diff options
Diffstat (limited to 'include/ruby/encoding.h')
| -rw-r--r-- | include/ruby/encoding.h | 263 |
1 files changed, 168 insertions, 95 deletions
diff --git a/include/ruby/encoding.h b/include/ruby/encoding.h index cced3d2688..e6ceb19cdf 100644 --- a/include/ruby/encoding.h +++ b/include/ruby/encoding.h @@ -20,83 +20,118 @@ extern "C" { #endif #include <stdarg.h> +#include "ruby/ruby.h" #include "ruby/oniguruma.h" -#if defined __GNUC__ && __GNUC__ >= 4 -#pragma GCC visibility push(default) -#endif +RUBY_SYMBOL_EXPORT_BEGIN -#define ENCODING_INLINE_MAX 1023 -#define ENCODING_SHIFT (FL_USHIFT+10) -#define ENCODING_MASK (((VALUE)ENCODING_INLINE_MAX)<<ENCODING_SHIFT) +enum ruby_encoding_consts { + RUBY_ENCODING_INLINE_MAX = 127, + RUBY_ENCODING_SHIFT = (RUBY_FL_USHIFT+10), + RUBY_ENCODING_MASK = (RUBY_ENCODING_INLINE_MAX<<RUBY_ENCODING_SHIFT + /* RUBY_FL_USER10..RUBY_FL_USER16 */), + RUBY_ENCODING_MAXNAMELEN = 42 +}; -#define ENCODING_SET_INLINED(obj,i) do {\ - RBASIC(obj)->flags &= ~ENCODING_MASK;\ - RBASIC(obj)->flags |= (VALUE)(i) << ENCODING_SHIFT;\ -} while (0) -#define ENCODING_SET(obj,i) do {\ - VALUE rb_encoding_set_obj = (obj); \ - int encoding_set_enc_index = (i); \ - if (encoding_set_enc_index < ENCODING_INLINE_MAX) \ - ENCODING_SET_INLINED(rb_encoding_set_obj, encoding_set_enc_index); \ - else \ - rb_enc_set_index(rb_encoding_set_obj, encoding_set_enc_index); \ +#define ENCODING_INLINE_MAX RUBY_ENCODING_INLINE_MAX +#define ENCODING_SHIFT RUBY_ENCODING_SHIFT +#define ENCODING_MASK RUBY_ENCODING_MASK + +#define RB_ENCODING_SET_INLINED(obj,i) do {\ + RBASIC(obj)->flags &= ~RUBY_ENCODING_MASK;\ + RBASIC(obj)->flags |= (VALUE)(i) << RUBY_ENCODING_SHIFT;\ } while (0) +#define RB_ENCODING_SET(obj,i) rb_enc_set_index((obj), (i)) -#define ENCODING_GET_INLINED(obj) (int)((RBASIC(obj)->flags & ENCODING_MASK)>>ENCODING_SHIFT) -#define ENCODING_GET(obj) \ - (ENCODING_GET_INLINED(obj) != ENCODING_INLINE_MAX ? \ - ENCODING_GET_INLINED(obj) : \ +#define RB_ENCODING_GET_INLINED(obj) \ + (int)((RBASIC(obj)->flags & RUBY_ENCODING_MASK)>>RUBY_ENCODING_SHIFT) +#define RB_ENCODING_GET(obj) \ + (RB_ENCODING_GET_INLINED(obj) != RUBY_ENCODING_INLINE_MAX ? \ + RB_ENCODING_GET_INLINED(obj) : \ rb_enc_get_index(obj)) -#define ENCODING_IS_ASCII8BIT(obj) (ENCODING_GET_INLINED(obj) == 0) - -#define ENCODING_MAXNAMELEN 42 +#define RB_ENCODING_IS_ASCII8BIT(obj) (RB_ENCODING_GET_INLINED(obj) == 0) + +#define ENCODING_SET_INLINED(obj,i) RB_ENCODING_SET_INLINED(obj,i) +#define ENCODING_SET(obj,i) RB_ENCODING_SET(obj,i) +#define ENCODING_GET_INLINED(obj) RB_ENCODING_GET_INLINED(obj) +#define ENCODING_GET(obj) RB_ENCODING_GET(obj) +#define