0, $4, &@$); + /*% ripper: params!(Qnil, $:1, $:3, Qnil, *$:4[0..2]) %*/ } - | f_optarg ',' f_rest_arg ',' f_arg opt_args_tail + | f_optarg(arg_value) ',' f_rest_arg ',' f_arg opt_args_tail(args_tail) { - $$ = new_args(p, Qnone, $1, $3, $5, $6, &@$); + $$ = new_args(p, 0, $1, $3, $5, $6, &@$); + /*% ripper: params!(Qnil, $:1, $:3, $:5, *$:6[0..2]) %*/ } - | f_optarg opt_args_tail + | f_optarg(arg_value) opt_args_tail(args_tail) { - $$ = new_args(p, Qnone, $1, Qnone, Qnone, $2, &@$); + $$ = new_args(p, 0, $1, 0, 0, $2, &@$); + /*% ripper: params!(Qnil, $:1, Qnil, Qnil, *$:2[0..2]) %*/ } - | f_optarg ',' f_arg opt_args_tail + | f_optarg(arg_value) ',' f_arg opt_args_tail(args_tail) { - $$ = new_args(p, Qnone, $1, Qnone, $3, $4, &@$); + $$ = new_args(p, 0, $1, 0, $3, $4, &@$); + /*% ripper: params!(Qnil, $:1, Qnil, $:3, *$:4[0..2]) %*/ } - | f_rest_arg opt_args_tail + | f_rest_arg opt_args_tail(args_tail) { - $$ = new_args(p, Qnone, Qnone, $1, Qnone, $2, &@$); + $$ = new_args(p, 0, 0, $1, 0, $2, &@$); + /*% ripper: params!(Qnil, Qnil, $:1, Qnil, *$:2[0..2]) %*/ } - | f_rest_arg ',' f_arg opt_args_tail + | f_rest_arg ',' f_arg opt_args_tail(args_tail) { - $$ = new_args(p, Qnone, Qnone, $1, $3, $4, &@$); + $$ = new_args(p, 0, 0, $1, $3, $4, &@$); + /*% ripper: params!(Qnil, Qnil, $:1, $:3, *$:4[0..2]) %*/ } | args_tail { - $$ = new_args(p, Qnone, Qnone, Qnone, Qnone, $1, &@$); + $$ = new_args(p, 0, 0, 0, 0, $1, &@$); + /*% ripper: params!(Qnil, Qnil, Qnil, Qnil, *$:1[0..2]) %*/ } | /* none */ { - $$ = new_args_tail(p, Qnone, Qnone, Qnone, &@0); - $$ = new_args(p, Qnone, Qnone, Qnone, Qnone, $$, &@0); + $$ = new_args_tail(p, 0, 0, 0, &@0); + $$ = new_args(p, 0, 0, 0, 0, $$, &@0); + /*% ripper: params!(Qnil, Qnil, Qnil, Qnil, Qnil, Qnil, Qnil) %*/ } ; args_forward : tBDOT3 { - /*%%%*/ #ifdef FORWARD_ARGS_WITH_RUBY2_KEYWORDS $$ = 0; #else $$ = idFWD_KWREST; #endif - /*% %*/ /*% ripper: args_forward! %*/ } ; @@ -5611,36 +6543,36 @@ f_bad_arg : tCONSTANT static const char mesg[] = "formal argument cannot be a constant"; /*%%%*/ yyerror1(&@1, mesg); - $$ = 0; /*% %*/ - /*% ripper[error]: param_error!(ERR_MESG(), $1) %*/ + $$ = 0; + /*% ripper[error]: param_error!(ERR_MESG(), $:1) %*/ } | tIVAR { static const char mesg[] = "formal argument cannot be an instance variable"; /*%%%*/ yyerror1(&@1, mesg); - $$ = 0; /*% %*/ - /*% ripper[error]: param_error!(ERR_MESG(), $1) %*/ + $$ = 0; + /*% ripper[error]: param_error!(ERR_MESG(), $:1) %*/ } | tGVAR { static const char mesg[] = "formal argument cannot be a global variable"; /*%%%*/ yyerror1(&@1, mesg); - $$ = 0; /*% %*/ - /*% ripper[error]: param_error!(ERR_MESG(), $1) %*/ + $$ = 0; + /*% ripper[error]: param_error!(ERR_MESG(), $:1) %*/ } | tCVAR { static const char mesg[] = "formal argument cannot be a class variable"; /*%%%*/ yyerror1(&@1, mesg); - $$ = 0; /*% %*/ - /*% ripper[error]: param_error!(ERR_MESG(), $1) %*/ + $$ = 0; + /*% ripper[error]: param_error!(ERR_MESG(), $:1) %*/ } ; @@ -5650,12 +6582,13 @@ f_norm_arg : f_bad_arg formal_argument(p, $1); p->max_numparam = ORDINAL_PARAM; $$ = $1; + /*% ripper: ripper_formal_argument(p, $1, $:1) %*/ } ; f_arg_asgn : f_norm_arg { - ID id = get_id($1); + ID id = $1; arg_var(p, id); p->cur_arg = id; $$ = $1; @@ -5665,14 +6598,11 @@ f_arg_asgn : f_norm_arg f_arg_item : f_arg_asgn { p->cur_arg = 0; - /*%%%*/ $$ = NEW_ARGS_AUX($1, 1, &NULL_LOC); - /*% %*/ - /*% ripper: get_value($1) %*/ + /*% ripper: $:1 %*/ } | tLPAREN f_margs rparen { - /*%%%*/ ID tid = internal_id(p); YYLTYPE loc; loc.beg_pos = @2.beg_pos; @@ -5685,23 +6615,20 @@ f_arg_item : f_arg_asgn $2->nd_value = NEW_LVAR(tid, &loc); } $$ = NEW_ARGS_AUX(tid, 1, &NULL_LOC); - $$->nd_next = $2; - /*% %*/ - /*% ripper: mlhs_paren!($2) %*/ + $$->nd_next = (NODE *)$2; + /*% ripper: mlhs_paren!($:2) %*/ } ; f_arg : f_arg_item - /*% ripper[brace]: rb_ary_new3(1, get_value($1)) %*/ + /*% ripper[brace]: rb_ary_new3(1, $:1) %*/ | f_arg ',' f_arg_item { - /*%%%*/ $$ = $1; $$->nd_plen++; $$->nd_next = block_append(p, $$->nd_next, $3->nd_next); - rb_discard_node(p, $3); - /*% %*/ - /*% ripper: rb_ary_push($1, get_value($3)) %*/ + rb_discard_node(p, (NODE *)$3); + /*% ripper: rb_ary_push($:1, $:3) %*/ } ; @@ -5709,10 +6636,11 @@ f_arg : f_arg_item f_label : tLABEL { arg_var(p, formal_argument(p, $1)); - p->cur_arg = get_id($1); + p->cur_arg = $1; p->max_numparam = ORDINAL_PARAM; p->ctxt.in_argdef = 0; $$ = $1; + /*% ripper: ripper_formal_argument(p, $1, $:1) %*/ } ; @@ -5720,70 +6648,29 @@ f_kw : f_label arg_value { p->cur_arg = 0; p->ctxt.in_argdef = 1; - /*%%%*/ $$ = new_kw_arg(p, assignable(p, $1, $2, &@$), &@$); - /*% %*/ - /*% ripper: rb_assoc_new(get_value(assignable(p, $1)), get_value($2)) %*/ + /*% ripper: [$:$, $:2] %*/ } | f_label { p->cur_arg = 0; p->ctxt.in_argdef = 1; - /*%%%*/ $$ = new_kw_arg(p, assignable(p, $1, NODE_SPECIAL_REQUIRED_KEYWORD, &@$), &@$); - /*% %*/ - /*% ripper: rb_assoc_new(get_value(assignable(p, $1)), 0) %*/ + /*% ripper: [$:$, 0] %*/ } ; f_block_kw : f_label primary_value { p->ctxt.in_argdef = 1; - /*%%%*/ $$ = new_kw_arg(p, assignable(p, $1, $2, &@$), &@$); - /*% %*/ - /*% ripper: rb_assoc_new(get_value(assignable(p, $1)), get_value($2)) %*/ + /*% ripper: [$:$, $:2] %*/ } | f_label { p->ctxt.in_argdef = 1; - /*%%%*/ $$ = new_kw_arg(p, assignable(p, $1, NODE_SPECIAL_REQUIRED_KEYWORD, &@$), &@$); - /*% %*/ - /*% ripper: rb_assoc_new(get_value(assignable(p, $1)), 0) %*/ - } - ; - -f_block_kwarg : f_block_kw - { - /*%%%*/ - $$ = $1; - /*% %*/ - /*% ripper: rb_ary_new3(1, get_value($1)) %*/ - } - | f_block_kwarg ',' f_block_kw - { - /*%%%*/ - $$ = kwd_append($1, $3); - /*% %*/ - /*% ripper: rb_ary_push($1, get_value($3)) %*/ - } - ; - - -f_kwarg : f_kw - { - /*%%%*/ - $$ = $1; - /*% %*/ - /*% ripper: rb_ary_new3(1, get_value($1)) %*/ - } - | f_kwarg ',' f_kw - { - /*%%%*/ - $$ = kwd_append($1, $3); - /*% %*/ - /*% ripper: rb_ary_push($1, get_value($3)) %*/ + /*% ripper: [$:$, 0] %*/ } ; @@ -5793,102 +6680,38 @@ kwrest_mark : tPOW f_no_kwarg : p_kwnorest { - /*%%%*/ - /*% %*/ /*% ripper: nokw_param!(Qnil) %*/ } ; f_kwrest : kwrest_mark tIDENTIFIER { - arg_var(p, shadowing_lvar(p, get_id($2))); - /*%%%*/ + arg_var(p, shadowing_lvar(p, $2)); $$ = $2; - /*% %*/ - /*% ripper: kwrest_param!($2) %*/ + /*% ripper: kwrest_param!($:2) %*/ } | kwrest_mark { arg_var(p, idFWD_KWREST); - /*%%%*/ $$ = idFWD_KWREST; - /*% %*/ /*% ripper: kwrest_param!(Qnil) %*/ } ; -f_opt : f_arg_asgn f_eq arg_value - { - p->cur_arg = 0; - p->ctxt.in_argdef = 1; - /*%%%*/ - $$ = NEW_OPT_ARG(0, assignable(p, $1, $3, &@$), &@$); - /*% %*/ - /*% ripper: rb_assoc_new(get_value(assignable(p, $1)), get_value($3)) %*/ - } - ; - -f_block_opt : f_arg_asgn f_eq primary_value - { - p->cur_arg = 0; - p->ctxt.in_argdef = 1; - /*%%%*/ - $$ = NEW_OPT_ARG(0, assignable(p, $1, $3, &@$), &@$); - /*% %*/ - /*% ripper: rb_assoc_new(get_value(assignable(p, $1)), get_value($3)) %*/ - } - ; - -f_block_optarg : f_block_opt - { - /*%%%*/ - $$ = $1; - /*% %*/ - /*% ripper: rb_ary_new3(1, get_value($1)) %*/ - } - | f_block_optarg ',' f_block_opt - { - /*%%%*/ - $$ = opt_arg_append($1, $3); - /*% %*/ - /*% ripper: rb_ary_push($1, get_value($3)) %*/ - } - ; - -f_optarg : f_opt - { - /*%%%*/ - $$ = $1; - /*% %*/ - /*% ripper: rb_ary_new3(1, get_value($1)) %*/ - } - | f_optarg ',' f_opt - { - /*%%%*/ - $$ = opt_arg_append($1, $3); - /*% %*/ - /*% ripper: rb_ary_push($1, get_value($3)) %*/ - } - ; - restarg_mark : '*' | tSTAR ; f_rest_arg : restarg_mark tIDENTIFIER { - arg_var(p, shadowing_lvar(p, get_id($2))); - /*%%%*/ + arg_var(p, shadowing_lvar(p, $2)); $$ = $2; - /*% %*/ - /*% ripper: rest_param!($2) %*/ + /*% ripper: rest_param!($:2) %*/ } | restarg_mark { arg_var(p, idFWD_REST); - /*%%%*/ $$ = idFWD_REST; - /*% %*/ /*% ripper: rest_param!(Qnil) %*/ } ; @@ -5899,18 +6722,14 @@ blkarg_mark : '&' f_block_arg : blkarg_mark tIDENTIFIER { - arg_var(p, shadowing_lvar(p, get_id($2))); - /*%%%*/ + arg_var(p, shadowing_lvar(p, $2)); $$ = $2; - /*% %*/ - /*% ripper: blockarg!($2) %*/ + /*% ripper: blockarg!($:2) %*/ } | blkarg_mark { arg_var(p, idFWD_BLOCK); - /*%%%*/ $$ = idFWD_BLOCK; - /*% %*/ /*% ripper: blockarg!(Qnil) %*/ } ; @@ -5918,10 +6737,12 @@ f_block_arg : blkarg_mark tIDENTIFIER opt_f_block_arg : ',' f_block_arg { $$ = $2; + /*% ripper: $:2 %*/ } | none { - $$ = Qnull; + $$ = 0; + /*% ripper: Qnil %*/ } ; @@ -5932,125 +6753,104 @@ singleton : var_ref } | '(' {SET_LEX_STATE(EXPR_BEG);} expr rparen { - /*%%%*/ - switch (nd_type($3)) { + NODE *expr = last_expr_node($3); + switch (nd_type(expr)) { case NODE_STR: case NODE_DSTR: case NODE_XSTR: case NODE_DXSTR: + case NODE_REGX: case NODE_DREGX: - case NODE_LIT: + case NODE_SYM: + case NODE_LINE: + case NODE_FILE: + case NODE_ENCODING: + case NODE_INTEGER: + case NODE_FLOAT: + case NODE_RATIONAL: + case NODE_IMAGINARY: + case NODE_DSYM: case NODE_LIST: case NODE_ZLIST: - yyerror1(&@3, "can't define singleton method for literals"); + yyerror1(&expr->nd_loc, "can't define singleton method for literals"); break; default: value_expr($3); break; } $$ = $3; - /*% %*/ - /*% ripper: paren!($3) %*/ + /*% ripper: paren!($:3) %*/ } ; assoc_list : none | assocs trailer { - /*%%%*/ $$ = $1; - /*% %*/ - /*% ripper: assoclist_from_args!($1) %*/ + /*% ripper: assoclist_from_args!($:1) %*/ } ; assocs : assoc - /*% ripper[brace]: rb_ary_new3(1, get_value($1)) %*/ + /*% ripper[brace]: rb_ary_new3(1, $:1) %*/ | assocs ',' assoc { - /*%%%*/ NODE *assocs = $1; NODE *tail = $3; if (!assocs) { assocs = tail; } else if (tail) { - if (assocs->nd_head && - !tail->nd_head && nd_type_p(tail->nd_next, NODE_LIST) && - nd_type_p(tail->nd_next->nd_head, NODE_HASH)) { - /* DSTAR */ - tail = tail->nd_next->nd_head->nd_head; + if (RNODE_LIST(assocs)->nd_head) { + NODE *n = RNODE_LIST(tail)->nd_next; + if (!RNODE_LIST(tail)->nd_head && nd_type_p(n, NODE_LIST) && + nd_type_p((n = RNODE_LIST(n)->nd_head), NODE_HASH)) { + /* DSTAR */ + tail = RNODE_HASH(n)->nd_head; + } + } + if (tail) { + assocs = list_concat(assocs, tail); } - assocs = list_concat(assocs, tail); } $$ = assocs; - /*% %*/ - /*% ripper: rb_ary_push($1, get_value($3)) %*/ + /*% ripper: rb_ary_push($:1, $:3) %*/ } ; assoc : arg_value tASSOC arg_value { - /*%%%*/ - if (nd_type_p($1, NODE_STR)) { - nd_set_type($1, NODE_LIT); - RB_OBJ_WRITE(p->ast, &$1->nd_lit, rb_fstring($1->nd_lit)); - } $$ = list_append(p, NEW_LIST($1, &@$), $3); - /*% %*/ - /*% ripper: assoc_new!($1, $3) %*/ + /*% ripper: assoc_new!($:1, $:3) %*/ } | tLABEL arg_value { - /*%%%*/ - $$ = list_append(p, NEW_LIST(NEW_LIT(ID2SYM($1), &@1), &@$), $2); - /*% %*/ - /*% ripper: assoc_new!($1, $2) %*/ + $$ = list_append(p, NEW_LIST(NEW_SYM(rb_id2str($1), &@1), &@$), $2); + /*% ripper: assoc_new!($:1, $:2) %*/ } | tLABEL { - /*%%%*/ NODE *val = gettable(p, $1, &@$); - if (!val) val = NEW_BEGIN(0, &@$); - $$ = list_append(p, NEW_LIST(NEW_LIT(ID2SYM($1), &@1), &@$), val); - /*% %*/ - /*% ripper: assoc_new!($1, Qnil) %*/ + if (!val) val = NEW_ERROR(&@$); + $$ = list_append(p, NEW_LIST(NEW_SYM(rb_id2str($1), &@1), &@$), val); + /*% ripper: assoc_new!($:1, Qnil) %*/ } | tSTRING_BEG string_contents tLABEL_END arg_value { - /*%%%*/ YYLTYPE loc = code_loc_gen(&@1, &@3); $$ = list_append(p, NEW_LIST(dsym_node(p, $2, &loc), &loc), $4); - /*% %*/ - /*% ripper: assoc_new!(dyna_symbol!($2), $4) %*/ + /*% ripper: assoc_new!(dyna_symbol!($:2), $:4) %*/ } | tDSTAR arg_value { - /*%%%*/ - if (nd_type_p($2, NODE_HASH) && - !($2->nd_head && $2->nd_head->nd_alen)) { - static VALUE empty_hash; - if (!empty_hash) { - empty_hash = rb_obj_freeze(rb_hash_new()); - rb_gc_register_mark_object(empty_hash); - } - $$ = list_append(p, NEW_LIST(0, &@$), NEW_LIT(empty_hash, &@$)); - } - else - $$ = list_append(p, NEW_LIST(0, &@$), $2); - /*% %*/ - /*% ripper: assoc_splat!($2) %*/ + $$ = list_append(p, NEW_LIST(0, &@$), $2); + /*% ripper: assoc_splat!($:2) %*/ } | tDSTAR { - if (!local_id(p, idFWD_KWREST) || - local_id(p, idFWD_ALL)) { - compile_error(p, "no anonymous keyword rest parameter"); - } - /*%%%*/ + forwarding_arg_check(p, idFWD_KWREST, idFWD_ALL, "keyword rest"); $$ = list_append(p, NEW_LIST(0, &@$), NEW_LVAR(idFWD_KWREST, &@$)); - /*% %*/ /*% ripper: assoc_splat!(Qnil) %*/ } ; @@ -6081,24 +6881,16 @@ call_op2 : call_op | tCOLON2 ; -opt_terms : /* none */ - | terms - ; - -opt_nl : /* none */ - | '\n' - ; - -rparen : opt_nl ')' +rparen : '\n'? ')' ; -rbracket : opt_nl ']' +rbracket : '\n'? ']' ; -rbrace : opt_nl '}' +rbrace : '\n'? '}' ; -trailer : opt_nl +trailer : '\n'? | ',' ; @@ -6116,7 +6908,7 @@ terms : term none : /* none */ { - $$ = Qnull; + $$ = 0; } ; %% @@ -6132,43 +6924,35 @@ static enum yytokentype parse_string(struct parser_params*,rb_strterm_literal_t* static enum yytokentype here_document(struct parser_params*,rb_strterm_heredoc_t*); #ifndef RIPPER +#define set_parser_s_value(x) (void)(x) +#else +#define set_parser_s_value(x) (p->s_value = (x)) +#endif + # define set_yylval_node(x) { \ YYLTYPE _cur_loc; \ rb_parser_set_location(p, &_cur_loc); \ yylval.node = (x); \ + set_parser_s_value(STR_NEW(p->lex.ptok, p->lex.pcur-p->lex.ptok)); \ } # define set_yylval_str(x) \ do { \ - set_yylval_node(NEW_STR(x, &_cur_loc)); \ - RB_OBJ_WRITTEN(p->ast, Qnil, x); \ + set_yylval_node(NEW_STR(rb_str_to_parser_string(p, x), &_cur_loc)); \ + set_parser_s_value(x); \ } while(0) -# define set_yylval_literal(x) \ -do { \ - set_yylval_node(NEW_LIT(x, &_cur_loc)); \ - RB_OBJ_WRITTEN(p->ast, Qnil, x); \ -} while(0) -# define set_yylval_num(x) (yylval.num = (x)) -# define set_yylval_id(x) (yylval.id = (x)) -# define set_yylval_name(x) (yylval.id = (x)) +# define set_yylval_num(x) { \ + yylval.num = (x); \ + set_parser_s_value(x); \ +} +# define set_yylval_id(x) (yylval.id = (x)) +# define set_yylval_name(x) { \ + (yylval.id = (x)); \ + set_parser_s_value(ID2SYM(x)); \ +} # define yylval_id() (yylval.id) -#else -static inline VALUE -ripper_yylval_id(struct parser_params *p, ID x) -{ - return ripper_new_yylval(p, x, ID2SYM(x), 0); -} -# define set_yylval_str(x) (yylval.val = add_mark_object(p, (x))) -# define set_yylval_num(x) (yylval.val = ripper_new_yylval(p, (x), 0, 0)) -# define set_yylval_id(x) (void)(x) -# define set_yylval_name(x) (void)(yylval.val = ripper_yylval_id(p, x)) -# define set_yylval_literal(x) add_mark_object(p, (x)) -# define set_yylval_node(x) (yylval.val = ripper_new_yylval(p, 0, 0, STR_NEW(p->lex.ptok, p->lex.pcur-p->lex.ptok))) -# define yylval_id() yylval.id -# define _cur_loc NULL_LOC /* dummy */ -#endif #define set_yylval_noname() set_yylval_id(keyword_nil) -#define has_delayed_token(p) (!NIL_P(p->delayed.token)) +#define has_delayed_token(p) (p->delayed.token != NULL) #ifndef RIPPER #define literal_flush(p, ptr) ((p)->lex.ptok = (ptr)) @@ -6186,35 +6970,100 @@ parser_has_token(struct parser_params *p) return pcur > ptok; } -static VALUE -code_loc_to_ary(const rb_code_location_t *loc) +static const char * +escaped_char(int c) { - VALUE ary = rb_ary_new_from_args(4, - INT2NUM(loc->beg_pos.lineno), INT2NUM(loc->beg_pos.column), - INT2NUM(loc->end_pos.lineno), INT2NUM(loc->end_pos.column)); - rb_obj_freeze(ary); - - return ary; + switch (c) { + case '"': return "\\\""; + case '\\': return "\\\\"; + case '\0': return "\\0"; + case '\n': return "\\n"; + case '\r': return "\\r"; + case '\t': return "\\t"; + case '\f': return "\\f"; + case '\013': return "\\v"; + case '\010': return "\\b"; + case '\007': return "\\a"; + case '\033': return "\\e"; + case '\x7f': return "\\c?"; + } + return NULL; } -static void -parser_append_tokens(struct parser_params *p, VALUE str, enum yytokentype t, int line) +static rb_parser_string_t * +rb_parser_str_escape(struct parser_params *p, rb_parser_string_t *str) { - VALUE ary; - int token_id; + rb_encoding *enc = p->enc; + const char *ptr = str->ptr; + const char *pend = ptr + str->len; + const char *prev = ptr; + char charbuf[5] = {'\\', 'x', 0, 0, 0}; + rb_parser_string_t * result = rb_parser_string_new(p, 0, 0); + int asciicompat = rb_enc_asciicompat(enc); + + while (ptr < pend) { + unsigned int c; + const char *cc; + int n = rb_enc_precise_mbclen(ptr, pend, enc); + if (!MBCLEN_CHARFOUND_P(n)) { + if (ptr > prev) rb_parser_str_buf_cat(p, result, prev, ptr - prev); + n = rb_enc_mbminlen(enc); + if (pend < ptr + n) + n = (int)(pend - ptr); + while (n--) { + c = *ptr & 0xf0 >> 4; + charbuf[2] = (c < 10) ? '0' + c : 'A' + c - 10; + c = *ptr & 0x0f; + charbuf[3] = (c < 10) ? '0' + c : 'A' + c - 10; + rb_parser_str_buf_cat(p, result, charbuf, 4); + prev = ++ptr; + } + continue; + } + n = MBCLEN_CHARFOUND_LEN(n); + c = rb_enc_mbc_to_codepoint(ptr, pend, enc); + ptr += n; + cc = escaped_char(c); + if (cc) { + if (ptr - n > prev) rb_parser_str_buf_cat(p, result, prev, ptr - n - prev); + rb_parser_str_buf_cat(p, result, cc, strlen(cc)); + prev = ptr; + } + else if (asciicompat && rb_enc_isascii(c, enc) && ISPRINT(c)) { + } + else { + if (ptr - n > prev) { + rb_parser_str_buf_cat(p, result, prev, ptr - n - prev); + prev = ptr - n; + } + rb_parser_str_buf_cat(p, result, prev, ptr - prev); + prev = ptr; + } + } + if (ptr > prev) rb_parser_str_buf_cat(p, result, prev, ptr - prev); - ary = rb_ary_new2(4); - token_id = p->token_id; - rb_ary_push(ary, INT2FIX(token_id)); - rb_ary_push(ary, ID2SYM(parser_token2id(t))); - rb_ary_push(ary, str); - rb_ary_push(ary, code_loc_to_ary(p->yylloc)); - rb_obj_freeze(ary); - rb_ary_push(p->tokens, ary); + return result; +} + +static void +parser_append_tokens(struct parser_params *p, rb_parser_string_t *str, enum yytokentype t, int line) +{ + rb_parser_ast_token_t *token = xcalloc(1, sizeof(rb_parser_ast_token_t)); + token->id = p->token_id; + token->type_name = parser_token2char(p, t); + token->str = str; + token->loc.beg_pos = p->yylloc->beg_pos; + token->loc.end_pos = p->yylloc->end_pos; + rb_parser_ary_push_ast_token(p, p->tokens, token); p->token_id++; if (p->debug) { - rb_parser_printf(p, "Append tokens (line: %d) %"PRIsVALUE"\n", line, ary); + rb_parser_string_t *str_escaped = rb_parser_str_escape(p, str); + rb_parser_printf(p, "Append tokens (line: %d) [%d, :%s, \"%s\", [%d, %d, %d, %d]]\n", + line, token->id, token->type_name, str_escaped->ptr, + token->loc.beg_pos.lineno, token->loc.beg_pos.column, + token->loc.end_pos.lineno, token->loc.end_pos.column); + rb_parser_string_free(p, str_escaped); } } @@ -6228,7 +7077,7 @@ parser_dispatch_scan_event(struct parser_params *p, enum yytokentype t, int line RUBY_SET_YYLLOC(*p->yylloc); if (p->keep_tokens) { - VALUE str = STR_NEW(p->lex.ptok, p->lex.pcur - p->lex.ptok); + rb_parser_string_t *str = rb_parser_encoding_string_new(p, p->lex.ptok, p->lex.pcur - p->lex.ptok, p->enc); parser_append_tokens(p, str, t, line); } @@ -6239,9 +7088,6 @@ parser_dispatch_scan_event(struct parser_params *p, enum yytokentype t, int line static void parser_dispatch_delayed_token(struct parser_params *p, enum yytokentype t, int line) { - int saved_line = p->ruby_sourceline; - const char *saved_tokp = p->lex.ptok; - debug_token_line(p, "parser_dispatch_delayed_token", line); if (!has_delayed_token(p)) return; @@ -6249,27 +7095,17 @@ parser_dispatch_delayed_token(struct parser_params *p, enum yytokentype t, int l RUBY_SET_YYLLOC_OF_DELAYED_TOKEN(*p->yylloc); if (p->keep_tokens) { - p->ruby_sourceline = p->delayed.beg_line; - p->lex.ptok = p->lex.pbeg + p->delayed.beg_col; + /* p->delayed.token is freed by rb_parser_tokens_free */ parser_append_tokens(p, p->delayed.token, t, line); - p->ruby_sourceline = saved_line; - p->lex.ptok = saved_tokp; + } else { + rb_parser_string_free(p, p->delayed.token); } - p->delayed.token = Qnil; + p->delayed.token = NULL; } #else #define literal_flush(p, ptr) ((void)(ptr)) -#define yylval_rval (*(RB_TYPE_P(yylval.val, T_NODE) ? &yylval.node->nd_rval : &yylval.val)) - -static inline VALUE -intern_sym(const char *name) -{ - ID id = rb_intern_const(name); - return ID2SYM(id); -} - static int ripper_has_scan_event(struct parser_params *p) { @@ -6291,21 +7127,27 @@ static void ripper_dispatch_scan_event(struct parser_params *p, enum yytokentype t) { if (!ripper_has_scan_event(p)) return; - add_mark_object(p, yylval_rval = ripper_scan_event_val(p, t)); + + set_parser_s_value(ripper_scan_event_val(p, t)); } #define dispatch_scan_event(p, t) ripper_dispatch_scan_event(p, t) static void ripper_dispatch_delayed_token(struct parser_params *p, enum yytokentype t) { + /* save and adjust the location to delayed token for callbacks */ int saved_line = p->ruby_sourceline; const char *saved_tokp = p->lex.ptok; + VALUE s_value, str; if (!has_delayed_token(p)) return; p->ruby_sourceline = p->delayed.beg_line; p->lex.ptok = p->lex.pbeg + p->delayed.beg_col; - add_mark_object(p, yylval_rval = ripper_dispatch1(p, ripper_token2eventid(t), p->delayed.token)); - p->delayed.token = Qnil; + str = rb_str_new_mutable_parser_string(p->delayed.token); + rb_parser_string_free(p, p->delayed.token); + s_value = ripper_dispatch1(p, ripper_token2eventid(t), str); + set_parser_s_value(s_value); + p->delayed.token = NULL; p->ruby_sourceline = saved_line; p->lex.ptok = saved_tokp; } @@ -6313,7 +7155,7 @@ ripper_dispatch_delayed_token(struct parser_params *p, enum yytokentype t) #endif /* RIPPER */ static inline int -is_identchar(const char *ptr, const char *MAYBE_UNUSED(ptr_end), rb_encoding *enc) +is_identchar(struct parser_params *p, const char *ptr, const char *MAYBE_UNUSED(ptr_end), rb_encoding *enc) { return rb_enc_isalnum((unsigned char)*ptr, enc) || *ptr == '_' || !ISASCII(*ptr); } @@ -6321,7 +7163,7 @@ is_identchar(const char *ptr, const char *MAYBE_UNUSED(ptr_end), rb_encoding *en static inline int parser_is_identchar(struct parser_params *p) { - return !(p)->eofp && is_identchar(p->lex.pcur-1, p->lex.pend, p->enc); + return !(p)->eofp && is_identchar(p, p->lex.pcur-1, p->lex.pend, p->enc); } static inline int @@ -6424,12 +7266,10 @@ parser_precise_mbclen(struct parser_params *p, const char *ptr) } #ifndef RIPPER -static void ruby_show_error_line(VALUE errbuf, const YYLTYPE *yylloc, int lineno, VALUE str); - static inline void parser_show_error_line(struct parser_params *p, const YYLTYPE *yylloc) { - VALUE str; + rb_parser_string_t *str; int lineno = p->ruby_sourceline; if (!yylloc) { return; @@ -6440,11 +7280,11 @@ parser_show_error_line(struct parser_params *p, const YYLTYPE *yylloc) else { return; } - ruby_show_error_line(p->error_buffer, yylloc, lineno, str); + ruby_show_error_line(p, p->error_buffer, yylloc, lineno, str); } static int -parser_yyerror(struct parser_params *p, const YYLTYPE *yylloc, const char *msg) +parser_yyerror(struct parser_params *p, const rb_code_location_t *yylloc, const char *msg) { #if 0 YYLTYPE current; @@ -6457,7 +7297,7 @@ parser_yyerror(struct parser_params *p, const YYLTYPE *yylloc, const char *msg) yylloc = 0; } #endif - compile_error(p, "%s", msg); + parser_compile_error(p, yylloc, "%s", msg); parser_show_error_line(p, yylloc); return 0; } @@ -6469,8 +7309,8 @@ parser_yyerror0(struct parser_params *p, const char *msg) return parser_yyerror(p, RUBY_SET_YYLLOC(current), msg); } -static void -ruby_show_error_line(VALUE errbuf, const YYLTYPE *yylloc, int lineno, VALUE str) +void +ruby_show_error_line(struct parser_params *p, VALUE errbuf, const YYLTYPE *yylloc, int lineno, rb_parser_string_t *str) { VALUE mesg; const int max_line_margin = 30; @@ -6478,13 +7318,13 @@ ruby_show_error_line(VALUE errbuf, const YYLTYPE *yylloc, int lineno, VALUE str) const char *pre = "", *post = "", *pend; const char *code = "", *caret = ""; const char *lim; - const char *const pbeg = RSTRING_PTR(str); + const char *const pbeg = PARSER_STRING_PTR(str); char *buf; long len; int i; if (!yylloc) return; - pend = RSTRING_END(str); + pend = rb_parser_string_end(str); if (pend > pbeg && pend[-1] == '\n') { if (--pend > pbeg && pend[-1] == '\r') --pend; } @@ -6505,11 +7345,11 @@ ruby_show_error_line(VALUE errbuf, const YYLTYPE *yylloc, int lineno, VALUE str) len = ptr_end - ptr; if (len > 4) { if (ptr > pbeg) { - ptr = rb_enc_prev_char(pbeg, ptr, pt, rb_enc_get(str)); + ptr = rb_enc_prev_char(pbeg, ptr, pt, rb_parser_str_get_encoding(str)); if (ptr > pbeg) pre = "..."; } if (ptr_end < pend) { - ptr_end = rb_enc_prev_char(pt, ptr_end, pend, rb_enc_get(str)); + ptr_end = rb_enc_prev_char(pt, ptr_end, pend, rb_parser_str_get_encoding(str)); if (ptr_end < pend) post = "..."; } } @@ -6528,7 +7368,7 @@ ruby_show_error_line(VALUE errbuf, const YYLTYPE *yylloc, int lineno, VALUE str) rb_str_cat_cstr(mesg, "\n"); } else { - mesg = rb_enc_str_new(0, 0, rb_enc_get(str)); + mesg = rb_enc_str_new(0, 0, rb_parser_str_get_encoding(str)); } if (!errbuf && rb_stderr_tty_p()) { #define CSI_BEGIN "\033[" @@ -6574,6 +7414,7 @@ ruby_show_error_line(VALUE errbuf, const YYLTYPE *yylloc, int lineno, VALUE str) if (!errbuf) rb_write_error_str(mesg); } #else + static int parser_yyerror(struct parser_params *p, const YYLTYPE *yylloc, const char *msg) { @@ -6607,7 +7448,6 @@ parser_show_error_line(struct parser_params *p, const YYLTYPE *yylloc) } #endif /* !RIPPER */ -#ifndef RIPPER static int vtable_size(const struct vtable *tbl) { @@ -6618,7 +7458,6 @@ vtable_size(const struct vtable *tbl) return 0; } } -#endif static struct vtable * vtable_alloc_gen(struct parser_params *p, int line, struct vtable *prev) @@ -6677,7 +7516,6 @@ vtable_add_gen(struct parser_params *p, int line, const char *name, } #define vtable_add(tbl, id) vtable_add_gen(p, __LINE__, #tbl, tbl, id) -#ifndef RIPPER static void vtable_pop_gen(struct parser_params *p, int line, const char *name, struct vtable *tbl, int n) @@ -6693,7 +7531,6 @@ vtable_pop_gen(struct parser_params *p, int line, const char *name, tbl->pos -= n; } #define vtable_pop(tbl, n) vtable_pop_gen(p, __LINE__, #tbl, tbl, n) -#endif static int vtable_included(const struct vtable * tbl, ID id) @@ -6712,60 +7549,29 @@ vtable_included(const struct vtable * tbl, ID id) static void parser_prepare(struct parser_params *p); -#ifndef RIPPER -static NODE *parser_append_options(struct parser_params *p, NODE *node); - -static VALUE -debug_lines(VALUE fname) -{ - ID script_lines; - CONST_ID(script_lines, "SCRIPT_LINES__"); - if (rb_const_defined_at(rb_cObject, script_lines)) { - VALUE hash = rb_const_get_at(rb_cObject, script_lines); - if (RB_TYPE_P(hash, T_HASH)) { - VALUE lines = rb_ary_new(); - rb_hash_aset(hash, fname, lines); - return lines; - } - } - return 0; -} - static int e_option_supplied(struct parser_params *p) { return strcmp(p->ruby_sourcefile, "-e") == 0; } +#ifndef RIPPER +static NODE *parser_append_options(struct parser_params *p, NODE *node); + static VALUE yycompile0(VALUE arg) { int n; NODE *tree; struct parser_params *p = (struct parser_params *)arg; - VALUE cov = Qfalse; + int cov = FALSE; - if (!compile_for_eval && !NIL_P(p->ruby_sourcefile_string)) { - p->debug_lines = debug_lines(p->ruby_sourcefile_string); - if (p->debug_lines && p->ruby_sourceline > 0) { - VALUE str = rb_default_rs; - n = p->ruby_sourceline; - do { - rb_ary_push(p->debug_lines, str); - } while (--n); - } - - if (!e_option_supplied(p)) { - cov = Qtrue; - } + if (!compile_for_eval && p->ruby_sourcefile_string && !e_option_supplied(p)) { + cov = TRUE; } - if (p->keep_script_lines || ruby_vm_keep_script_lines) { - if (!p->debug_lines) { - p->debug_lines = rb_ary_new(); - } - - RB_OBJ_WRITE(p->ast, &p->ast->body.script_lines, p->debug_lines); + if (p->debug_lines) { + p->ast->body.script_lines = p->debug_lines; } parser_prepare(p); @@ -6776,14 +7582,16 @@ yycompile0(VALUE arg) RUBY_DTRACE_PARSE_HOOK(BEGIN); n = yyparse(p); RUBY_DTRACE_PARSE_HOOK(END); + p->debug_lines = 0; + xfree(p->lex.strterm); p->lex.strterm = 0; p->lex.pcur = p->lex.pbeg = p->lex.pend = 0; if (n || p->error_p) { VALUE mesg = p->error_buffer; if (!mesg) { - mesg = rb_class_new_instance(0, 0, rb_eSyntaxError); + mesg = syntax_error_new(); } if (!p->error_tolerant) { rb_set_errinfo(mesg); @@ -6795,45 +7603,64 @@ yycompile0(VALUE arg) tree = NEW_NIL(&NULL_LOC); } else { - VALUE opt = p->compile_option; - VALUE tokens = p->tokens; + rb_parser_ary_t *tokens = p->tokens; NODE *prelude; - NODE *body = parser_append_options(p, tree->nd_body); - if (!opt) opt = rb_obj_hide(rb_ident_hash_new()); - rb_hash_aset(opt, rb_sym_intern_ascii_cstr("coverage_enabled"), cov); + NODE *body = parser_append_options(p, RNODE_SCOPE(tree)->nd_body); prelude = block_append(p, p->eval_tree_begin, body); - tree->nd_body = prelude; - RB_OBJ_WRITE(p->ast, &p->ast->body.compile_option, opt); + RNODE_SCOPE(tree)->nd_body = prelude; + p->ast->body.frozen_string_literal = p->frozen_string_literal; + p->ast->body.coverage_enabled = cov; if (p->keep_tokens) { - rb_obj_freeze(tokens); - rb_ast_set_tokens(p->ast, tokens); + p->ast->node_buffer->tokens = tokens; + p->tokens = NULL; } } p->ast->body.root = tree; - if (!p->ast->body.script_lines) p->ast->body.script_lines = INT2FIX(p->line_count); + p->ast->body.line_count = p->line_count; return TRUE; } +static void +set_arg_error(struct parser_params *p, const char *err) +{ + VALUE