diff options
Diffstat (limited to 're.c')
| -rw-r--r-- | re.c | 2633 |
1 files changed, 1593 insertions, 1040 deletions
@@ -16,6 +16,9 @@ #include "encindex.h" #include "hrtime.h" #include "internal.h" +#include "internal/bignum.h" +#include "internal/encoding.h" +#include "internal/error.h" #include "internal/hash.h" #include "internal/imemo.h" #include "internal/re.h" @@ -27,15 +30,13 @@ #include "ruby/encoding.h" #include "ruby/re.h" #include "ruby/util.h" +#include "ractor_core.h" VALUE rb_eRegexpError, rb_eRegexpTimeoutError; typedef char onig_errmsg_buffer[ONIG_MAX_ERROR_MESSAGE_LEN]; #define errcpy(err, msg) strlcpy((err), (msg), ONIG_MAX_ERROR_MESSAGE_LEN) -#define BEG(no) (regs->beg[(no)]) -#define END(no) (regs->end[(no)]) - #if 'a' == 97 /* it's ascii */ static const char casetable[] = { '\000', '\001', '\002', '\003', '\004', '\005', '\006', '\007', @@ -87,6 +88,9 @@ static const char casetable[] = { # error >>> "You lose. You will need a translation table for your character set." <<< #endif +// The process-global timeout for regexp matching +rb_hrtime_t rb_reg_match_time_limit = 0; + int rb_memcicmp(const void *x, const void *y, long len) { @@ -94,22 +98,22 @@ rb_memcicmp(const void *x, const void *y, long len) int tmp; while (len--) { - if ((tmp = casetable[(unsigned)*p1++] - casetable[(unsigned)*p2++])) - return tmp; + if ((tmp = casetable[(unsigned)*p1++] - casetable[(unsigned)*p2++])) + return tmp; } return 0; } -#ifdef HAVE_MEMMEM +#if defined(HAVE_MEMMEM) && !defined(__APPLE__) static inline long rb_memsearch_ss(const unsigned char *xs, long m, const unsigned char *ys, long n) { const unsigned char *y; if ((y = memmem(ys, n, xs, m)) != NULL) - return y - ys; + return y - ys; else - return -1; + return -1; } #else static inline long @@ -121,26 +125,26 @@ rb_memsearch_ss(const unsigned char *xs, long m, const unsigned char *ys, long n VALUE hx, hy, mask = VALUE_MAX >> ((SIZEOF_VALUE - m) * CHAR_BIT); if (m > SIZEOF_VALUE) - rb_bug("!!too long pattern string!!"); + rb_bug("!!too long pattern string!!"); if (!(y = memchr(y, *x, n - m + 1))) - return -1; + return -1; /* Prepare hash value */ for (hx = *x++, hy = *y++; x < xe; ++x, ++y) { - hx <<= CHAR_BIT; - hy <<= CHAR_BIT; - hx |= *x; - hy |= *y; + hx <<= CHAR_BIT; + hy <<= CHAR_BIT; + hx |= *x; + hy |= *y; } /* Searching */ while (hx != hy) { - if (y == ye) - return -1; - hy <<= CHAR_BIT; - hy |= *y; - hy &= mask; - y++; + if (y == ye) + return -1; + hy <<= CHAR_BIT; + hy |= *y; + hy &= mask; + y++; } return y - ys - m; } @@ -155,13 +159,13 @@ rb_memsearch_qs(const unsigned char *xs, long m, const unsigned char *ys, long n /* Preprocessing */ for (i = 0; i < 256; ++i) - qstable[i] = m + 1; + qstable[i] = m + 1; for (; x < xe; ++x) - qstable[*x] = xe - x; + qstable[*x] = xe - x; /* Searching */ for (; y + m <= ys + n; y += *(qstable + y[m])) { - if (*xs == *y && memcmp(xs, y, m) == 0) - return y - ys; + if (*xs == *y && memcmp(xs, y, m) == 0) + return y - ys; } return -1; } @@ -172,28 +176,28 @@ rb_memsearch_qs_utf8_hash(const unsigned char *x) register const unsigned