ENCODING_IS_ASCII8BIT(obj) RB_ENCODING_IS_ASCII8BIT(obj) +#define ENCODING_MAXNAMELEN RUBY_ENCODING_MAXNAMELEN + +enum ruby_coderange_type { + RUBY_ENC_CODERANGE_UNKNOWN = 0, + RUBY_ENC_CODERANGE_7BIT = ((int)RUBY_FL_USER8), + RUBY_ENC_CODERANGE_VALID = ((int)RUBY_FL_USER9), + RUBY_ENC_CODERANGE_BROKEN = ((int)(RUBY_FL_USER8|RUBY_FL_USER9)), + RUBY_ENC_CODERANGE_MASK = (RUBY_ENC_CODERANGE_7BIT| + RUBY_ENC_CODERANGE_VALID| + RUBY_ENC_CODERANGE_BROKEN) +}; -#define ENC_CODERANGE_MASK ((int)(FL_USER8|FL_USER9)) -#define ENC_CODERANGE_UNKNOWN 0 -#define ENC_CODERANGE_7BIT ((int)FL_USER8) -#define ENC_CODERANGE_VALID ((int)FL_USER9) -#define ENC_CODERANGE_BROKEN ((int)(FL_USER8|FL_USER9)) -#define ENC_CODERANGE(obj) ((int)RBASIC(obj)->flags & ENC_CODERANGE_MASK) -#define ENC_CODERANGE_ASCIIONLY(obj) (ENC_CODERANGE(obj) == ENC_CODERANGE_7BIT) -#define ENC_CODERANGE_SET(obj,cr) (RBASIC(obj)->flags = \ - (RBASIC(obj)->flags & ~ENC_CODERANGE_MASK) | (cr)) -#define ENC_CODERANGE_CLEAR(obj) ENC_CODERANGE_SET((obj),0) +static inline int +rb_enc_coderange_clean_p(int cr) +{ + return (cr ^ (cr >> 1)) & RUBY_ENC_CODERANGE_7BIT; +} +#define RB_ENC_CODERANGE_CLEAN_P(cr) rb_enc_coderange_clean_p(cr) +#define RB_ENC_CODERANGE(obj) ((int)RBASIC(obj)->flags & RUBY_ENC_CODERANGE_MASK) +#define RB_ENC_CODERANGE_ASCIIONLY(obj) (RB_ENC_CODERANGE(obj) == RUBY_ENC_CODERANGE_7BIT) +#define RB_ENC_CODERANGE_SET(obj,cr) (\ + RBASIC(obj)->flags = \ + (RBASIC(obj)->flags & ~RUBY_ENC_CODERANGE_MASK) | (cr)) +#define RB_ENC_CODERANGE_CLEAR(obj) RB_ENC_CODERANGE_SET((obj),0) /* assumed ASCII compatibility */ -#define ENC_CODERANGE_AND(a, b) \ - ((a) == ENC_CODERANGE_7BIT ? (b) : \ - (a) == ENC_CODERANGE_VALID ? ((b) == ENC_CODERANGE_7BIT ? ENC_CODERANGE_VALID : (b)) : \ - ENC_CODERANGE_UNKNOWN) +#define RB_ENC_CODERANGE_AND(a, b) \ + ((a) == RUBY_ENC_CODERANGE_7BIT ? (b) : \ + (a) != RUBY_ENC_CODERANGE_VALID ? RUBY_ENC_CODERANGE_UNKNOWN : \ + (b) == RUBY_ENC_CODERANGE_7BIT ? RUBY_ENC_CODERANGE_VALID : (b)) -#define ENCODING_CODERANGE_SET(obj, encindex, cr) \ +#define RB_ENCODING_CODERANGE_SET(obj, encindex, cr) \ do { \ VALUE rb_encoding_coderange_obj = (obj); \ - ENCODING_SET(rb_encoding_coderange_obj, (encindex)); \ - ENC_CODERANGE_SET(rb_encoding_coderange_obj, (cr)); \ + RB_ENCODING_SET(rb_encoding_coderange_obj, (encindex)); \ + RB_ENC_CODERANGE_SET(rb_encoding_coderange_obj, (cr)); \ } while (0) -typedef OnigEncodingType rb_encoding; +#define ENC_CODERANGE_MASK RUBY_ENC_CODERANGE_MASK +#define ENC_CODERANGE_UNKNOWN RUBY_ENC_CODERANGE_UNKNOWN +#define ENC_CODERANGE_7BIT RUBY_ENC_CODERANGE_7BIT +#define ENC_CODERANGE_VALID