excargs[3]; + + excargs[0] = rb_eArgError; + excargs[1] = rb_str_new_cstr(err); + excargs[2] = rb_make_backtrace(); + rb_set_errinfo(rb_make_exception(3, excargs)); +} + static rb_ast_t * -yycompile(VALUE vparser, struct parser_params *p, VALUE fname, int line) +yycompile(struct parser_params *p, const char *fname_ptr, long fname_len, rb_encoding *fname_enc, int line) { rb_ast_t *ast; - if (NIL_P(fname)) { - p->ruby_sourcefile_string = Qnil; + if (!fname_ptr) { + p->ruby_sourcefile_string = NULL; p->ruby_sourcefile = "(none)"; } else { - p->ruby_sourcefile_string = rb_fstring(fname); - p->ruby_sourcefile = StringValueCStr(fname); + int w; + if (cstr_null_check(p, fname_ptr, fname_len, fname_enc, &w)) { + if (w) { + set_arg_error(p, "string contains null char"); + } + else { + set_arg_error(p, "string contains null byte"); + } + return rb_ast_new(); + } + + p->ruby_sourcefile_string = rb_parser_encoding_string_new(p, fname_ptr, fname_len, fname_enc); + p->ruby_sourcefile = fname_ptr; } p->ruby_sourceline = line - 1; p->lvtbl = NULL; p->ast = ast = rb_ast_new(); - rb_suppress_tracing(yycompile0, (VALUE)p); + compile_callback(yycompile0, (VALUE)p); p->ast = 0; - RB_GC_GUARD(vparser); /* prohibit tail call optimization */ while (p->lvtbl) { local_pop(p); @@ -6844,118 +7671,36 @@ yycompile(VALUE vparser, struct parser_params *p, VALUE fname, int line) #endif /* !RIPPER */ static rb_encoding * -must_be_ascii_compatible(VALUE s) +must_be_ascii_compatible(struct parser_params *p, rb_parser_string_t *s) { - rb_encoding *enc = rb_enc_get(s); + rb_encoding *enc = rb_parser_str_get_encoding(s); if (!rb_enc_asciicompat(enc)) { rb_raise(rb_eArgError, "invalid source encoding"); } return enc; } -static VALUE -lex_get_str(struct parser_params *p, VALUE s) -{ - char *beg, *end, *start; - long len; - - beg = RSTRING_PTR(s); - len = RSTRING_LEN(s); - start = beg; - if (p->lex.gets_.ptr) { - if (len == p->lex.gets_.ptr) return Qnil; - beg += p->lex.gets_.ptr; - len -= p->lex.gets_.ptr; - } - end = memchr(beg, '\n', len); - if (end) len = ++end - beg; - p->lex.gets_.ptr += len; - return rb_str_subseq(s, beg - start, len); -} - -static VALUE +static rb_parser_string_t * lex_getline(struct parser_params *p) { - VALUE line = (*p->lex.gets)(p, p->lex.input); - if (NIL_P(line)) return line; - must_be_ascii_compatible(line); - if (RB_OBJ_FROZEN(line)) line = rb_str_dup(line); // needed for RubyVM::AST.of because script_lines in iseq is deep-frozen + rb_parser_string_t *line = (*p->lex.gets)(p, p->lex.input, p->line_count); + if (!line) return 0; p->line_count++; + string_buffer_append(p, line); + must_be_ascii_compatible(p, line); return line; } -static const rb_data_type_t parser_data_type; - #ifndef RIPPER -static rb_ast_t* -parser_compile_string(VALUE vparser, VALUE fname, VALUE s, int line) -{ - struct parser_params *p; - - TypedData_Get_Struct(vparser, struct parser_params, &parser_data_type, p); - - p->lex.gets = lex_get_str; - p->lex.gets_.ptr = 0; - p->lex.input = rb_str_new_frozen(s); - p->lex.pbeg = p->lex.pcur = p->lex.pend = 0; - - return yycompile(vparser, p, fname, line); -} - -rb_ast_t* -rb_parser_compile_string(VALUE vparser, const char *f, VALUE s, int line) -{ - return rb_parser_compile_string_path(vparser, rb_filesystem_str_new_cstr(f), s, line); -} - rb_ast_t* -rb_parser_compile_string_path(VALUE vparser, VALUE f, VALUE s, int line) -{ - must_be_ascii_compatible(s); - return parser_compile_string(vparser, f, s, line); -} - -VALUE rb_io_gets_internal(VALUE io); - -static VALUE -lex_io_gets(struct parser_params *p, VALUE io) +rb_parser_compile(rb_parser_t *p, rb_parser_lex_gets_func *gets, + const char *fname_ptr, long fname_len, rb_encoding *fname_enc, rb_parser_input_data input, int line) { - return rb_io_gets_internal(io); -} - -rb_ast_t* -rb_parser_compile_file_path(VALUE vparser, VALUE fname, VALUE file, int start) -{ - struct parser_params *p; - - TypedData_Get_Struct(vparser, struct parser_params, &parser_data_type, p); - - p->lex.gets = lex_io_gets; - p->lex.input = file; - p->lex.pbeg = p->lex.pcur = p->lex.pend = 0; - - return yycompile(vparser, p, fname, start); -} - -static VALUE -lex_generic_gets(struct parser_params *p, VALUE input) -{ - return (*p->lex.gets_.call)(input, p->line_count); -} - -rb_ast_t* -rb_parser_compile_generic(VALUE vparser, VALUE (*lex_gets)(VALUE, int), VALUE fname, VALUE input, int start) -{ - struct parser_params *p; - - TypedData_Get_Struct(vparser, struct parser_params, &parser_data_type, p); - - p->lex.gets = lex_generic_gets; - p->lex.gets_.call = lex_gets; + p->lex.gets = gets; p->lex.input = input; p->lex.pbeg = p->lex.pcur = p->lex.pend = 0; - return yycompile(vparser, p, fname, start); + return yycompile(p, fname_ptr, fname_len, fname_enc, line); } #endif /* !RIPPER */ @@ -6982,7 +7727,7 @@ enum string_type { }; static VALUE -parser_str_new(const char *ptr, long len, rb_encoding *enc, int func, rb_encoding *enc0) +parser_str_new(struct parser_params *p, const char *ptr, long len, rb_encoding *enc, int func, rb_encoding *enc0) { VALUE str; @@ -6990,7 +7735,7 @@ parser_str_new(const char *ptr, long len, rb_encoding *enc, int func, rb_encodin if (!(func & STR_FUNC_REGEXP) && rb_enc_asciicompat(enc)) { if (is_ascii_string(str)) { } - else if (rb_is_usascii_enc(enc0) && enc != rb_utf8_encoding()) { + else if (rb_is_usascii_enc((void *)enc0) && enc != rb_utf8_encoding()) { rb_enc_associate(str, rb_ascii8bit_encoding()); } } @@ -6998,6 +7743,30 @@ parser_str_new(const char *ptr, long len, rb_encoding *enc, int func, rb_encodin return str; } +static int +strterm_is_heredoc(rb_strterm_t *strterm) +{ + return strterm->heredoc; +} + +static rb_strterm_t * +new_strterm(struct parser_params *p, int func, int term, int paren) +{ + rb_strterm_t *strterm = ZALLOC(rb_strterm_t); + strterm->u.literal.func = func; + strterm->u.literal.term = term; + strterm->u.literal.paren = paren; + return strterm; +} + +static rb_strterm_t * +new_heredoc(struct parser_params *p) +{ + rb_strterm_t *strterm = ZALLOC(rb_strterm_t); + strterm->heredoc = true; + return strterm; +} + #define peek(p,c) peek_n(p, (c), 0) #define peek_n(p,c,n) (!lex_eol_n_p(p, n) && (c) == (unsigned char)(p)->lex.pcur[n]) #define peekc(p) peekc_n(p, 0) @@ -7011,13 +7780,21 @@ add_delayed_token(struct parser_params *p, const char *tok, const char *end, int #endif if (tok < end) { + if (has_delayed_token(p)) { + bool next_line = parser_string_end_with_newline_p(p, p->delayed.token); + int end_line = (next_line ? 1 : 0) + p->delayed.end_line; + int end_col = (next_line ? 0 : p->delayed.end_col); + if (end_line != p->ruby_sourceline || end_col != tok - p->lex.pbeg) { + dispatch_delayed_token(p, tSTRING_CONTENT); + } + } if (!has_delayed_token(p)) { - p->delayed.token = rb_str_buf_new(end - tok); - rb_enc_associate(p->delayed.token, p->enc); + p->delayed.token = rb_parser_string_new(p, 0, 0); + rb_parser_enc_associate(p, p->delayed.token, p->enc); p->delayed.beg_line = p->ruby_sourceline; p->delayed.beg_col = rb_long2int(tok - p->lex.pbeg); } - rb_str_buf_cat(p->delayed.token, tok, end - tok); + rb_parser_str_buf_cat(p, p->delayed.token, tok, end - tok); p->delayed.end_line = p->ruby_sourceline; p->delayed.end_col = rb_long2int(end - p->lex.pbeg); p->lex.ptok = end; @@ -7025,27 +7802,27 @@ add_delayed_token(struct parser_params *p, const char *tok, const char *end, int } static void -set_lastline(struct parser_params *p, VALUE v) +set_lastline(struct parser_params *p, rb_parser_string_t *str) { - p->lex.pbeg = p->lex.pcur = RSTRING_PTR(v); - p->lex.pend = p->lex.pcur + RSTRING_LEN(v); - p->lex.lastline = v; + p->lex.pbeg = p->lex.pcur = PARSER_STRING_PTR(str); + p->lex.pend = p->lex.pcur + PARSER_STRING_LEN(str); + p->lex.lastline = str; } static int nextline(struct parser_params *p, int set_encoding) { - VALUE v = p->lex.nextline; + rb_parser_string_t *str = p->lex.nextline; p->lex.nextline = 0; - if (!v) { + if (!str) { if (p->eofp) return -1; - if (p->lex.pend > p->lex.pbeg && *(p->lex.pend-1) != '\n') { + if (!lex_eol_ptr_p(p, p->lex.pbeg) && *(p->lex.pend-1) != '\n') { goto end_of_input; } - if (!p->lex.input || NIL_P(v = lex_getline(p))) { + if (!p->lex.input || !(str = lex_getline(p))) { end_of_input: p->eofp = 1; lex_goto_eol(p); @@ -7053,13 +7830,14 @@ nextline(struct parser_params *p, int set_encoding) } #ifndef RIPPER if (p->debug_lines) { - if (set_encoding) rb_enc_associate(v, p->enc); - rb_ary_push(p->debug_lines, v); + if (set_encoding) rb_parser_enc_associate(p, str, p->enc); + rb_parser_string_t *copy = rb_parser_string_deep_copy(p, str); + rb_parser_ary_push_script_line(p, p->debug_lines, copy); } #endif p->cr_seen = FALSE; } - else if (NIL_P(v)) { + else if (str == AFTER_HEREDOC_WITHOUT_TERMINTOR) { /* after here-document without terminator */ goto end_of_input; } @@ -7069,7 +7847,7 @@ nextline(struct parser_params *p, int set_encoding) p->heredoc_end = 0; } p->ruby_sourceline++; - set_lastline(p, v); + set_lastline(p, str); token_flush(p); return 0; } @@ -7089,7 +7867,7 @@ nextc0(struct parser_params *p, int set_encoding) { int c; - if (UNLIKELY((p->lex.pcur == p->lex.pend) || p->eofp || RTEST(p->lex.nextline))) { + if (UNLIKELY(lex_eol_p(p) || p->eofp || p->lex.nextline > AFTER_HEREDOC_WITHOUT_TERMINTOR)) { if (nextline(p, set_encoding)) return -1; } c = (unsigned char)*p->lex.pcur++; @@ -7105,6 +7883,7 @@ static void pushback(struct parser_params *p, int c) { if (c == -1) return; + p->eofp = 0; p->lex.pcur--; if (p->lex.pcur > p->lex.pbeg && p->lex.pcur[0] == '\n' && p->lex.pcur[-1] == '\r') { p->lex.pcur--; @@ -7121,7 +7900,7 @@ static int looking_at_eol_p(struct parser_params *p) { const char *ptr = p->lex.pcur; - while (ptr < p->lex.pend) { + while (!lex_eol_ptr_p(p, ptr)) { int c = (unsigned char)*ptr++; int eol = (c == '\n' || c == '#'); if (eol || !ISSPACE(c)) { @@ -7135,7 +7914,6 @@ static char* newtok(struct parser_params *p) { p->tokidx = 0; - p->tokline = p->ruby_sourceline; if (!p->tokenbuf) { p->toksiz = 60; p->tokenbuf = ALLOC_N(char, 60); @@ -7174,10 +7952,10 @@ tok_hex(struct parser_params *p, size_t *numlen) { int c; - c = scan_hex(p->lex.pcur, 2, numlen); + c = (int)ruby_scan_hex(p->lex.pcur, 2, numlen); if (!*numlen) { yyerror0("invalid hex escape"); - token_flush(p); + dispatch_scan_event(p, tSTRING_CONTENT); return 0; } p->lex.pcur += *numlen; @@ -7214,18 +7992,18 @@ escaped_control_code(int c) } #define WARN_SPACE_CHAR(c, prefix) \ - rb_warn1("invalid character syntax; use "prefix"\\%c", WARN_I(c2)) + rb_warn1("invalid character syntax; use "prefix"\\%c", WARN_I(c)) static int tokadd_codepoint(struct parser_params *p, rb_encoding **encp, int regexp_literal, int wide) { size_t numlen; - int codepoint = scan_hex(p->lex.pcur, wide ? p->lex.pend - p->lex.pcur : 4, &numlen); + int codepoint = (int)ruby_scan_hex(p->lex.pcur, wide ? p->lex.pend - p->lex.pcur : 4, &numlen); p->lex.pcur += numlen; if (p->lex.strterm == NULL || - (p->lex.strterm->flags & STRTERM_HEREDOC) || - (p->lex.strterm->u.literal.u1.func != str_regexp)) { + strterm_is_heredoc(p->lex.strterm) || + (p->lex.strterm->u.literal.func != str_regexp)) { if (wide ? (numlen == 0 || numlen > 6) : (numlen < 4)) { literal_flush(p, p->lex.pcur); yyerror0("invalid Unicode escape"); @@ -7262,6 +8040,18 @@ tokadd_codepoint(struct parser_params *p, rb_encoding **encp, return TRUE; } +static int tokadd_mbchar(struct parser_params *p, int c); + +static int +tokskip_mbchar(struct parser_params *p) +{ + int len = parser_precise_mbclen(p, p->lex.pcur-1); + if (len > 0) { + p->lex.pcur += len - 1; + } + return len; +} + /* return value is for ?\u3042 */ static void tokadd_utf8(struct parser_params *p, rb_encoding **encp, @@ -7279,44 +8069,71 @@ tokadd_utf8(struct parser_params *p, rb_encoding **encp, if (regexp_literal) { tokadd(p, '\\'); tokadd(p, 'u'); } if (peek(p, open_brace)) { /* handle \u{...} form */ - const char *second = NULL; - int c, last = nextc(p); - if (p->lex.pcur >= p->lex.pend) goto unterminated; - while (ISSPACE(c = *p->lex.pcur) && ++p->lex.pcur < p->lex.pend); - while (c != close_brace) { - if (c == term) goto unterminated; - if (second == multiple_codepoints) - second = p->lex.pcur; - if (regexp_literal) tokadd(p, last); - if (!tokadd_codepoint(p, encp, regexp_literal, TRUE)) { - break; - } - while (ISSPACE(c = *p->lex.pcur)) { - if (++p->lex.pcur >= p->lex.pend) goto unterminated; - last = c; + if (regexp_literal && p->lex.strterm->u.literal.func == str_regexp) { + /* + * Skip parsing validation code and copy bytes as-is until term or + * closing brace, in order to correctly handle extended regexps where + * invalid unicode escapes are allowed in comments. The regexp parser + * does its own validation and will catch any issues. + */ + tokadd(p, open_brace); + while (!lex_eol_ptr_p(p, ++p->lex.pcur)) { + int c = peekc(p); + if (c == close_brace) { + tokadd(p, c); + ++p->lex.pcur; + break; + } + else if (c == term) { + break; + } + if (c == '\\' && !lex_eol_n_p(p, 1)) { + tokadd(p, c); + c = *++p->lex.pcur; + } + tokadd_mbchar(p, c); } - if (term == -1 && !second) - second = multiple_codepoints; } + else { + const char *second = NULL; + int c, last = nextc(p); + if (lex_eol_p(p)) goto unterminated; + while (ISSPACE(c = peekc(p)) && !lex_eol_ptr_p(p, ++p->lex.pcur)); + while (c != close_brace) { + if (c == term) goto unterminated; + if (second == multiple_codepoints) + second = p->lex.pcur; + if (regexp_literal) tokadd(p, last); + if (!tokadd_codepoint(p, encp, regexp_literal, TRUE)) { + break; + } + while (ISSPACE(c = peekc(p))) { + if (lex_eol_ptr_p(p, ++p->lex.pcur)) goto unterminated; + last = c; + } + if (term == -1 && !second) + second = multiple_codepoints; + } - if (c != close_brace) { - unterminated: - token_flush(p); - yyerror0("unterminated Unicode escape"); - return; - } - if (second && second != multiple_codepoints) { - const char *pcur = p->lex.pcur; - p->lex.pcur = second; - dispatch_scan_event(p, tSTRING_CONTENT); - token_flush(p); - p->lex.pcur = pcur; - yyerror0(multiple_codepoints); - token_flush(p); - } + if (c != close_brace) { + unterminated: + token_flush(p); + yyerror0("unterminated Unicode escape"); + return; + } + if (second && second != multiple_codepoints) { + const char *pcur = p->lex.pcur; + p->lex.pcur = second; + dispatch_scan_event(p, tSTRING_CONTENT); + token_flush(p); + p->lex.pcur = pcur; + yyerror0(multiple_codepoints); + token_flush(p); + } - if (regexp_literal) tokadd(p, close_brace); - nextc(p); + if (regexp_literal) tokadd(p, close_brace); + nextc(p); + } } else { /* handle \uxxxx form */ if (!tokadd_codepoint(p, encp, regexp_literal, FALSE)) { @@ -7330,7 +8147,7 @@ tokadd_utf8(struct parser_params *p, rb_encoding **encp, #define ESCAPE_META 2 static int -read_escape(struct parser_params *p, int flags, rb_encoding **encp) +read_escape(struct parser_params *p, int flags) { int c; size_t numlen; @@ -7363,7 +8180,7 @@ read_escape(struct parser_params *p, int flags, rb_encoding **encp) case '0': case '1': case '2': case '3': /* octal constant */ case '4': case '5': case '6': case '7': pushback(p, c); - c = scan_oct(p->lex.pcur, 3, &numlen); + c = (int)ruby_scan_oct(p->lex.pcur, 3, &numlen); p->lex.pcur += numlen; return c; @@ -7389,7 +8206,7 @@ read_escape(struct parser_params *p, int flags, rb_encoding **encp) nextc(p); goto eof; } - return read_escape(p, flags|ESCAPE_META, encp) | 0x80; + return read_escape(p, flags|ESCAPE_META) | 0x80; } else if (c == -1 || !ISASCII(c)) goto eof; else { @@ -7418,11 +8235,15 @@ read_escape(struct parser_params *p, int flags, rb_encoding **encp) nextc(p); goto eof; } - c = read_escape(p, flags|ESCAPE_CONTROL, encp); + c = read_escape(p, flags|ESCAPE_CONTROL); } else if (c == '?') return 0177; - else if (c == -1 || !ISASCII(c)) goto eof; + else if (c == -1) goto eof; + else if (!ISASCII(c)) { + tokskip_mbchar(p); + goto eof; + } else { int c2 = escaped_control_code(c); if (c2) { @@ -7450,7 +8271,7 @@ read_escape(struct parser_params *p, int flags, rb_encoding **encp) eof: case -1: yyerror0("Invalid escape character syntax"); - token_flush(p); + dispatch_scan_event(p, tSTRING_CONTENT); return '\0'; default: @@ -7466,7 +8287,7 @@ tokaddmbc(struct parser_params *p, int c, rb_encoding *enc) } static int -tokadd_escape(struct parser_params *p, rb_encoding **encp) +tokadd_escape(struct parser_params *p) { int c; size_t numlen; @@ -7507,6 +8328,61 @@ tokadd_escape(struct parser_params *p, rb_encoding **encp) } static int +char_to_option(int c) +{ + int val; + + switch (c) { + case 'i': + val = RE_ONIG_OPTION_IGNORECASE; + break; + case 'x': + val = RE_ONIG_OPTION_EXTEND; + break; + case 'm': + val = RE_ONIG_OPTION_MULTILINE; + break; + default: + val = 0; + break; + } + return val; +} + +#define ARG_ENCODING_FIXED 16 +#define ARG_ENCODING_NONE 32 +#define ENC_ASCII8BIT 1 +#define ENC_EUC_JP 2 +#define ENC_Windows_31J 3 +#define ENC_UTF8 4 + +static int +char_to_option_kcode(int c, int *option, int *kcode) +{ + *option = 0; + + switch (c) { + case 'n': + *kcode = ENC_ASCII8BIT; + return (*option = ARG_ENCODING_NONE); + case 'e': + *kcode = ENC_EUC_JP; + break; + case 's': + *kcode = ENC_Windows_31J; + break; + case 'u': + *kcode = ENC_UTF8; + break; + default: + *kcode = -1; + return (*option = char_to_option(c)); + } + *option = ARG_ENCODING_FIXED; + return 1; +} + +static int regx_options(struct parser_params *p) { int kcode = 0; @@ -7519,9 +8395,9 @@ regx_options(struct parser_params *p) if (c == 'o') { options |= RE_OPTION_ONCE; } - else if (rb_char_to_option_kcode(c, &opt, &kc)) { + else if (char_to_option_kcode(c, &opt, &kc)) { if (kc >= 0) { - if (kc != rb_ascii8bit_encindex()) kcode = c; + if (kc != ENC_ASCII8BIT) kcode = c; kopt = opt; } else { @@ -7590,6 +8466,10 @@ parser_update_heredoc_indent(struct parser_params *p, int c) } p->heredoc_line_indent = -1; } + else { + /* Whitespace only line has no indentation */ + p->heredoc_line_indent = 0; + } } return FALSE; } @@ -7612,6 +8492,13 @@ parser_mixed_escape(struct parser_params *p, const char *beg, rb_encoding *enc1, p->lex.pcur = pos; } +static inline char +nibble_char_upper(unsigned int c) +{ + c &= 0xf; + return c + (c < 10 ? '0' : 'A' - 10); +} + static int tokadd_string(struct parser_params *p, int func, int term, int paren, long *nest, @@ -7650,8 +8537,8 @@ tokadd_string(struct parser_params *p, } --*nest; } - else if ((func & STR_FUNC_EXPAND) && c == '#' && p->lex.pcur < p->lex.pend) { - int c2 = *p->lex.pcur; + else if ((func & STR_FUNC_EXPAND) && c == '#' && !lex_eol_p(p)) { + unsigned char c2 = *p->lex.pcur; if (c2 == '$' || c2 == '@' || c2 == '{') { pushback(p, c); break; @@ -7699,14 +8586,13 @@ tokadd_string(struct parser_params *p, case 'C': case 'M': { pushback(p, c); - c = read_escape(p, 0, enc); + c = read_escape(p, 0); - int i; - char escbuf[5]; - snprintf(escbuf, sizeof(escbuf), "\\x%02X", c); - for (i = 0; i < 4; i++) { - tokadd(p, escbuf[i]); - } + char *t = tokspace(p, rb_strlen_lit("\\x00")); + *t++ = '\\'; + *t++ = 'x'; + *t++ = nibble_char_upper(c >> 4); + *t++ = nibble_char_upper(c); continue; } } @@ -7716,7 +8602,7 @@ tokadd_string(struct parser_params *p, continue; } pushback(p, c); - if ((c = tokadd_escape(p, enc)) < 0) + if ((c = tokadd_escape(p)) < 0) return -1; if (*enc && *enc != *encp) { mixed_escape(p->lex.ptok+2, *enc, *encp); @@ -7726,7 +8612,7 @@ tokadd_string(struct parser_params *p, else if (func & STR_FUNC_EXPAND) { pushback(p, c); if (func & STR_FUNC_ESCAPE) tokadd(p, '\\'); - c = read_escape(p, 0, enc); + c = read_escape(p, 0); } else if ((func & STR_FUNC_QWORDS) && ISSPACE(c)) { /* ignore backslashed spaces in %w */ @@ -7773,45 +8659,15 @@ tokadd_string(struct parser_params *p, return c; } -static inline rb_strterm_t * -new_strterm(VALUE v1, VALUE v2, VALUE v3, VALUE v0) -{ - return (rb_strterm_t*)rb_imemo_new(imemo_parser_strterm, v1, v2, v3, v0); -} - -/* imemo_parser_strterm for literal */ -#define NEW_STRTERM(func, term, paren) \ - new_strterm((VALUE)(func), (VALUE)(paren), (VALUE)(term), 0) +#define NEW_STRTERM(func, term, paren) new_strterm(p, func, term, paren) -#ifdef RIPPER static void flush_string_content(struct parser_params *p, rb_encoding *enc) { - VALUE content = yylval.val; - if (!ripper_is_node_yylval(content)) - content = ripper_new_yylval(p, 0, 0, content); if (has_delayed_token(p)) { ptrdiff_t len = p->lex.pcur - p->lex.ptok; if (len > 0) { - rb_enc_str_buf_cat(p->delayed.token, p->lex.ptok, len, enc); - } - dispatch_delayed_token(p, tSTRING_CONTENT); - p->lex.ptok = p->lex.pcur; - RNODE(content)->nd_rval = yylval.val; - } - dispatch_scan_event(p, tSTRING_CONTENT); - if (yylval.val != content) - RNODE(content)->nd_rval = yylval.val; - yylval.val = content; -} -#else -static void -flush_string_content(struct parser_params *p, rb_encoding *enc) -{ - if (has_delayed_token(p)) { - ptrdiff_t len = p->lex.pcur - p->lex.ptok; - if (len > 0) { - rb_enc_str_buf_cat(p->delayed.token, p->lex.ptok, len, enc); + rb_parser_enc_str_buf_cat(p, p->delayed.token, p->lex.ptok, len, enc); p->delayed.end_line = p->ruby_sourceline; p->delayed.end_col = rb_long2int(p->lex.pcur - p->lex.pbeg); } @@ -7820,9 +8676,8 @@ flush_string_content(struct parser_params *p, rb_encoding *enc) } dispatch_scan_event(p, tSTRING_CONTENT); } -#endif -RUBY_FUNC_EXPORTED const unsigned int ruby_global_name_punct_bits[(0x7e - 0x20 + 31) / 32]; +RUBY_FUNC_EXPORTED const uint_least32_t ruby_global_name_punct_bits[(0x7e - 0x20 + 31) / 32]; /* this can be shared with ripper, since it's independent from struct * parser_params. */ #ifndef RIPPER @@ -7834,7 +8689,7 @@ RUBY_FUNC_EXPORTED const unsigned int ruby_global_name_punct_bits[(0x7e - 0x20 + BIT(':', idx) | BIT('<', idx) | BIT('>', idx) | BIT('\"', idx) | \ BIT('&', idx) | BIT('`', idx) | BIT('\'', idx) | BIT('+', idx) | \ BIT('0', idx)) -const unsigned int ruby_global_name_punct_bits[] = { +const uint_least32_t ruby_global_name_punct_bits[] = { SPECIAL_PUNCT(0), SPECIAL_PUNCT(1), SPECIAL_PUNCT(2), @@ -7849,12 +8704,12 @@ parser_peek_variable_name(struct parser_params *p) int c; const char *ptr = p->lex.pcur; - if (ptr + 1 >= p->lex.pend) return 0; + if (lex_eol_ptr_n_p(p, ptr, 1)) return 0; c = *ptr++; switch (c) { case '$': if ((c = *ptr) == '-') { - if (++ptr >= p->lex.pend) return 0; + if (lex_eol_ptr_p(p, ++ptr)) return 0; c = *ptr; } else if (is_global_name_punct(c) || ISDIGIT(c)) { @@ -7863,13 +8718,14 @@ parser_peek_variable_name(struct parser_params *p) break; case '@': if ((c = *ptr) == '@') { - if (++ptr >= p->lex.pend) return 0; + if (lex_eol_ptr_p(p, ++ptr)) return 0; c = *ptr; } break; case '{': p->lex.pcur = ptr; p->command_start = TRUE; + yylval.state = p->lex.state; return tSTRING_DBEG; default: return 0; @@ -7892,6 +8748,7 @@ parser_peek_variable_name(struct parser_params *p) static inline enum yytokentype parser_string_term(struct parser_params *p, int func) { + xfree(p->lex.strterm); p->lex.strterm = 0; if (func & STR_FUNC_REGEXP) { set_yylval_num(regx_options(p)); @@ -7911,9 +8768,9 @@ parser_string_term(struct parser_params *p, int func) static enum yytokentype parse_string(struct parser_params *p, rb_strterm_literal_t *quote) { - int func = (int)quote->u1.func; - int term = (int)quote->u3.term; - int paren = (int)quote->u2.paren; + int func = quote->func; + int term = quote->term; + int paren = quote->paren; int c, space = 0; rb_encoding *enc = p->enc; rb_encoding *base_enc = 0; @@ -7922,6 +8779,7 @@ parse_string(struct parser_params *p, rb_strterm_literal_t *quote) if (func & STR_FUNC_TERM) { if (func & STR_FUNC_QWORDS) nextc(p); /* delayed term */ SET_LEX_STATE(EXPR_END); + xfree(p->lex.strterm); p->lex.strterm = 0; return func & STR_FUNC_REGEXP ? tREGEXP_END : tSTRING_END; } @@ -7931,12 +8789,12 @@ parse_string(struct parser_params *p, rb_strterm_literal_t *quote) space = 1; } if (func & STR_FUNC_LIST) { - quote->u1.func &= ~STR_FUNC_LIST; + quote->func &= ~STR_FUNC_LIST; space = 1; } - if (c == term && !quote->u0.nest) { + if (c == term && !quote->nest) { if (func & STR_FUNC_QWORDS) { - quote->u1.func |= STR_FUNC_TERM; + quote->func |= STR_FUNC_TERM; pushback(p, c); /* dispatch the term at tSTRING_END */ add_delayed_token(p, p->lex.ptok, p->lex.pcur, __LINE__); return ' '; @@ -7950,24 +8808,25 @@ parse_string(struct parser_params *p, rb_strterm_literal_t *quote) } newtok(p); if ((func & STR_FUNC_EXPAND) && c == '#') { - int t = parser_peek_variable_name(p); + enum yytokentype t = parser_peek_variable_name(p); if (t) return t; tokadd(p, '#'); c = nextc(p); } pushback(p, c); - if (tokadd_string(p, func, term, paren, "e->u0.nest, + if (tokadd_string(p, func, term, paren, "e->nest, &enc, &base_enc) == -1) { if (p->eofp) { #ifndef RIPPER # define unterminated_literal(mesg) yyerror0(mesg) #else -# define unterminated_literal(mesg) compile_error(p, mesg) +# define unterminated_literal(mesg) compile_error(p, mesg) #endif literal_flush(p, p->lex.pcur); if (func & STR_FUNC_QWORDS) { /* no content to add, bailing out here */ unterminated_literal("unterminated list meets end of file"); + xfree(p->lex.strterm); p->lex.strterm = 0; return tSTRING_END; } @@ -7977,7 +8836,7 @@ parse_string(struct parser_params *p, rb_strterm_literal_t *quote) else { unterminated_literal("unterminated string meets end of file"); } - quote->u1.func |= STR_FUNC_TERM; + quote->func |= STR_FUNC_TERM; } } @@ -8058,14 +8917,14 @@ heredoc_identifier(struct parser_params *p) dispatch_scan_event(p, tHEREDOC_BEG); lex_goto_eol(p); - p->lex.strterm = new_strterm(0, 0, 0, p->lex.lastline); - p->lex.strterm->flags |= STRTERM_HEREDOC; + p->lex.strterm = new_heredoc(p); rb_strterm_heredoc_t *here = &p->lex.strterm->u.heredoc; here->offset = offset; here->sourceline = p->ruby_sourceline; - here->length = (int)len; + here->length = (unsigned)len; here->quote = quote; here->func = func; + here->lastline = p->lex.lastline; token_flush(p); p->heredoc_indent = indent; @@ -8076,29 +8935,28 @@ heredoc_identifier(struct parser_params *p) static void heredoc_restore(struct parser_params *p, rb_strterm_heredoc_t *here) { - VALUE line; + rb_parser_string_t *line; + rb_strterm_t *term = p->lex.strterm; p->lex.strterm = 0; line = here->lastline; p->lex.lastline = line; - p->lex.pbeg = RSTRING_PTR(line); - p->lex.pend = p->lex.pbeg + RSTRING_LEN(line); + p->lex.pbeg = PARSER_STRING_PTR(line); + p->lex.pend = p->lex.pbeg + PARSER_STRING_LEN(line); p->lex.pcur = p->lex.pbeg + here->offset + here->length + here->quote; p->lex.ptok = p->lex.pbeg + here->offset - here->quote; p->heredoc_end = p->ruby_sourceline; p->ruby_sourceline = (int)here->sourceline; - if (p->eofp) p->lex.nextline = Qnil; + if (p->eofp) p->lex.nextline = AFTER_HEREDOC_WITHOUT_TERMINTOR; p->eofp = 0; + xfree(term); } static int -dedent_string(VALUE string, int width) +dedent_string_column(const char *str, long len, int width) { - char *str; - long len; int i, col = 0; - RSTRING_GETMEM(string, str, len); for (i = 0; i < len && col < width; i++) { if (str[i] == ' ') { col++; @@ -8112,35 +8970,49 @@ dedent_string(VALUE string, int width) break; } } + + return i; +} + +static int +dedent_string(struct parser_params *p, rb_parser_string_t *string, int width) +{ + char *str; + long len; + int i; + + len = PARSER_STRING_LEN(string); + str = PARSER_STRING_PTR(string); + + i = dedent_string_column(str, len, width); if (!i) return 0; - rb_str_modify(string); - str = RSTRING_PTR(string); - if (RSTRING_LEN(string) != len) - rb_fatal("literal string changed: %+"PRIsVALUE, string); + + rb_parser_str_modify(string); + str = PARSER_STRING_PTR(string); + if (PARSER_STRING_LEN(string) != len) + rb_fatal("literal string changed: %s", PARSER_STRING_PTR(string)); MEMMOVE(str, str + i, char, len - i); - rb_str_set_len(string, len - i); + rb_parser_str_set_len(p, string, len - i); return i; } -#ifndef RIPPER static NODE * heredoc_dedent(struct parser_params *p, NODE *root) { NODE *node, *str_node, *prev_node; int indent = p->heredoc_indent; - VALUE prev_lit = 0; + rb_parser_string_t *prev_lit = 0; if (indent <= 0) return root; - p->heredoc_indent = 0; if (!root) return root; prev_node = node = str_node = root; - if (nd_type_p(root, NODE_LIST)) str_node = root->nd_head; + if (nd_type_p(root, NODE_LIST)) str_node = RNODE_LIST(root)->nd_head; while (str_node) { - VALUE lit = str_node->nd_lit; - if (str_node->flags & NODE_FL_NEWLINE) { - dedent_string(lit, indent); + rb_parser_string_t *lit = RNODE_STR(str_node)->string; + if (nd_fl_newline(str_node)) { + dedent_string(p, lit, indent); } if (!prev_lit) { prev_lit = lit; @@ -8149,22 +9021,22 @@ heredoc_dedent(struct parser_params *p, NODE *root) return 0; } else { - NODE *end = node->nd_end; - node = prev_node->nd_next = node->nd_next; + NODE *end = RNODE_LIST(node)->as.nd_end; + node = RNODE_LIST(prev_node)->nd_next = RNODE_LIST(node)->nd_next; if (!node) { if (nd_type_p(prev_node, NODE_DSTR)) nd_set_type(prev_node, NODE_STR); break; } - node->nd_end = end; + RNODE_LIST(node)->as.nd_end = end; goto next_str; } str_node = 0; - while ((node = (prev_node = node)->nd_next) != 0) { + while ((nd_type_p(node, NODE_LIST) || nd_type_p(node, NODE_DSTR)) && (node = RNODE_LIST(prev_node = node)->nd_next) != 0) { next_str: if (!nd_type_p(node, NODE_LIST)) break; - if ((str_node = node->nd_head) != 0) { + if ((str_node = RNODE_LIST(node)->nd_head) != 0) { enum node_type type = nd_type(str_node); if (type == NODE_STR || type == NODE_DSTR) break; prev_lit = 0; @@ -8174,38 +9046,6 @@ heredoc_dedent(struct parser_params *p, NODE *root) } return root; } -#else /* RIPPER */ -static VALUE -heredoc_dedent(struct parser_params *p, VALUE array) -{ - int indent = p->heredoc_indent; - - if (indent <= 0) return array; - p->heredoc_indent = 0; - dispatch2(heredoc_dedent, array, INT2NUM(indent)); - return array; -} - -/* - * call-seq: - * Ripper.dedent_string(input, width) -> Integer - * - * USE OF RIPPER LIBRARY ONLY. - * - * Strips up to +width+ leading whitespaces from +input+, - * and returns the stripped column width. - */ -static VALUE -parser_dedent_string(VALUE self, VALUE input, VALUE width) -{ - int wid, col; - - StringValue(input); - wid = NUM2UINT(width); - col = dedent_string(input, wid); - return INT2NUM(col); -} -#endif static int whole_match_p(struct parser_params *p, const char *eos, long len, int indent) @@ -8229,7 +9069,7 @@ static int word_match_p(struct parser_params *p, const char *word, long len) { if (strncmp(p->lex.pcur, word, len)) return 0; - if (p->lex.pcur + len == p->lex.pend) return 1; + if (lex_eol_n_p(p, len)) return 1; int c = (unsigned char)p->lex.pcur[len]; if (ISSPACE(c)) return 1; switch (c) { @@ -8273,27 +9113,41 @@ number_literal_suffix(struct parser_params *p, int mask) } static enum yytokentype -set_number_literal(struct parser_params *p, VALUE v, - enum yytokentype type, int suffix) +set_number_literal(struct parser_params *p, enum yytokentype type, int suffix, int base, int seen_point) { - if (suffix & NUM_SUFFIX_I) { - v = rb_complex_raw(INT2FIX(0), v); - type = tIMAGINARY; + enum rb_numeric_type numeric_type = integer_literal; + + if (type == tFLOAT) { + numeric_type = float_literal; } - set_yylval_literal(v); - SET_LEX_STATE(EXPR_END); - return type; -} -static enum yytokentype -set_integer_literal(struct parser_params *p, VALUE v, int suffix) -{ - enum yytokentype type = tINTEGER; if (suffix & NUM_SUFFIX_R) { - v = rb_rational_raw1(v); type = tRATIONAL; + numeric_type = rational_literal; + } + if (suffix & NUM_SUFFIX_I) { + type = tIMAGINARY; + } + + switch (type) { + case tINTEGER: + set_yylval_node(NEW_INTEGER(strdup(tok(p)), base, &_cur_loc)); + break; + case tFLOAT: + set_yylval_node(NEW_FLOAT(strdup(tok(p)), &_cur_loc)); + break; + case tRATIONAL: + set_yylval_node(NEW_RATIONAL(strdup(tok(p)), base, seen_point, &_cur_loc)); + break; + case tIMAGINARY: + set_yylval_node(NEW_IMAGINARY(strdup(tok(p)), base, seen_point, numeric_type, &_cur_loc)); + (void)numeric_type; /* for ripper */ + break; + default: + rb_bug("unexpected token: %d", type); } - return set_number_literal(p, v, type, suffix); + SET_LEX_STATE(EXPR_END); + return type; } #ifdef RIPPER @@ -8319,7 +9173,7 @@ parser_dispatch_heredoc_end(struct parser_params *p, int line) dispatch_delayed_token(p, tSTRING_CONTENT); if (p->keep_tokens) { - VALUE str = STR_NEW(p->lex.ptok, p->lex.pend - p->lex.ptok); + rb_parser_string_t *str = rb_parser_encoding_string_new(p, p->lex.ptok, p->lex.pend - p->lex.ptok, p->enc); RUBY_SET_YYLLOC_OF_HEREDOC_END(*p->yylloc); parser_append_tokens(p, str, tHEREDOC_END, line); } @@ -8340,8 +9194,11 @@ here_document(struct parser_params *p, rb_strterm_heredoc_t *here) rb_encoding *enc = p->enc; rb_encoding *base_enc = 0; int bol; +#ifdef RIPPER + VALUE s_value; +#endif - eos = RSTRING_PTR(here->lastline) + here->offset; + eos = PARSER_STRING_PTR(here->lastline) + here->offset; len = here->length; indent = (func = here->func) & STR_FUNC_INDENT; @@ -8362,7 +9219,7 @@ here_document(struct parser_params *p, rb_strterm_heredoc_t *here) enc = rb_ascii8bit_encoding(); } } - rb_enc_str_buf_cat(p->delayed.token, p->lex.ptok, len, enc); + rb_parser_enc_str_buf_cat(p, p->delayed.token, p->lex.ptok, len, enc); } dispatch_delayed_token(p, tSTRING_CONTENT); } @@ -8372,7 +9229,6 @@ here_document(struct parser_params *p, rb_strterm_heredoc_t *here) compile_error(p, "can't find string \"%.