int mix = 8353; register unsigned int h = *x; if (h < 0xC0) { - return h + 256; + return h + 256; } else if (h < 0xE0) { - h *= mix; - h += x[1]; + h *= mix; + h += x[1]; } else if (h < 0xF0) { - h *= mix; - h += x[1]; - h *= mix; - h += x[2]; + h *= mix; + h += x[1]; + h *= mix; + h += x[2]; } else if (h < 0xF5) { - h *= mix; - h += x[1]; - h *= mix; - h += x[2]; - h *= mix; - h += x[3]; + h *= mix; + h += x[1]; + h *= mix; + h += x[2]; + h *= mix; + h += x[3]; } else { - return h + 256; + return h + 256; } return (unsigned char)h; } @@ -207,43 +211,41 @@ rb_memsearch_qs_utf8(const unsigned char *xs, long m, const unsigned char *ys, l /* Preprocessing */ for (i = 0; i < 512; ++i) { - qstable[i] = m + 1; + qstable[i] = m + 1; } for (; x < xe; ++x) { - qstable[rb_memsearch_qs_utf8_hash(x)] = xe - x; + qstable[rb_memsearch_qs_utf8_hash(x)] = xe - x; } /* Searching */ for (; y + m <= ys + n; y += qstable[rb_memsearch_qs_utf8_hash(y+m)]) { - if (*xs == *y && memcmp(xs, y, m) == 0) - return y - ys; + if (*xs == *y && memcmp(xs, y, m) == 0) + return y - ys; } return -1; } static inline long -rb_memsearch_wchar(const unsigned char *xs, long m, const unsigned char *ys, long n) +rb_memsearch_with_char_size(const unsigned char *xs, long m, const unsigned char *ys, long n, int char_size) { const unsigned char *x = xs, x0 = *xs, *y = ys; - enum {char_size = 2}; for (n -= m; n >= 0; n -= char_size, y += char_size) { - if (x0 == *y && memcmp(x+1, y+1, m-1) == 0) - return y - ys; + if (x0 == *y && memcmp(x+1, y+1, m-1) == 0) + return y - ys; } return -1; } static inline long -rb_memsearch_qchar(const unsigned char *xs, long m, const unsigned char *ys, long n) +rb_memsearch_wchar(const unsigned char *xs, long m, const unsigned char *ys, long n) { - const unsigned char *x = xs, x0 = *xs, *y = ys; - enum {char_size = 4}; + return rb_memsearch_with_char_size(xs, m, ys, n, 2); +} - for (n -= m; n >= 0; n -= char_size, y += char_size) { - if (x0 == *y && memcmp(x+1, y+1, m-1) == 0) - return y - ys; - } - return -1; +static inline long +rb_memsearch_qchar(const unsigned char *xs, long m, const unsigned char *ys, long n) +{ + return rb_memsearch_with_char_size(xs, m, ys, n, 4); } long @@ -253,46 +255,40 @@ rb_memsearch(const void *x0, long m, const void *y0, long n, rb_encoding *enc) if (m > n) return -1; else if (m == n) { - return memcmp(x0, y0, m) == 0 ? 0 : -1; + return memcmp(x0, y0, m) == 0 ? 0 : -1; } else if (m < 1) { - return 0; + return 0; } else if (m == 1) { - const unsigned char *ys = memchr(y, *x, n); + const unsigned char *ys = memchr(y, *x, n); - if (ys) - return ys - y; - else - return -1; + if (ys) + return ys - y; + else + return -1; } else if (LIKELY(rb_enc_mbminlen(enc) == 1)) { - if (m <= SIZEOF_VALUE) { - return rb_memsearch_ss(x0, m, y0, n); - } - else if (enc == rb_utf8_encoding()){ - return rb_memsearch_qs_utf8(x0, m, y0, n); - } + if (m <= SIZEOF_VALUE) { + return rb_memsearch_ss(x0, m, y0, n); + } + else if (enc == rb_utf8_encoding()){ + return rb_memsearch_qs_utf8(x0, m, y0, n); + } } else if (LIKELY(rb_enc_mbminlen(enc) == 