RUBY_ENC_CODERANGE_VALID +#define ENC_CODERANGE_BROKEN RUBY_ENC_CODERANGE_BROKEN +#define ENC_CODERANGE_CLEAN_P(cr) RB_ENC_CODERANGE_CLEAN_P(cr) +#define ENC_CODERANGE(obj) RB_ENC_CODERANGE(obj) +#define ENC_CODERANGE_ASCIIONLY(obj) RB_ENC_CODERANGE_ASCIIONLY(obj) +#define ENC_CODERANGE_SET(obj,cr) RB_ENC_CODERANGE_SET(obj,cr) +#define ENC_CODERANGE_CLEAR(obj) RB_ENC_CODERANGE_CLEAR(obj) +#define ENC_CODERANGE_AND(a, b) RB_ENC_CODERANGE_AND(a, b) +#define ENCODING_CODERANGE_SET(obj, encindex, cr) RB_ENCODING_CODERANGE_SET(obj, encindex, cr) + +typedef const OnigEncodingType rb_encoding; int rb_char_to_option_kcode(int c, int *option, int *kcode); int rb_enc_replicate(const char *, rb_encoding *); int rb_define_dummy_encoding(const char *); -#define rb_enc_to_index(enc) ((enc) ? ENC_TO_ENCINDEX(enc) : 0) +PUREFUNC(int rb_enc_dummy_p(rb_encoding *enc)); +PUREFUNC(int rb_enc_to_index(rb_encoding *enc)); int rb_enc_get_index(VALUE obj); void rb_enc_set_index(VALUE obj, int encindex); int rb_enc_find_index(const char *name); int rb_to_encoding_index(VALUE); -rb_encoding* rb_to_encoding(VALUE); -rb_encoding* rb_enc_get(VALUE); -rb_encoding* rb_enc_compatible(VALUE,VALUE); -rb_encoding* rb_enc_check(VALUE,VALUE); +rb_encoding *rb_to_encoding(VALUE); +rb_encoding *rb_find_encoding(VALUE); +rb_encoding *rb_enc_get(VALUE); +rb_encoding *rb_enc_compatible(VALUE,VALUE); +rb_encoding *rb_enc_check(VALUE,VALUE); VALUE rb_enc_associate_index(VALUE, int); VALUE rb_enc_associate(VALUE, rb_encoding*); void rb_enc_copy(VALUE dst, VALUE src); VALUE rb_enc_str_new(const char*, long, rb_encoding*); +VALUE rb_enc_str_new_cstr(const char*, rb_encoding*); +VALUE rb_enc_str_new_static(const char*, long, rb_encoding*); VALUE rb_enc_reg_new(const char*, long, rb_encoding*, int); PRINTF_ARGS(VALUE rb_enc_sprintf(rb_encoding *, const char*, ...), 2, 3); VALUE rb_enc_vsprintf(rb_encoding *, const char*, va_list); @@ -111,11 +146,26 @@ VALUE rb_str_export_to_enc(VALUE, rb_encoding *); VALUE rb_str_conv_enc(VALUE str, rb_encoding *from, rb_encoding *to); VALUE rb_str_conv_enc_opts(VALUE str, rb_encoding *from, rb_encoding *to, int ecflags, VALUE ecopts); +#ifdef HAVE_BUILTIN___BUILTIN_CONSTANT_P +#define rb_enc_str_new(str, len, enc) RB_GNUC_EXTENSION_BLOCK( \ + (__builtin_constant_p(str) && __builtin_constant_p(len)) ? \ + rb_enc_str_new_static((str), (len), (enc)) : \ + rb_enc_str_new((str), (len), (enc)) \ +) +#define rb_enc_str_new_cstr(str, enc) RB_GNUC_EXTENSION_BLOCK( \ + (__builtin_constant_p(str)) ? \ + rb_enc_str_new_static((str), (long)strlen(str), (enc)) : \ + rb_enc_str_new_cstr((str), (enc)) \ +) +#endif + +PRINTF_ARGS(NORETURN(void