*s\" anywhere before EOF", (int)len, eos); token_flush(p); - p->lex.strterm = 0; SET_LEX_STATE(EXPR_END); return tSTRING_END; } @@ -8392,14 +9248,13 @@ here_document(struct parser_params *p, rb_strterm_heredoc_t *here) restore: heredoc_restore(p, &p->lex.strterm->u.heredoc); token_flush(p); - p->lex.strterm = 0; SET_LEX_STATE(EXPR_END); return tSTRING_END; } if (!(func & STR_FUNC_EXPAND)) { do { - ptr = RSTRING_PTR(p->lex.lastline); + ptr = PARSER_STRING_PTR(p->lex.lastline); ptr_end = p->lex.pend; if (ptr_end > ptr) { switch (ptr_end[-1]) { @@ -8424,7 +9279,7 @@ here_document(struct parser_params *p, rb_strterm_heredoc_t *here) rb_str_cat(str, ptr, ptr_end - ptr); else str = STR_NEW(ptr, ptr_end - ptr); - if (ptr_end < p->lex.pend) rb_str_cat(str, "\n", 1); + if (!lex_eol_ptr_p(p, ptr_end)) rb_str_cat(str, "\n", 1); lex_goto_eol(p); if (p->heredoc_indent > 0) { goto flush_str; @@ -8441,7 +9296,7 @@ here_document(struct parser_params *p, rb_strterm_heredoc_t *here) /* int mb = ENC_CODERANGE_7BIT, *mbp = &mb;*/ newtok(p); if (c == '#') { - int t = parser_peek_variable_name(p); + enum yytokentype t = parser_peek_variable_name(p); if (p->heredoc_line_indent != -1) { if (p->heredoc_indent > p->heredoc_line_indent) { p->heredoc_indent = p->heredoc_line_indent; @@ -8466,7 +9321,7 @@ here_document(struct parser_params *p, rb_strterm_heredoc_t *here) flush_str: set_yylval_str(str); #ifndef RIPPER - if (bol) yylval.node->flags |= NODE_FL_NEWLINE; + if (bol) nd_set_fl_newline(yylval.node); #endif flush_string_content(p, enc); return tSTRING_CONTENT; @@ -8482,16 +9337,20 @@ here_document(struct parser_params *p, rb_strterm_heredoc_t *here) str = STR_NEW3(tok(p), toklen(p), enc, func); } dispatch_heredoc_end(p); -#ifdef RIPPER - str = ripper_new_yylval(p, ripper_token2eventid(tSTRING_CONTENT), - yylval.val, str); -#endif heredoc_restore(p, &p->lex.strterm->u.heredoc); token_flush(p); p->lex.strterm = NEW_STRTERM(func | STR_FUNC_TERM, 0, 0); +#ifdef RIPPER + /* Preserve s_value for set_yylval_str */ + s_value = p->s_value; +#endif set_yylval_str(str); +#ifdef RIPPER + set_parser_s_value(s_value); +#endif + #ifndef RIPPER - if (bol) yylval.node->flags |= NODE_FL_NEWLINE; + if (bol) nd_set_fl_newline(yylval.node); #endif return tSTRING_CONTENT; } @@ -8503,10 +9362,10 @@ arg_ambiguous(struct parser_params *p, char c) { #ifndef RIPPER if (c == '/') { - rb_warning1("ambiguity between regexp and two divisions: wrap regexp in parentheses or add a space after `%c' operator", WARN_I(c)); + rb_warning1("ambiguity between regexp and two divisions: wrap regexp in parentheses or add a space after '%c' operator", WARN_I(c)); } else { - rb_warning1("ambiguous first argument; put parentheses or a space even after `%c' operator", WARN_I(c)); + rb_warning1("ambiguous first argument; put parentheses or a space even after '%c' operator", WARN_I(c)); } #else dispatch1(arg_ambiguous, rb_usascii_str_new(&c, 1)); @@ -8515,22 +9374,12 @@ arg_ambiguous(struct parser_params *p, char c) } static ID -#ifndef RIPPER -formal_argument(struct parser_params *p, ID lhs) -#else -formal_argument(struct parser_params *p, VALUE lhs) -#endif +formal_argument(struct parser_params *p, ID id) { - ID id = get_id(lhs); - switch (id_type(id)) { case ID_LOCAL: break; -#ifndef RIPPER -# define ERR(mesg) yyerror0(mesg) -#else -# define ERR(mesg) (dispatch2(param_error, WARN_S(mesg), lhs), ripper_error(p)) -#endif +#define ERR(mesg) yyerror0(mesg) case ID_CONST: ERR("formal argument cannot be a constant"); return 0; @@ -8549,8 +9398,47 @@ formal_argument(struct parser_params *p, VALUE lhs) #undef ERR } shadowing_lvar(p, id); + +/* + * Workaround for Prism::ParseTest#test_filepath for "unparser/corpus/literal/def.txt" + * + * See the discussion on https://github.com/ruby/ruby/pull/9923 + */ +#ifndef RIPPER + return id; +#else + return 0; +#endif +} + +#ifdef RIPPER +static VALUE +ripper_formal_argument(struct parser_params *p, ID id, VALUE lhs) +{ + switch (id_type(id)) { + case ID_LOCAL: + break; +#define ERR(mesg) (dispatch2(param_error, WARN_S(mesg), lhs), ripper_error(p)) + case ID_CONST: + ERR("formal argument cannot be a constant"); + break; + case ID_INSTANCE: + ERR("formal argument cannot be an instance variable"); + break; + case ID_GLOBAL: + ERR("formal argument cannot be a global variable"); + break; + case ID_CLASS: + ERR("formal argument cannot be a class variable"); + break; + default: + ERR("formal argument must be local variable"); + break; +#undef ERR + } return lhs; } +#endif static int lvar_defined(struct parser_params *p, ID id) @@ -8582,17 +9470,29 @@ parser_encode_length(struct parser_params *p, const char *name, long len) static void parser_set_encode(struct parser_params *p, const char *name) { - int idx = rb_enc_find_index(name); rb_encoding *enc; VALUE excargs[3]; + int idx = 0; + const char *wrong = 0; + switch (*name) { + case 'e': case 'E': wrong = "external"; break; + case 'i': case 'I': wrong = "internal"; break; + case 'f': case 'F': wrong = "filesystem"; break; + case 'l': case 'L': wrong = "locale"; break; + } + if (wrong && STRCASECMP(name, wrong) == 0) goto unknown; + idx = rb_enc_find_index(name); if (idx < 0) { + unknown: excargs[1] = rb_sprintf("unknown encoding name: %s", name); error: excargs[0] = rb_eArgError; excargs[2] = rb_make_backtrace(); - rb_ary_unshift(excargs[2], rb_sprintf("%"PRIsVALUE":%d", p->ruby_sourcefile_string, p->ruby_sourceline)); - rb_exc_raise(rb_make_exception(3, excargs)); + rb_ary_unshift(excargs[2], rb_sprintf("%"PRIsVALUE":%d", rb_str_new_mutable_parser_string(p->ruby_sourcefile_string), p->ruby_sourceline)); + VALUE exc = rb_make_exception(3, excargs); + ruby_show_error_line(p, exc, &(YYLTYPE)RUBY_INIT_YYLLOC(), p->ruby_sourceline, p->lex.lastline); + rb_exc_raise(exc); } enc = rb_enc_from_index(idx); if (!rb_enc_asciicompat(enc)) { @@ -8602,25 +9502,19 @@ parser_set_encode(struct parser_params *p, const char *name) p->enc = enc; #ifndef RIPPER if (p->debug_lines) { - VALUE lines = p->debug_lines; - long i, n = RARRAY_LEN(lines); - for (i = 0; i < n; ++i) { - rb_enc_associate_index(RARRAY_AREF(lines, i), idx); + long i; + for (i = 0; i < p->debug_lines->len; i++) { + rb_parser_enc_associate(p, p->debug_lines->data[i], enc); } } #endif } -static int +static bool comment_at_top(struct parser_params *p) { - const char *ptr = p->lex.pbeg, *ptr_end = p->lex.pcur - 1; - if (p->line_count != (p->has_shebang ? 2 : 1)) return 0; - while (ptr < ptr_end) { - if (!ISSPACE(*ptr)) return 0; - ptr++; - } - return 1; + if (p->token_seen) return false; + return (p->line_count == (p->has_shebang ? 2 : 1)); } typedef long (*rb_magic_comment_length_t)(struct parser_params *p, const char *name, long len); @@ -8670,22 +9564,19 @@ parser_set_token_info(struct parser_params *p, const char *name, const char *val } static void -parser_set_compile_option_flag(struct parser_params *p, const char *name, const char *val) +parser_set_frozen_string_literal(struct parser_params *p, const char *name, const char *val) { int b; if (p->token_seen) { - rb_warning1("`%s' is ignored after any tokens", WARN_S(name)); + rb_warning1("'%s' is ignored after any tokens", WARN_S(name)); return; } b = parser_get_bool(p, name, val); if (b < 0) return; - if (!p->compile_option) - p->compile_option = rb_obj_hide(rb_ident_hash_new()); - rb_hash_aset(p->compile_option, ID2SYM(rb_intern(name)), - RBOOL(b)); + p->frozen_string_literal = b; } static void @@ -8694,30 +9585,30 @@ parser_set_shareable_constant_value(struct parser_params *p, const char *name, c for (const char *s = p->lex.pbeg, *e = p->lex.pcur; s < e; ++s) { if (*s == ' ' || *s == '\t') continue; if (*s == '#') break; - rb_warning1("`%s' is ignored unless in comment-only line", WARN_S(name)); + rb_warning1("'%s' is ignored unless in comment-only line", WARN_S(name)); return; } switch (*val) { case 'n': case 'N': if (STRCASECMP(val, "none") == 0) { - p->ctxt.shareable_constant_value = shareable_none; + p->ctxt.shareable_constant_value = rb_parser_shareable_none; return; } break; case 'l': case 'L': if (STRCASECMP(val, "literal") == 0) { - p->ctxt.shareable_constant_value = shareable_literal; + p->ctxt.shareable_constant_value = rb_parser_shareable_literal; return; } break; case 'e': case 'E': if (STRCASECMP(val, "experimental_copy") == 0) { - p->ctxt.shareable_constant_value = shareable_copy; + p->ctxt.shareable_constant_value = rb_parser_shareable_copy; return; } if (STRCASECMP(val, "experimental_everything") == 0) { - p->ctxt.shareable_constant_value = shareable_everything; + p->ctxt.shareable_constant_value = rb_parser_shareable_everything; return; } break; @@ -8743,7 +9634,7 @@ struct magic_comment { static const struct magic_comment magic_comments[] = { {"coding", magic_comment_encoding, parser_encode_length}, {"encoding", magic_comment_encoding, parser_encode_length}, - {"frozen_string_literal", parser_set_compile_option_flag}, + {"frozen_string_literal", parser_set_frozen_string_literal}, {"shareable_constant_value", parser_set_shareable_constant_value}, {"warn_indent", parser_set_token_info}, # if WARN_PAST_SCOPE @@ -8837,6 +9728,7 @@ parser_magic_comment(struct parser_params *p, const char *str, long len) do str++; while (--len > 0 && ISSPACE(*str)); if (!len) break; + const char *tok_beg = str; if (*str == '"') { for (vbeg = ++str; --len > 0 && *str != '"'; str++) { if (*str == '\\') { @@ -8854,6 +9746,7 @@ parser_magic_comment(struct parser_params *p, const char *str, long len) for (vbeg = str; len > 0 && *str != '"' && *str != ';' && !ISSPACE(*str); --len, str++); vend = str; } + const char *tok_end = str; if (indicator) { while (len > 0 && (*str == ';' || ISSPACE(*str))) --len, str++; } @@ -8875,6 +9768,8 @@ parser_magic_comment(struct parser_params *p, const char *str, long len) n = (*mc->length)(p, vbeg, n); } str_copy(val, vbeg, n); + p->lex.ptok = tok_beg; + p->lex.pcur = tok_end; (*mc->func)(p, mc->name, RSTRING_PTR(val)); break; } @@ -8928,6 +9823,8 @@ set_file_encoding(struct parser_params *p, const char *str, const char *send) beg = str; while ((*str == '-' || *str == '_' || ISALNUM(*str)) && ++str < send); s = rb_str_new(beg, parser_encode_length(p, beg, str - beg)); + p->lex.ptok = beg; + p->lex.pcur = str; parser_set_encode(p, RSTRING_PTR(s)); rb_str_resize(s, 0); } @@ -8942,30 +9839,31 @@ parser_prepare(struct parser_params *p) if (peek(p, '!')) p->has_shebang = 1; break; case 0xef: /* UTF-8 BOM marker */ - if (p->lex.pend - p->lex.pcur >= 2 && + if (!lex_eol_n_p(p, 2) && (unsigned char)p->lex.pcur[0] == 0xbb && (unsigned char)p->lex.pcur[1] == 0xbf) { p->enc = rb_utf8_encoding(); p->lex.pcur += 2; #ifndef RIPPER if (p->debug_lines) { - rb_enc_associate(p->lex.lastline, p->enc); + rb_parser_string_set_encoding(p->lex.lastline, p->enc); } #endif p->lex.pbeg = p->lex.pcur; + token_flush(p); return; } break; - case EOF: + case -1: /* end of script. */ return; } pushback(p, c); - p->enc = rb_enc_get(p->lex.lastline); + p->enc = rb_parser_str_get_encoding(p->lex.lastline); } #ifndef RIPPER #define ambiguous_operator(tok, op, syn) ( \ - rb_warning0("`"op"' after local variable or literal is interpreted as binary operator"), \ + rb_warning0("'"op"' after local variable or literal is interpreted as binary operator"), \ rb_warning0("even though it seems like "syn"")) #else #define ambiguous_operator(tok, op, syn) \ @@ -8977,24 +9875,13 @@ parser_prepare(struct parser_params *p) (ambiguous_operator(tok, op, syn), 0)), \ (enum yytokentype)(tok)) -static VALUE -parse_rational(struct parser_params *p, char *str, int len, int seen_point) -{ - VALUE v; - char *point = &str[seen_point]; - size_t fraclen = len-seen_point-1; - memmove(point, point+1, fraclen+1); - v = rb_cstr_to_inum(str, 10, FALSE); - return rb_rational_new(v, rb_int_positive_pow(10, fraclen)); -} - static enum yytokentype no_digits(struct parser_params *p) { yyerror0("numeric literal without digits"); if (peek(p, '_')) nextc(p); /* dummy 0, for tUMINUS_NUM at numeric */ - return set_integer_literal(p, INT2FIX(0), 0); + return set_number_literal(p, tINTEGER, 0, 10, 0); } static enum yytokentype @@ -9035,7 +9922,7 @@ parse_numeric(struct parser_params *p, int c) } else if (nondigit) goto trailing_uc; suffix = number_literal_suffix(p, NUM_SUFFIX_ALL); - return set_integer_literal(p, rb_cstr_to_inum(tok(p), 16, FALSE), suffix); + return set_number_literal(p, tINTEGER, suffix, 16, 0); } if (c == 'b' || c == 'B') { /* binary */ @@ -9059,7 +9946,7 @@ parse_numeric(struct parser_params *p, int c) } else if (nondigit) goto trailing_uc; suffix = number_literal_suffix(p, NUM_SUFFIX_ALL); - return set_integer_literal(p, rb_cstr_to_inum(tok(p), 2, FALSE), suffix); + return set_number_literal(p, tINTEGER, suffix, 2, 0); } if (c == 'd' || c == 'D') { /* decimal */ @@ -9083,7 +9970,7 @@ parse_numeric(struct parser_params *p, int c) } else if (nondigit) goto trailing_uc; suffix = number_literal_suffix(p, NUM_SUFFIX_ALL); - return set_integer_literal(p, rb_cstr_to_inum(tok(p), 10, FALSE), suffix); + return set_number_literal(p, tINTEGER, suffix, 10, 0); } if (c == '_') { /* 0_0 */ @@ -9093,6 +9980,7 @@ parse_numeric(struct parser_params *p, int c) /* prefixed octal */ c = nextc(p); if (c == -1 || c == '_' || !ISDIGIT(c)) { + tokfix(p); return no_digits(p); } } @@ -9115,7 +10003,7 @@ parse_numeric(struct parser_params *p, int c) tokfix(p); if (nondigit) goto trailing_uc; suffix = number_literal_suffix(p, NUM_SUFFIX_ALL); - return set_integer_literal(p, rb_cstr_to_inum(tok(p), 8, FALSE), suffix); + return set_number_literal(p, tINTEGER, suffix, 8, 0); } if (nondigit) { pushback(p, c); @@ -9131,8 +10019,9 @@ parse_numeric(struct parser_params *p, int c) } else { pushback(p, c); + tokfix(p); suffix = number_literal_suffix(p, NUM_SUFFIX_ALL); - return set_integer_literal(p, INT2FIX(0), suffix); + return set_number_literal(p, tINTEGER, suffix, 10, 0); } } @@ -9178,6 +10067,7 @@ parse_numeric(struct parser_params *p, int c) c = nextc(p); if (c != '-' && c != '+' && !ISDIGIT(c)) { pushback(p, c); + c = nondigit; nondigit = 0; goto decode_num; } @@ -9205,31 +10095,28 @@ parse_numeric(struct parser_params *p, int c) trailing_uc: literal_flush(p, p->lex.pcur - 1); YYLTYPE loc = RUBY_INIT_YYLLOC(); - compile_error(p, "trailing `%c' in number", nondigit); + compile_error(p, "trailing '%c' in number", nondigit); parser_show_error_line(p, &loc); } tokfix(p); if (is_float) { enum yytokentype type = tFLOAT; - VALUE v; suffix = number_literal_suffix(p, seen_e ? NUM_SUFFIX_I : NUM_SUFFIX_ALL); if (suffix & NUM_SUFFIX_R) { type = tRATIONAL; - v = parse_rational(p, tok(p), toklen(p), seen_point); } else { - double d = strtod(tok(p), 0); + strtod(tok(p), 0); if (errno == ERANGE) { rb_warning1("Float %s out of range", WARN_S(tok(p))); errno = 0; } - v = DBL2NUM(d); } - return set_number_literal(p, v, type, suffix); + return set_number_literal(p, type, suffix, 0, seen_point); } suffix = number_literal_suffix(p, NUM_SUFFIX_ALL); - return set_integer_literal(p, rb_cstr_to_inum(tok(p), 10, FALSE), suffix); + return set_number_literal(p, tINTEGER, suffix, 10, 0); } static enum yytokentype @@ -9266,16 +10153,16 @@ parse_qmark(struct parser_params *p, int space_seen) if (tokadd_mbchar(p, c) == -1) return 0; } else if ((rb_enc_isalnum(c, p->enc) || c == '_') && - p->lex.pcur < p->lex.pend && is_identchar(p->lex.pcur, p->lex.pend, p->enc)) { + !lex_eol_p(p) && is_identchar(p, p->lex.pcur, p->lex.pend, p->enc)) { if (space_seen) { const char *start = p->lex.pcur - 1, *ptr = start; do { int n = parser_precise_mbclen(p, ptr); if (n < 0) return -1; ptr += n; - } while (ptr < p->lex.pend && is_identchar(ptr, p->lex.pend, p->enc)); - rb_warn2("`?' just followed by `%.*s' is interpreted as" \ - " a conditional operator, put a space after `?'", + } while (!lex_eol_ptr_p(p, ptr) && is_identchar(p, ptr, p->lex.pend, p->enc)); + rb_warn2("'?' just followed by '%.*s' is interpreted as" \ + " a conditional operator, put a space after '?'", WARN_I((int)(ptr - start)), WARN_S_L(start, (ptr - start))); } goto ternary; @@ -9286,12 +10173,12 @@ parse_qmark(struct parser_params *p, int space_seen) enc = rb_utf8_encoding(); tokadd_utf8(p, &enc, -1, 0, 0); } - else if (!lex_eol_p(p) && !(c = *p->lex.pcur, ISASCII(c))) { + else if (!ISASCII(c = peekc(p))) { nextc(p); if (tokadd_mbchar(p, c) == -1) return 0; } else { - c = read_escape(p, 0, &enc); + c = read_escape(p, 0); tokadd(p, c); } } @@ -9438,7 +10325,7 @@ parse_numvar(struct parser_params *p) if (overflow || n > nth_ref_max) { /* compile_error()? */ - rb_warn1("`%s' is too big for a number variable, always nil", WARN_S(tok(p))); + rb_warn1("'%s' is too big for a number variable, always nil", WARN_S(tok(p))); return 0; /* $0 is $PROGRAM_NAME, not NTH_REF */ } else { @@ -9467,22 +10354,22 @@ parse_gvar(struct parser_params *p, const enum lex_state_e last_state) pushback(p, c); c = '_'; /* fall through */ - case '~': /* $~: match-data */ - case '*': /* $*: argv */ - case '$': /* $$: pid */ - case '?': /* $?: last status */ - case '!': /* $!: error string */ - case '@': /* $@: error position */ - case '/': /* $/: input record separator */ - case '\\': /* $\: output record separator */ - case ';': /* $;: field separator */ - case ',': /* $,: output field separator */ - case '.': /* $.: last read line number */ - case '=': /* $=: ignorecase */ - case ':': /* $:: load path */ - case '<': /* $<: reading filename */ - case '>': /* $>: default output handle */ - case '\"': /* $": already loaded files */ + case '~': /* $~: match-data */ + case '*': /* $*: argv */ + case '$': /* $$: pid */ + case '?': /* $?: last status */ + case '!': /* $!: error string */ + case '@': /* $@: error position */ + case '/': /* $/: input record separator */ + case '\\': /* $\: output record separator */ + case ';': /* $;: field separator */ + case ',': /* $,: output field separator */ + case '.': /* $.: last read line number */ + case '=': /* $=: ignorecase */ + case ':': /* $:: load path */ + case '<': /* $<: default input handle */ + case '>': /* $>: default output handle */ + case '\"': /* $": already loaded files */ tokadd(p, '$'); tokadd(p, c); goto gvar; @@ -9503,10 +10390,10 @@ parse_gvar(struct parser_params *p, const enum lex_state_e last_state) set_yylval_name(TOK_INTERN()); return tGVAR; - case '&': /* $&: last match */ - case '`': /* $`: string before last match */ - case '\'': /* $': string after last match */ - case '+': /* $+: string matches last paren. */ + case '&': /* $&: last match */ + case '`': /* $`: string before last match */ + case '\'': /* $': string after last match */ + case '+': /* $+: string matches last paren. */ if (IS_lex_state_for(last_state, EXPR_FNAME)) { tokadd(p, '$'); tokadd(p, c); @@ -9534,11 +10421,11 @@ parse_gvar(struct parser_params *p, const enum lex_state_e last_state) if (!parser_is_identchar(p)) { YYLTYPE loc = RUBY_INIT_YYLLOC(); if (c == -1 || ISSPACE(c)) { - compile_error(p, "`$' without identifiers is not allowed as a global variable name"); + compile_error(p, "'$' without identifiers is not allowed as a global variable name"); } else { pushback(p, c); - compile_error(p, "`$%c' is not allowed as a global variable name", c); + compile_error(p, "'$%c' is not allowed as a global variable name", c); } parser_show_error_line(p, &loc); set_yylval_noname(); @@ -9551,11 +10438,16 @@ parse_gvar(struct parser_params *p, const enum lex_state_e last_state) if (tokadd_ident(p, c)) return 0; SET_LEX_STATE(EXPR_END); - tokenize_ident(p); + if (VALID_SYMNAME_P(tok(p), toklen(p), p->enc, ID_GLOBAL)) { + tokenize_ident(p); + } + else { + compile_error(p, "'%.*s' is not allowed as a global variable name", toklen(p), tok(p)); + set_yylval_noname(); + } return tGVAR; } -#ifndef RIPPER static bool parser_numbered_param(struct parser_params *p, int n) { @@ -9577,7 +10469,6 @@ parser_numbered_param(struct parser_params *p, int n) } return true; } -#endif static enum yytokentype parse_atmark(struct parser_params *p, const enum lex_state_e last_state) @@ -9600,10 +10491,10 @@ parse_atmark(struct parser_params *p, const enum lex_state_e last_state) pushback(p, c); RUBY_SET_YYLLOC(loc); if (result == tIVAR) { - compile_error(p, "`@' without identifiers is not allowed as an instance variable name"); + compile_error(p, "'@' without identifiers is not allowed as an instance variable name"); } else { - compile_error(p, "`@@' without identifiers is not allowed as a class variable name"); + compile_error(p, "'@@' without identifiers is not allowed as a class variable name"); } parser_show_error_line(p, &loc); set_yylval_noname(); @@ -9614,10 +10505,10 @@ parse_atmark(struct parser_params *p, const enum lex_state_e last_state) pushback(p, c); RUBY_SET_YYLLOC(loc); if (result == tIVAR) { - compile_error(p, "`@%c' is not allowed as an instance variable name", c); + compile_error(p, "'@%c' is not allowed as an instance variable name", c); } else { - compile_error(p, "`@@%c' is not allowed as a class variable name", c); + compile_error(p, "'@@%c' is not allowed as a class variable name", c); } parser_show_error_line(p, &loc); set_yylval_noname(); @@ -9634,13 +10525,13 @@ static enum yytokentype parse_ident(struct parser_params *p, int c, int cmd_state) { enum yytokentype result; - int mb = ENC_CODERANGE_7BIT; + bool is_ascii = true; const enum lex_state_e last_state = p->lex.state; ID ident; int enforce_keyword_end = 0; do { - if (!ISASCII(c)) mb = ENC_CODERANGE_UNKNOWN; + if (!ISASCII(c)) is_ascii = false; if (tokadd_mbchar(p, c) == -1) return 0; c = nextc(p); } while (parser_is_identchar(p)); @@ -9669,14 +10560,14 @@ parse_ident(struct parser_params *p, int c, int cmd_state) } #ifndef RIPPER - if (!NIL_P(peek_end_expect_token_locations(p))) { - VALUE end_loc; + if (peek_end_expect_token_locations(p)) { + const rb_code_position_t *end_pos; int lineno, column; int beg_pos = (int)(p->lex.ptok - p->lex.pbeg); - end_loc = peek_end_expect_token_locations(p); - lineno = NUM2INT(rb_ary_entry(end_loc, 0)); - column = NUM2INT(rb_ary_entry(end_loc, 1)); + end_pos = peek_end_expect_token_locations(p)->pos; + lineno = end_pos->lineno; + column = end_pos->column; if (p->debug) { rb_parser_printf(p, "enforce_keyword_end check. current: (%d, %d), peek: (%d, %d)\n", @@ -9694,7 +10585,7 @@ parse_ident(struct parser_params *p, int c, int cmd_state) } #endif - if (mb == ENC_CODERANGE_7BIT && (!IS_lex_state(EXPR_DOT) || enforce_keyword_end)) { + if (is_ascii && (!IS_lex_state(EXPR_DOT) || enforce_keyword_end)) { const struct kwtable *kw; /* See if it is a reserved word. */ @@ -9777,7 +10668,7 @@ parser_yylex(struct parser_params *p) int token_seen = p->token_seen; if (p->lex.strterm) { - if (p->lex.strterm->flags & STRTERM_HEREDOC) { + if (strterm_is_heredoc(p->lex.strterm)) { token_flush(p); return here_document(p, &p->lex.strterm->u.heredoc); } @@ -9799,9 +10690,9 @@ parser_yylex(struct parser_params *p) case '\004': /* ^D */ case '\032': /* ^Z */ case -1: /* end of script. */ - p->eofp = 1; + p->eofp = 1; #ifndef RIPPER - if (!NIL_P(p->end_expect_token_locations) && RARRAY_LEN(p->end_expect_token_locations) > 0) { + if (p->end_expect_token_locations) { pop_end_expect_token_locations(p); RUBY_SET_YYLLOC_OF_DUMMY_END(*p->yylloc); return tDUMNY_END; @@ -9840,19 +10731,21 @@ parser_yylex(struct parser_params *p) case '#': /* it's a comment */ p->token_seen = token_seen; + const char *const pcur = p->lex.pcur, *const ptok = p->lex.ptok; /* no magic_comment in shebang line */ if (!parser_magic_comment(p, p->lex.pcur, p->lex.pend - p->lex.pcur)) { if (comment_at_top(p)) { set_file_encoding(p, p->lex.pcur, p->lex.pend); } } + p->lex.pcur = pcur, p->lex.ptok = ptok; lex_goto_eol(p); dispatch_scan_event(p, tCOMMENT); fallthru = TRUE; /* fall through */ case '\n': p->token_seen = token_seen; - VALUE prevline = p->lex.lastline; + rb_parser_string_t *prevline = p->lex.lastline; c = (IS_lex_state(EXPR_BEG|EXPR_CLASS|EXPR_FNAME|EXPR_DOT) && !IS_lex_state(EXPR_LABELED)); if (c || IS_lex_state_all(EXPR_ARG|EXPR_LABELED)) { @@ -9914,7 +10807,7 @@ parser_yylex(struct parser_params *p) } pushback(p, c); if (IS_SPCARG(c)) { - rb_warning0("`**' interpreted as argument prefix"); + rb_warning0("'**' interpreted as argument prefix"); c = tDSTAR; } else if (IS_BEG()) { @@ -9932,7 +10825,7 @@ parser_yylex(struct parser_params *p) } pushback(p, c); if (IS_SPCARG(c)) { - rb_warning0("`*' interpreted as argument prefix"); + rb_warning0("'*' interpreted as argument prefix"); c = tSTAR; } else if (IS_BEG()) { @@ -10018,7 +10911,7 @@ parser_yylex(struct parser_params *p) !IS_lex_state(EXPR_DOT | EXPR_CLASS) && !IS_END() && (!IS_ARG() || IS_lex_state(EXPR_LABELED) || space_seen)) { - int token = heredoc_identifier(p); + enum yytokentype token = heredoc_identifier(p); if (token) return token < 0 ? 