rb_enc_raise(rb_encoding *, VALUE, const char*, ...)), 3, 4); + /* index -> rb_encoding */ -rb_encoding* rb_enc_from_index(int idx); +rb_encoding *rb_enc_from_index(int idx); /* name -> rb_encoding */ -rb_encoding * rb_enc_find(const char *name); +rb_encoding *rb_enc_find(const char *name); /* rb_encoding * -> name */ #define rb_enc_name(enc) (enc)->name @@ -153,6 +203,9 @@ unsigned int rb_enc_codepoint(const char *p, const char *e, rb_encoding *enc); /* -> codelen>0 or raise exception */ int rb_enc_codelen(int code, rb_encoding *enc); +/* -> 0 for invalid codepoint */ +int rb_enc_code_to_mbclen(int code, rb_encoding *enc); +#define rb_enc_code_to_mbclen(c, enc) ONIGENC_CODE_TO_MBCLEN((enc), (c)); /* code,ptr,encoding -> write buf */ #define rb_enc_mbcput(c,buf,enc) ONIGENC_CODE_TO_MBC((enc),(c),(UChar*)(buf)) @@ -178,11 +231,16 @@ int rb_enc_codelen(int code, rb_encoding *enc); #define rb_enc_isspace(c,enc) ONIGENC_IS_CODE_SPACE((enc),(c)) #define rb_enc_isdigit(c,enc) ONIGENC_IS_CODE_DIGIT((enc),(c)) -#define rb_enc_asciicompat(enc) (rb_enc_mbminlen(enc)==1 && !rb_enc_dummy_p(enc)) +static inline int +rb_enc_asciicompat_inline(rb_encoding *enc) +{ + return rb_enc_mbminlen(enc)==1 && !rb_enc_dummy_p(enc); +} +#define rb_enc_asciicompat(enc) rb_enc_asciicompat_inline(enc) int rb_enc_casefold(char *to, const char *p, const char *e, rb_encoding *enc); -int rb_enc_toupper(int c, rb_encoding *enc); -int rb_enc_tolower(int c, rb_encoding *enc); +CONSTFUNC(int rb_enc_toupper(int c, rb_encoding *enc)); +CONSTFUNC(int rb_enc_tolower(int c, rb_encoding *enc)); ID rb_intern3(const char*, long, rb_encoding*); ID rb_interned_id_p(const char *, long, rb_encoding *); int rb_enc_symname_p(const char*, rb_encoding*); @@ -192,7 +250,7 @@ long rb_str_coderange_scan_restartable(const char*, const char*, rb_encoding*, i int rb_enc_str_asciionly_p(VALUE); #define rb_enc_str_asciicompat_p(str) rb_enc_asciicompat(rb_enc_get(str)) VALUE rb_enc_from_encoding(rb_encoding *enc); -int rb_enc_unicode_p(rb_encoding *enc); +PUREFUNC(int rb_enc_unicode_p(rb_encoding *enc)); rb_encoding *rb_ascii8bit_encoding(void); rb_encoding *rb_utf8_encoding(void); rb_encoding *rb_usascii_encoding(void); @@ -200,9 +258,15 @@ rb_encoding *rb_locale_encoding(void); rb_encoding *rb_filesystem_encoding(void); rb_encoding *rb_default_external_encoding(void); rb_encoding *rb_default_internal_encoding(void); -int rb_ascii8bit_encindex(void); -int rb_utf8_encindex(void); -int rb_usascii_encindex(void); +#ifndef rb_ascii8bit_encindex +CONSTFUNC(int rb_ascii8bit_encindex(void)); +#endif +#ifndef rb_utf8_encindex +CONSTFUNC(int rb_utf8_encindex(void)); +#endif +#ifndef rb_usascii_encindex +CONSTFUNC(int rb_usascii_encindex(void)); +#endif int rb_locale_encindex(void); int rb_filesystem_encindex(void); VALUE rb_enc_default_external(void); @@ -217,21 +281,10 @@ char *rb_enc_path_last_separator(const char *,const char *,rb_encoding*); char *rb_enc_path_end(const char *,const char *,rb_encoding*); const char *ruby_enc_find_basename(const char *name, long *baselen, long *alllen, rb_encoding *enc); const char *ruby_enc_find_extname(const char *name, long *len, rb_encoding *enc); +ID rb_check_id_cstr(const char *ptr, long len, rb_encoding *enc); +VALUE rb_check_symbol_cstr(const char *ptr, long len, rb_encoding *enc); RUBY_EXTERN VALUE rb_cEncoding; -#define ENC_DUMMY_FLAG (1<<24) -#define ENC_INDEX_MASK (~(~0U<<24)) - -#define ENC_TO_ENCINDEX(enc) (int)((enc)->ruby_encoding_index & ENC_INDEX_MASK) - -#define ENC_DUMMY_P(enc) ((enc)->ruby_encoding_index & ENC_DUMMY_FLAG) -#define ENC_SET_DUMMY(enc) ((enc)->ruby_encoding_index |= ENC_DUMMY_FLAG) - -static inline int -rb_enc_dummy_p(rb_encoding *enc) -{ - return ENC_DUMMY_P(enc) != 0; -} /* econv stuff */ @@ -295,50 +348,70 @@ VALUE rb_econv_str_convert(rb_econv_t *ec, VALUE src, int flags); VALUE rb_econv_substr_convert(rb_econv_t *ec, VALUE src, long byteoff, long bytesize, int flags); VALUE rb_econv_str_append(rb_econv_t *ec, VALUE src, VALUE dst, int flags); VALUE rb_econv_substr_append(rb_econv_t *ec, VALUE src, long byteoff, long bytesize, VALUE dst, int flags); +VALUE rb_econv_append(rb_econv_t *ec, const char *bytesrc, long bytesize, VALUE dst, int flags); void rb_econv_binmode(rb_econv_t *ec); +enum ruby_econv_flag_type { /* flags for rb_econv_open */ + RUBY_ECONV_ERROR_HANDLER_MASK = 0x000000ff, -#define ECONV_ERROR_HANDLER_MASK 0x000000ff + RUBY_ECONV_INVALID_MASK = 0x0000000f, + RUBY_ECONV_INVALID_REPLACE = 0x00000002, -#define ECONV_INVALID_MASK 0x0000000f -#define ECONV_INVALID_REPLACE 0x00000002 + RUBY_ECONV_UNDEF_MASK = 0x000000f0, + RUBY_ECONV_UNDEF_REPLACE = 0x00000020, + RUBY_ECONV_UNDEF_HEX_CHARREF = 0x00000030, -#define ECONV_UNDEF_MASK 0x000000f0 -#define ECONV_UNDEF_REPLACE 0x00000020 -#define ECONV_UNDEF_HEX_CHARREF 0x00000030 + RUBY_ECONV_DECORATOR_MASK = 0x0000ff00, + RUBY_ECONV_NEWLINE_DECORATOR_MASK = 0x00003f00, + RUBY_ECONV_NEWLINE_DECORATOR_READ_MASK = 0x00000f00, + RUBY_ECONV_NEWLINE_DECORATOR_WRITE_MASK = 0x00003000, -#define ECONV_DECORATOR_MASK 0x0000ff00 -#define ECONV_NEWLINE_DECORATOR_MASK 0x00003f00 -#define ECONV_NEWLINE_DECORATOR_READ_MASK 0x00000f00 -#define ECONV_NEWLINE_DECORATOR_WRITE_MASK 0x00003000 + RUBY_ECONV_UNIVERSAL_NEWLINE_DECORATOR = 0x00000100, + RUBY_ECONV_CRLF_NEWLINE_DECORATOR = 0x00001000, + RUBY_ECONV_CR_NEWLINE_DECORATOR = 0x00002000, + RUBY_ECONV_XML_TEXT_DECORATOR = 0x00004000, + RUBY_ECONV_XML_ATTR_CONTENT_DECORATOR = 0x00008000, -#define ECONV_UNIVERSAL_NEWLINE_DECORATOR 0x00000100 -#define ECONV_CRLF_NEWLINE_DECORATOR 0x00001000 -#define ECONV_CR_NEWLINE_DECORATOR 0x00002000 -#define ECONV_XML_TEXT_DECORATOR 0x00004000 -#define ECONV_XML_ATTR_CONTENT_DECORATOR 0x00008000 - -#define ECONV_STATEFUL_DECORATOR_MASK 0x00f00000 -#define ECONV_XML_ATTR_QUOTE_DECORATOR 0x00100000 + RUBY_ECONV_STATEFUL_DECORATOR_MASK = 0x00f00000, + RUBY_ECONV_XML_ATTR_QUOTE_DECORATOR = 0x00100000, + RUBY_ECONV_DEFAULT_NEWLINE_DECORATOR = #if defined(RUBY_TEST_CRLF_ENVIRONMENT) || defined(_WIN32) -#define ECONV_DEFAULT_NEWLINE_DECORATOR ECONV_CRLF_NEWLINE_DECORATOR + RUBY_ECONV_CRLF_NEWLINE_DECORATOR, #else -#define ECONV_DEFAULT_NEWLINE_DECORATOR 0 + 0, #endif - +#define ECONV_ERROR_HANDLER_MASK RUBY_ECONV_ERROR_HANDLER_MASK +#define ECONV_INVALID_MASK RUBY_ECONV_INVALID_MASK +#define ECONV_INVALID_REPLACE RUBY_ECONV_INVALID_REPLACE +#define ECONV_UNDEF_MASK RUBY_ECONV_UNDEF_MASK +#define ECONV_UNDEF_REPLACE RUBY_ECONV_UNDEF_REPLACE +#define ECONV_UNDEF_HEX_CHARREF RUBY_ECONV_UNDEF_HEX_CHARREF +#define ECONV_DECORATOR_MASK RUBY_ECONV_DECORATOR_MASK +#define ECONV_NEWLINE_DECORATOR_MASK RUBY_ECONV_NEWLINE_DECORATOR_MASK +#define ECONV_NEWLINE_DECORATOR_READ_MASK RUBY_ECONV_NEWLINE_DECORATOR_READ_MASK +#define ECONV_NEWLINE_DECORATOR_WRITE_MASK RUBY_ECONV_NEWLINE_DECORATOR_WRITE_MASK +#define ECONV_UNIVERSAL_NEWLINE_DECORATOR RUBY_ECONV_UNIVERSAL_NEWLINE_DECORATOR +#define ECONV_CRLF_NEWLINE_DECORATOR RUBY_ECONV_CRLF_NEWLINE_DECORATOR +#define ECONV_CR_NEWLINE_DECORATOR RUBY_ECONV_CR_NEWLINE_DECORATOR +#define ECONV_XML_TEXT_DECORATOR RUBY_ECONV_XML_TEXT_DECORATOR +#define ECONV_XML_ATTR_CONTENT_DECORATOR RUBY_ECONV_XML_ATTR_CONTENT_DECORATOR +#define ECONV_STATEFUL_DECORATOR_MASK RUBY_ECONV_STATEFUL_DECORATOR_MASK +#define ECONV_XML_ATTR_QUOTE_DECORATOR RUBY_ECONV_XML_ATTR_QUOTE_DECORATOR +#define ECONV_DEFAULT_NEWLINE_DECORATOR RUBY_ECONV_DEFAULT_NEWLINE_DECORATOR /* end of flags for rb_econv_open */ /* flags for rb_econv_convert */ -#define ECONV_PARTIAL_INPUT 0x00010000 -#define ECONV_AFTER_OUTPUT 0x00020000 + RUBY_ECONV_PARTIAL_INPUT = 0x00010000, + RUBY_ECONV_AFTER_OUTPUT = 0x00020000, +#define ECONV_PARTIAL_INPUT RUBY_ECONV_PARTIAL_INPUT +#define ECONV_AFTER_OUTPUT RUBY_ECONV_AFTER_OUTPUT /* end of flags for rb_econv_convert */ +RUBY_ECONV_FLAGS_PLACEHOLDER}; -#if defined __GNUC__ && __GNUC__ >= 4 -#pragma GCC visibility pop -#endif +RUBY_SYMBOL_EXPORT_END #if defined(__cplusplus) #if 0 |