0 : token; } if (IS_AFTER_OPERATOR()) { @@ -10121,8 +11014,8 @@ parser_yylex(struct parser_params *p) if ((c != ':') || (c = peekc_n(p, 1)) == -1 || !(c == '\'' || c == '"' || - is_identchar((p->lex.pcur+1), p->lex.pend, p->enc))) { - rb_warning0("`&' interpreted as argument prefix"); + is_identchar(p, (p->lex.pcur+1), p->lex.pend, p->enc))) { + rb_warning0("'&' interpreted as argument prefix"); } c = tAMPER; } @@ -10204,6 +11097,7 @@ parser_yylex(struct parser_params *p) } if (c == '>') { SET_LEX_STATE(EXPR_ENDFN); + yylval.num = p->lex.lpar_beg; return tLAMBDA; } if (IS_BEG() || (IS_SPCARG(c) && arg_ambiguous(p, '-'))) { @@ -10473,7 +11367,7 @@ parser_yylex(struct parser_params *p) default: if (!parser_is_identchar(p)) { - compile_error(p, "Invalid char `\\x%02X' in expression", c); + compile_error(p, "Invalid char '\\x%02X' in expression", c); token_flush(p); goto retry; } @@ -10491,7 +11385,7 @@ yylex(YYSTYPE *lval, YYLTYPE *yylloc, struct parser_params *p) enum yytokentype t; p->lval = lval; - lval->val = Qundef; + lval->node = 0; p->yylloc = yylloc; t = parser_yylex(p); @@ -10507,30 +11401,1170 @@ yylex(YYSTYPE *lval, YYLTYPE *yylloc, struct parser_params *p) #define LVAR_USED ((ID)1 << (sizeof(ID) * CHAR_BIT - 1)) static NODE* -node_new_temporal(struct parser_params *p, enum node_type type, VALUE a0, VALUE a1, VALUE a2) +node_new_internal(struct parser_params *p, enum node_type type, size_t size, size_t alignment) { - NODE *n = rb_ast_newnode(p->ast, type); + NODE *n = rb_ast_newnode(p->ast, type, size, alignment); - rb_node_init(n, type, a0, a1, a2); + rb_node_init(n, type); return n; } +static NODE * +nd_set_loc(NODE *nd, const YYLTYPE *loc) +{ + nd->nd_loc = *loc; + nd_set_line(nd, loc->beg_pos.lineno); + return nd; +} + static NODE* -node_newnode(struct parser_params *p, enum node_type type, VALUE a0, VALUE a1, VALUE a2, const rb_code_location_t *loc) +node_newnode(struct parser_params *p, enum node_type type, size_t size, size_t alignment, const rb_code_location_t *loc) { - NODE *n = node_new_temporal(p, type, a0, a1, a2); + NODE *n = node_new_internal(p, type, size, alignment); nd_set_loc(n, loc); nd_set_node_id(n, parser_get_node_id(p)); return n; } -static NODE * -nd_set_loc(NODE *nd, const YYLTYPE *loc) +#define NODE_NEWNODE(node_type, type, loc) (type *)(node_newnode(p, node_type, sizeof(type), RUBY_ALIGNOF(type), loc)) + +static rb_node_scope_t * +rb_node_scope_new(struct parser_params *p, rb_node_args_t *nd_args, NODE *nd_body, const YYLTYPE *loc) { - nd->nd_loc = *loc; - nd_set_line(nd, loc->beg_pos.lineno); - return nd; + rb_ast_id_table_t *nd_tbl; + nd_tbl = local_tbl(p); + rb_node_scope_t *n = NODE_NEWNODE(NODE_SCOPE, rb_node_scope_t, loc); + n->nd_tbl = nd_tbl; + n->nd_body = nd_body; + n->nd_args = nd_args; + + return n; +} + +static rb_node_scope_t * +rb_node_scope_new2(struct parser_params *p, rb_ast_id_table_t *nd_tbl, rb_node_args_t *nd_args, NODE *nd_body, const YYLTYPE *loc) +{ + rb_node_scope_t *n = NODE_NEWNODE(NODE_SCOPE, rb_node_scope_t, loc); + n->nd_tbl = nd_tbl; + n->nd_body = nd_body; + n->nd_args = nd_args; + + return n; +} + +static rb_node_defn_t * +rb_node_defn_new(struct parser_params *p, ID nd_mid, NODE *nd_defn, const YYLTYPE *loc) +{ + rb_node_defn_t *n = NODE_NEWNODE(NODE_DEFN, rb_node_defn_t, loc); + n->nd_mid = nd_mid; + n->nd_defn = nd_defn; + + return n; +} + +static rb_node_defs_t * +rb_node_defs_new(struct parser_params *p, NODE *nd_recv, ID nd_mid, NODE *nd_defn, const YYLTYPE *loc) +{ + rb_node_defs_t *n = NODE_NEWNODE(NODE_DEFS, rb_node_defs_t, loc); + n->nd_recv = nd_recv; + n->nd_mid = nd_mid; + n->nd_defn = nd_defn; + + return n; +} + +static rb_node_block_t * +rb_node_block_new(struct parser_params *p, NODE *nd_head, const YYLTYPE *loc) +{ + rb_node_block_t *n = NODE_NEWNODE(NODE_BLOCK, rb_node_block_t, loc); + n->nd_head = nd_head; + n->nd_end = (NODE *)n; + n->nd_next = 0; + + return n; +} + +static rb_node_for_t * +rb_node_for_new(struct parser_params *p, NODE *nd_iter, NODE *nd_body, const YYLTYPE *loc) +{ + rb_node_for_t *n = NODE_NEWNODE(NODE_FOR, rb_node_for_t, loc); + n->nd_body = nd_body; + n->nd_iter = nd_iter; + + return n; +} + +static rb_node_for_masgn_t * +rb_node_for_masgn_new(struct parser_params *p, NODE *nd_var, const YYLTYPE *loc) +{ + rb_node_for_masgn_t *n = NODE_NEWNODE(NODE_FOR_MASGN, rb_node_for_masgn_t, loc); + n->nd_var = nd_var; + + return n; +} + +static rb_node_retry_t * +rb_node_retry_new(struct parser_params *p, const YYLTYPE *loc) +{ + rb_node_retry_t *n = NODE_NEWNODE(NODE_RETRY, rb_node_retry_t, loc); + + return n; +} + +static rb_node_begin_t * +rb_node_begin_new(struct parser_params *p, NODE *nd_body, const YYLTYPE *loc) +{ + rb_node_begin_t *n = NODE_NEWNODE(NODE_BEGIN, rb_node_begin_t, loc); + n->nd_body = nd_body; + + return n; +} + +static rb_node_rescue_t * +rb_node_rescue_new(struct parser_params *p, NODE *nd_head, NODE *nd_resq, NODE *nd_else, const YYLTYPE *loc) +{ + rb_node_rescue_t *n = NODE_NEWNODE(NODE_RESCUE, rb_node_rescue_t, loc); + n->nd_head = nd_head; + n->nd_resq = nd_resq; + n->nd_else = nd_else; + + return n; +} + +static rb_node_resbody_t * +rb_node_resbody_new(struct parser_params *p, NODE *nd_args, NODE *nd_body, NODE *nd_next, const YYLTYPE *loc) +{ + rb_node_resbody_t *n = NODE_NEWNODE(NODE_RESBODY, rb_node_resbody_t, loc); + n->nd_args = nd_args; + n->nd_body = nd_body; + n->nd_next = nd_next; + + return n; +} + +static rb_node_ensure_t * +rb_node_ensure_new(struct parser_params *p, NODE *nd_head, NODE *nd_ensr, const YYLTYPE *loc) +{ + rb_node_ensure_t *n = NODE_NEWNODE(NODE_ENSURE, rb_node_ensure_t, loc); + n->nd_head = nd_head; + n->nd_ensr = nd_ensr; + + return n; +} + +static rb_node_and_t * +rb_node_and_new(struct parser_params *p, NODE *nd_1st, NODE *nd_2nd, const YYLTYPE *loc) +{ + rb_node_and_t *n = NODE_NEWNODE(NODE_AND, rb_node_and_t, loc); + n->nd_1st = nd_1st; + n->nd_2nd = nd_2nd; + + return n; +} + +static rb_node_or_t * +rb_node_or_new(struct parser_params *p, NODE *nd_1st, NODE *nd_2nd, const YYLTYPE *loc) +{ + rb_node_or_t *n = NODE_NEWNODE(NODE_OR, rb_node_or_t, loc); + n->nd_1st = nd_1st; + n->nd_2nd = nd_2nd; + + return n; +} + +static rb_node_return_t * +rb_node_return_new(struct parser_params *p, NODE *nd_stts, const YYLTYPE *loc) +{ + rb_node_return_t *n = NODE_NEWNODE(NODE_RETURN, rb_node_return_t, loc); + n->nd_stts = nd_stts; + return n; +} + +static rb_node_yield_t * +rb_node_yield_new(struct parser_params *p, NODE *nd_head, const YYLTYPE *loc) +{ + rb_node_yield_t *n = NODE_NEWNODE(NODE_YIELD, rb_node_yield_t, loc); + n->nd_head = nd_head; + + return n; +} + +static rb_node_if_t * +rb_node_if_new(struct parser_params *p, NODE *nd_cond, NODE *nd_body, NODE *nd_else, const YYLTYPE *loc) +{ + rb_node_if_t *n = NODE_NEWNODE(NODE_IF, rb_node_if_t, loc); + n->nd_cond = nd_cond; + n->nd_body = nd_body; + n->nd_else = nd_else; + + return n; +} + +static rb_node_unless_t * +rb_node_unless_new(struct parser_params *p, NODE *nd_cond, NODE *nd_body, NODE *nd_else, const YYLTYPE *loc) +{ + rb_node_unless_t *n = NODE_NEWNODE(NODE_UNLESS, rb_node_unless_t, loc); + n->nd_cond = nd_cond; + n->nd_body = nd_body; + n->nd_else = nd_else; + + return n; +} + +static rb_node_class_t * +rb_node_class_new(struct parser_params *p, NODE *nd_cpath, NODE *nd_body, NODE *nd_super, const YYLTYPE *loc) +{ + /* Keep the order of node creation */ + NODE *scope = NEW_SCOPE(0, nd_body, loc); + rb_node_class_t *n = NODE_NEWNODE(NODE_CLASS, rb_node_class_t, loc); + n->nd_cpath = nd_cpath; + n->nd_body = scope; + n->nd_super = nd_super; + + return n; +} + +static rb_node_sclass_t * +rb_node_sclass_new(struct parser_params *p, NODE *nd_recv, NODE *nd_body, const YYLTYPE *loc) +{ + /* Keep the order of node creation */ + NODE *scope = NEW_SCOPE(0, nd_body, loc); + rb_node_sclass_t *n = NODE_NEWNODE(NODE_SCLASS, rb_node_sclass_t, loc); + n->nd_recv = nd_recv; + n->nd_body = scope; + + return n; +} + +static rb_node_module_t * +rb_node_module_new(struct parser_params *p, NODE *nd_cpath, NODE *nd_body, const YYLTYPE *loc) +{ + /* Keep the order of node creation */ + NODE *scope = NEW_SCOPE(0, nd_body, loc); + rb_node_module_t *n = NODE_NEWNODE(NODE_MODULE, rb_node_module_t, loc); + n->nd_cpath = nd_cpath; + n->nd_body = scope; + + return n; +} + +static rb_node_iter_t * +rb_node_iter_new(struct parser_params *p, rb_node_args_t *nd_args, NODE *nd_body, const YYLTYPE *loc) +{ + /* Keep the order of node creation */ + NODE *scope = NEW_SCOPE(nd_args, nd_body, loc); + rb_node_iter_t *n = NODE_NEWNODE(NODE_ITER, rb_node_iter_t, loc); + n->nd_body = scope; + n->nd_iter = 0; + + return n; +} + +static rb_node_lambda_t * +rb_node_lambda_new(struct parser_params *p, rb_node_args_t *nd_args, NODE *nd_body, const YYLTYPE *loc) +{ + /* Keep the order of node creation */ + NODE *scope = NEW_SCOPE(nd_args, nd_body, loc); + rb_node_lambda_t *n = NODE_NEWNODE(NODE_LAMBDA, rb_node_lambda_t, loc); + n->nd_body = scope; + + return n; +} + +static rb_node_case_t * +rb_node_case_new(struct parser_params *p, NODE *nd_head, NODE *nd_body, const YYLTYPE *loc) +{ + rb_node_case_t *n = NODE_NEWNODE(NODE_CASE, rb_node_case_t, loc); + n->nd_head = nd_head; + n->nd_body = nd_body; + + return n; +} + +static rb_node_case2_t * +rb_node_case2_new(struct parser_params *p, NODE *nd_body, const YYLTYPE *loc) +{ + rb_node_case2_t *n = NODE_NEWNODE(NODE_CASE2, rb_node_case2_t, loc); + n->nd_head = 0; + n->nd_body = nd_body; + + return n; +} + +static rb_node_case3_t * +rb_node_case3_new(struct parser_params *p, NODE *nd_head, NODE *nd_body, const YYLTYPE *loc) +{ + rb_node_case3_t *n = NODE_NEWNODE(NODE_CASE3, rb_node_case3_t, loc); + n->nd_head = nd_head; + n->nd_body = nd_body; + + return n; +} + +static rb_node_when_t * +rb_node_when_new(struct parser_params *p, NODE *nd_head, NODE *nd_body, NODE *nd_next, const YYLTYPE *loc) +{ + rb_node_when_t *n = NODE_NEWNODE(NODE_WHEN, rb_node_when_t, loc); + n->nd_head = nd_head; + n->nd_body = nd_body; + n->nd_next = nd_next; + + return n; +} + +static rb_node_in_t * +rb_node_in_new(struct parser_params *p, NODE *nd_head, NODE *nd_body, NODE *nd_next, const YYLTYPE *loc) +{ + rb_node_in_t *n = NODE_NEWNODE(NODE_IN, rb_node_in_t, loc); + n->nd_head = nd_head; + n->nd_body = nd_body; + n->nd_next = nd_next; + + return n; +} + +static rb_node_while_t * +rb_node_while_new(struct parser_params *p, NODE *nd_cond, NODE *nd_body, long nd_state, const YYLTYPE *loc) +{ + rb_node_while_t *n = NODE_NEWNODE(NODE_WHILE, rb_node_while_t, loc); + n->nd_cond = nd_cond; + n->nd_body = nd_body; + n->nd_state = nd_state; + + return n; +} + +static rb_node_until_t * +rb_node_until_new(struct parser_params *p, NODE *nd_cond, NODE *nd_body, long nd_state, const YYLTYPE *loc) +{ + rb_node_until_t *n = NODE_NEWNODE(NODE_UNTIL, rb_node_until_t, loc); + n->nd_cond = nd_cond; + n->nd_body = nd_body; + n->nd_state = nd_state; + + return n; +} + +static rb_node_colon2_t * +rb_node_colon2_new(struct parser_params *p, NODE *nd_head, ID nd_mid, const YYLTYPE *loc) +{ + rb_node_colon2_t *n = NODE_NEWNODE(NODE_COLON2, rb_node_colon2_t, loc); + n->nd_head = nd_head; + n->nd_mid = nd_mid; + + return n; +} + +static rb_node_colon3_t * +rb_node_colon3_new(struct parser_params *p, ID nd_mid, const YYLTYPE *loc) +{ + rb_node_colon3_t *n = NODE_NEWNODE(NODE_COLON3, rb_node_colon3_t, loc); + n->nd_mid = nd_mid; + + return n; +} + +static rb_node_dot2_t * +rb_node_dot2_new(struct parser_params *p, NODE *nd_beg, NODE *nd_end, const YYLTYPE *loc) +{ + rb_node_dot2_t *n = NODE_NEWNODE(NODE_DOT2, rb_node_dot2_t, loc); + n->nd_beg = nd_beg; + n->nd_end = nd_end; + + return n; +} + +static rb_node_dot3_t * +rb_node_dot3_new(struct parser_params *p, NODE *nd_beg, NODE *nd_end, const YYLTYPE *loc) +{ + rb_node_dot3_t *n = NODE_NEWNODE(NODE_DOT3, rb_node_dot3_t, loc); + n->nd_beg = nd_beg; + n->nd_end = nd_end; + + return n; +} + +static rb_node_self_t * +rb_node_self_new(struct parser_params *p, const YYLTYPE *loc) +{ + rb_node_self_t *n = NODE_NEWNODE(NODE_SELF, rb_node_self_t, loc); + n->nd_state = 1; + + return n; +} + +static rb_node_nil_t * +rb_node_nil_new(struct parser_params *p, const YYLTYPE *loc) +{ + rb_node_nil_t *n = NODE_NEWNODE(NODE_NIL, rb_node_nil_t, loc); + + return n; +} + +static rb_node_true_t * +rb_node_true_new(struct parser_params *p, const YYLTYPE *loc) +{ + rb_node_true_t *n = NODE_NEWNODE(NODE_TRUE, rb_node_true_t, loc); + + return n; +} + +static rb_node_false_t * +rb_node_false_new(struct parser_params *p, const YYLTYPE *loc) +{ + rb_node_false_t *n = NODE_NEWNODE(NODE_FALSE, rb_node_false_t, loc); + + return n; +} + +static rb_node_super_t * +rb_node_super_new(struct parser_params *p, NODE *nd_args, const YYLTYPE *loc) +{ + rb_node_super_t *n = NODE_NEWNODE(NODE_SUPER, rb_node_super_t, loc); + n->nd_args = nd_args; + + return n; +} + +static rb_node_zsuper_t * +rb_node_zsuper_new(struct parser_params *p, const YYLTYPE *loc) +{ + rb_node_zsuper_t *n = NODE_NEWNODE(NODE_ZSUPER, rb_node_zsuper_t, loc); + + return n; +} + +static rb_node_match2_t * +rb_node_match2_new(struct parser_params *p, NODE *nd_recv, NODE *nd_value, const YYLTYPE *loc) +{ + rb_node_match2_t *n = NODE_NEWNODE(NODE_MATCH2, rb_node_match2_t, loc); + n->nd_recv = nd_recv; + n->nd_value = nd_value; + n->nd_args = 0; + + return n; +} + +static rb_node_match3_t * +rb_node_match3_new(struct parser_params *p, NODE *nd_recv, NODE *nd_value, const YYLTYPE *loc) +{ + rb_node_match3_t *n = NODE_NEWNODE(NODE_MATCH3, rb_node_match3_t, loc); + n->nd_recv = nd_recv; + n->nd_value = nd_value; + + return n; +} + +/* TODO: Use union for NODE_LIST2 */ +static rb_node_list_t * +rb_node_list_new(struct parser_params *p, NODE *nd_head, const YYLTYPE *loc) +{ + rb_node_list_t *n = NODE_NEWNODE(NODE_LIST, rb_node_list_t, loc); + n->nd_head = nd_head; + n->as.nd_alen = 1; + n->nd_next = 0; + + return n; +} + +static rb_node_list_t * +rb_node_list_new2(struct parser_params *p, NODE *nd_head, long nd_alen, NODE *nd_next, const YYLTYPE *loc) +{ + rb_node_list_t *n = NODE_NEWNODE(NODE_LIST, rb_node_list_t, loc); + n->nd_head = nd_head; + n->as.nd_alen = nd_alen; + n->nd_next = nd_next; + + return n; +} + +static rb_node_zlist_t * +rb_node_zlist_new(struct parser_params *p, const YYLTYPE *loc) +{ + rb_node_zlist_t *n = NODE_NEWNODE(NODE_ZLIST, rb_node_zlist_t, loc); + + return n; +} + +static rb_node_hash_t * +rb_node_hash_new(struct parser_params *p, NODE *nd_head, const YYLTYPE *loc) +{ + rb_node_hash_t *n = NODE_NEWNODE(NODE_HASH, rb_node_hash_t, loc); + n->nd_head = nd_head; + n->nd_brace = 0; + + return n; +} + +static rb_node_masgn_t * +rb_node_masgn_new(struct parser_params *p, NODE *nd_head, NODE *nd_args, const YYLTYPE *loc) +{ + rb_node_masgn_t *n = NODE_NEWNODE(NODE_MASGN, rb_node_masgn_t, loc); + n->nd_head = nd_head; + n->nd_value = 0; + n->nd_args = nd_args; + + return n; +} + +static rb_node_gasgn_t * +rb_node_gasgn_new(struct parser_params *p, ID nd_vid, NODE *nd_value, const YYLTYPE *loc) +{ + rb_node_gasgn_t *n = NODE_NEWNODE(NODE_GASGN, rb_node_gasgn_t, loc); + n->nd_vid = nd_vid; + n->nd_value = nd_value; + + return n; +} + +static rb_node_lasgn_t * +rb_node_lasgn_new(struct parser_params *p, ID nd_vid, NODE *nd_value, const YYLTYPE *loc) +{ + rb_node_lasgn_t *n = NODE_NEWNODE(NODE_LASGN, rb_node_lasgn_t, loc); + n->nd_vid = nd_vid; + n->nd_value = nd_value; + + return n; +} + +static rb_node_dasgn_t * +rb_node_dasgn_new(struct parser_params *p, ID nd_vid, NODE *nd_value, const YYLTYPE *loc) +{ + rb_node_dasgn_t *n = NODE_NEWNODE(NODE_DASGN, rb_node_dasgn_t, loc); + n->nd_vid = nd_vid; + n->nd_value = nd_value; + + return n; +} + +static rb_node_iasgn_t * +rb_node_iasgn_new(struct parser_params *p, ID nd_vid, NODE *nd_value, const YYLTYPE *loc) +{ + rb_node_iasgn_t *n = NODE_NEWNODE(NODE_IASGN, rb_node_iasgn_t, loc); + n->nd_vid = nd_vid; + n->nd_value = nd_value; + + return n; +} + +static rb_node_cvasgn_t * +rb_node_cvasgn_new(struct parser_params *p, ID nd_vid, NODE *nd_value, const YYLTYPE *loc) +{ + rb_node_cvasgn_t *n = NODE_NEWNODE(NODE_CVASGN, rb_node_cvasgn_t, loc); + n->nd_vid = nd_vid; + n->nd_value = nd_value; + + return n; +} + +static rb_node_op_asgn1_t * +rb_node_op_asgn1_new(struct parser_params *p, NODE *nd_recv, ID nd_mid, NODE *index, NODE *rvalue, const YYLTYPE *loc) +{ + rb_node_op_asgn1_t *n = NODE_NEWNODE(NODE_OP_ASGN1, rb_node_op_asgn1_t, loc); + n->nd_recv = nd_recv; + n->nd_mid = nd_mid; + n->nd_index = index; + n->nd_rvalue = rvalue; + + return n; +} + +static rb_node_op_asgn2_t * +rb_node_op_asgn2_new(struct parser_params *p, NODE *nd_recv, NODE *nd_value, ID nd_vid, ID nd_mid, bool nd_aid, const YYLTYPE *loc) +{ + rb_node_op_asgn2_t *n = NODE_NEWNODE(NODE_OP_ASGN2, rb_node_op_asgn2_t, loc); + n->nd_recv = nd_recv; + n->nd_value = nd_value; + n->nd_vid = nd_vid; + n->nd_mid = nd_mid; + n->nd_aid = nd_aid; + + return n; +} + +static rb_node_op_asgn_or_t * +rb_node_op_asgn_or_new(struct parser_params *p, NODE *nd_head, NODE *nd_value, const YYLTYPE *loc) +{ + rb_node_op_asgn_or_t *n = NODE_NEWNODE(NODE_OP_ASGN_OR, rb_node_op_asgn_or_t, loc); + n->nd_head = nd_head; + n->nd_value = nd_value; + + return n; +} + +static rb_node_op_asgn_and_t * +rb_node_op_asgn_and_new(struct parser_params *p, NODE *nd_head, NODE *nd_value, const YYLTYPE *loc) +{ + rb_node_op_asgn_and_t *n = NODE_NEWNODE(NODE_OP_ASGN_AND, rb_node_op_asgn_and_t, loc); + n->nd_head = nd_head; + n->nd_value = nd_value; + + return n; +} + +static rb_node_gvar_t * +rb_node_gvar_new(struct parser_params *p, ID nd_vid, const YYLTYPE *loc) +{ + rb_node_gvar_t *n = NODE_NEWNODE(NODE_GVAR, rb_node_gvar_t, loc); + n->nd_vid = nd_vid; + + return n; +} + +static rb_node_lvar_t * +rb_node_lvar_new(struct parser_params *p, ID nd_vid, const YYLTYPE *loc) +{ + rb_node_lvar_t *n = NODE_NEWNODE(NODE_LVAR, rb_node_lvar_t, loc); + n->nd_vid = nd_vid; + + return n; +} + +static rb_node_dvar_t * +rb_node_dvar_new(struct parser_params *p, ID nd_vid, const YYLTYPE *loc) +{ + rb_node_dvar_t *n = NODE_NEWNODE(NODE_DVAR, rb_node_dvar_t, loc); + n->nd_vid = nd_vid; + + return n; +} + +static rb_node_ivar_t * +rb_node_ivar_new(struct parser_params *p, ID nd_vid, const YYLTYPE *loc) +{ + rb_node_ivar_t *n = NODE_NEWNODE(NODE_IVAR, rb_node_ivar_t, loc); + n->nd_vid = nd_vid; + + return n; +} + +static rb_node_const_t * +rb_node_const_new(struct parser_params *p, ID nd_vid, const YYLTYPE *loc) +{ + rb_node_const_t *n = NODE_NEWNODE(NODE_CONST, rb_node_const_t, loc); + n->nd_vid = nd_vid; + + return n; +} + +static rb_node_cvar_t * +rb_node_cvar_new(struct parser_params *p, ID nd_vid, const YYLTYPE *loc) +{ + rb_node_cvar_t *n = NODE_NEWNODE(NODE_CVAR, rb_node_cvar_t, loc); + n->nd_vid = nd_vid; + + return n; +} + +static rb_node_nth_ref_t * +rb_node_nth_ref_new(struct parser_params *p, long nd_nth, const YYLTYPE *loc) +{ + rb_node_nth_ref_t *n = NODE_NEWNODE(NODE_NTH_REF, rb_node_nth_ref_t, loc); + n->nd_nth = nd_nth; + + return n; +} + +static rb_node_back_ref_t * +rb_node_back_ref_new(struct parser_params *p, long nd_nth, const YYLTYPE *loc) +{ + rb_node_back_ref_t *n = NODE_NEWNODE(NODE_BACK_REF, rb_node_back_ref_t, loc); + n->nd_nth = nd_nth; + + return n; +} + +static rb_node_integer_t * +rb_node_integer_new(struct parser_params *p, char* val, int base, const YYLTYPE *loc) +{ + rb_node_integer_t *n = NODE_NEWNODE(NODE_INTEGER, rb_node_integer_t, loc); + n->val = val; + n->minus = FALSE; + n->base = base; + + return n; +} + +static rb_node_float_t * +rb_node_float_new(struct parser_params *p, char* val, const YYLTYPE *loc) +{ + rb_node_float_t *n = NODE_NEWNODE(NODE_FLOAT, rb_node_float_t, loc); + n->val = val; + n->minus = FALSE; + + return n; +} + +static rb_node_rational_t * +rb_node_rational_new(struct parser_params *p, char* val, int base, int seen_point, const YYLTYPE *loc) +{ + rb_node_rational_t *n = NODE_NEWNODE(NODE_RATIONAL, rb_node_rational_t, loc); + n->val = val; + n->minus = FALSE; + n->base = base; + n->seen_point = seen_point; + + return n; +} + +static rb_node_imaginary_t * +rb_node_imaginary_new(struct parser_params *p, char* val, int base, int seen_point, enum rb_numeric_type numeric_type, const YYLTYPE *loc) +{ + rb_node_imaginary_t *n = NODE_NEWNODE(NODE_IMAGINARY, rb_node_imaginary_t, loc); + n->val = val; + n->minus = FALSE; + n->base = base; + n->seen_point = seen_point; + n->type = numeric_type; + + return n; +} + +static rb_node_str_t * +rb_node_str_new(struct parser_params *p, rb_parser_string_t *string, const YYLTYPE *loc) +{ + rb_node_str_t *n = NODE_NEWNODE(NODE_STR, rb_node_str_t, loc); + n->string = string; + + return n; +} + +/* TODO; Use union for NODE_DSTR2 */ +static rb_node_dstr_t * +rb_node_dstr_new0(struct parser_params *p, rb_parser_string_t *string, long nd_alen, NODE *nd_next, const YYLTYPE *loc) +{ + rb_node_dstr_t *n = NODE_NEWNODE(NODE_DSTR, rb_node_dstr_t, loc); + n->string = string; + n->as.nd_alen = nd_alen; + n->nd_next = (rb_node_list_t *)nd_next; + + return n; +} + +static rb_node_dstr_t * +rb_node_dstr_new(struct parser_params *p, rb_parser_string_t *string, const YYLTYPE *loc) +{ + return rb_node_dstr_new0(p, string, 1, 0, loc); +} + +static rb_node_xstr_t * +rb_node_xstr_new(struct parser_params *p, rb_parser_string_t *string, const YYLTYPE *loc) +{ + rb_node_xstr_t *n = NODE_NEWNODE(NODE_XSTR, rb_node_xstr_t, loc); + n->string = string; + + return n; +} + +static rb_node_dxstr_t * +rb_node_dxstr_new(struct parser_params *p, rb_parser_string_t *string, long nd_alen, NODE *nd_next, const YYLTYPE *loc) +{ + rb_node_dxstr_t *n = NODE_NEWNODE(NODE_DXSTR, rb_node_dxstr_t, loc); + n->string = string; + n->nd_alen = nd_alen; + n->nd_next = (rb_node_list_t *)nd_next; + + return n; +} + +static rb_node_sym_t * +rb_node_sym_new(struct parser_params *p, VALUE str, const YYLTYPE *loc) +{ + rb_node_sym_t *n = NODE_NEWNODE(NODE_SYM, rb_node_sym_t, loc); + n->string = rb_str_to_parser_string(p, str); + + return n; +} + +static rb_node_dsym_t * +rb_node_dsym_new(struct parser_params *p, rb_parser_string_t *string, long nd_alen, NODE *nd_next, const YYLTYPE *loc) +{ + rb_node_dsym_t *n = NODE_NEWNODE(NODE_DSYM, rb_node_dsym_t, loc); + n->string = string; + n->nd_alen = nd_alen; + n->nd_next = (rb_node_list_t *)nd_next; + + return n; +} + +static rb_node_evstr_t * +rb_node_evstr_new(struct parser_params *p, NODE *nd_body, const YYLTYPE *loc) +{ + rb_node_evstr_t *n = NODE_NEWNODE(NODE_EVSTR, rb_node_evstr_t, loc); + n->nd_body = nd_body; + + return n; +} + +static rb_node_regx_t * +rb_node_regx_new(struct parser_params *p, rb_parser_string_t *string, int options, const YYLTYPE *loc) +{ + rb_node_regx_t *n = NODE_NEWNODE(NODE_REGX, rb_node_regx_t, loc); + n->string = string; + n->options = options & RE_OPTION_MASK; + + return n; +} + +static rb_node_call_t * +rb_node_call_new(struct parser_params *p, NODE *nd_recv, ID nd_mid, NODE *nd_args, const YYLTYPE *loc) +{ + rb_node_call_t *n = NODE_NEWNODE(NODE_CALL, rb_node_call_t, loc); + n->nd_recv = nd_recv; + n->nd_mid = nd_mid; + n->nd_args = nd_args; + + return n; +} + +static rb_node_opcall_t * +rb_node_opcall_new(struct parser_params *p, NODE *nd_recv, ID nd_mid, NODE *nd_args, const YYLTYPE *loc) +{ + rb_node_opcall_t *n = NODE_NEWNODE(NODE_OPCALL, rb_node_opcall_t, loc); + n->nd_recv = nd_recv; + n->nd_mid = nd_mid; + n->nd_args = nd_args; + + return n; +} + +static rb_node_fcall_t * +rb_node_fcall_new(struct parser_params *p, ID nd_mid, NODE *nd_args, const YYLTYPE *loc) +{ + rb_node_fcall_t *n = NODE_NEWNODE(NODE_FCALL, rb_node_fcall_t, loc); + n->nd_mid = nd_mid; + n->nd_args = nd_args; + + return n; +} + +static rb_node_qcall_t * +rb_node_qcall_new(struct parser_params *p, NODE *nd_recv, ID nd_mid, NODE *nd_args, const YYLTYPE *loc) +{ + rb_node_qcall_t *n = NODE_NEWNODE(NODE_QCALL, rb_node_qcall_t, loc); + n->nd_recv = nd_recv; + n->nd_mid = nd_mid; + n->nd_args = nd_args; + + return n; +} + +static rb_node_vcall_t * +rb_node_vcall_new(struct parser_params *p, ID nd_mid, const YYLTYPE *loc) +{ + rb_node_vcall_t *n = NODE_NEWNODE(NODE_VCALL, rb_node_vcall_t, loc); + n->nd_mid = nd_mid; + + return n; +} + +static rb_node_once_t * +rb_node_once_new(struct parser_params *p, NODE *nd_body, const YYLTYPE *loc) +{ + rb_node_once_t *n = NODE_NEWNODE(NODE_ONCE, rb_node_once_t, loc); + n->nd_body = nd_body; + + return n; +} + +static rb_node_args_t * +rb_node_args_new(struct parser_params *p, const YYLTYPE *loc) +{ + rb_node_args_t *n = NODE_NEWNODE(NODE_ARGS, rb_node_args_t, loc); + MEMZERO(&n->nd_ainfo, struct rb_args_info, 1); + + return n; +} + +static rb_node_args_aux_t * +rb_node_args_aux_new(struct parser_params *p, ID nd_pid, int nd_plen, const YYLTYPE *loc) +{ + rb_node_args_aux_t *n = NODE_NEWNODE(NODE_ARGS_AUX, rb_node_args_aux_t, loc); + n->nd_pid = nd_pid; + n->nd_plen = nd_plen; + n->nd_next = 0; + + return n; +} + +static rb_node_opt_arg_t * +rb_node_opt_arg_new(struct parser_params *p, NODE *nd_body, const YYLTYPE *loc) +{ + rb_node_opt_arg_t *n = NODE_NEWNODE(NODE_OPT_ARG, rb_node_opt_arg_t, loc); + n->nd_body = nd_body; + n->nd_next = 0; + + return n; +} + +static rb_node_kw_arg_t * +rb_node_kw_arg_new(struct parser_params *p, NODE *nd_body, const YYLTYPE *loc) +{ + rb_node_kw_arg_t *n = NODE_NEWNODE(NODE_KW_ARG, rb_node_kw_arg_t, loc); + n->nd_body = nd_body; + n->nd_next = 0; + + return n; +} + +static rb_node_postarg_t * +rb_node_postarg_new(struct parser_params *p, NODE *nd_1st, NODE *nd_2nd, const YYLTYPE *loc) +{ + rb_node_postarg_t *n = NODE_NEWNODE(NODE_POSTARG, rb_node_postarg_t, loc); + n->nd_1st = nd_1st; + n->nd_2nd = nd_2nd; + + return n; +} + +static rb_node_argscat_t * +rb_node_argscat_new(struct parser_params *p, NODE *nd_head, NODE *nd_body, const YYLTYPE *loc) +{ + rb_node_argscat_t *n = NODE_NEWNODE(NODE_ARGSCAT, rb_node_argscat_t, loc); + n->nd_head = nd_head; + n->nd_body = nd_body; + + return n; +} + +static rb_node_argspush_t * +rb_node_argspush_new(struct parser_params *p, NODE *nd_head, NODE *nd_body, const YYLTYPE *loc) +{ + rb_node_argspush_t *n = NODE_NEWNODE(NODE_ARGSPUSH, rb_node_argspush_t, loc); + n->nd_head = nd_head; + n->nd_body = nd_body; + + return n; +} + +static rb_node_splat_t * +rb_node_splat_new(struct parser_params *p, NODE *nd_head, const YYLTYPE *loc) +{ + rb_node_splat_t *n = NODE_NEWNODE(NODE_SPLAT, rb_node_splat_t, loc); + n->nd_head = nd_head; + + return n; +} + +static rb_node_block_pass_t * +rb_node_block_pass_new(struct parser_params *p, NODE *nd_body, const YYLTYPE *loc) +{ + rb_node_block_pass_t *n = NODE_NEWNODE(NODE_BLOCK_PASS, rb_node_block_pass_t, loc); + n->nd_head = 0; + n->nd_body = nd_body; + + return n; +} + +static rb_node_alias_t * +rb_node_alias_new(struct parser_params *p, NODE *nd_1st, NODE *nd_2nd, const YYLTYPE *loc) +{ + rb_node_alias_t *n = NODE_NEWNODE(NODE_ALIAS, rb_node_alias_t, loc); + n->nd_1st = nd_1st; + n->nd_2nd = nd_2nd; + + return n; +} + +static rb_node_valias_t * +rb_node_valias_new(struct parser_params *p, ID nd_alias, ID nd_orig, const YYLTYPE *loc) +{ + rb_node_valias_t *n = NODE_NEWNODE(NODE_VALIAS, rb_node_valias_t, loc); + n->nd_alias = nd_alias; + n->nd_orig = nd_orig; + + return n; +} + +static rb_node_undef_t * +rb_node_undef_new(struct parser_params *p, NODE *nd_undef, const YYLTYPE *loc) +{ + rb_node_undef_t *n = NODE_NEWNODE(NODE_UNDEF, rb_node_undef_t, loc); + n->nd_undef = nd_undef; + + return n; +} + +static rb_node_errinfo_t * +rb_node_errinfo_new(struct parser_params *p, const YYLTYPE *loc) +{ + rb_node_errinfo_t *n = NODE_NEWNODE(NODE_ERRINFO, rb_node_errinfo_t, loc); + + return n; +} + +static rb_node_defined_t * +rb_node_defined_new(struct parser_params *p, NODE *nd_head, const YYLTYPE *loc) +{ + rb_node_defined_t *n = NODE_NEWNODE(NODE_DEFINED, rb_node_defined_t, loc); + n->nd_head = nd_head; + + return n; +} + +static rb_node_postexe_t * +rb_node_postexe_new(struct parser_params *p, NODE *nd_body, const YYLTYPE *loc) +{ + rb_node_postexe_t *n = NODE_NEWNODE(NODE_POSTEXE, rb_node_postexe_t, loc); + n->nd_body = nd_body; + + return n; +} + +static rb_node_attrasgn_t * +rb_node_attrasgn_new(struct parser_params *p, NODE *nd_recv, ID nd_mid, NODE *nd_args, const YYLTYPE *loc) +{ + rb_node_attrasgn_t *n = NODE_NEWNODE(NODE_ATTRASGN, rb_node_attrasgn_t, loc); + n->nd_recv = nd_recv; + n->nd_mid = nd_mid; + n->nd_args = nd_args; + + return n; +} + +static rb_node_aryptn_t * +rb_node_aryptn_new(struct parser_params *p, NODE *pre_args, NODE *rest_arg, NODE *post_args, const YYLTYPE *loc) +{ + rb_node_aryptn_t *n = NODE_NEWNODE(NODE_ARYPTN, rb_node_aryptn_t, loc); + n->nd_pconst = 0; + n->pre_args = pre_args; + n->rest_arg = rest_arg; + n->post_args = post_args; + + return n; +} + +static rb_node_hshptn_t * +rb_node_hshptn_new(struct parser_params *p, NODE *nd_pconst, NODE *nd_pkwargs, NODE *nd_pkwrestarg, const YYLTYPE *loc) +{ + rb_node_hshptn_t *n = NODE_NEWNODE(NODE_HSHPTN, rb_node_hshptn_t, loc); + n->nd_pconst = nd_pconst; + n->nd_pkwargs = nd_pkwargs; + n->nd_pkwrestarg = nd_pkwrestarg; + + return n; +} + +static rb_node_fndptn_t * +rb_node_fndptn_new(struct parser_params *p, NODE *pre_rest_arg, NODE *args, NODE *post_rest_arg, const YYLTYPE *loc) +{ + rb_node_fndptn_t *n = NODE_NEWNODE(NODE_FNDPTN, rb_node_fndptn_t, loc); + n->nd_pconst = 0; + n->pre_rest_arg = pre_rest_arg; + n->args = args; + n->post_rest_arg = post_rest_arg; + + return n; +} + +static rb_node_line_t * +rb_node_line_new(struct parser_params *p, const YYLTYPE *loc) +{ + rb_node_line_t *n = NODE_NEWNODE(NODE_LINE, rb_node_line_t, loc); + + return n; +} + +static rb_node_file_t * +rb_node_file_new(struct parser_params *p, rb_parser_string_t *str, const YYLTYPE *loc) +{ + rb_node_file_t *n = NODE_NEWNODE(NODE_FILE, rb_node_file_t, loc); + n->path = str; + + return n; +} + +static rb_node_encoding_t * +rb_node_encoding_new(struct parser_params *p, const YYLTYPE *loc) +{ + rb_node_encoding_t *n = NODE_NEWNODE(NODE_ENCODING, rb_node_encoding_t, loc); + n->enc = p->enc; + + return n; +} + +static rb_node_cdecl_t * +rb_node_cdecl_new(struct parser_params *p, ID nd_vid, NODE *nd_value, NODE *nd_else, enum rb_parser_shareability shareability, const YYLTYPE *loc) +{ + rb_node_cdecl_t *n = NODE_NEWNODE(NODE_CDECL, rb_node_cdecl_t, loc); + n->nd_vid = nd_vid; + n->nd_value = nd_value; + n->nd_else = nd_else; + n->shareability = shareability; + + return n; +} + +static rb_node_op_cdecl_t * +rb_node_op_cdecl_new(struct parser_params *p, NODE *nd_head, NODE *nd_value, ID nd_aid, enum rb_parser_shareability shareability, const YYLTYPE *loc) +{ + rb_node_op_cdecl_t *n = NODE_NEWNODE(NODE_OP_CDECL, rb_node_op_cdecl_t, loc); + n->nd_head = nd_head; + n->nd_value = nd_value; + n->nd_aid = nd_aid; + n->shareability = shareability; + + return n; +} + +static rb_node_error_t * +rb_node_error_new(struct parser_params *p, const YYLTYPE *loc) +{ + rb_node_error_t *n = NODE_NEWNODE(NODE_ERROR, rb_node_error_t, loc); + + return n; +} + +static rb_node_break_t * +rb_node_break_new(struct parser_params *p, NODE *nd_stts, const YYLTYPE *loc) +{ + rb_node_break_t *n = NODE_NEWNODE(NODE_BREAK, rb_node_break_t, loc); + n->nd_stts = nd_stts; + n->nd_chain = 0; + + return n; +} + +static rb_node_next_t * +rb_node_next_new(struct parser_params *p, NODE *nd_stts, const YYLTYPE *loc) +{ + rb_node_next_t *n = NODE_NEWNODE(NODE_NEXT, rb_node_next_t, loc); + n->nd_stts = nd_stts; + n->nd_chain = 0; + + return n; +} + +static rb_node_redo_t * +rb_node_redo_new(struct parser_params *p, const YYLTYPE *loc) +{ + rb_node_redo_t *n = NODE_NEWNODE(NODE_REDO, rb_node_redo_t, loc); + n->nd_chain = 0; + + return n; +} + +static rb_node_def_temp_t * +rb_node_def_temp_new(struct parser_params *p, const YYLTYPE *loc) +{ + rb_node_def_temp_t *n = NODE_NEWNODE((enum node_type)NODE_DEF_TEMP, rb_node_def_temp_t, loc); + n->save.cur_arg = p->cur_arg; + n->save.numparam_save = 0; + n->save.max_numparam = 0; + n->save.ctxt = p->ctxt; + n->nd_def = 0; + n->nd_mid = 0; + + return n; +} + +static rb_node_def_temp_t * +def_head_save(struct parser_params *p, rb_node_def_temp_t *n) +{ + n->save.numparam_save = numparam_push(p); + n->save.max_numparam = p->max_numparam; + return n; } #ifndef RIPPER @@ -10545,13 +12579,14 @@ nodeline(NODE *node) { return nd_line(node); } +#endif static NODE* newline_node(NODE *node) { if (node) { node = remove_begin(node); - node->flags |= NODE_FL_NEWLINE; + nd_set_fl_newline(node); } return node; } @@ -10564,18 +12599,6 @@ fixpos(NODE *node, NODE *orig) nd_set_line(node, nd_line(orig)); } -static void -parser_warning(struct parser_params *p, NODE *node, const char *mesg) -{ - rb_compile_warning(p->ruby_sourcefile, nd_line(node), "%s", mesg); -} - -static void -parser_warn(struct parser_params *p, NODE *node, const char *mesg) -{ - rb_compile_warn(p->ruby_sourcefile, nd_line(node), "%s", mesg); -} - static NODE* block_append(struct parser_params *p, NODE *head, NODE *tail) { @@ -10585,34 +12608,23 @@ block_append(struct parser_params *p, NODE *head, NODE *tail) if (h == 0) return tail; switch (nd_type(h)) { - case NODE_LIT: - case NODE_STR: - case NODE_SELF: - case NODE_TRUE: - case NODE_FALSE: - case NODE_NIL: - parser_warning(p, h, "unused literal ignored"); - return tail; default: h = end = NEW_BLOCK(head, &head->nd_loc); - end->nd_end = end; head = end; break; case NODE_BLOCK: - end = h->nd_end; + end = RNODE_BLOCK(h)->nd_end; break; } - nd = end->nd_head; + nd = RNODE_BLOCK(end)->nd_head; switch (nd_type(nd)) { case NODE_RETURN: case NODE_BREAK: case NODE_NEXT: case NODE_REDO: case NODE_RETRY: - if (RTEST(ruby_verbose)) { - parser_warning(p, tail, "statement not reached"); - } + rb_warning0L(nd_line(tail), "statement not reached"); break; default: @@ -10621,10 +12633,9 @@ block_append(struct parser_params *p, NODE *head, NODE *tail) if (!nd_type_p(tail, NODE_BLOCK)) { tail = NEW_BLOCK(tail, &tail->nd_loc); - tail->nd_end = tail; } - end->nd_next = tail; - h->nd_end = tail->nd_end; + RNODE_BLOCK(end)->nd_next = tail; + RNODE_BLOCK(h)->nd_end = RNODE_BLOCK(tail)->nd_end; nd_set_last_loc(head, nd_last_loc(tail)); return head; } @@ -10636,16 +12647,16 @@ list_append(struct parser_params *p, NODE *list, NODE *item) NODE *last; if (list == 0) return NEW_LIST(item, &item->nd_loc); - if (list->nd_next) { - last = list->nd_next->nd_end; + if (RNODE_LIST(list)->nd_next) { + last = RNODE_LIST(RNODE_LIST(list)->nd_next)->as.nd_end; } else { last = list; } - list->nd_alen += 1; - last->nd_next = NEW_LIST(item, &item->nd_loc); - list->nd_next->nd_end = last->nd_next; + RNODE_LIST(list)->as.nd_alen += 1; + RNODE_LIST(last)->nd_next = NEW_LIST(item, &item->nd_loc); + RNODE_LIST(RNODE_LIST(list)->nd_next)->as.nd_end = RNODE_LIST(last)->nd_next; nd_set_last_loc(list, nd_last_loc(item)); @@ -10658,20 +12669,20 @@ list_concat(NODE *head, NODE *tail) { NODE *last; - if (head->nd_next) { - last = head->nd_next->nd_end; + if (RNODE_LIST(head)->nd_next) { + last = RNODE_LIST(RNODE_LIST(head)->nd_next)->as.nd_end; } else { last = head; } - head->nd_alen += tail->nd_alen; - last->nd_next = tail; - if (tail->nd_next) { - head->nd_next->nd_end = tail->nd_next->nd_end; + RNODE_LIST(head)->as.nd_alen += RNODE_LIST(tail)->as.nd_alen; + RNODE_LIST(last)->nd_next = tail; + if (RNODE_LIST(tail)->nd_next) { + RNODE_LIST(RNODE_LIST(head)->nd_next)->as.nd_end = RNODE_LIST(RNODE_LIST(tail)->nd_next)->as.nd_end; } else { - head->nd_next->nd_end = tail; + RNODE_LIST(RNODE_LIST(head)->nd_next)->as.nd_end = tail; } nd_set_last_loc(head, nd_last_loc(tail)); @@ -10680,40 +12691,51 @@ list_concat(NODE *head, NODE *tail) } static int -literal_concat0(struct parser_params *p, VALUE head, VALUE tail) +literal_concat0(struct parser_params *p, rb_parser_string_t *head, rb_parser_string_t *tail) { - if (NIL_P(tail)) return 1; - if (!rb_enc_compatible(head, tail)) { + if (!tail) return 1; + if (!rb_parser_enc_compatible(p, head, tail)) { compile_error(p, "string literal encodings differ (%s / %s)", - rb_enc_name(rb_enc_get(head)), - rb_enc_name(rb_enc_get(tail))); - rb_str_resize(head, 0); - rb_str_resize(tail, 0); + rb_enc_name(rb_parser_str_get_encoding(head)), + rb_enc_name(rb_parser_str_get_encoding(tail))); + rb_parser_str_resize(p, head, 0); + rb_parser_str_resize(p, tail, 0); return 0; } - rb_str_buf_append(head, tail); + rb_parser_str_buf_append(p, head, tail); return 1; } -static VALUE -string_literal_head(enum node_type htype, NODE *head) +static rb_parser_string_t * +string_literal_head(struct parser_params *p, enum node_type htype, NODE *head) { - if (htype != NODE_DSTR) return Qfalse; - if (head->nd_next) { - head = head->nd_next->nd_end->nd_head; - if (!head || !nd_type_p(head, NODE_STR)) return Qfalse; + if (htype != NODE_DSTR) return NULL; + if (RNODE_DSTR(head)->nd_next) { + head = RNODE_LIST(RNODE_LIST(RNODE_DSTR(head)->nd_next)->as.nd_end)->nd_head; + if (!head || !nd_type_p(head, NODE_STR)) return NULL; } - const VALUE lit = head->nd_lit; - ASSUME(lit != Qfalse); + rb_parser_string_t *lit = RNODE_DSTR(head)->string; + ASSUME(lit); return lit; } +static rb_parser_string_t * +rb_parser_string_deep_copy(struct parser_params *p, const rb_parser_string_t *orig) +{ + rb_parser_string_t *copy; + if (!orig) return NULL; + copy = rb_parser_string_new(p, PARSER_STRING_PTR(orig), PARSER_STRING_LEN(orig)); + copy->coderange = orig->coderange; + copy->enc = orig->enc; + return copy; +} + /* concat two string literals */ static NODE * literal_concat(struct parser_params *p, NODE *head, NODE *tail, const YYLTYPE *loc) { enum node_type htype; - VALUE lit; + rb_parser_string_t *lit; if (!head) return tail; if (!tail) return head; @@ -10726,7 +12748,7 @@ literal_concat(struct parser_params *p, NODE *head, NODE *tail, const YYLTYPE *l if (p->heredoc_indent > 0) { switch (htype) { case NODE_STR: - nd_set_type(head, NODE_DSTR); + head = str2dstr(p, head); case NODE_DSTR: return list_append(p, head, tail); default: @@ -10735,14 +12757,14 @@ literal_concat(struct parser_params *p, NODE *head, NODE *tail, const YYLTYPE *l } switch (nd_type(tail)) { case NODE_STR: - if ((lit = string_literal_head(htype, head)) != Qfalse) { + if ((lit = string_literal_head(p, htype, head)) != false) { htype = NODE_STR; } else { - lit = head->nd_lit; + lit = RNODE_DSTR(head)->string; } if (htype == NODE_STR) { - if (!literal_concat0(p, lit, tail->nd_lit)) { + if (!literal_concat0(p, lit, RNODE_STR(tail)->string)) { error: rb_discard_node(p, head); rb_discard_node(p, tail); @@ -10757,39 +12779,43 @@ literal_concat(struct parser_params *p, NODE *head, NODE *tail, const YYLTYPE *l case NODE_DSTR: if (htype == NODE_STR) { - if (!literal_concat0(p, head->nd_lit, tail->nd_lit)) + if (!literal_concat0(p, RNODE_STR(head)->string, RNODE_DSTR(tail)->string)) goto error; - tail->nd_lit = head->nd_lit; + rb_parser_string_free(p, RNODE_DSTR(tail)->string); + RNODE_DSTR(tail)->string = RNODE_STR(head)->string; + RNODE_STR(head)->string = NULL; rb_discard_node(p, head); head = tail; } - else if (NIL_P(tail->nd_lit)) { + else if (!RNODE_DSTR(tail)->string) { append: - head->nd_alen += tail->nd_alen - 1; - if (!head->nd_next) { - head->nd_next = tail->nd_next; + RNODE_DSTR(head)->as.nd_alen += RNODE_DSTR(tail)->as.nd_alen - 1; + if (!RNODE_DSTR(head)->nd_next) { + RNODE_DSTR(head)->nd_next = RNODE_DSTR(tail)->nd_next; } - else if (tail->nd_next) { - head->nd_next->nd_end->nd_next = tail->nd_next; - head->nd_next->nd_end = tail->nd_next->nd_end; + else if (RNODE_DSTR(tail)->nd_next) { + RNODE_DSTR(RNODE_DSTR(RNODE_DSTR(head)->nd_next)->as.nd_end)->nd_next = RNODE_DSTR(tail)->nd_next; + RNODE_DSTR(RNODE_DSTR(head)->nd_next)->as.nd_end = RNODE_DSTR(RNODE_DSTR(tail)->nd_next)->as.nd_end; } rb_discard_node(p, tail); } - else if ((lit = string_literal_head(htype, head)) != Qfalse) { - if (!literal_concat0(p, lit, tail->nd_lit)) + else if ((lit = string_literal_head(p, htype, head)) != false) { + if (!literal_concat0(p, lit, RNODE_DSTR(tail)->string)) goto error; - tail->nd_lit = Qnil; + rb_parser_string_free(p, RNODE_DSTR(tail)->string); + RNODE_DSTR(tail)->string = 0; goto append; } else { - list_concat(head, NEW_NODE(NODE_LIST, NEW_STR(tail->nd_lit, loc), tail->nd_alen, tail->nd_next, loc)); + list_concat(head, NEW_LIST2(NEW_STR(RNODE_DSTR(tail)->string, loc), RNODE_DSTR(tail)->as.nd_alen, (NODE *)RNODE_DSTR(tail)->nd_next, loc)); + RNODE_DSTR(tail)->string = 0; } break; case NODE_EVSTR: if (htype == NODE_STR) { - nd_set_type(head, NODE_DSTR); - head->nd_alen = 1; + head = str2dstr(p, head); + RNODE_DSTR(head)->as.nd_alen = 1; } list_append(p, head, tail); break; @@ -10797,6 +12823,40 @@ literal_concat(struct parser_params *p, NODE *head, NODE *tail, const YYLTYPE *l return head; } +static void +nd_copy_flag(NODE *new_node, NODE *old_node) +{ + if (nd_fl_newline(old_node)) nd_set_fl_newline(new_node); + nd_set_line(new_node, nd_line(old_node)); + new_node->nd_loc = old_node->nd_loc; + new_node->node_id = old_node->node_id; +} + +static NODE * +str2dstr(struct parser_params *p, NODE *node) +{ + NODE *new_node = (NODE *)NODE_NEW_INTERNAL(NODE_DSTR, rb_node_dstr_t); + nd_copy_flag(new_node, node); + RNODE_DSTR(new_node)->string = RNODE_STR(node)->string; + RNODE_DSTR(new_node)->as.nd_alen = 0; + RNODE_DSTR(new_node)->nd_next = 0; + RNODE_STR(node)->string = 0; + + return new_node; +} + +static NODE * +str2regx(struct parser_params *p, NODE *node, int options) +{ + NODE *new_node = (NODE *)NODE_NEW_INTERNAL(NODE_REGX, rb_node_regx_t); + nd_copy_flag(new_node, node); + RNODE_REGX(new_node)->string = RNODE_STR(node)->string; + RNODE_REGX(new_node)->options = options; + RNODE_STR(node)->string = 0; + + return new_node; +} + static NODE * evstr2dstr(struct parser_params *p, NODE *node) { @@ -10814,8 +12874,7 @@ new_evstr(struct parser_params *p, NODE *node, const YYLTYPE *loc) if (node) { switch (nd_type(node)) { case NODE_STR: - nd_set_type(node, NODE_DSTR); - return node; + return str2dstr(p, node); case NODE_DSTR: break; case NODE_EVSTR: @@ -10828,9 +12887,7 @@ new_evstr(struct parser_params *p, NODE *node, const YYLTYPE *loc) static NODE * new_dstr(struct parser_params *p, NODE *node, const YYLTYPE *loc) { - VALUE lit = STR_NEW0(); - NODE *dstr = NEW_DSTR(lit, loc); - RB_OBJ_WRITTEN(p->ast, Qnil, lit); + NODE *dstr = NEW_DSTR(STRING_NEW0(), loc); return list_append(p, dstr, node); } @@ -10875,7 +12932,15 @@ new_command_qcall(struct parser_params* p, ID atype, NODE *recv, ID mid, NODE *a return ret; } -#define nd_once_body(node) (nd_type_p((node), NODE_ONCE) ? (node)->nd_body : node) +#define nd_once_body(node) (nd_type_p((node), NODE_ONCE) ? RNODE_ONCE(node)->nd_body : node) + +static NODE* +last_expr_once_body(NODE *node) +{ + if (!node) return 0; + return nd_once_body(node); +} + static NODE* match_op(struct parser_params *p, NODE *node1, NODE *node2, const YYLTYPE *op_loc, const YYLTYPE *loc) { @@ -10884,7 +12949,8 @@ match_op(struct parser_params *p, NODE *node1, NODE *node2, const YYLTYPE *op_lo value_expr(node1); value_expr(node2); - if (node1 && (n = nd_once_body(node1)) != 0) { + + if ((n = last_expr_once_body(node1)) != 0) { switch (nd_type(n)) { case NODE_DREGX: { @@ -10893,24 +12959,23 @@ match_op(struct parser_params *p, NODE *node1, NODE *node2, const YYLTYPE *op_lo return match; } - case NODE_LIT: - if (RB_TYPE_P(n->nd_lit, T_REGEXP)) { - const VALUE lit = n->nd_lit; - NODE *match = NEW_MATCH2(node1, node2, loc); - match->nd_args = reg_named_capture_assign(p, lit, loc); - nd_set_line(match, line); - return match; + case NODE_REGX: + { + const VALUE lit = rb_node_regx_string_val(n); + if (!NIL_P(lit)) { + NODE *match = NEW_MATCH2(node1, node2, loc); + RNODE_MATCH2(match)->nd_args = reg_named_capture_assign(p, lit, loc); + nd_set_line(match, line); + return match; + } } } } - if (node2 && (n = nd_once_body(node2)) != 0) { + if ((n = last_expr_once_body(node2)) != 0) { NODE *match3; switch (nd_type(n)) { - case NODE_LIT: - if (!RB_TYPE_P(n->nd_lit, T_REGEXP)) break; - /* fallthru */ case NODE_DREGX: match3 = NEW_MATCH3(node2, node1, loc); return match3; @@ -10953,6 +13018,34 @@ numparam_nested_p(struct parser_params *p) return 0; } +static int +numparam_used_p(struct parser_params *p) +{ + NODE *numparam = p->lvtbl->numparam.current; + if (numparam) { + compile_error(p, "numbered parameter is already used in\n" + "%s:%d: current block here", + p->ruby_sourcefile, nd_line(numparam)); + parser_show_error_line(p, &numparam->nd_loc); + return 1; + } + return 0; +} + +static int +it_used_p(struct parser_params *p) +{ + NODE *it = p->lvtbl->it; + if (it) { + compile_error(p, "'it' is already used in\n" + "%s:%d: current block here", + p->ruby_sourcefile, nd_line(it)); + parser_show_error_line(p, &it->nd_loc); + return 1; + } + return 0; +} + static NODE* gettable(struct parser_params *p, ID id, const YYLTYPE *loc) { @@ -10969,27 +13062,26 @@ gettable(struct parser_params *p, ID id, const YYLTYPE *loc) return NEW_FALSE(loc); case keyword__FILE__: { - VALUE file = p->ruby_sourcefile_string; - if (NIL_P(file)) - file = rb_str_new(0, 0); - else - file = rb_str_dup(file); - node = NEW_STR(file, loc); - RB_OBJ_WRITTEN(p->ast, Qnil, file); + rb_parser_string_t *file; + if (p->ruby_sourcefile_string) { + file = rb_parser_string_deep_copy(p, p->ruby_sourcefile_string); + } + else { + file = STRING_NEW0(); + } + node = NEW_FILE(file, loc); } return node; case keyword__LINE__: - return NEW_LIT(INT2FIX(p->tokline), loc); + return NEW_LINE(loc); case keyword__ENCODING__: - node = NEW_LIT(rb_enc_from_encoding(p->enc), loc); - RB_OBJ_WRITTEN(p->ast, Qnil, node->nd_lit); - return node; + return NEW_ENCODING(loc); } switch (id_type(id)) { case ID_LOCAL: if (dyna_in_block(p) && dvar_defined_ref(p, id, &vidp)) { - if (NUMPARAM_ID_P(id) && numparam_nested_p(p)) return 0; + if (NUMPARAM_ID_P(id) && (numparam_nested_p(p) || it_used_p(p))) return 0; if (id == p->cur_arg) { compile_error(p, "circular argument reference - %"PRIsWARN, rb_id2str(id)); return 0; @@ -11009,7 +13101,7 @@ gettable(struct parser_params *p, ID id, const YYLTYPE *loc) } if (dyna_in_block(p) && NUMPARAM_ID_P(id) && parser_numbered_param(p, NUMPARAM_ID_TO_IDX(id))) { - if (numparam_nested_p(p)) return 0; + if (numparam_nested_p(p) || it_used_p(p)) return 0; node = NEW_DVAR(id, loc); struct local_vars *local = p->lvtbl; if (!local->numparam.current) local->numparam.current = node; @@ -11021,6 +13113,20 @@ gettable(struct parser_params *p, ID id, const YYLTYPE *loc) } # endif /* method call without arguments */ + if (dyna_in_block(p) && id == rb_intern("it") && !(DVARS_TERMINAL_P(p->lvtbl->args) || DVARS_TERMINAL_P(p->lvtbl->args->prev))) { + if (numparam_used_p(p)) return 0; + if (p->max_numparam == ORDINAL_PARAM) { + compile_error(p, "ordinary parameter is defined"); + return 0; + } + if (!p->it_id) { + p->it_id = internal_id(p); + vtable_add(p->lvtbl->args, p->it_id); + } + NODE *node = NEW_DVAR(p->it_id, loc); + if (!p->lvtbl->it) p->lvtbl->it = node; + return node; + } return NEW_VCALL(id, loc); case ID_GLOBAL: return NEW_GVAR(id, loc); @@ -11035,26 +13141,27 @@ gettable(struct parser_params *p, ID id, const YYLTYPE *loc) return 0; } -static NODE * -opt_arg_append(NODE *opt_list, NODE *opt) +static rb_node_opt_arg_t * +opt_arg_append(rb_node_opt_arg_t *opt_list, rb_node_opt_arg_t *opt) { - NODE *opts = opt_list; - opts->nd_loc.end_pos = opt->nd_loc.end_pos; + rb_node_opt_arg_t *opts = opt_list; + RNODE(opts)->nd_loc.end_pos = RNODE(opt)->nd_loc.end_pos; while (opts->nd_next) { opts = opts->nd_next; - opts->nd_loc.end_pos = opt->nd_loc.end_pos; + RNODE(opts)->nd_loc.end_pos = RNODE(opt)->nd_loc.end_pos; } opts->nd_next = opt; return opt_list; } -static NODE * -kwd_append(NODE *kwlist, NODE *kw) +static rb_node_kw_arg_t * +kwd_append(rb_node_kw_arg_t *kwlist, rb_node_kw_arg_t *kw) { if (kwlist) { - opt_arg_append(kwlist, kw); + /* Assume rb_node_kw_arg_t and rb_node_opt_arg_t has same structure */ + opt_arg_append(RNODE_OPT_ARG(kwlist), RNODE_OPT_ARG(kw)); } return kwlist; } @@ -11062,7 +13169,34 @@ kwd_append(NODE *kwlist, NODE *kw) static NODE * new_defined(struct parser_params *p, NODE *expr, const YYLTYPE *loc) { - return NEW_DEFINED(remove_begin_all(expr), loc); + NODE *n = expr; + while (n) { + if (nd_type_p(n, NODE_BEGIN)) { + n = RNODE_BEGIN(n)->nd_body; + } + else if (nd_type_p(n, NODE_BLOCK) && RNODE_BLOCK(n)->nd_end == n) { + n = RNODE_BLOCK(n)->nd_head; + } + else { + break; + } + } + return NEW_DEFINED(n, loc); +} + +static NODE* +str_to_sym_node(struct parser_params *p, NODE *node, const YYLTYPE *loc) +{ + VALUE lit; + rb_parser_string_t *str = RNODE_STR(node)->string; + if (rb_parser_enc_str_coderange(p, str) == RB_PARSER_ENC_CODERANGE_BROKEN) { + yyerror1(loc, "invalid symbol"); + lit = STR_NEW0(); + } + else { + lit = rb_str_new_parser_string(str); + } + return NEW_SYM(lit, loc); } static NODE* @@ -11074,11 +13208,10 @@ symbol_append(struct parser_params *p, NODE *symbols, NODE *symbol) nd_set_type(symbol, NODE_DSYM); break; case NODE_STR: - nd_set_type(symbol, NODE_LIT); - RB_OBJ_WRITTEN(p->ast, Qnil, symbol->nd_lit = rb_str_intern(symbol->nd_lit)); + symbol = str_to_sym_node(p, symbol, &RNODE(symbol)->nd_loc); break; default: - compile_error(p, "unexpected node as symbol: %s", ruby_node_name(type)); + compile_error(p, "unexpected node as symbol: %s", parser_node_name(type)); } return list_append(p, symbols, symbol); } @@ -11086,84 +13219,80 @@ symbol_append(struct parser_params *p, NODE *symbols, NODE *symbol) static NODE * new_regexp(struct parser_params *p, NODE *node, int options, const YYLTYPE *loc) { - NODE *list, *prev; - VALUE lit; + struct RNode_LIST *list; + NODE *prev; if (!node) { - node = NEW_LIT(reg_compile(p, STR_NEW0(), options), loc); - RB_OBJ_WRITTEN(p->ast, Qnil, node->nd_lit); + /* Check string is valid regex */ + rb_parser_string_t *str = STRING_NEW0(); + reg_compile(p, str, options); + node = NEW_REGX(str, options, loc); return node; } switch (nd_type(node)) { case NODE_STR: { - VALUE src = node->nd_lit; - nd_set_type(node, NODE_LIT); - nd_set_loc(node, loc); - RB_OBJ_WRITTEN(p->ast, Qnil, node->nd_lit = reg_compile(p, src, options)); + /* Check string is valid regex */ + reg_compile(p, RNODE_STR(node)->string, options); + node = str2regx(p, node, options); } break; default: - lit = STR_NEW0(); - node = NEW_NODE(NODE_DSTR, lit, 1, NEW_LIST(node, loc), loc); - RB_OBJ_WRITTEN(p->ast, Qnil, lit); + node = NEW_DSTR0(STRING_NEW0(), 1, NEW_LIST(node, loc), loc); /* fall through */ case NODE_DSTR: nd_set_type(node, NODE_DREGX); nd_set_loc(node, loc); - node->nd_cflag = options & RE_OPTION_MASK; - if (!NIL_P(node->nd_lit)) reg_fragment_check(p, node->nd_lit, options); - for (list = (prev = node)->nd_next; list; list = list->nd_next) { + RNODE_DREGX(node)->nd_cflag = options & RE_OPTION_MASK; + if (RNODE_DREGX(node)->string) reg_fragment_check(p, RNODE_DREGX(node)->string, options); + for (list = RNODE_DREGX(prev = node)->nd_next; list; list = RNODE_LIST(list->nd_next)) { NODE *frag = list->nd_head; enum node_type type = nd_type(frag); - if (type == NODE_STR || (type == NODE_DSTR && !frag->nd_next)) { - VALUE tail = frag->nd_lit; - if (reg_fragment_check(p, tail, options) && prev && !NIL_P(prev->nd_lit)) { - VALUE lit = prev == node ? prev->nd_lit : prev->nd_head->nd_lit; + if (type == NODE_STR || (type == NODE_DSTR && !RNODE_DSTR(frag)->nd_next)) { + rb_parser_string_t *tail = RNODE_STR(frag)->string; + if (reg_fragment_check(p, tail, options) && prev && RNODE_DREGX(prev)->string) { + rb_parser_string_t *lit = prev == node ? RNODE_DREGX(prev)->string : RNODE_STR(RNODE_LIST(prev)->nd_head)->string; if (!literal_concat0(p, lit, tail)) { return NEW_NIL(loc); /* dummy node on error */ } - rb_str_resize(tail, 0); - prev->nd_next = list->nd_next; + rb_parser_str_resize(p, tail, 0); + RNODE_LIST(prev)->nd_next = list->nd_next; rb_discard_node(p, list->nd_head); - rb_discard_node(p, list); - list = prev; + rb_discard_node(p, (NODE *)list); + list = RNODE_LIST(prev); } else { - prev = list; + prev = (NODE *)list; } } else { prev = 0; } } - if (!node->nd_next) { - VALUE src = node->nd_lit; - nd_set_type(node, NODE_LIT); - RB_OBJ_WRITTEN(p->ast, Qnil, node->nd_lit = reg_compile(p, src, options)); + if (!RNODE_DREGX(node)->nd_next) { + /* Check string is valid regex */ + reg_compile(p, RNODE_DREGX(node)->string, options); } if (options & RE_OPTION_ONCE) { - node = NEW_NODE(NODE_ONCE, 0, node, 0, loc); + node = NEW_ONCE(node, loc); } break; } return node; } -static NODE * +static rb_node_kw_arg_t * new_kw_arg(struct parser_params *p, NODE *k, const YYLTYPE *loc) { if (!k) return 0; - return NEW_KW_ARG(0, (k), loc); + return NEW_KW_ARG((k), loc); } static NODE * new_xstring(struct parser_params *p, NODE *node, const YYLTYPE *loc) { if (!node) { - VALUE lit = STR_NEW0(); - NODE *xstr = NEW_XSTR(lit, loc); - RB_OBJ_WRITTEN(p->ast, Qnil, lit); + NODE *xstr = NEW_XSTR(STRING_NEW0(), loc); return xstr; } switch (nd_type(node)) { @@ -11176,40 +13305,42 @@ new_xstring(struct parser_params *p, NODE *node, const YYLTYPE *loc) nd_set_loc(node, loc); break; default: - node = NEW_NODE(NODE_DXSTR, Qnil, 1, NEW_LIST(node, loc), loc); + node = NEW_DXSTR(0, 1, NEW_LIST(node, loc), loc); break; } return node; } +static const +struct st_hash_type literal_type = { + literal_cmp, + literal_hash, +}; + +static int nd_type_st_key_enable_p(NODE *node); + static void check_literal_when(struct parser_params *p, NODE *arg, const YYLTYPE *loc) { - VALUE lit; - + /* See https://bugs.ruby-lang.org/issues/20331 for discussion about what is warned. */ if (!arg || !p->case_labels) return; + if (!nd_type_st_key_enable_p(arg)) return; - lit = rb_node_case_when_optimizable_literal(arg); - if (UNDEF_P(lit)) return; - if (nd_type_p(arg, NODE_STR)) { - RB_OBJ_WRITTEN(p->ast, Qnil, arg->nd_lit = lit); - } - - if (NIL_P(p->case_labels)) { - p->case_labels = rb_obj_hide(rb_hash_new()); + if (p->case_labels == CHECK_LITERAL_WHEN) { + p->case_labels = st_init_table(&literal_type); } else { - VALUE line = rb_hash_lookup(p->case_labels, lit); - if (!NIL_P(line)) { - rb_warning1("duplicated `when' clause with line %d is ignored", - WARN_IVAL(line)); + st_data_t line; + if (st_lookup(p->case_labels, (st_data_t)arg, &line)) { + rb_warning1("duplicated 'when' clause with line %d is ignored", + WARN_I((int)line)); return; } } - rb_hash_aset(p->case_labels, lit, INT2NUM(p->ruby_sourceline)); + st_insert(p->case_labels, (st_data_t)arg, (st_data_t)p->ruby_sourceline); } -#else /* !RIPPER */ +#ifdef RIPPER static int id_is_var(struct parser_params *p, ID id) { @@ -11229,26 +13360,7 @@ id_is_var(struct parser_params *p, ID id) compile_error(p, "identifier %"PRIsVALUE" is not valid to get", rb_id2str(id)); return 0; } - -static VALUE -new_regexp(struct parser_params *p, VALUE re, VALUE opt, const YYLTYPE *loc) -{ - VALUE src = 0, err; - int options = 0; - if (ripper_is_node_yylval(re)) { - src = RNODE(re)->nd_cval; - re = RNODE(re)->nd_rval; - } - if (ripper_is_node_yylval(opt)) { - options = (int)RNODE(opt)->nd_tag; - opt = RNODE(opt)->nd_rval; - } - if (src && NIL_P(parser_reg_compile(p, src, options, &err))) { - compile_error(p, "%"PRIsVALUE, err); - } - return dispatch2(regexp_literal, re, opt); -} -#endif /* !RIPPER */ +#endif static inline enum lex_state_e parser_set_lex_state(struct parser_params *p, enum lex_state_e ls, int line) @@ -11260,6 +13372,20 @@ parser_set_lex_state(struct parser_params *p, enum lex_state_e ls, int line) } #ifndef RIPPER +static void +flush_debug_buffer(struct parser_params *p, VALUE out, VALUE str) +{ + VALUE mesg = p->debug_buffer; + + if (!NIL_P(mesg) && RSTRING_LEN(mesg)) { + p->debug_buffer = Qnil; + rb_io_puts(1, &mesg, out); + } + if (!NIL_P(str) && RSTRING_LEN(str)) { + rb_io_write(p->debug_output, str); + } +} + static const char rb_parser_lex_state_names[][8] = { "BEG", "END", "ENDARG", "ENDFN", "ARG", "CMDARG", "MID", "FNAME", "DOT", "CLASS", @@ -11267,7 +13393,7 @@ static const char rb_parser_lex_state_names[][8] = { }; static VALUE -append_lex_state_name(enum lex_state_e state, VALUE buf) +append_lex_state_name(struct parser_params *p, enum lex_state_e state, VALUE buf) { int i, sep = 0; unsigned int mask = 1; @@ -11288,42 +13414,28 @@ append_lex_state_name(enum lex_state_e state, VALUE buf) return buf; } -static void -flush_debug_buffer(struct parser_params *p, VALUE out, VALUE str) -{ - VALUE mesg = p->debug_buffer; - - if (!NIL_P(mesg) && RSTRING_LEN(mesg)) { - p->debug_buffer = Qnil; - rb_io_puts(1, &mesg, out); - } - if (!NIL_P(str) && RSTRING_LEN(str)) { - rb_io_write(p->debug_output, str); - } -} - enum lex_state_e rb_parser_trace_lex_state(struct parser_params *p, enum lex_state_e from, enum lex_state_e to, int line) { VALUE mesg; mesg = rb_str_new_cstr("lex_state: "); - append_lex_state_name(from, mesg); + append_lex_state_name(p, from, mesg); rb_str_cat_cstr(mesg, " -> "); - append_lex_state_name(to, mesg); + append_lex_state_name(p, to, mesg); rb_str_catf(mesg, " at line %d\n", line); flush_debug_buffer(p, p->debug_output, mesg); return to; } VALUE -rb_parser_lex_state_name(enum lex_state_e state) +rb_parser_lex_state_name(struct parser_params *p, enum lex_state_e state) { - return rb_fstring(append_lex_state_name(state, rb_str_new(0, 0))); + return rb_str_to_interned_str(append_lex_state_name(p, state, rb_str_new(0, 0))); } static void -append_bitstack_value(stack_type stack, VALUE mesg) +append_bitstack_value(struct parser_params *p, stack_type stack, VALUE mesg) { if (stack == 0) { rb_str_cat_cstr(mesg, "0"); @@ -11340,7 +13452,7 @@ rb_parser_show_bitstack(struct parser_params *p, stack_type stack, const char *name, int line) { VALUE mesg = rb_sprintf("%s: ", name); - append_bitstack_value(stack, mesg); + append_bitstack_value(p, stack, mesg); rb_str_catf(mesg, " at line %d\n", line); flush_debug_buffer(p, p->debug_output, mesg); } @@ -11358,13 +13470,13 @@ rb_parser_fatal(struct parser_params *p, const char *fmt, ...) RB_GC_GUARD(mesg); mesg = rb_str_new(0, 0); - append_lex_state_name(p->lex.state, mesg); + append_lex_state_name(p, p->lex.state, mesg); compile_error(p, "lex.state: %"PRIsVALUE, mesg); rb_str_resize(mesg, 0); - append_bitstack_value(p->cond_stack, mesg); + append_bitstack_value(p, p->cond_stack, mesg); compile_error(p, "cond_stack: %"PRIsVALUE, mesg); rb_str_resize(mesg, 0); - append_bitstack_value(p->cmdarg_stack, mesg); + append_bitstack_value(p, p->cmdarg_stack, mesg); compile_error(p, "cmdarg_stack: %"PRIsVALUE, mesg); if (p->debug_output == rb_ractor_stdout()) p->debug_output = rb_ractor_stderr(); @@ -11498,7 +13610,6 @@ assignable0(struct parser_params *p, ID id, const char **err) return -1; } -#ifndef RIPPER static NODE* assignable(struct parser_params *p, ID id, NODE *val, const YYLTYPE *loc) { @@ -11509,25 +13620,20 @@ assignable(struct parser_params *p, ID id, NODE *val, const YYLTYPE *loc) case NODE_LASGN: return NEW_LASGN(id, val, loc); case NODE_GASGN: return NEW_GASGN(id, val, loc); case NODE_IASGN: return NEW_IASGN(id, val, loc); - case NODE_CDECL: return NEW_CDECL(id, val, 0, loc); + case NODE_CDECL: return NEW_CDECL(id, val, 0, p->ctxt.shareable_constant_value, loc); case NODE_CVASGN: return NEW_CVASGN(id, val, loc); } +/* TODO: FIXME */ +#ifndef RIPPER if (err) yyerror1(loc, err); - return NEW_BEGIN(0, loc); -} #else -static VALUE -assignable(struct parser_params *p, VALUE lhs) -{ - const char *err = 0; - assignable0(p, get_id(lhs), &err); - if (err) lhs = assign_error(p, err, lhs); - return lhs; -} + if (err) set_value(assign_error(p, err, p->s_lvalue)); #endif + return NEW_ERROR(loc); +} static int -is_private_local_id(ID name) +is_private_local_id(struct parser_params *p, ID name) { VALUE s; if (name == idUScore) return 1; @@ -11542,7 +13648,7 @@ shadowing_lvar_0(struct parser_params *p, ID name) { if (dyna_in_block(p)) { if (dvar_curr(p, name)) { - if (is_private_local_id(name)) return 1; + if (is_private_local_id(p, name)) return 1; yyerror0("duplicated argument name"); } else if (dvar_defined(p, name) || local_id(p, name)) { @@ -11555,7 +13661,7 @@ shadowing_lvar_0(struct parser_params *p, ID name) } else { if (local_id(p, name)) { - if (is_private_local_id(name)) return 1; + if (is_private_local_id(p, name)) return 1; yyerror0("duplicated argument name"); } } @@ -11580,12 +13686,44 @@ new_bv(struct parser_params *p, ID name) } if (!shadowing_lvar_0(p, name)) return; dyna_var(p, name); + ID *vidp = 0; + if (dvar_defined_ref(p, name, &vidp)) { + if (vidp) *vidp |= LVAR_USED; + } +} + +static void +aryset_check(struct parser_params *p, NODE *args) +{ + NODE *block = 0, *kwds = 0; + if (args && nd_type_p(args, NODE_BLOCK_PASS)) { + block = RNODE_BLOCK_PASS(args)->nd_body; + args = RNODE_BLOCK_PASS(args)->nd_head; + } + if (args && nd_type_p(args, NODE_ARGSCAT)) { + args = RNODE_ARGSCAT(args)->nd_body; + } + if (args && nd_type_p(args, NODE_ARGSPUSH)) { + kwds = RNODE_ARGSPUSH(args)->nd_body; + } + else { + for (NODE *next = args; next && nd_type_p(next, NODE_LIST); + next = RNODE_LIST(next)->nd_next) { + kwds = RNODE_LIST(next)->nd_head; + } + } + if (kwds && nd_type_p(kwds, NODE_HASH) && !RNODE_HASH(kwds)->nd_brace) { + yyerror1(&kwds->nd_loc, "keyword arg given in index"); + } + if (block) { + yyerror1(&block->nd_loc, "block arg given in index"); + } } -#ifndef RIPPER static NODE * aryset(struct parser_params *p, NODE *recv, NODE *idx, const YYLTYPE *loc) { + aryset_check(p, idx); return NEW_ATTRASGN(recv, tASET, idx, loc); } @@ -11609,24 +13747,31 @@ rb_backref_error(struct parser_params *p, NODE *node) { switch (nd_type(node)) { case NODE_NTH_REF: - compile_error(p, "Can't set variable $%ld", node->nd_nth); + compile_error(p, "Can't set variable $%ld", RNODE_NTH_REF(node)->nd_nth); break; case NODE_BACK_REF: - compile_error(p, "Can't set variable $%c", (int)node->nd_nth); + compile_error(p, "Can't set variable $%c", (int)RNODE_BACK_REF(node)->nd_nth); break; } } -#else + +#ifdef RIPPER static VALUE -backref_error(struct parser_params *p, NODE *ref, VALUE expr) +backref_error(struct parser_params *p, NODE *node, VALUE expr) { VALUE mesg = rb_str_new_cstr("Can't set variable "); - rb_str_append(mesg, ref->nd_cval); + switch (nd_type(node)) { + case NODE_NTH_REF: + rb_str_catf(mesg, "$%ld", RNODE_NTH_REF(node)->nd_nth); + break; + case NODE_BACK_REF: + rb_str_catf(mesg, "$%c", (int)RNODE_BACK_REF(node)->nd_nth); + break; + } return dispatch2(assign_error, mesg, expr); } #endif -#ifndef RIPPER static NODE * arg_append(struct parser_params *p, NODE *node1, NODE *node2, const YYLTYPE *loc) { @@ -11635,18 +13780,18 @@ arg_append(struct parser_params *p, NODE *node1, NODE *node2, const YYLTYPE *loc case NODE_LIST: return list_append(p, node1, node2); case NODE_BLOCK_PASS: - node1->nd_head = arg_append(p, node1->nd_head, node2, loc); - node1->nd_loc.end_pos = node1->nd_head->nd_loc.end_pos; + RNODE_BLOCK_PASS(node1)->nd_head = arg_append(p, RNODE_BLOCK_PASS(node1)->nd_head, node2, loc); + node1->nd_loc.end_pos = RNODE_BLOCK_PASS(node1)->nd_head->nd_loc.end_pos; return node1; case NODE_ARGSPUSH: - node1->nd_body = list_append(p, NEW_LIST(node1->nd_body, &node1->nd_body->nd_loc), node2); - node1->nd_loc.end_pos = node1->nd_body->nd_loc.end_pos; + RNODE_ARGSPUSH(node1)->nd_body = list_append(p, NEW_LIST(RNODE_ARGSPUSH(node1)->nd_body, &RNODE_ARGSPUSH(node1)->nd_body->nd_loc), node2); + node1->nd_loc.end_pos = RNODE_ARGSPUSH(node1)->nd_body->nd_loc.end_pos; nd_set_type(node1, NODE_ARGSCAT); return node1; case NODE_ARGSCAT: - if (!nd_type_p(node1->nd_body, NODE_LIST)) break; - node1->nd_body = list_append(p, node1->nd_body, node2); - node1->nd_loc.end_pos = node1->nd_body->nd_loc.end_pos; + if (!nd_type_p(RNODE_ARGSCAT(node1)->nd_body, NODE_LIST)) break; + RNODE_ARGSCAT(node1)->nd_body = list_append(p, RNODE_ARGSCAT(node1)->nd_body, node2); + node1->nd_loc.end_pos = RNODE_ARGSCAT(node1)->nd_body->nd_loc.end_pos; return node1; } return NEW_ARGSPUSH(node1, node2, loc); @@ -11658,20 +13803,20 @@ arg_concat(struct parser_params *p, NODE *node1, NODE *node2, const YYLTYPE *loc if (!node2) return node1; switch (nd_type(node1)) { case NODE_BLOCK_PASS: - if (node1->nd_head) - node1->nd_head = arg_concat(p, node1->nd_head, node2, loc); + if (RNODE_BLOCK_PASS(node1)->nd_head) + RNODE_BLOCK_PASS(node1)->nd_head = arg_concat(p, RNODE_BLOCK_PASS(node1)->nd_head, node2, loc); else - node1->nd_head = NEW_LIST(node2, loc); + RNODE_LIST(node1)->nd_head = NEW_LIST(node2, loc); return node1; case NODE_ARGSPUSH: if (!nd_type_p(node2, NODE_LIST)) break; - node1->nd_body = list_concat(NEW_LIST(node1->nd_body, loc), node2); + RNODE_ARGSPUSH(node1)->nd_body = list_concat(NEW_LIST(RNODE_ARGSPUSH(node1)->nd_body, loc), node2); nd_set_type(node1, NODE_ARGSCAT); return node1; case NODE_ARGSCAT: if (!nd_type_p(node2, NODE_LIST) || - !nd_type_p(node1->nd_body, NODE_LIST)) break; - node1->nd_body = list_concat(node1->nd_body, node2); + !nd_type_p(RNODE_ARGSCAT(node1)->nd_body, NODE_LIST)) break; + RNODE_ARGSCAT(node1)->nd_body = list_concat(RNODE_ARGSCAT(node1)->nd_body, node2); return node1; } return NEW_ARGSCAT(node1, node2, loc); @@ -11700,7 +13845,7 @@ rest_arg_append(struct parser_params *p, NODE *args, NODE *rest_arg, const YYLTY static NODE * splat_array(NODE* node) { - if (nd_type_p(node, NODE_SPLAT)) node = node->nd_head; + if (nd_type_p(node, NODE_SPLAT)) node = RNODE_SPLAT(node)->nd_head; if (nd_type_p(node, NODE_LIST)) return node; return 0; } @@ -11712,12 +13857,12 @@ mark_lvar_used(struct parser_params *p, NODE *rhs) if (!rhs) return; switch (nd_type(rhs)) { case NODE_LASGN: - if (local_id_ref(p, rhs->nd_vid, &vidp)) { + if (local_id_ref(p, RNODE_LASGN(rhs)->nd_vid, &vidp)) { if (vidp) *vidp |= LVAR_USED; } break; case NODE_DASGN: - if (dvar_defined_ref(p, rhs->nd_vid, &vidp)) { + if (dvar_defined_ref(p, RNODE_DASGN(rhs)->nd_vid, &vidp)) { if (vidp) *vidp |= LVAR_USED; } break; @@ -11731,252 +13876,8 @@ mark_lvar_used(struct parser_params *p, NODE *rhs) } } -static NODE * -const_decl_path(struct parser_params *p, NODE **dest) -{ - NODE *n = *dest; - if (!nd_type_p(n, NODE_CALL)) { - const YYLTYPE *loc = &n->nd_loc; - VALUE path; - if (n->nd_vid) { - path = rb_id2str(n->nd_vid); - } - else { - n = n->nd_else; - path = rb_ary_new(); - for (; n && nd_type_p(n, NODE_COLON2); n = n->nd_head) { - rb_ary_push(path, rb_id2str(n->nd_mid)); - } - if (n && nd_type_p(n, NODE_CONST)) { - // Const::Name - rb_ary_push(path, rb_id2str(n->nd_vid)); - } - else if (n && nd_type_p(n, NODE_COLON3)) { - // ::Const::Name - rb_ary_push(path, rb_str_new(0, 0)); - } - else { - // expression::Name - rb_ary_push(path, rb_str_new_cstr("...")); - } - path = rb_ary_join(rb_ary_reverse(path), rb_str_new_cstr("::")); - path = rb_fstring(path); - } - *dest = n = NEW_LIT(path, loc); - RB_OBJ_WRITTEN(p->ast, Qnil, n->nd_lit); - } - return n; -} - -extern VALUE rb_mRubyVMFrozenCore; - -static NODE * -make_shareable_node(struct parser_params *p, NODE *value, bool copy, const YYLTYPE *loc) -{ - NODE *fcore = NEW_LIT(rb_mRubyVMFrozenCore, loc); - - if (copy) { - return NEW_CALL(fcore, rb_intern("make_shareable_copy"), - NEW_LIST(value, loc), loc); - } - else { - return NEW_CALL(fcore, rb_intern("make_shareable"), - NEW_LIST(value, loc), loc); - } -} - -static NODE * -ensure_shareable_node(struct parser_params *p, NODE **dest, NODE *value, const YYLTYPE *loc) -{ - NODE *fcore = NEW_LIT(rb_mRubyVMFrozenCore, loc); - NODE *args = NEW_LIST(value, loc); - args = list_append(p, args, const_decl_path(p, dest)); - return NEW_CALL(fcore, rb_intern("ensure_shareable"), args, loc); -} - static int is_static_content(NODE *node); -static VALUE -shareable_literal_value(NODE *node) -{ - if (!node) return Qnil; - enum node_type type = nd_type(node); - switch (type) { - case NODE_TRUE: - return Qtrue; - case NODE_FALSE: - return Qfalse; - case NODE_NIL: - return Qnil; - case NODE_LIT: - return node->nd_lit; - default: - return Qundef; - } -} - -#ifndef SHAREABLE_BARE_EXPRESSION -#define SHAREABLE_BARE_EXPRESSION 1 -#endif - -static NODE * -shareable_literal_constant(struct parser_params *p, enum shareability shareable, - NODE **dest, NODE *value, const YYLTYPE *loc, size_t level) -{ -# define shareable_literal_constant_next(n) \ - shareable_literal_constant(p, shareable, dest, (n), &(n)->nd_loc, level+1) - VALUE lit = Qnil; - - if (!value) return 0; - enum node_type type = nd_type(value); - switch (type) { - case NODE_TRUE: - case NODE_FALSE: - case NODE_NIL: - case NODE_LIT: - return value; - - case NODE_DSTR: - if (shareable == shareable_literal) { - value = NEW_CALL(value, idUMinus, 0, loc); - } - return value; - - case NODE_STR: - lit = rb_fstring(value->nd_lit); - nd_set_type(value, NODE_LIT); - RB_OBJ_WRITE(p->ast, &value->nd_lit, lit); - return value; - - case NODE_ZLIST: - lit = rb_ary_new(); - OBJ_FREEZE_RAW(lit); - NODE *n = NEW_LIT(lit, loc); - RB_OBJ_WRITTEN(p->ast, Qnil, n->nd_lit); - return n; - - case NODE_LIST: - lit = rb_ary_new(); - for (NODE *n = value; n; n = n->nd_next) { - NODE *elt = n->nd_head; - if (elt) { - elt = shareable_literal_constant_next(elt); - if (elt) { - n->nd_head = elt; - } - else if (RTEST(lit)) { - rb_ary_clear(lit); - lit = Qfalse; - } - } - if (RTEST(lit)) { - VALUE e = shareable_literal_value(elt); - if (!UNDEF_P(e)) { - rb_ary_push(lit, e); - } - else { - rb_ary_clear(lit); - lit = Qnil; /* make shareable at runtime */ - } - } - } - break; - - case NODE_HASH: - if (!value->nd_brace) return 0; - lit = rb_hash_new(); - for (NODE *n = value->nd_head; n; n = n->nd_next->nd_next) { - NODE *key = n->nd_head; - NODE *val = n->nd_next->nd_head; - if (key) { - key = shareable_literal_constant_next(key); - if (key) { - n->nd_head = key; - } - else if (RTEST(lit)) { - rb_hash_clear(lit); - lit = Qfalse; - } - } - if (val) { - val = shareable_literal_constant_next(val); - if (val) { - n->nd_next->nd_head = val; - } - else if (RTEST(lit)) { - rb_hash_clear(lit); - lit = Qfalse; - } - } - if (RTEST(lit)) { - VALUE k = shareable_literal_value(key); - VALUE v = shareable_literal_value(val); - if (!UNDEF_P(k) && !UNDEF_P(v)) { - rb_hash_aset(lit, k, v); - } - else { - rb_hash_clear(lit); - lit = Qnil; /* make shareable at runtime */ - } - } - } - break; - - default: - if (shareable == shareable_literal && - (SHAREABLE_BARE_EXPRESSION || level > 0)) { - return ensure_shareable_node(p, dest, value, loc); - } - return 0; - } - - /* Array or Hash */ - if (!lit) return 0; - if (NIL_P(lit)) { - // if shareable_literal, all elements should have been ensured - // as shareable - value = make_shareable_node(p, value, false, loc); - } - else { - value = NEW_LIT(rb_ractor_make_shareable(lit), loc); - RB_OBJ_WRITTEN(p->ast, Qnil, value->nd_lit); - } - - return value; -# undef shareable_literal_constant_next -} - -static NODE * -shareable_constant_value(struct parser_params *p, enum shareability shareable, - NODE *lhs, NODE *value, const YYLTYPE *loc) -{ - if (!value) return 0; - switch (shareable) { - case shareable_none: - return value; - - case shareable_literal: - { - NODE *lit = shareable_literal_constant(p, shareable, &lhs, value, loc, 0); - if (lit) return lit; - return value; - } - break; - - case shareable_copy: - case shareable_everything: - { - NODE *lit = shareable_literal_constant(p, shareable, &lhs, value, loc, 0); - if (lit) return lit; - return make_shareable_node(p, value, shareable == shareable_copy, loc); - } - break; - - default: - UNREACHABLE_RETURN(0); - } -} - static NODE * node_assign(struct parser_params *p, NODE *lhs, NODE *rhs, struct lex_context ctxt, const YYLTYPE *loc) { @@ -11984,21 +13885,18 @@ node_assign(struct parser_params *p, NODE *lhs, NODE *rhs, struct lex_context ct switch (nd_type(lhs)) { case NODE_CDECL: - rhs = shareable_constant_value(p, ctxt.shareable_constant_value, lhs, rhs, loc); - /* fallthru */ - case NODE_GASGN: case NODE_IASGN: case NODE_LASGN: case NODE_DASGN: case NODE_MASGN: case NODE_CVASGN: - lhs->nd_value = rhs; + set_nd_value(p, lhs, rhs); nd_set_loc(lhs, loc); break; case NODE_ATTRASGN: - lhs->nd_args = arg_append(p, lhs->nd_args, rhs, loc); + RNODE_ATTRASGN(lhs)->nd_args = arg_append(p, RNODE_ATTRASGN(lhs)->nd_args, rhs, loc); nd_set_loc(lhs, loc); break; @@ -12020,52 +13918,81 @@ value_expr_check(struct parser_params *p, NODE *node) } while (node) { switch (nd_type(node)) { + case NODE_ENSURE: + vn = RNODE_ENSURE(node)->nd_head; + node = RNODE_ENSURE(node)->nd_ensr; + /* nd_ensr should not be NULL, check it out next */ + if (vn && (vn = value_expr_check(p, vn))) { + goto found; + } + break; + + case NODE_RESCUE: + /* void only if all children are void */ + vn = RNODE_RESCUE(node)->nd_head; + if (!vn || !(vn = value_expr_check(p, vn))) return NULL; + if (!void_node) void_node = vn; + for (NODE *r = RNODE_RESCUE(node)->nd_resq; r; r = RNODE_RESBODY(r)->nd_next) { + if (!nd_type_p(r, NODE_RESBODY)) { + compile_error(p, "unexpected node"); + return NULL; + } + if (!(vn = value_expr_check(p, RNODE_RESBODY(r)->nd_body))) { + void_node = 0; + break; + } + if (!void_node) void_node = vn; + } + node = RNODE_RESCUE(node)->nd_else; + if (!node) return void_node; + break; + case NODE_RETURN: case NODE_BREAK: case NODE_NEXT: case NODE_REDO: case NODE_RETRY: - return void_node ? void_node : node; + goto found; case NODE_CASE3: - if (!node->nd_body || !nd_type_p(node->nd_body, NODE_IN)) { + if (!RNODE_CASE3(node)->nd_body || !nd_type_p(RNODE_CASE3(node)->nd_body, NODE_IN)) { compile_error(p, "unexpected node"); return NULL; } - if (node->nd_body->nd_body) { + if (RNODE_IN(RNODE_CASE3(node)->nd_body)->nd_body) { return NULL; } - /* single line pattern matching */ - return void_node ? void_node : node; + /* single line pattern matching with "=>" operator */ + goto found; case NODE_BLOCK: - while (node->nd_next) { - node = node->nd_next; + while (RNODE_BLOCK(node)->nd_next) { + node = RNODE_BLOCK(node)->nd_next; } - node = node->nd_head; + node = RNODE_BLOCK(node)->nd_head; break; case NODE_BEGIN: - node = node->nd_body; + node = RNODE_BEGIN(node)->nd_body; break; case NODE_IF: case NODE_UNLESS: - if (!node->nd_body) { + if (!RNODE_IF(node)->nd_body) { return NULL; } - else if (!node->nd_else) { + else if (!RNODE_IF(node)->nd_else) { return NULL; } - vn = value_expr_check(p, node->nd_body); + vn = value_expr_check(p, RNODE_IF(node)->nd_body); if (!vn) return NULL; if (!void_node) void_node = vn; - node = node->nd_else; + node = RNODE_IF(node)->nd_else; break; case NODE_AND: case NODE_OR: - node = node->nd_1st; + node = RNODE_AND(node)->nd_1st; break; case NODE_LASGN: @@ -12080,6 +14007,10 @@ value_expr_check(struct parser_params *p, NODE *node) } return NULL; + + found: + /* return the first found node */ + return void_node ? void_node : node; } static int @@ -12093,6 +14024,7 @@ value_expr_gen(struct parser_params *p, NODE *node) } return TRUE; } + static void void_expr(struct parser_params *p, NODE *node) { @@ -12103,7 +14035,7 @@ void_expr(struct parser_params *p, NODE *node) if (!node || !(node = nd_once_body(node))) return; switch (nd_type(node)) { case NODE_OPCALL: - switch (node->nd_mid) { + switch (RNODE_OPCALL(node)->nd_mid) { case '+': case '-': case '*': @@ -12122,7 +14054,7 @@ void_expr(struct parser_params *p, NODE *node) case tLEQ: case tEQ: case tNEQ: - useless = rb_id2name(node->nd_mid); + useless = rb_id2name(RNODE_OPCALL(node)->nd_mid); break; } break; @@ -12139,9 +14071,17 @@ void_expr(struct parser_params *p, NODE *node) case NODE_CONST: useless = "a constant"; break; - case NODE_LIT: + case NODE_SYM: + case NODE_LINE: + case NODE_FILE: + case NODE_ENCODING: + case NODE_INTEGER: + case NODE_FLOAT: + case NODE_RATIONAL: + case NODE_IMAGINARY: case NODE_STR: case NODE_DSTR: + case NODE_REGX: case NODE_DREGX: useless = "a literal"; break; @@ -12185,9 +14125,9 @@ void_stmts(struct parser_params *p, NODE *node) if (!node) return n; if (!nd_type_p(node, NODE_BLOCK)) return n; - while (node->nd_next) { - void_expr(p, node->nd_head); - node = node->nd_next; + while (RNODE_BLOCK(node)->nd_next) { + void_expr(p, RNODE_BLOCK(node)->nd_head); + node = RNODE_BLOCK(node)->nd_next; } return n; } @@ -12196,18 +14136,8 @@ static NODE * remove_begin(NODE *node) { NODE **n = &node, *n1 = node; - while (n1 && nd_type_p(n1, NODE_BEGIN) && n1->nd_body) { - *n = n1 = n1->nd_body; - } - return node; -} - -static NODE * -remove_begin_all(NODE *node) -{ - NODE **n = &node, *n1 = node; - while (n1 && nd_type_p(n1, NODE_BEGIN)) { - *n = n1 = n1->nd_body; + while (n1 && nd_type_p(n1, NODE_BEGIN) && RNODE_BEGIN(n1)->nd_body) { + *n = n1 = RNODE_BEGIN(n1)->nd_body; } return node; } @@ -12221,54 +14151,55 @@ reduce_nodes(struct parser_params *p, NODE **body) *body = NEW_NIL(&NULL_LOC); return; } -#define subnodes(n1, n2) \ - ((!node->n1) ? (node->n2 ? (body = &node->n2, 1) : 0) : \ - (!node->n2) ? (body = &node->n1, 1) : \ - (reduce_nodes(p, &node->n1), body = &node->n2, 1)) +#define subnodes(type, n1, n2) \ + ((!type(node)->n1) ? (type(node)->n2 ? (body = &type(node)->n2, 1) : 0) : \ + (!type(node)->n2) ? (body = &type(node)->n1, 1) : \ + (reduce_nodes(p, &type(node)->n1), body = &type(node)->n2, 1)) while (node) { - int newline = (int)(node->flags & NODE_FL_NEWLINE); + int newline = (int)(nd_fl_newline(node)); switch (nd_type(node)) { end: case NODE_NIL: *body = 0; return; case NODE_RETURN: - *body = node = node->nd_stts; - if (newline && node) node->flags |= NODE_FL_NEWLINE; + *body = node = RNODE_RETURN(node)->nd_stts; + if (newline && node) nd_set_fl_newline(node); continue; case NODE_BEGIN: - *body = node = node->nd_body; - if (newline && node) node->flags |= NODE_FL_NEWLINE; + *body = node = RNODE_BEGIN(node)->nd_body; + if (newline && node) nd_set_fl_newline(node); continue; case NODE_BLOCK: - body = &node->nd_end->nd_head; + body = &RNODE_BLOCK(RNODE_BLOCK(node)->nd_end)->nd_head; break; case NODE_IF: case NODE_UNLESS: - if (subnodes(nd_body, nd_else)) break; + if (subnodes(RNODE_IF, nd_body, nd_else)) break; return; case NODE_CASE: - body = &node->nd_body; + body = &RNODE_CASE(node)->nd_body; break; case NODE_WHEN: - if (!subnodes(nd_body, nd_next)) goto end; + if (!subnodes(RNODE_WHEN, nd_body, nd_next)) goto end; break; case NODE_ENSURE: - if (!subnodes(nd_head, nd_resq)) goto end; + body = &RNODE_ENSURE(node)->nd_head; break; case NODE_RESCUE: - if (node->nd_else) { - body = &node->nd_resq; + newline = 0; // RESBODY should not be a NEWLINE + if (RNODE_RESCUE(node)->nd_else) { + body = &RNODE_RESCUE(node)->nd_resq; break; } - if (!subnodes(nd_head, nd_resq)) goto end; + if (!subnodes(RNODE_RESCUE, nd_head, nd_resq)) goto end; break; default: return; } node = *body; - if (newline && node) node->flags |= NODE_FL_NEWLINE; + if (newline && node) nd_set_fl_newline(node); } #undef subnodes @@ -12280,12 +14211,20 @@ is_static_content(NODE *node) if (!node) return 1; switch (nd_type(node)) { case NODE_HASH: - if (!(node = node->nd_head)) break; + if (!(node = RNODE_HASH(node)->nd_head)) break; case NODE_LIST: do { - if (!is_static_content(node->nd_head)) return 0; - } while ((node = node->nd_next) != 0); - case NODE_LIT: + if (!is_static_content(RNODE_LIST(node)->nd_head)) return 0; + } while ((node = RNODE_LIST(node)->nd_next) != 0); + case NODE_SYM: + case NODE_REGX: + case NODE_LINE: + case NODE_FILE: + case NODE_ENCODING: + case NODE_INTEGER: + case NODE_FLOAT: + case NODE_RATIONAL: + case NODE_IMAGINARY: case NODE_STR: case NODE_NIL: case NODE_TRUE: @@ -12307,16 +14246,18 @@ assign_in_cond(struct parser_params *p, NODE *node) case NODE_DASGN: case NODE_GASGN: case NODE_IASGN: + case NODE_CVASGN: + case NODE_CDECL: break; default: return 0; } - if (!node->nd_value) return 1; - if (is_static_content(node->nd_value)) { + if (!get_nd_value(p, node)) return 1; + if (is_static_content(get_nd_value(p, node))) { /* reports always */ - parser_warn(p, node->nd_value, "found `= literal' in conditional, should be =="); + rb_warn0L(nd_line(get_nd_value(p, node)), "found '= literal' in conditional, should be =="); } return 1; } @@ -12327,14 +14268,15 @@ enum cond_type { COND_IN_FF }; -#define SWITCH_BY_COND_TYPE(t, w, arg) \ +#define SWITCH_BY_COND_TYPE(t, w, arg) do { \ switch (t) { \ case COND_IN_OP: break; \ case COND_IN_COND: rb_##w##0(arg "literal in condition"); break; \ case COND_IN_FF: rb_##w##0(arg "literal in flip-flop"); break; \ - } + } \ +} while (0) -static NODE *cond0(struct parser_params*,NODE*,enum cond_type,const YYLTYPE*); +static NODE *cond0(struct parser_params*,NODE*,enum cond_type,const YYLTYPE*,bool); static NODE* range_op(struct parser_params *p, NODE *node, const YYLTYPE *loc) @@ -12345,67 +14287,87 @@ range_op(struct parser_params *p, NODE *node, const YYLTYPE *loc) type = nd_type(node); value_expr(node); - if (type == NODE_LIT && FIXNUM_P(node->nd_lit)) { - if (!e_option_supplied(p)) parser_warn(p, node, "integer literal in flip-flop"); + if (type == NODE_INTEGER) { + if (!e_option_supplied(p)) rb_warn0L(nd_line(node), "integer literal in flip-flop"); ID lineno = rb_intern("$."); return NEW_CALL(node, tEQ, NEW_LIST(NEW_GVAR(lineno, loc), loc), loc); } - return cond0(p, node, COND_IN_FF, loc); + return cond0(p, node, COND_IN_FF, loc, true); } static NODE* -cond0(struct parser_params *p, NODE *node, enum cond_type type, const YYLTYPE *loc) +cond0(struct parser_params *p, NODE *node, enum cond_type type, const YYLTYPE *loc, bool top) { if (node == 0) return 0; if (!(node = nd_once_body(node))) return 0; assign_in_cond(p, node); switch (nd_type(node)) { + case NODE_BEGIN: + RNODE_BEGIN(node)->nd_body = cond0(p, RNODE_BEGIN(node)->nd_body, type, loc, top); + break; + case NODE_DSTR: case NODE_EVSTR: case NODE_STR: - SWITCH_BY_COND_TYPE(type, warn, "string ") + case NODE_FILE: + SWITCH_BY_COND_TYPE(type, warn, "string "); + break; + + case NODE_REGX: + if (!e_option_supplied(p)) SWITCH_BY_COND_TYPE(type, warn, "regex "); + nd_set_type(node, NODE_MATCH); break; case NODE_DREGX: - if (!e_option_supplied(p)) SWITCH_BY_COND_TYPE(type, warning, "regex ") + if (!e_option_supplied(p)) SWITCH_BY_COND_TYPE(type, warning, "regex "); return NEW_MATCH2(node, NEW_GVAR(idLASTLINE, loc), loc); + case NODE_BLOCK: + { + NODE *end = RNODE_BLOCK(node)->nd_end; + NODE **expr = &RNODE_BLOCK(end)->nd_head; + if (top) top = node == end; + *expr = cond0(p, *expr, type, loc, top); + } + break; + case NODE_AND: case NODE_OR: - node->nd_1st = cond0(p, node->nd_1st, COND_IN_COND, loc); - node->nd_2nd = cond0(p, node->nd_2nd, COND_IN_COND, loc); + RNODE_AND(node)->nd_1st = cond0(p, RNODE_AND(node)->nd_1st, COND_IN_COND, loc, true); + RNODE_AND(node)->nd_2nd = cond0(p, RNODE_AND(node)->nd_2nd, COND_IN_COND, loc, true); break; case NODE_DOT2: case NODE_DOT3: - node->nd_beg = range_op(p, node->nd_beg, loc); - node->nd_end = range_op(p, node->nd_end, loc); + if (!top) break; + RNODE_DOT2(node)->nd_beg = range_op(p, RNODE_DOT2(node)->nd_beg, loc); + RNODE_DOT2(node)->nd_end = range_op(p, RNODE_DOT2(node)->nd_end, loc); if (nd_type_p(node, NODE_DOT2)) nd_set_type(node,NODE_FLIP2); else if (nd_type_p(node, NODE_DOT3)) nd_set_type(node, NODE_FLIP3); break; + case NODE_SYM: case NODE_DSYM: - warn_symbol: - SWITCH_BY_COND_TYPE(type, warning, "symbol ") + SWITCH_BY_COND_TYPE(type, warning, "symbol "); + break; + + case NODE_LINE: + SWITCH_BY_COND_TYPE(type, warning, ""); + break; + + case NODE_ENCODING: + SWITCH_BY_COND_TYPE(type, warning, ""); + break; + + case NODE_INTEGER: + case NODE_FLOAT: + case NODE_RATIONAL: + case NODE_IMAGINARY: + SWITCH_BY_COND_TYPE(type, warning, ""); break; - case NODE_LIT: - if (RB_TYPE_P(node->nd_lit, T_REGEXP)) { - if (!e_option_supplied(p)) SWITCH_BY_COND_TYPE(type, warn, "regex ") - nd_set_type(node, NODE_MATCH); - } - else if (node->nd_lit == Qtrue || - node->nd_lit == Qfalse) { - /* booleans are OK, e.g., while true */ - } - else if (SYMBOL_P(node->nd_lit)) { - goto warn_symbol; - } - else { - SWITCH_BY_COND_TYPE(type, warning, "") - } default: break; } @@ -12416,14 +14378,14 @@ static NODE* cond(struct parser_params *p, NODE *node, const YYLTYPE *loc) { if (node == 0) return 0; - return cond0(p, node, COND_IN_COND, loc); + return cond0(p, node, COND_IN_COND, loc, true); } static NODE* method_cond(struct parser_params *p, NODE *node, const YYLTYPE *loc) { if (node == 0) return 0; - return cond0(p, node, COND_IN_OP, loc); + return cond0(p, node, COND_IN_OP, loc, true); } static NODE* @@ -12437,7 +14399,7 @@ static NODE* new_if(struct parser_params *p, NODE *cc, NODE *left, NODE *right, const YYLTYPE *loc) { if (!cc) return right; - cc = cond0(p, cc, COND_IN_COND, loc); + cc = cond0(p, cc, COND_IN_COND, loc, true); return newline_node(NEW_IF(cc, left, right, loc)); } @@ -12445,10 +14407,12 @@ static NODE* new_unless(struct parser_params *p, NODE *cc, NODE *left, NODE *right, const YYLTYPE *loc) { if (!cc) return right; - cc = cond0(p, cc, COND_IN_COND, loc); + cc = cond0(p, cc, COND_IN_COND, loc, true); return newline_node(NEW_UNLESS(cc, left, right, loc)); } +#define NEW_AND_OR(type, f, s, loc) (type == NODE_AND ? NEW_AND(f,s,loc) : NEW_OR(f,s,loc)) + static NODE* logop(struct parser_params *p, ID id, NODE *left, NODE *right, const YYLTYPE *op_loc, const YYLTYPE *loc) @@ -12458,19 +14422,21 @@ logop(struct parser_params *p, ID id, NODE *left, NODE *right, value_expr(left); if (left && nd_type_p(left, type)) { NODE *node = left, *second; - while ((second = node->nd_2nd) != 0 && nd_type_p(second, type)) { + while ((second = RNODE_AND(node)->nd_2nd) != 0 && nd_type_p(second, type)) { node = second; } - node->nd_2nd = NEW_NODE(type, second, right, 0, loc); - nd_set_line(node->nd_2nd, op_loc->beg_pos.lineno); + RNODE_AND(node)->nd_2nd = NEW_AND_OR(type, second, right, loc); + nd_set_line(RNODE_AND(node)->nd_2nd, op_loc->beg_pos.lineno); left->nd_loc.end_pos = loc->end_pos; return left; } - op = NEW_NODE(type, left, right, 0, loc); + op = NEW_AND_OR(type, left, right, loc); nd_set_line(op, op_loc->beg_pos.lineno); return op; } +#undef NEW_AND_OR + static void no_blockarg(struct parser_params *p, NODE *node) { @@ -12484,13 +14450,8 @@ ret_args(struct parser_params *p, NODE *node) { if (node) { no_blockarg(p, node); - if (nd_type_p(node, NODE_LIST)) { - if (node->nd_next == 0) { - node = node->nd_head; - } - else { - nd_set_type(node, NODE_VALUES); - } + if (nd_type_p(node, NODE_LIST) && !RNODE_LIST(node)->nd_next) { + node = RNODE_LIST(node)->nd_head; } } return node; @@ -12504,53 +14465,35 @@ new_yield(struct parser_params *p, NODE *node, const YYLTYPE *loc) return NEW_YIELD(node, loc); } -static VALUE -negate_lit(struct parser_params *p, VALUE lit) +static NODE* +negate_lit(struct parser_params *p, NODE* node) { - if (FIXNUM_P(lit)) { - return LONG2FIX(-FIX2LONG(lit)); - } - if (SPECIAL_CONST_P(lit)) { -#if USE_FLONUM - if (FLONUM_P(lit)) { - return DBL2NUM(-RFLOAT_VALUE(lit)); - } -#endif - goto unknown; - } - switch (BUILTIN_TYPE(lit)) { - case T_BIGNUM: - BIGNUM_NEGATE(lit); - lit = rb_big_norm(lit); - break; - case T_RATIONAL: - RATIONAL_SET_NUM(lit, negate_lit(p, RRATIONAL(lit)->num)); + switch (nd_type(node)) { + case NODE_INTEGER: + RNODE_INTEGER(node)->minus = TRUE; break; - case T_COMPLEX: - RCOMPLEX_SET_REAL(lit, negate_lit(p, RCOMPLEX(lit)->real)); - RCOMPLEX_SET_IMAG(lit, negate_lit(p, RCOMPLEX(lit)->imag)); + case NODE_FLOAT: + RNODE_FLOAT(node)->minus = TRUE; break; - case T_FLOAT: - lit = DBL2NUM(-RFLOAT_VALUE(lit)); + case NODE_RATIONAL: + RNODE_RATIONAL(node)->minus = TRUE; break; - unknown: - default: - rb_parser_fatal(p, "unknown literal type (%s) passed to negate_lit", - rb_builtin_class_name(lit)); + case NODE_IMAGINARY: + RNODE_IMAGINARY(node)->minus = TRUE; break; } - return lit; + return node; } static NODE * -arg_blk_pass(NODE *node1, NODE *node2) +arg_blk_pass(NODE *node1, rb_node_block_pass_t *node2) { if (node2) { - if (!node1) return node2; + if (!node1) return (NODE *)node2; node2->nd_head = node1; nd_set_first_lineno(node2, nd_first_lineno(node1)); nd_set_first_column(node2, nd_first_column(node1)); - return node2; + return (NODE *)node2; } return node1; } @@ -12568,24 +14511,23 @@ args_info_empty_p(struct rb_args_info *args) return true; } -static NODE* -new_args(struct parser_params *p, NODE *pre_args, NODE *opt_args, ID rest_arg, NODE *post_args, NODE *tail, const YYLTYPE *loc) +static rb_node_args_t * +new_args(struct parser_params *p, rb_node_args_aux_t *pre_args, rb_node_opt_arg_t *opt_args, ID rest_arg, rb_node_args_aux_t *post_args, rb_node_args_t *tail, const YYLTYPE *loc) { - int saved_line = p->ruby_sourceline; - struct rb_args_info *args = tail->nd_ainfo; + struct rb_args_info *args = &tail->nd_ainfo; if (args->forwarding) { if (rest_arg) { - yyerror1(&tail->nd_loc, "... after rest argument"); + yyerror1(&RNODE(tail)->nd_loc, "... after rest argument"); return tail; } rest_arg = idFWD_REST; } - args->pre_args_num = pre_args ? rb_long2int(pre_args->nd_plen) : 0; + args->pre_args_num = pre_args ? pre_args->nd_plen : 0; args->pre_init = pre_args ? pre_args->nd_next : 0; - args->post_args_num = post_args ? rb_long2int(post_args->nd_plen) : 0; + args->post_args_num = post_args ? post_args->nd_plen : 0; args->post_init = post_args ? post_args->nd_next : 0; args->first_post_arg = post_args ? post_args->nd_pid : 0; @@ -12599,23 +14541,16 @@ new_args(struct parser_params *p, NODE *pre_args, NODE *opt_args, ID rest_arg, N args->ruby2_keywords = 0; #endif - p->ruby_sourceline = saved_line; - nd_set_loc(tail, loc); + nd_set_loc(RNODE(tail), loc); return tail; } -static NODE* -new_args_tail(struct parser_params *p, NODE *kw_args, ID kw_rest_arg, ID block, const YYLTYPE *kw_rest_loc) +static rb_node_args_t * +new_args_tail(struct parser_params *p, rb_node_kw_arg_t *kw_args, ID kw_rest_arg, ID block, const YYLTYPE *kw_rest_loc) { - int saved_line = p->ruby_sourceline; - NODE *node; - VALUE tmpbuf = rb_imemo_tmpbuf_auto_free_pointer(); - struct rb_args_info *args = ZALLOC(struct rb_args_info); - rb_imemo_tmpbuf_set_ptr(tmpbuf, args); - args->imemo = tmpbuf; - node = NEW_NODE(NODE_ARGS, 0, 0, args, &NULL_LOC); - RB_OBJ_WRITTEN(p->ast, Qnil, tmpbuf); + rb_node_args_t *node = NEW_ARGS(&NULL_LOC); + struct rb_args_info *args = &node->nd_ainfo; if (p->error_p) return node; args->block_arg = block; @@ -12630,21 +14565,21 @@ new_args_tail(struct parser_params *p, NODE *kw_args, ID kw_rest_arg, ID block, */ ID kw_bits = internal_id(p), *required_kw_vars, *kw_vars; struct vtable *vtargs = p->lvtbl->args; - NODE *kwn = kw_args; + rb_node_kw_arg_t *kwn = kw_args; if (block) block = vtargs->tbl[vtargs->pos-1]; vtable_pop(vtargs, !!block + !!kw_rest_arg); required_kw_vars = kw_vars = &vtargs->tbl[vtargs->pos]; while (kwn) { - if (!NODE_REQUIRED_KEYWORD_P(kwn->nd_body)) + if (!NODE_REQUIRED_KEYWORD_P(get_nd_value(p, kwn->nd_body))) --kw_vars; --required_kw_vars; kwn = kwn->nd_next; } for (kwn = kw_args; kwn; kwn = kwn->nd_next) { - ID vid = kwn->nd_body->nd_vid; - if (NODE_REQUIRED_KEYWORD_P(kwn->nd_body)) { + ID vid = get_nd_vid(p, kwn->nd_body); + if (NODE_REQUIRED_KEYWORD_P(get_nd_value(p, kwn->nd_body))) { *required_kw_vars++ = vid; } else { @@ -12657,7 +14592,6 @@ new_args_tail(struct parser_params *p, NODE *kw_args, ID kw_rest_arg, ID block, if (block) arg_var(p, block); args->kw_rest_arg = NEW_DVAR(kw_rest_arg, kw_rest_loc); - args->kw_rest_arg->nd_cflag = kw_bits; } else if (kw_rest_arg == idNil) { args->no_kwarg = 1; @@ -12666,20 +14600,19 @@ new_args_tail(struct parser_params *p, NODE *kw_args, ID kw_rest_arg, ID block, args->kw_rest_arg = NEW_DVAR(kw_rest_arg, kw_rest_loc); } - p->ruby_sourceline = saved_line; return node; } -static NODE * -args_with_numbered(struct parser_params *p, NODE *args, int max_numparam) +static rb_node_args_t * +args_with_numbered(struct parser_params *p, rb_node_args_t *args, int max_numparam, ID it_id) { - if (max_numparam > NO_PARAM) { + if (max_numparam > NO_PARAM || it_id) { if (!args) { YYLTYPE loc = RUBY_INIT_YYLLOC(); args = new_args_tail(p, 0, 0, 0, 0); - nd_set_loc(args, &loc); + nd_set_loc(RNODE(args), &loc); } - args->nd_ainfo->pre_args_num = max_numparam; + args->nd_ainfo.pre_args_num = it_id ? 1 : max_numparam; } return args; } @@ -12687,17 +14620,15 @@ args_with_numbered(struct parser_params *p, NODE *args, int max_numparam) static NODE* new_array_pattern(struct parser_params *p, NODE *constant, NODE *pre_arg, NODE *aryptn, const YYLTYPE *loc) { - struct rb_ary_pattern_info *apinfo = aryptn->nd_apinfo; - - aryptn->nd_pconst = constant; + RNODE_ARYPTN(aryptn)->nd_pconst = constant; if (pre_arg) { NODE *pre_args = NEW_LIST(pre_arg, loc); - if (apinfo->pre_args) { - apinfo->pre_args = list_concat(pre_args, apinfo->pre_args); + if (RNODE_ARYPTN(aryptn)->pre_args) { + RNODE_ARYPTN(aryptn)->pre_args = list_concat(pre_args, RNODE_ARYPTN(aryptn)->pre_args); } else { - apinfo->pre_args = pre_args; + RNODE_ARYPTN(aryptn)->pre_args = pre_args; } } return aryptn; @@ -12706,33 +14637,21 @@ new_array_pattern(struct parser_params *p, NODE *constant, NODE *pre_arg, NODE * static NODE* new_array_pattern_tail(struct parser_params *p, NODE *pre_args, int has_rest, NODE *rest_arg, NODE *post_args, const YYLTYPE *loc) { - int saved_line = p->ruby_sourceline; - NODE *node; - VALUE tmpbuf = rb_imemo_tmpbuf_auto_free_pointer(); - struct rb_ary_pattern_info *apinfo = ZALLOC(struct rb_ary_pattern_info); - rb_imemo_tmpbuf_set_ptr(tmpbuf, apinfo); - node = NEW_NODE(NODE_ARYPTN, 0, tmpbuf, apinfo, loc); - RB_OBJ_WRITTEN(p->ast, Qnil, tmpbuf); - - apinfo->pre_args = pre_args; - if (has_rest) { - apinfo->rest_arg = rest_arg ? rest_arg : NODE_SPECIAL_NO_NAME_REST; + rest_arg = rest_arg ? rest_arg : NODE_SPECIAL_NO_NAME_REST; } else { - apinfo->rest_arg = NULL; + rest_arg = NULL; } + NODE *node = NEW_ARYPTN(pre_args, rest_arg, post_args, loc); - apinfo->post_args = post_args; - - p->ruby_sourceline = saved_line; return node; } static NODE* new_find_pattern(struct parser_params *p, NODE *constant, NODE *fndptn, const YYLTYPE *loc) { - fndptn->nd_pconst = constant; + RNODE_FNDPTN(fndptn)->nd_pconst = constant; return fndptn; } @@ -12740,33 +14659,23 @@ new_find_pattern(struct parser_params *p, NODE *constant, NODE *fndptn, const YY static NODE* new_find_pattern_tail(struct parser_params *p, NODE *pre_rest_arg, NODE *args, NODE *post_rest_arg, const YYLTYPE *loc) { - int saved_line = p->ruby_sourceline; - NODE *node; - VALUE tmpbuf = rb_imemo_tmpbuf_auto_free_pointer(); - struct rb_fnd_pattern_info *fpinfo = ZALLOC(struct rb_fnd_pattern_info); - rb_imemo_tmpbuf_set_ptr(tmpbuf, fpinfo); - node = NEW_NODE(NODE_FNDPTN, 0, tmpbuf, fpinfo, loc); - RB_OBJ_WRITTEN(p->ast, Qnil, tmpbuf); - - fpinfo->pre_rest_arg = pre_rest_arg ? pre_rest_arg : NODE_SPECIAL_NO_NAME_REST; - fpinfo->args = args; - fpinfo->post_rest_arg = post_rest_arg ? post_rest_arg : NODE_SPECIAL_NO_NAME_REST; + pre_rest_arg = pre_rest_arg ? pre_rest_arg : NODE_SPECIAL_NO_NAME_REST; + post_rest_arg = post_rest_arg ? post_rest_arg : NODE_SPECIAL_NO_NAME_REST; + NODE *node = NEW_FNDPTN(pre_rest_arg, args, post_rest_arg, loc); - p->ruby_sourceline = saved_line; return node; } static NODE* new_hash_pattern(struct parser_params *p, NODE *constant, NODE *hshptn, const YYLTYPE *loc) { - hshptn->nd_pconst = constant; + RNODE_HSHPTN(hshptn)->nd_pconst = constant; return hshptn; } static NODE* new_hash_pattern_tail(struct parser_params *p, NODE *kw_args, ID kw_rest_arg, const YYLTYPE *loc) { - int saved_line = p->ruby_sourceline; NODE *node, *kw_rest_arg_node; if (kw_rest_arg == idNil) { @@ -12779,19 +14688,16 @@ new_hash_pattern_tail(struct parser_params *p, NODE *kw_args, ID kw_rest_arg, co kw_rest_arg_node = NULL; } - node = NEW_NODE(NODE_HSHPTN, 0, kw_args, kw_rest_arg_node, loc); + node = NEW_HSHPTN(0, kw_args, kw_rest_arg_node, loc); - p->ruby_sourceline = saved_line; return node; } static NODE* dsym_node(struct parser_params *p, NODE *node, const YYLTYPE *loc) { - VALUE lit; - if (!node) { - return NEW_LIT(ID2SYM(idNULL), loc); + return NEW_SYM(STR_NEW0(), loc); } switch (nd_type(node)) { @@ -12800,117 +14706,108 @@ dsym_node(struct parser_params *p, NODE *node, const YYLTYPE *loc) nd_set_loc(node, loc); break; case NODE_STR: - lit = node->nd_lit; - RB_OBJ_WRITTEN(p->ast, Qnil, node->nd_lit = ID2SYM(rb_intern_str(lit))); - nd_set_type(node, NODE_LIT); - nd_set_loc(node, loc); + node = str_to_sym_node(p, node, loc); break; default: - node = NEW_NODE(NODE_DSYM, Qnil, 1, NEW_LIST(node, loc), loc); + node = NEW_DSYM(0, 1, NEW_LIST(node, loc), loc); break; } return node; } static int -append_literal_keys(st_data_t k, st_data_t v, st_data_t h) -{ - NODE *node = (NODE *)v; - NODE **result = (NODE **)h; - node->nd_alen = 2; - node->nd_next->nd_end = node->nd_next; - node->nd_next->nd_next = 0; - if (*result) - list_concat(*result, node); - else - *result = node; - return ST_CONTINUE; -} - -static bool -hash_literal_key_p(VALUE k) +nd_type_st_key_enable_p(NODE *node) { - switch (OBJ_BUILTIN_TYPE(k)) { - case T_NODE: - return false; - default: + switch (nd_type(node)) { + case NODE_INTEGER: + case NODE_FLOAT: + case NODE_RATIONAL: + case NODE_IMAGINARY: + case NODE_STR: + case NODE_SYM: + case NODE_REGX: + case NODE_LINE: + case NODE_FILE: + case NODE_ENCODING: return true; + default: + return false; } } -static int -literal_cmp(VALUE val, VALUE lit) +static VALUE +nd_value(struct parser_params *p, NODE *node) { - if (val == lit) return 0; - if (!hash_literal_key_p(val) || !hash_literal_key_p(lit)) return -1; - return rb_iseq_cdhash_cmp(val, lit); + switch (nd_type(node)) { + case NODE_STR: + return rb_node_str_string_val(node); + case NODE_INTEGER: + return rb_node_integer_literal_val(node); + case NODE_FLOAT: + return rb_node_float_literal_val(node); + case NODE_RATIONAL: + return rb_node_rational_literal_val(node); + case NODE_IMAGINARY: + return rb_node_imaginary_literal_val(node); + case NODE_SYM: + return rb_node_sym_string_val(node); + case NODE_REGX: + return rb_node_regx_string_val(node); + case NODE_LINE: + return rb_node_line_lineno_val(node); + case NODE_ENCODING: + return rb_node_encoding_val(node); + case NODE_FILE: + return rb_node_file_path_val(node); + default: + rb_bug("unexpected node: %s", ruby_node_name(nd_type(node))); + UNREACHABLE_RETURN(0); + } } -static st_index_t -literal_hash(VALUE a) -{ - if (!hash_literal_key_p(a)) return (st_index_t)a; - return rb_iseq_cdhash_hash(a); -} +static void +warn_duplicate_keys(struct parser_params *p, NODE *hash) +{ + /* See https://bugs.ruby-lang.org/issues/20331 for discussion about what is warned. */ + st_table *literal_keys = st_init_table_with_size(&literal_type, RNODE_LIST(hash)->as.nd_alen / 2); + while (hash && RNODE_LIST(hash)->nd_next) { + NODE *head = RNODE_LIST(hash)->nd_head; + NODE *value = RNODE_LIST(hash)->nd_next; + NODE *next = RNODE_LIST(value)->nd_next; + st_data_t key; + st_data_t data; -static const struct st_hash_type literal_type = { - literal_cmp, - literal_hash, -}; + /* keyword splat, e.g. {k: 1, **z, k: 2} */ + if (!head) { + head = value; + } -static NODE * -remove_duplicate_keys(struct parser_params *p, NODE *hash) -{ - st_table *literal_keys = st_init_table_with_size(&literal_type, hash->nd_alen / 2); - NODE *result = 0; - NODE *last_expr = 0; - rb_code_location_t loc = hash->nd_loc; - while (hash && hash->nd_head && hash->nd_next) { - NODE *head = hash->nd_head; - NODE *value = hash->nd_next; - NODE *next = value->nd_next; - st_data_t key = (st_data_t)head; - st_data_t data; - value->nd_next = 0; - if (nd_type_p(head, NODE_LIT) && - st_delete(literal_keys, (key = (st_data_t)head->nd_lit, &key), &data)) { - NODE *dup_value = ((NODE *)data)->nd_next; - rb_compile_warn(p->ruby_sourcefile, nd_line((NODE *)data), - "key %+"PRIsVALUE" is duplicated and overwritten on line %d", - head->nd_lit, nd_line(head)); - if (dup_value == last_expr) { - value->nd_head = block_append(p, dup_value->nd_head, value->nd_head); - } - else { - last_expr->nd_head = block_append(p, dup_value->nd_head, last_expr->nd_head); + if (nd_type_st_key_enable_p(head)) { + key = (st_data_t)head; + + if (st_delete(literal_keys, &key, &data)) { + rb_warn2L(nd_line((NODE *)data), + "key %+"PRIsWARN" is duplicated and overwritten on line %d", + nd_value(p, head), WARN_I(nd_line(head))); } + st_insert(literal_keys, (st_data_t)key, (st_data_t)hash); } - st_insert(literal_keys, (st_data_t)key, (st_data_t)hash); - last_expr = nd_type_p(head, NODE_LIT) ? value : head; hash = next; } - st_foreach(literal_keys, append_literal_keys, (st_data_t)&result); st_free_table(literal_keys); - if (hash) { - if (!result) result = hash; - else list_concat(result, hash); - } - result->nd_loc = loc; - return result; } static NODE * new_hash(struct parser_params *p, NODE *hash, const YYLTYPE *loc) { - if (hash) hash = remove_duplicate_keys(p, hash); + if (hash) warn_duplicate_keys(p, hash); return NEW_HASH(hash, loc); } -#endif static void error_duplicate_pattern_variable(struct parser_params *p, ID id, const YYLTYPE *loc) { - if (is_private_local_id(id)) { + if (is_private_local_id(p, id)) { return; } if (st_is_member(p->pvtbl, id)) { @@ -12934,69 +14831,39 @@ error_duplicate_pattern_key(struct parser_params *p, VALUE key, const YYLTYPE *l st_insert(p->pktbl, (st_data_t)key, 0); } -#ifndef RIPPER static NODE * new_unique_key_hash(struct parser_params *p, NODE *hash, const YYLTYPE *loc) { return NEW_HASH(hash, loc); } -#endif /* !RIPPER */ -#ifndef RIPPER static NODE * new_op_assign(struct parser_params *p, NODE *lhs, ID op, NODE *rhs, struct lex_context ctxt, const YYLTYPE *loc) { NODE *asgn; if (lhs) { - ID vid = lhs->nd_vid; + ID vid = get_nd_vid(p, lhs); YYLTYPE lhs_loc = lhs->nd_loc; - int shareable = ctxt.shareable_constant_value; - if (shareable) { - switch (nd_type(lhs)) { - case NODE_CDECL: - case NODE_COLON2: - case NODE_COLON3: - break; - default: - shareable = 0; - break; - } - } if (op == tOROP) { - rhs = shareable_constant_value(p, shareable, lhs, rhs, &rhs->nd_loc); - lhs->nd_value = rhs; + set_nd_value(p, lhs, rhs); nd_set_loc(lhs, loc); asgn = NEW_OP_ASGN_OR(gettable(p, vid, &lhs_loc), lhs, loc); - if (is_notop_id(vid)) { - switch (id_type(vid)) { - case ID_GLOBAL: - case ID_INSTANCE: - case ID_CLASS: - asgn->nd_aid = vid; - } - } } else if (op == tANDOP) { - if (shareable) { - rhs = shareable_constant_value(p, shareable, lhs, rhs, &rhs->nd_loc); - } - lhs->nd_value = rhs; + set_nd_value(p, lhs, rhs); nd_set_loc(lhs, loc); asgn = NEW_OP_ASGN_AND(gettable(p, vid, &lhs_loc), lhs, loc); } else { asgn = lhs; rhs = NEW_CALL(gettable(p, vid, &lhs_loc), op, NEW_LIST(rhs, &rhs->nd_loc), loc); - if (shareable) { - rhs = shareable_constant_value(p, shareable, lhs, rhs, &rhs->nd_loc); - } - asgn->nd_value = rhs; + set_nd_value(p, asgn, rhs); nd_set_loc(asgn, loc); } } else { - asgn = NEW_BEGIN(0, loc); + asgn = NEW_ERROR(loc); } return asgn; } @@ -13007,14 +14874,9 @@ new_ary_op_assign(struct parser_params *p, NODE *ary, { NODE *asgn; + aryset_check(p, args); args = make_list(args, args_loc); - if (nd_type_p(args, NODE_BLOCK_PASS)) { - args = NEW_ARGSCAT(args, rhs, loc); - } - else { - args = arg_concat(p, args, rhs, loc); - } - asgn = NEW_OP_ASGN1(ary, op, args, loc); + asgn = NEW_OP_ASGN1(ary, op, args, rhs, loc); fixpos(asgn, ary); return asgn; } @@ -13036,11 +14898,10 @@ new_const_op_assign(struct parser_params *p, NODE *lhs, ID op, NODE *rhs, struct NODE *asgn; if (lhs) { - rhs = shareable_constant_value(p, ctxt.shareable_constant_value, lhs, rhs, loc); - asgn = NEW_OP_CDECL(lhs, op, rhs, loc); + asgn = NEW_OP_CDECL(lhs, op, rhs, ctxt.shareable_constant_value, loc); } else { - asgn = NEW_BEGIN(0, loc); + asgn = NEW_ERROR(loc); } fixpos(asgn, lhs); return asgn; @@ -13050,20 +14911,16 @@ static NODE * const_decl(struct parser_params *p, NODE *path, const YYLTYPE *loc) { if (p->ctxt.in_def) { +#ifndef RIPPER yyerror1(loc, "dynamic constant assignment"); - } - return NEW_CDECL(0, 0, (path), loc); -} #else -static VALUE -const_decl(struct parser_params *p, VALUE path) -{ - if (p->ctxt.in_def) { - path = assign_error(p, "dynamic constant assignment", path); + set_value(assign_error(p, "dynamic constant assignment", p->s_lvalue)); +#endif } - return path; + return NEW_CDECL(0, 0, (path), p->ctxt.shareable_constant_value, loc); } +#ifdef RIPPER static VALUE assign_error(struct parser_params *p, const char *mesg, VALUE a) { @@ -13071,15 +14928,8 @@ assign_error(struct parser_params *p, const char *mesg, VALUE a) ripper_error(p); return a; } - -static VALUE -var_field(struct parser_params *p, VALUE a) -{ - return ripper_new_yylval(p, get_id(a), dispatch1(var_field, a), 0); -} #endif -#ifndef RIPPER static NODE * new_bodystmt(struct parser_params *p, NODE *head, NODE *rescue, NODE *rescue_else, NODE *ensure, const YYLTYPE *loc) { @@ -13100,7 +14950,6 @@ new_bodystmt(struct parser_params *p, NODE *head, NODE *rescue, NODE *rescue_els fixpos(result, head); return result; } -#endif static void warn_unused_var(struct parser_params *p, struct local_vars *local) @@ -13117,7 +14966,7 @@ warn_unused_var(struct parser_params *p, struct local_vars *local) ID *u = local->used->tbl; for (int i = 0; i < cnt; ++i) { if (!v[i] || (u[i] & LVAR_USED)) continue; - if (is_private_local_id(v[i])) continue; + if (is_private_local_id(p, v[i])) continue; rb_warn1L((int)u[i], "assigned but unused variable - %"PRIsWARN, rb_id2str(v[i])); } #endif @@ -13137,10 +14986,11 @@ local_push(struct parser_params *p, int toplevel_scope) #ifndef RIPPER if (toplevel_scope && compile_for_eval) warn_unused_vars = 0; if (toplevel_scope && e_option_supplied(p)) warn_unused_vars = 0; +#endif local->numparam.outer = 0; local->numparam.inner = 0; local->numparam.current = 0; -#endif + local->it = 0; local->used = warn_unused_vars ? vtable_alloc(0) : 0; # if WARN_PAST_SCOPE @@ -13152,29 +15002,45 @@ local_push(struct parser_params *p, int toplevel_scope) } static void +vtable_chain_free(struct parser_params *p, struct vtable *table) +{ + while (!DVARS_TERMINAL_P(table)) { + struct vtable *cur_table = table; + table = cur_table->prev; + vtable_free(cur_table); + } +} + +static void +local_free(struct parser_params *p, struct local_vars *local) +{ + vtable_chain_free(p, local->used); + +# if WARN_PAST_SCOPE + vtable_chain_free(p, local->past); +# endif + + vtable_chain_free(p, local->args); + vtable_chain_free(p, local->vars); + + ruby_sized_xfree(local, sizeof(struct local_vars)); +} + +static void local_pop(struct parser_params *p) { struct local_vars *local = p->lvtbl->prev; if (p->lvtbl->used) { warn_unused_var(p, p->lvtbl); - vtable_free(p->lvtbl->used); - } -# if WARN_PAST_SCOPE - while (p->lvtbl->past) { - struct vtable *past = p->lvtbl->past; - p->lvtbl->past = past->prev; - vtable_free(past); } -# endif - vtable_free(p->lvtbl->args); - vtable_free(p->lvtbl->vars); + + local_free(p, p->lvtbl); + p->lvtbl = local; + CMDARG_POP(); COND_POP(); - ruby_sized_xfree(p->lvtbl, sizeof(*p->lvtbl)); - p->lvtbl = local; } -#ifndef RIPPER static rb_ast_id_table_t * local_tbl(struct parser_params *p) { @@ -13201,19 +15067,6 @@ local_tbl(struct parser_params *p) return tbl; } -static NODE* -node_newnode_with_locals(struct parser_params *p, enum node_type type, VALUE a1, VALUE a2, const rb_code_location_t *loc) -{ - rb_ast_id_table_t *a0; - NODE *n; - - a0 = local_tbl(p); - n = NEW_NODE(type, a0, a1, a2, loc); - return n; -} - -#endif - static void numparam_name(struct parser_params *p, ID id) { @@ -13239,6 +15092,14 @@ local_var(struct parser_params *p, ID id) } } +#ifndef RIPPER +int +rb_parser_local_defined(struct parser_params *p, ID id, const struct rb_iseq_struct *iseq) +{ + return rb_local_defined(id, iseq); +} +#endif + static int local_id_ref(struct parser_params *p, ID id, ID **vidrefp) { @@ -13255,7 +15116,7 @@ local_id_ref(struct parser_params *p, ID id, ID **vidrefp) } if (vars && vars->prev == DVARS_INHERIT) { - return rb_local_defined(id, p->parent_iseq); + return rb_parser_local_defined(p, id, p->parent_iseq); } else if (vtable_included(args, id)) { return 1; @@ -13292,7 +15153,40 @@ add_forwarding_args(struct parser_params *p) arg_var(p, idFWD_ALL); } -#ifndef RIPPER +static void +forwarding_arg_check(struct parser_params *p, ID arg, ID all, const char *var) +{ + bool conflict = false; + + struct vtable *vars, *args; + + vars = p->lvtbl->vars; + args = p->lvtbl->args; + + while (vars && !DVARS_TERMINAL_P(vars->prev)) { + conflict |= (vtable_included(args, arg) && !(all && vtable_included(args, all))); + vars = vars->prev; + args = args->prev; + } + + bool found = false; + if (vars && vars->prev == DVARS_INHERIT && !found) { + found = (rb_parser_local_defined(p, arg, p->parent_iseq) && + !(all && rb_parser_local_defined(p, all, p->parent_iseq))); + } + else { + found = (vtable_included(args, arg) && + !(all && vtable_included(args, all))); + } + + if (!found) { + compile_error(p, "no anonymous %s parameter", var); + } + else if (conflict) { + compile_error(p, "anonymous %s parameter is also used within block", var); + } +} + static NODE * new_args_forward_call(struct parser_params *p, NODE *leading, const YYLTYPE *loc, const YYLTYPE *argsloc) { @@ -13300,19 +15194,17 @@ new_args_forward_call(struct parser_params *p, NODE *leading, const YYLTYPE *loc #ifndef FORWARD_ARGS_WITH_RUBY2_KEYWORDS NODE *kwrest = list_append(p, NEW_LIST(0, loc), NEW_LVAR(idFWD_KWREST, loc)); #endif - NODE *block = NEW_BLOCK_PASS(NEW_LVAR(idFWD_BLOCK, loc), loc); + rb_node_block_pass_t *block = NEW_BLOCK_PASS(NEW_LVAR(idFWD_BLOCK, loc), loc); NODE *args = leading ? rest_arg_append(p, leading, rest, argsloc) : NEW_SPLAT(rest, loc); #ifndef FORWARD_ARGS_WITH_RUBY2_KEYWORDS args = arg_append(p, args, new_hash(p, kwrest, loc), loc); #endif return arg_blk_pass(args, block); } -#endif static NODE * numparam_push(struct parser_params *p) { -#ifndef RIPPER struct local_vars *local = p->lvtbl; NODE *inner = local->numparam.inner; if (!local->numparam.outer) { @@ -13320,16 +15212,13 @@ numparam_push(struct parser_params *p) } local->numparam.inner = 0; local->numparam.current = 0; + local->it = 0; return inner; -#else - return 0; -#endif } static void numparam_pop(struct parser_params *p, NODE *prev_inner) { -#ifndef RIPPER struct local_vars *local = p->lvtbl; if (prev_inner) { /* prefer first one */ @@ -13348,7 +15237,7 @@ numparam_pop(struct parser_params *p, NODE *prev_inner) /* no numbered parameter */ local->numparam.current = 0; } -#endif + local->it = 0; } static const struct vtable * @@ -13411,7 +15300,8 @@ dyna_in_block(struct parser_params *p) return !DVARS_TERMINAL_P(p->lvtbl->vars) && p->lvtbl->vars->prev != DVARS_TOPSCOPE; } -static int +#ifndef RIPPER +int dvar_defined_ref(struct parser_params *p, ID id, ID **vidrefp) { struct vtable *vars, *args, *used; @@ -13441,6 +15331,7 @@ dvar_defined_ref(struct parser_params *p, ID id, ID **vidrefp) return 0; } +#endif static int dvar_defined(struct parser_params *p, ID id) @@ -13456,43 +15347,83 @@ dvar_curr(struct parser_params *p, ID id) } static void -reg_fragment_enc_error(struct parser_params* p, VALUE str, int c) +reg_fragment_enc_error(struct parser_params* p, rb_parser_string_t *str, int c) { compile_error(p, "regexp encoding option '%c' differs from source encoding '%s'", - c, rb_enc_name(rb_enc_get(str))); + c, rb_enc_name(rb_parser_str_get_encoding(str))); } #ifndef RIPPER +static rb_encoding * +find_enc(struct parser_params* p, const char *name) +{ + int idx = rb_enc_find_index(name); + if (idx < 0) { + rb_bug("unknown encoding name: %s", name); + } + + return rb_enc_from_index(idx); +} + +static rb_encoding * +kcode_to_enc(struct parser_params* p, int kcode) +{ + rb_encoding *enc; + + switch (kcode) { + case ENC_ASCII8BIT: + enc = rb_ascii8bit_encoding(); + break; + case ENC_EUC_JP: + enc = find_enc(p, "EUC-JP"); + break; + case ENC_Windows_31J: + enc = find_enc(p, "Windows-31J"); + break; + case ENC_UTF8: + enc = rb_utf8_encoding(); + break; + default: + enc = NULL; + break; + } + + return enc; +} + int -rb_reg_fragment_setenc(struct parser_params* p, VALUE str, int options) +rb_reg_fragment_setenc(struct parser_params* p, rb_parser_string_t *str, int options) { int c = RE_OPTION_ENCODING_IDX(options); if (c) { int opt, idx; - rb_char_to_option_kcode(c, &opt, &idx); - if (idx != ENCODING_GET(str) && - !is_ascii_string(str)) { + rb_encoding *enc; + + char_to_option_kcode(c, &opt, &idx); + enc = kcode_to_enc(p, idx); + if (enc != rb_parser_str_get_encoding(str) && + !rb_parser_is_ascii_string(p, str)) { goto error; } - ENCODING_SET(str, idx); + rb_parser_string_set_encoding(str, enc); } else if (RE_OPTION_ENCODING_NONE(options)) { - if (!ENCODING_IS_ASCII8BIT(str) && - !is_ascii_string(str)) { + if (!PARSER_ENCODING_IS_ASCII8BIT(p, str) && + !rb_parser_is_ascii_string(p, str)) { c = 'n'; goto error; } - rb_enc_associate(str, rb_ascii8bit_encoding()); + rb_parser_enc_associate(p, str, rb_ascii8bit_encoding()); } else if (rb_is_usascii_enc(p->enc)) { - if (!is_ascii_string(str)) { + if (!rb_parser_is_ascii_string(p, str)) { /* raise in re.c */ - rb_enc_associate(str, rb_usascii_encoding()); + rb_parser_enc_associate(p, str, rb_usascii_encoding()); } else { - rb_enc_associate(str, rb_ascii8bit_encoding()); + rb_parser_enc_associate(p, str, rb_ascii8bit_encoding()); } } return 0; @@ -13500,20 +15431,24 @@ rb_reg_fragment_setenc(struct parser_params* p, VALUE str, int options) error: return c; } +#endif static void -reg_fragment_setenc(struct parser_params* p, VALUE str, int options) +reg_fragment_setenc(struct parser_params* p, rb_parser_string_t *str, int options) { int c = rb_reg_fragment_setenc(p, str, options); if (c) reg_fragment_enc_error(p, str, c); } -static int -reg_fragment_check(struct parser_params* p, VALUE str, int options) +#ifndef RIPPER +int +reg_fragment_check(struct parser_params* p, rb_parser_string_t *str, int options) { - VALUE err; + VALUE err, str2; reg_fragment_setenc(p, str, options); - err = rb_reg_check_preprocess(str); + /* TODO */ + str2 = rb_str_new_parser_string(str); + err = rb_reg_check_preprocess(str2); if (err != Qnil) { err = rb_obj_as_string(err); compile_error(p, "%"PRIsVALUE, err); @@ -13521,7 +15456,9 @@ reg_fragment_check(struct parser_params* p, VALUE str, int options) } return 1; } +#endif +#ifndef UNIVERSAL_PARSER typedef struct { struct parser_params* parser; rb_encoding *enc; @@ -13538,23 +15475,8 @@ reg_named_capture_assign_iter(const OnigUChar *name, const OnigUChar *name_end, rb_encoding *enc = arg->enc; long len = name_end - name; const char *s = (const char *)name; - ID var; - NODE *node, *succ; - if (!len) return ST_CONTINUE; - if (rb_enc_symname_type(s, len, enc, (1U<<ID_LOCAL)) != ID_LOCAL) - return ST_CONTINUE; - - var = intern_cstr(s, len, enc); - if (len < MAX_WORD_LENGTH && rb_reserved_word(s, (int)len)) { - if (!lvar_defined(p, var)) return ST_CONTINUE; - } - node = node_assign(p, assignable(p, var, 0, arg->loc), NEW_LIT(ID2SYM(var), arg->loc), NO_LEX_CTXT, arg->loc); - succ = arg->succ_block; - if (!succ) succ = NEW_BEGIN(0, arg->loc); - succ = block_append(p, succ, node); - arg->succ_block = succ; - return ST_CONTINUE; + return rb_reg_named_capture_assign_iter_impl(p, s, len, enc, &arg->succ_block, arg->loc); } static NODE * @@ -13569,24 +15491,54 @@ reg_named_capture_assign(struct parser_params* p, VALUE regexp, const YYLTYPE *l onig_foreach_name(RREGEXP_PTR(regexp), reg_named_capture_assign_iter, &arg); if (!arg.succ_block) return 0; - return arg.succ_block->nd_next; + return RNODE_BLOCK(arg.succ_block)->nd_next; } +#endif + +#ifndef RIPPER +int +rb_reg_named_capture_assign_iter_impl(struct parser_params *p, const char *s, long len, + rb_encoding *enc, NODE **succ_block, const rb_code_location_t *loc) +{ + ID var; + NODE *node, *succ; + + if (!len) return ST_CONTINUE; + if (!VALID_SYMNAME_P(s, len, enc, ID_LOCAL)) + return ST_CONTINUE; + + var = intern_cstr(s, len, enc); + if (len < MAX_WORD_LENGTH && rb_reserved_word(s, (int)len)) { + if (!lvar_defined(p, var)) return ST_CONTINUE; + } + node = node_assign(p, assignable(p, var, 0, loc), NEW_SYM(rb_id2str(var), loc), NO_LEX_CTXT, loc); + succ = *succ_block; + if (!succ) succ = NEW_ERROR(loc); + succ = block_append(p, succ, node); + *succ_block = succ; + return ST_CONTINUE; +} +#endif static VALUE -parser_reg_compile(struct parser_params* p, VALUE str, int options) +parser_reg_compile(struct parser_params* p, rb_parser_string_t *str, int options) { + VALUE str2; reg_fragment_setenc(p, str, options); - return rb_parser_reg_compile(p, str, options); + str2 = rb_str_new_parser_string(str); + return rb_parser_reg_compile(p, str2, options); } +#ifndef RIPPER VALUE rb_parser_reg_compile(struct parser_params* p, VALUE str, int options) { return rb_reg_compile(str, options & RE_OPTION_MASK, p->ruby_sourcefile, p->ruby_sourceline); } +#endif static VALUE -reg_compile(struct parser_params* p, VALUE str, int options) +reg_compile(struct parser_params* p, rb_parser_string_t *str, int options) { VALUE re; VALUE err; @@ -13601,30 +15553,11 @@ reg_compile(struct parser_params* p, VALUE str, int options) } return re; } -#else -static VALUE -parser_reg_compile(struct parser_params* p, VALUE str, int options, VALUE *errmsg) -{ - VALUE err = rb_errinfo(); - VALUE re; - str = ripper_is_node_yylval(str) ? RNODE(str)->nd_cval : str; - int c = rb_reg_fragment_setenc(p, str, options); - if (c) reg_fragment_enc_error(p, str, c); - re = rb_parser_reg_compile(p, str, options); - if (NIL_P(re)) { - *errmsg = rb_attr_get(rb_errinfo(), idMesg); - rb_set_errinfo(err); - } - return re; -} -#endif #ifndef RIPPER void -rb_parser_set_options(VALUE vparser, int print, int loop, int chomp, int split) +rb_ruby_parser_set_options(struct parser_params *p, int print, int loop, int chomp, int split) { - struct parser_params *p; - TypedData_Get_Struct(vparser, struct parser_params, &parser_data_type, p); p->do_print = print; p->do_loop = loop; p->do_chomp = chomp; @@ -13638,7 +15571,7 @@ parser_append_options(struct parser_params *p, NODE *node) const YYLTYPE *const LOC = &default_location; if (p->do_print) { - NODE *print = NEW_FCALL(rb_intern("print"), + NODE *print = (NODE *)NEW_FCALL(rb_intern("print"), NEW_LIST(NEW_GVAR(idLASTLINE, LOC), LOC), LOC); node = block_append(p, node, print); @@ -13658,12 +15591,12 @@ parser_append_options(struct parser_params *p, NODE *node) node = block_append(p, split, node); } if (p->do_chomp) { - NODE *chomp = NEW_LIT(ID2SYM(rb_intern("chomp")), LOC); + NODE *chomp = NEW_SYM(rb_str_new_cstr("chomp"), LOC); chomp = list_append(p, NEW_LIST(chomp, LOC), NEW_TRUE(LOC)); irs = list_append(p, irs, NEW_HASH(chomp, LOC)); } - node = NEW_WHILE(NEW_FCALL(idGets, irs, LOC), node, 1, LOC); + node = NEW_WHILE((NODE *)NEW_FCALL(idGets, irs, LOC), node, 1, LOC); } return node; @@ -13677,7 +15610,7 @@ rb_init_parse(void) (void)nodeline; } -static ID +ID internal_id(struct parser_params *p) { return rb_make_temporary_id(vtable_size(p->lvtbl->args) + vtable_size(p->lvtbl->vars)); @@ -13689,74 +15622,80 @@ parser_initialize(struct parser_params *p) { /* note: we rely on TypedData_Make_Struct to set most fields to 0 */ p->command_start = TRUE; - p->ruby_sourcefile_string = Qnil; + p->ruby_sourcefile_string = NULL; p->lex.lpar_beg = -1; /* make lambda_beginning_p() == FALSE at first */ + string_buffer_init(p); p->node_id = 0; - p->delayed.token = Qnil; -#ifdef RIPPER - p->result = Qnil; - p->parsing_thread = Qnil; -#else + p->delayed.token = NULL; + p->frozen_string_literal = -1; /* not specified */ +#ifndef RIPPER p->error_buffer = Qfalse; - p->end_expect_token_locations = Qnil; + p->end_expect_token_locations = NULL; p->token_id = 0; - p->tokens = Qnil; + p->tokens = NULL; +#else + p->result = Qnil; + p->parsing_thread = Qnil; + p->s_value = Qnil; + p->s_lvalue = Qnil; + p->s_value_stack = rb_ary_new(); #endif p->debug_buffer = Qnil; p->debug_output = rb_ractor_stdout(); p->enc = rb_utf8_encoding(); + p->exits = 0; } #ifdef RIPPER -#define parser_mark ripper_parser_mark -#define parser_free ripper_parser_free +#define rb_ruby_parser_mark ripper_parser_mark +#define rb_ruby_parser_free ripper_parser_free +#define rb_ruby_parser_memsize ripper_parser_memsize #endif -static void -parser_mark(void *ptr) +void +rb_ruby_parser_mark(void *ptr) { struct parser_params *p = (struct parser_params*)ptr; - rb_gc_mark(p->lex.input); - rb_gc_mark(p->lex.lastline); - rb_gc_mark(p->lex.nextline); - rb_gc_mark(p->ruby_sourcefile_string); - rb_gc_mark((VALUE)p->lex.strterm); - rb_gc_mark((VALUE)p->ast); - rb_gc_mark(p->case_labels); - rb_gc_mark(p->delayed.token); #ifndef RIPPER - rb_gc_mark(p->debug_lines); - rb_gc_mark(p->compile_option); rb_gc_mark(p->error_buffer); - rb_gc_mark(p->end_expect_token_locations); - rb_gc_mark(p->tokens); #else rb_gc_mark(p->value); rb_gc_mark(p->result); rb_gc_mark(p->parsing_thread); + rb_gc_mark(p->s_value); + rb_gc_mark(p->s_lvalue); + rb_gc_mark(p->s_value_stack); #endif rb_gc_mark(p->debug_buffer); rb_gc_mark(p->debug_output); -#ifdef YYMALLOC - rb_gc_mark((VALUE)p->heap); -#endif } -static void -parser_free(void *ptr) +void +rb_ruby_parser_free(void *ptr) { struct parser_params *p = (struct parser_params*)ptr; struct local_vars *local, *prev; +#ifndef RIPPER + if (p->tokens) { + rb_parser_ary_free(p, p->tokens); + } +#endif + if (p->tokenbuf) { ruby_sized_xfree(p->tokenbuf, p->toksiz); } + + if (p->ruby_sourcefile_string) { + rb_parser_string_free(p, p->ruby_sourcefile_string); + } + for (local = p->lvtbl; local; local = prev) { - if (local->vars) xfree(local->vars); prev = local->prev; - xfree(local); + local_free(p, local); } + { token_info *ptinfo; while ((ptinfo = p->token_info) != 0) { @@ -13764,11 +15703,21 @@ parser_free(void *ptr) xfree(ptinfo); } } + string_buffer_free(p); + + if (p->pvtbl) { + st_free_table(p->pvtbl); + } + + if (CASE_LABELS_ENABLED_P(p->case_labels)) { + st_free_table(p->case_labels); + } + xfree(ptr); } -static size_t -parser_memsize(const void *ptr) +size_t +rb_ruby_parser_memsize(const void *ptr) { struct parser_params *p = (struct parser_params*)ptr; struct local_vars *local; @@ -13782,20 +15731,6 @@ parser_memsize(const void *ptr) return size; } -static const rb_data_type_t parser_data_type = { -#ifndef RIPPER - "parser", -#else - "ripper", -#endif - { - parser_mark, - parser_free, - parser_memsize, - }, - 0, 0, RUBY_TYPED_FREE_IMMEDIATELY -}; - #ifndef RIPPER #undef rb_reserved_word @@ -13805,245 +15740,260 @@ rb_reserved_word(const char *str, unsigned int len) return reserved_word(str, len); } -VALUE -rb_parser_new(void) +#ifdef UNIVERSAL_PARSER +rb_parser_t * +rb_ruby_parser_allocate(const rb_parser_config_t *config) +{ + /* parser_initialize expects fields to be set to 0 */ + rb_parser_t *p = (rb_parser_t *)config->calloc(1, sizeof(rb_parser_t)); + p->config = config; + return p; +} + +rb_parser_t * +rb_ruby_parser_new(const rb_parser_config_t *config) { - struct parser_params *p; - VALUE parser = TypedData_Make_Struct(0, struct parser_params, - &parser_data_type, p); + /* parser_initialize expects fields to be set to 0 */ + rb_parser_t *p = rb_ruby_parser_allocate(config); parser_initialize(p); - return parser; + return p; +} +#else +rb_parser_t * +rb_ruby_parser_allocate(void) +{ + /* parser_initialize expects fields to be set to 0 */ + rb_parser_t *p = (rb_parser_t *)ruby_xcalloc(1, sizeof(rb_parser_t)); + return p; } -VALUE -rb_parser_set_context(VALUE vparser, const struct rb_iseq_struct *base, int main) +rb_parser_t * +rb_ruby_parser_new(void) { - struct parser_params *p; + /* parser_initialize expects fields to be set to 0 */ + rb_parser_t *p = rb_ruby_parser_allocate(); + parser_initialize(p); + return p; +} +#endif - TypedData_Get_Struct(vparser, struct parser_params, &parser_data_type, p); +rb_parser_t * +rb_ruby_parser_set_context(rb_parser_t *p, const struct rb_iseq_struct *base, int main) +{ p->error_buffer = main ? Qfalse : Qnil; p->parent_iseq = base; - return vparser; + return p; } void -rb_parser_keep_script_lines(VALUE vparser) +rb_ruby_parser_set_script_lines(rb_parser_t *p) { - struct parser_params *p; - - TypedData_Get_Struct(vparser, struct parser_params, &parser_data_type, p); - p->keep_script_lines = 1; + p->debug_lines = rb_parser_ary_new_capa_for_script_line(p, 10); } void -rb_parser_error_tolerant(VALUE vparser) +rb_ruby_parser_error_tolerant(rb_parser_t *p) { - struct parser_params *p; - - TypedData_Get_Struct(vparser, struct parser_params, &parser_data_type, p); p->error_tolerant = 1; - p->end_expect_token_locations = rb_ary_new(); } void -rb_parser_keep_tokens(VALUE vparser) +rb_ruby_parser_keep_tokens(rb_parser_t *p) { - struct parser_params *p; - - TypedData_Get_Struct(vparser, struct parser_params, &parser_data_type, p); p->keep_tokens = 1; - p->tokens = rb_ary_new(); + p->tokens = rb_parser_ary_new_capa_for_ast_token(p, 10); } -#endif +rb_encoding * +rb_ruby_parser_encoding(rb_parser_t *p) +{ + return p->enc; +} -#ifdef RIPPER -#define rb_parser_end_seen_p ripper_parser_end_seen_p -#define rb_parser_encoding ripper_parser_encoding -#define rb_parser_get_yydebug ripper_parser_get_yydebug -#define rb_parser_set_yydebug ripper_parser_set_yydebug -#define rb_parser_get_debug_output ripper_parser_get_debug_output -#define rb_parser_set_debug_output ripper_parser_set_debug_output -static VALUE ripper_parser_end_seen_p(VALUE vparser); -static VALUE ripper_parser_encoding(VALUE vparser); -static VALUE ripper_parser_get_yydebug(VALUE self); -static VALUE ripper_parser_set_yydebug(VALUE self, VALUE flag); -static VALUE ripper_parser_get_debug_output(VALUE self); -static VALUE ripper_parser_set_debug_output(VALUE self, VALUE output); +int +rb_ruby_parser_end_seen_p(rb_parser_t *p) +{ + return p->ruby__end__seen; +} -/* - * call-seq: - * ripper.error? -> Boolean - * - * Return true if parsed source has errors. - */ -static VALUE -ripper_error_p(VALUE vparser) +int +rb_ruby_parser_set_yydebug(rb_parser_t *p, int flag) { - struct parser_params *p; + p->debug = flag; + return flag; +} +#endif /* !RIPPER */ - TypedData_Get_Struct(vparser, struct parser_params, &parser_data_type, p); - return RBOOL(p->error_p); +#ifdef RIPPER +int +rb_ruby_parser_get_yydebug(rb_parser_t *p) +{ + return p->debug; } -#endif -/* - * call-seq: - * ripper.end_seen? -> Boolean - * - * Return true if parsed source ended by +\_\_END\_\_+. - */ -VALUE -rb_parser_end_seen_p(VALUE vparser) +void +rb_ruby_parser_set_value(rb_parser_t *p, VALUE value) { - struct parser_params *p; + p->value = value; +} - TypedData_Get_Struct(vparser, struct parser_params, &parser_data_type, p); - return RBOOL(p->ruby__end__seen); +int +rb_ruby_parser_error_p(rb_parser_t *p) +{ + return p->error_p; } -/* - * call-seq: - * ripper.encoding -> encoding - * - * Return encoding of the source. - */ VALUE -rb_parser_encoding(VALUE vparser) +rb_ruby_parser_debug_output(rb_parser_t *p) { - struct parser_params *p; + return p->debug_output; +} - TypedData_Get_Struct(vparser, struct parser_params, &parser_data_type, p); - return rb_enc_from_encoding(p->enc); +void +rb_ruby_parser_set_debug_output(rb_parser_t *p, VALUE output) +{ + p->debug_output = output; } -#ifdef RIPPER -/* - * call-seq: - * ripper.yydebug -> true or false - * - * Get yydebug. - */ VALUE -rb_parser_get_yydebug(VALUE self) +rb_ruby_parser_parsing_thread(rb_parser_t *p) { - struct parser_params *p; - - TypedData_Get_Struct(self, struct parser_params, &parser_data_type, p); - return RBOOL(p->debug); + return p->parsing_thread; } -#endif -/* - * call-seq: - * ripper.yydebug = flag - * - * Set yydebug. - */ -VALUE -rb_parser_set_yydebug(VALUE self, VALUE flag) +void +rb_ruby_parser_set_parsing_thread(rb_parser_t *p, VALUE parsing_thread) { - struct parser_params *p; + p->parsing_thread = parsing_thread; +} - TypedData_Get_Struct(self, struct parser_params, &parser_data_type, p); - p->debug = RTEST(flag); - return flag; +void +rb_ruby_parser_ripper_initialize(rb_parser_t *p, rb_parser_lex_gets_func *gets, rb_parser_input_data input, VALUE sourcefile_string, const char *sourcefile, int sourceline) +{ + p->lex.gets = gets; + p->lex.input = input; + p->eofp = 0; + p->ruby_sourcefile_string = rb_str_to_parser_string(p, sourcefile_string); + p->ruby_sourcefile = sourcefile; + p->ruby_sourceline = sourceline; } -/* - * call-seq: - * ripper.debug_output -> obj - * - * Get debug output. - */ VALUE -rb_parser_get_debug_output(VALUE self) +rb_ruby_parser_result(rb_parser_t *p) { - struct parser_params *p; + return p->result; +} - TypedData_Get_Struct(self, struct parser_params, &parser_data_type, p); - return p->debug_output; +rb_encoding * +rb_ruby_parser_enc(rb_parser_t *p) +{ + return p->enc; } -/* - * call-seq: - * ripper.debug_output = obj - * - * Set debug output. - */ VALUE -rb_parser_set_debug_output(VALUE self, VALUE output) +rb_ruby_parser_ruby_sourcefile_string(rb_parser_t *p) { - struct parser_params *p; + return rb_str_new_parser_string(p->ruby_sourcefile_string); +} - TypedData_Get_Struct(self, struct parser_params, &parser_data_type, p); - return p->debug_output = output; +int +rb_ruby_parser_ruby_sourceline(rb_parser_t *p) +{ + return p->ruby_sourceline; } -#ifndef RIPPER -#ifdef YYMALLOC -#define HEAPCNT(n, size) ((n) * (size) / sizeof(YYSTYPE)) -/* Keep the order; NEWHEAP then xmalloc and ADD2HEAP to get rid of - * potential memory leak */ -#define NEWHEAP() rb_imemo_tmpbuf_parser_heap(0, p->heap, 0) -#define ADD2HEAP(new, cnt, ptr) ((p->heap = (new))->ptr = (ptr), \ - (new)->cnt = (cnt), (ptr)) +int +rb_ruby_parser_lex_state(rb_parser_t *p) +{ + return p->lex.state; +} -void * -rb_parser_malloc(struct parser_params *p, size_t size) +void +rb_ruby_ripper_parse0(rb_parser_t *p) { - size_t cnt = HEAPCNT(1, size); - rb_imemo_tmpbuf_t *n = NEWHEAP(); - void *ptr = xmalloc(size); + parser_prepare(p); + p->ast = rb_ast_new(); + ripper_yyparse((void*)p); + rb_ast_free(p->ast); + p->ast = 0; + p->eval_tree = 0; + p->eval_tree_begin = 0; +} - return ADD2HEAP(n, cnt, ptr); +int +rb_ruby_ripper_dedent_string(rb_parser_t *p, VALUE string, int width) +{ + char *str; + long len; + int i; + + RSTRING_GETMEM(string, str, len); + i = dedent_string_column(str, len, width); + if (!i) return 0; + + rb_str_modify(string); + str = RSTRING_PTR(string); + if (RSTRING_LEN(string) != len) + rb_fatal("literal string changed: %+"PRIsVALUE, string); + MEMMOVE(str, str + i, char, len - i); + rb_str_set_len(string, len - i); + return i; } -void * -rb_parser_calloc(struct parser_params *p, size_t nelem, size_t size) +int +rb_ruby_ripper_initialized_p(rb_parser_t *p) { - size_t cnt = HEAPCNT(nelem, size); - rb_imemo_tmpbuf_t *n = NEWHEAP(); - void *ptr = xcalloc(nelem, size); + return p->lex.input != 0; +} - return ADD2HEAP(n, cnt, ptr); +void +rb_ruby_ripper_parser_initialize(rb_parser_t *p) +{ + parser_initialize(p); } -void * -rb_parser_realloc(struct parser_params *p, void *ptr, size_t size) +long +rb_ruby_ripper_column(rb_parser_t *p) { - rb_imemo_tmpbuf_t *n; - size_t cnt = HEAPCNT(1, size); + return p->lex.ptok - p->lex.pbeg; +} - if (ptr && (n = p->heap) != NULL) { - do { - if (n->ptr == ptr) { - n->ptr = ptr = xrealloc(ptr, size); - if (n->cnt) n->cnt = cnt; - return ptr; - } - } while ((n = n->next) != NULL); - } - n = NEWHEAP(); - ptr = xrealloc(ptr, size); - return ADD2HEAP(n, cnt, ptr); +long +rb_ruby_ripper_token_len(rb_parser_t *p) +{ + return p->lex.pcur - p->lex.ptok; } -void -rb_parser_free(struct parser_params *p, void *ptr) +rb_parser_string_t * +rb_ruby_ripper_lex_lastline(rb_parser_t *p) { - rb_imemo_tmpbuf_t **prev = &p->heap, *n; + return p->lex.lastline; +} - while ((n = *prev) != NULL) { - if (n->ptr == ptr) { - *prev = n->next; - break; - } - prev = &n->next; - } +VALUE +rb_ruby_ripper_lex_state_name(struct parser_params *p, int state) +{ + return rb_parser_lex_state_name(p, (enum lex_state_e)state); +} + +#ifdef UNIVERSAL_PARSER +rb_parser_t * +rb_ripper_parser_params_allocate(const rb_parser_config_t *config) +{ + rb_parser_t *p = (rb_parser_t *)config->calloc(1, sizeof(rb_parser_t)); + p->config = config; + return p; } #endif +struct parser_params* +rb_ruby_ripper_parser_allocate(void) +{ + return (struct parser_params *)ruby_xcalloc(1, sizeof(struct parser_params)); +} +#endif /* RIPPER */ + +#ifndef RIPPER void rb_parser_printf(struct parser_params *p, const char *fmt, ...) { @@ -14054,25 +16004,34 @@ rb_parser_printf(struct parser_params *p, const char *fmt, ...) va_start(ap, fmt); rb_str_vcatf(mesg, fmt, ap); va_end(ap); - if (RSTRING_END(mesg)[-1] == '\n') { + if (end_with_newline_p(p, mesg)) { rb_io_write(p->debug_output, mesg); p->debug_buffer = Qnil; } } static void -parser_compile_error(struct parser_params *p, const char *fmt, ...) +parser_compile_error(struct parser_params *p, const rb_code_location_t *loc, const char *fmt, ...) { va_list ap; + int lineno, column; + + if (loc) { + lineno = loc->end_pos.lineno; + column = loc->end_pos.column; + } + else { + lineno = p->ruby_sourceline; + column = rb_long2int(p->lex.pcur - p->lex.pbeg); + } rb_io_flush(p->debug_output); p->error_p = 1; va_start(ap, fmt); p->error_buffer = rb_syntax_error_append(p->error_buffer, - p->ruby_sourcefile_string, - p->ruby_sourceline, - rb_long2int(p->lex.pcur - p->lex.pbeg), + p->ruby_sourcefile_string ? rb_str_new_parser_string(p->ruby_sourcefile_string) : Qnil, + lineno, column, p->enc, fmt, ap); va_end(ap); } @@ -14100,31 +16059,30 @@ rb_yytnamerr(struct parser_params *p, char *yyres, const char *yystr) while (*++yyp) { switch (*yyp) { - case '`': + case '\'': if (!bquote) { - bquote = count_char(yyp+1, '`') + 1; + bquote = count_char(yyp+1, '\'') + 1; if (yyres) memcpy(&yyres[yyn], yyp, bquote); yyn += bquote; yyp += bquote - 1; break; } - goto default_char; - - case '\'': - if (bquote && count_char(yyp+1, '\'') + 1 == bquote) { - if (yyres) memcpy(yyres + yyn, yyp, bquote); - yyn += bquote; - yyp += bquote - 1; - bquote = 0; - break; - } - if (yyp[1] && yyp[1] != '\'' && yyp[2] == '\'') { - if (yyres) memcpy(yyres + yyn, yyp, 3); - yyn += 3; - yyp += 2; - break; + else { + if (bquote && count_char(yyp+1, '\'') + 1 == bquote) { + if (yyres) memcpy(yyres + yyn, yyp, bquote); + yyn += bquote; + yyp += bquote - 1; + bquote = 0; + break; + } + if (yyp[1] && yyp[1] != '\'' && yyp[2] == '\'') { + if (yyres) memcpy(yyres + yyn, yyp, 3); + yyn += 3; + yyp += 2; + break; + } + goto do_not_strip_quotes; } - goto do_not_strip_quotes; case ',': goto do_not_strip_quotes; @@ -14133,7 +16091,6 @@ rb_yytnamerr(struct parser_params *p, char *yyres, const char *yystr) if (*++yyp != '\\') goto do_not_strip_quotes; /* Fall through. */ - default_char: default: if (yyres) yyres[yyn] = *yyp; @@ -14157,46 +16114,7 @@ rb_yytnamerr(struct parser_params *p, char *yyres, const char *yystr) #endif #ifdef RIPPER -#ifdef RIPPER_DEBUG -/* :nodoc: */ -static VALUE -ripper_validate_object(VALUE self, VALUE x) -{ - if (x == Qfalse) return x; - if (x == Qtrue) return x; - if (NIL_P(x)) return x; - if (UNDEF_P(x)) - rb_raise(rb_eArgError, "Qundef given"); - if (FIXNUM_P(x)) return x; - if (SYMBOL_P(x)) return x; - switch (BUILTIN_TYPE(x)) { - case T_STRING: - case T_OBJECT: - case T_ARRAY: - case T_BIGNUM: - case T_FLOAT: - case T_COMPLEX: - case T_RATIONAL: - break; - case T_NODE: - if (!nd_type_p((NODE *)x, NODE_RIPPER)) { - rb_raise(rb_eArgError, "NODE given: %p", (void *)x); - } - x = ((NODE *)x)->nd_rval; - break; - default: - rb_raise(rb_eArgError, "wrong type of ruby object: %p (%s)", - (void *)x, rb_obj_classname(x)); - } - if (!RBASIC_CLASS(x)) { - rb_raise(rb_eArgError, "hidden ruby object: %p (%s)", - (void *)x, rb_builtin_type_name(TYPE(x))); - } - return x; -} -#endif - -#define validate(x) ((x) = get_value(x)) +#define validate(x) (void)(x) static VALUE ripper_dispatch0(struct parser_params *p, ID mid) @@ -14262,377 +16180,21 @@ ripper_dispatch7(struct parser_params *p, ID mid, VALUE a, VALUE b, VALUE c, VAL return rb_funcall(p->value, mid, 7, a, b, c, d, e, f, g); } -static ID -ripper_get_id(VALUE v) -{ - NODE *nd; - if (!RB_TYPE_P(v, T_NODE)) return 0; - nd = (NODE *)v; - if (!nd_type_p(nd, NODE_RIPPER)) return 0; - return nd->nd_vid; -} - -static VALUE -ripper_get_value(VALUE v) -{ - NODE *nd; - if (UNDEF_P(v)) return Qnil; - if (!RB_TYPE_P(v, T_NODE)) return v; - nd = (NODE *)v; - if (!nd_type_p(nd, NODE_RIPPER)) return Qnil; - return nd->nd_rval; -} - -static void +void ripper_error(struct parser_params *p) { p->error_p = TRUE; } -static void -ripper_compile_error(struct parser_params *p, const char *fmt, ...) -{ - VALUE str; - va_list args; - - va_start(args, fmt); - str = rb_vsprintf(fmt, args); - va_end(args); - rb_funcall(p->value, rb_intern("compile_error"), 1, str); - ripper_error(p); -} - -static VALUE -ripper_lex_get_generic(struct parser_params *p, VALUE src) -{ - VALUE line = rb_funcallv_public(src, id_gets, 0, 0); - if (!NIL_P(line) && !RB_TYPE_P(line, T_STRING)) { - rb_raise(rb_eTypeError, - "gets returned %"PRIsVALUE" (expected String or nil)", - rb_obj_class(line)); - } - return line; -} - -static VALUE -ripper_lex_io_get(struct parser_params *p, VALUE src) -{ - return rb_io_gets(src); -} - -static VALUE -ripper_s_allocate(VALUE klass) -{ - struct parser_params *p; - VALUE self = TypedData_Make_Struct(klass, struct parser_params, - &parser_data_type, p); - p->value = self; - return self; -} - -#define ripper_initialized_p(r) ((r)->lex.input != 0) - -/* - * call-seq: - * Ripper.new(src, filename="(ripper)", lineno=1) -> ripper - * - * Create a new Ripper object. - * _src_ must be a String, an IO, or an Object which has #gets method. - * - * This method does not starts parsing. - * See also Ripper#parse and Ripper.parse. - */ -static VALUE -ripper_initialize(int argc, VALUE *argv, VALUE self) -{ - struct parser_params *p; - VALUE src, fname, lineno; - - TypedData_Get_Struct(self, struct parser_params, &parser_data_type, p); - rb_scan_args(argc, argv, "12", &src, &fname, &lineno); - if (RB_TYPE_P(src, T_FILE)) { - p->lex.gets = ripper_lex_io_get; - } - else if (rb_respond_to(src, id_gets)) { - p->lex.gets = ripper_lex_get_generic; - } - else { - StringValue(src); - p->lex.gets = lex_get_str; - } - p->lex.input = src; - p->eofp = 0; - if (NIL_P(fname)) { - fname = STR_NEW2("(ripper)"); - OBJ_FREEZE(fname); - } - else { - StringValueCStr(fname); - fname = rb_str_new_frozen(fname); - } - parser_initialize(p); - - p->ruby_sourcefile_string = fname; - p->ruby_sourcefile = RSTRING_PTR(fname); - p->ruby_sourceline = NIL_P(lineno) ? 0 : NUM2INT(lineno) - 1; - - return Qnil; -} - -static VALUE -ripper_parse0(VALUE parser_v) -{ - struct parser_params *p; - - TypedData_Get_Struct(parser_v, struct parser_params, &parser_data_type, p); - parser_prepare(p); - p->ast = rb_ast_new(); - ripper_yyparse((void*)p); - rb_ast_dispose(p->ast); - p->ast = 0; - return p->result; -} - -static VALUE -ripper_ensure(VALUE parser_v) -{ - struct parser_params *p; - - TypedData_Get_Struct(parser_v, struct parser_params, &parser_data_type, p); - p->parsing_thread = Qnil; - return Qnil; -} - -/* - * call-seq: - * ripper.parse - * - * Start parsing and returns the value of the root action. - */ -static VALUE -ripper_parse(VALUE self) -{ - struct parser_params *p; - - TypedData_Get_Struct(self, struct parser_params, &parser_data_type, p); - if (!ripper_initialized_p(p)) { - rb_raise(rb_eArgError, "method called for uninitialized object"); - } - if (!NIL_P(p->parsing_thread)) { - if (p->parsing_thread == rb_thread_current()) - rb_raise(rb_eArgError, "Ripper#parse is not reentrant"); - else - rb_raise(rb_eArgError, "Ripper#parse is not multithread-safe"); - } - p->parsing_thread = rb_thread_current(); - rb_ensure(ripper_parse0, self, ripper_ensure, self); - - return p->result; -} - -/* - * call-seq: - * ripper.column -> Integer - * - * Return column number of current parsing line. - * This number starts from 0. - */ -static VALUE -ripper_column(VALUE self) -{ - struct parser_params *p; - long col; - - TypedData_Get_Struct(self, struct parser_params, &parser_data_type, p); - if (!ripper_initialized_p(p)) { - rb_raise(rb_eArgError, "method called for uninitialized object"); - } - if (NIL_P(p->parsing_thread)) return Qnil; - col = p->lex.ptok - p->lex.pbeg; - return LONG2NUM(col); -} - -/* - * call-seq: - * ripper.filename -> String - * - * Return current parsing filename. - */ -static VALUE -ripper_filename(VALUE self) -{ - struct parser_params *p; - - TypedData_Get_Struct(self, struct parser_params, &parser_data_type, p); - if (!ripper_initialized_p(p)) { - rb_raise(rb_eArgError, "method called for uninitialized object"); - } - return p->ruby_sourcefile_string; -} - -/* - * call-seq: - * ripper.lineno -> Integer - * - * Return line number of current parsing line. - * This number starts from 1. - */ -static VALUE -ripper_lineno(VALUE self) -{ - struct parser_params *p; - - TypedData_Get_Struct(self, struct parser_params, &parser_data_type, p); - if (!ripper_initialized_p(p)) { - rb_raise(rb_eArgError, "method called for uninitialized object"); - } - if (NIL_P(p->parsing_thread)) return Qnil; - return INT2NUM(p->ruby_sourceline); -} - -/* - * call-seq: - * ripper.state -> Integer - * - * Return scanner state of current token. - */ -static VALUE -ripper_state(VALUE self) -{ - struct parser_params *p; - - TypedData_Get_Struct(self, struct parser_params, &parser_data_type, p); - if (!ripper_initialized_p(p)) { - rb_raise(rb_eArgError, "method called for uninitialized object"); - } - if (NIL_P(p->parsing_thread)) return Qnil; - return INT2NUM(p->lex.state); -} - -/* - * call-seq: - * ripper.token -> String - * - * Return the current token string. - */ -static VALUE -ripper_token(VALUE self) -{ - struct parser_params *p; - long pos, len; - - TypedData_Get_Struct(self, struct parser_params, &parser_data_type, p); - if (!ripper_initialized_p(p)) { - rb_raise(rb_eArgError, "method called for uninitialized object"); - } - if (NIL_P(p->parsing_thread)) return Qnil; - pos = p->lex.ptok - p->lex.pbeg; - len = p->lex.pcur - p->lex.ptok; - return rb_str_subseq(p->lex.lastline, pos, len); -} - -#ifdef RIPPER_DEBUG -/* :nodoc: */ -static VALUE -ripper_assert_Qundef(VALUE self, VALUE obj, VALUE msg) -{ - StringValue(msg); - if (UNDEF_P(obj)) { - rb_raise(rb_eArgError, "%"PRIsVALUE, msg); - } - return Qnil; -} - -/* :nodoc: */ -static VALUE -ripper_value(VALUE self, VALUE obj) -{ - return ULONG2NUM(obj); -} -#endif - -/* - * call-seq: - * Ripper.lex_state_name(integer) -> string - * - * Returns a string representation of lex_state. - */ -static VALUE -ripper_lex_state_name(VALUE self, VALUE state) -{ - return rb_parser_lex_state_name(NUM2INT(state)); -} - -void -Init_ripper(void) +VALUE +ripper_value(struct parser_params *p) { - ripper_init_eventids1(); - ripper_init_eventids2(); - id_warn = rb_intern_const("warn"); - id_warning = rb_intern_const("warning"); - id_gets = rb_intern_const("gets"); - id_assoc = rb_intern_const("=>"); - (void)yystpcpy; /* may not used in newer bison */ - InitVM(ripper); + return p->value; } -void -InitVM_ripper(void) -{ - VALUE Ripper; - - Ripper = rb_define_class("Ripper", rb_cObject); - /* version of Ripper */ - rb_define_const(Ripper, "Version", rb_usascii_str_new2(RIPPER_VERSION)); - rb_define_alloc_func(Ripper, ripper_s_allocate); - rb_define_method(Ripper, "initialize", ripper_initialize, -1); - rb_define_method(Ripper, "parse", ripper_parse, 0); - rb_define_method(Ripper, "column", ripper_column, 0); - rb_define_method(Ripper, "filename", ripper_filename, 0); - rb_define_method(Ripper, "lineno", ripper_lineno, 0); - rb_define_method(Ripper, "state", ripper_state, 0); - rb_define_method(Ripper, "token", ripper_token, 0); - rb_define_method(Ripper, "end_seen?", rb_parser_end_seen_p, 0); - rb_define_method(Ripper, "encoding", rb_parser_encoding, 0); - rb_define_method(Ripper, "yydebug", rb_parser_get_yydebug, 0); - rb_define_method(Ripper, "yydebug=", rb_parser_set_yydebug, 1); - rb_define_method(Ripper, "debug_output", rb_parser_get_debug_output, 0); - rb_define_method(Ripper, "debug_output=", rb_parser_set_debug_output, 1); - rb_define_method(Ripper, "error?", ripper_error_p, 0); -#ifdef RIPPER_DEBUG - rb_define_method(Ripper, "assert_Qundef", ripper_assert_Qundef, 2); - rb_define_method(Ripper, "rawVALUE", ripper_value, 1); - rb_define_method(Ripper, "validate_object", ripper_validate_object, 1); -#endif - - rb_define_singleton_method(Ripper, "dedent_string", parser_dedent_string, 2); - rb_define_private_method(Ripper, "dedent_string", parser_dedent_string, 2); - - rb_define_singleton_method(Ripper, "lex_state_name", ripper_lex_state_name, 1); - -<% @exprs.each do |expr, desc| -%> - /* <%=desc%> */ - rb_define_const(Ripper, "<%=expr%>", INT2NUM(<%=expr%>)); -<% end %> - ripper_init_eventids1_table(Ripper); - ripper_init_eventids2_table(Ripper); - -# if 0 - /* Hack to let RDoc document SCRIPT_LINES__ */ - - /* - * When a Hash is assigned to +SCRIPT_LINES__+ the contents of files loaded - * after the assignment will be added as an Array of lines with the file - * name as the key. - */ - rb_define_global_const("SCRIPT_LINES__", Qnil); -#endif - -} #endif /* RIPPER */ - /* * Local variables: * mode: c |