2)) { - return rb_memsearch_wchar(x0, m, y0, n); + return rb_memsearch_wchar(x0, m, y0, n); } else if (LIKELY(rb_enc_mbminlen(enc) == 4)) { - return rb_memsearch_qchar(x0, m, y0, n); + return rb_memsearch_qchar(x0, m, y0, n); } return rb_memsearch_qs(x0, m, y0, n); } -#define REG_LITERAL FL_USER5 #define REG_ENCODING_NONE FL_USER6 #define KCODE_FIXED FL_USER4 -#define ARG_REG_OPTION_MASK \ - (ONIG_OPTION_IGNORECASE|ONIG_OPTION_MULTILINE|ONIG_OPTION_EXTEND) -#define ARG_ENCODING_FIXED 16 -#define ARG_ENCODING_NONE 32 - static int char_to_option(int c) { @@ -300,17 +296,17 @@ char_to_option(int c) switch (c) { case 'i': - val = ONIG_OPTION_IGNORECASE; - break; + val = ONIG_OPTION_IGNORECASE; + break; case 'x': - val = ONIG_OPTION_EXTEND; - break; + val = ONIG_OPTION_EXTEND; + break; case 'm': - val = ONIG_OPTION_MULTILINE; - break; + val = ONIG_OPTION_MULTILINE; + break; default: - val = 0; - break; + val = 0; + break; } return val; } @@ -338,17 +334,17 @@ rb_char_to_option_kcode(int c, int *option, int *kcode) *kcode = rb_ascii8bit_encindex(); return (*option = ARG_ENCODING_NONE); case 'e': - *kcode = ENCINDEX_EUC_JP; - break; + *kcode = ENCINDEX_EUC_JP; + break; case 's': - *kcode = ENCINDEX_Windows_31J; - break; + *kcode = ENCINDEX_Windows_31J; + break; case 'u': - *kcode = rb_utf8_encindex(); - break; + *kcode = rb_utf8_encindex(); + break; default: - *kcode = -1; - return (*option = char_to_option(c)); + *kcode = -1; + return (*option = char_to_option(c)); } *option = ARG_ENCODING_FIXED; return 1; @@ -358,13 +354,13 @@ static void rb_reg_check(VALUE re) { if (!RREGEXP_PTR(re) || !RREGEXP_SRC(re) || !RREGEXP_SRC_PTR(re)) { - rb_raise(rb_eTypeError, "uninitialized Regexp"); + rb_raise(rb_eTypeError, "uninitialized Regexp"); } } static void rb_reg_expr_str(VALUE str, const char *s, long len, - rb_encoding *enc, rb_encoding *resenc, int term) + rb_encoding *enc, rb_encoding *resenc, int term) { const char *p, *pend; int cr = ENC_CODERANGE_UNKNOWN; @@ -374,85 +370,85 @@ rb_reg_expr_str(VALUE str, const char *s, long len, p = s; pend = p + len; rb_str_coderange_scan_restartable(p, pend, enc, &cr); if (rb_enc_asciicompat(enc) && ENC_CODERANGE_CLEAN_P(cr)) { - while (p < pend) { - c = rb_enc_ascget(p, pend, &clen, enc); - if (c == -1) { - if (enc == resenc) { - p += mbclen(p, pend, enc); - } - else { - need_escape = 1; - break; - } - } - else if (c != term && rb_enc_isprint(c, enc)) { - p += clen; - } - else { - need_escape = 1; - break; - } - } + while (p < pend) { + c = rb_enc_ascget(p, pend, &clen, enc); + if (c == -1) { + if (enc == resenc) { + p += mbclen(p, pend, enc); + } + else { + need_escape = 1; + break; + } + } + else if (c != term && rb_enc_isprint(c, enc)) { + p += clen; + } + else { + need_escape = 1; + break; + } + } } else { - need_escape = 1; + need_escape = 1; } if (!need_escape) { - rb_str_buf_cat(str, s, len); + rb_str_buf_cat(str, s, len); } else { - int unicode_p = rb_enc_unicode_p(enc); - p = s; - while (p<pend) { + int unicode_p = rb_enc_unicode_p(enc); + p = s; + while (p<pend) { c = rb_enc_ascget(p, pend, &clen, enc); - if (c == '\\' && p+clen < pend) { - int n = clen + mbclen(p+clen, pend, enc); - rb_str_buf_cat(str, p, n); - p += n; - continue; - } - else if (c == -1) { - clen = rb_enc_precise_mbclen(p, pend, enc); - if (!MBCLEN_CHARFOUND_P(clen)) { - c = (unsigned char)*p; - clen = 1; - goto hex; - } - if (resenc) { - unsigned int c = rb_enc_mbc_to_codepoint(p, pend, enc); - rb_str_buf_cat_escaped_char(str, c, unicode_p); - } - else { - clen = MBCLEN_CHARFOUND_LEN(clen); - rb_str_buf_cat(str, p, clen); - } - } - else if (c == term) { - char c = '\\'; - rb_str_buf_cat(str, &c, 1); - rb_str_buf_cat(str, p, clen); - } - else if (rb_enc_isprint(c, enc)) { - rb_str_buf_cat(str, p, clen); - } - else if (!rb_enc_isspace(c, enc)) { - char b[8]; - - hex: - snprintf(b, sizeof(b), "\\x%02X", c); - rb_str_buf_cat(str, b, 4); - } - else { - rb_str_buf_cat(str, p, clen); - } - p += clen; - } + if (c == '\\' && p+clen < pend) { + int n = clen + mbclen(p+clen, pend, enc); + rb_str_buf_cat(str, p, n); + p += n; + continue; + } + else if (c == -1) { + clen = rb_enc_precise_mbclen(p, pend, enc); + if (!MBCLEN_CHARFOUND_P(clen)) { + c = (unsigned char)*p; + clen = 1; + goto hex; + } + if (resenc) { + unsigned int c = rb_enc_mbc_to_codepoint(p, pend, enc); + rb_str_buf_cat_escaped_char(str, c, unicode_p); + } + else { + clen = MBCLEN_CHARFOUND_LEN(clen); + rb_str_buf_cat(str, p, clen); + } + } + else if (c == term) { + char c = '\\'; + rb_str_buf_cat(str, &c, 1); + rb_str_buf_cat(str, p, clen); + } + else if (rb_enc_isprint(c, enc)) { + rb_str_buf_cat(str, p, clen); + } + else if (!rb_enc_isspace(c, enc)) { + char b[8]; + + hex: + snprintf(b, sizeof(b), "\\x%02X", c); + rb_str_buf_cat(str, b, 4); + } + else { + rb_str_buf_cat(str, p, clen); + } + p += clen; + } } } static VALUE -rb_reg_desc(const char *s, long len, VALUE re) +rb_reg_desc(VALUE re) { rb_encoding *enc = rb_enc_get(re); VALUE str = rb_str_buf_new2("/"); @@ -460,20 +456,24 @@ rb_reg_desc(const char *s, long len, VALUE re) if (resenc == NULL) resenc = rb_default_external_encoding(); if (re && rb_enc_asciicompat(enc)) { - rb_enc_copy(str, re); + rb_enc_copy(str, re); } else { - rb_enc_associate(str, rb_usascii_encoding()); + rb_enc_associate(str, rb_usascii_encoding()); } - rb_reg_expr_str(str, s, len, enc, resenc, '/'); + + VALUE src_str = RREGEXP_SRC(re); + rb_reg_expr_str(str, RSTRING_PTR(src_str), RSTRING_LEN(src_str), enc, resenc, '/'); + RB_GC_GUARD(src_str); + rb_str_buf_cat2(str, "/"); if (re) { - char opts[OPTBUF_SIZE]; - rb_reg_check(re); - if (*option_to_str(opts, RREGEXP_PTR(re)->options)) - rb_str_buf_cat2(str, opts); - if (RBASIC(re)->flags & REG_ENCODING_NONE) - rb_str_buf_cat2(str, "n"); + char opts[OPTBUF_SIZE]; + rb_reg_check(re); + if (*option_to_str(opts, RREGEXP_PTR(re)->options)) + rb_str_buf_cat2(str, opts); + if (RBASIC(re)->flags & REG_ENCODING_NONE) + rb_str_buf_cat2(str, "n"); } return str; } @@ -487,10 +487,14 @@ rb_reg_desc(const char *s, long len, VALUE re) * * /ab+c/ix.source # => "ab+c" * - * Note that escape sequences are retained as is: + * Regexp escape sequences are retained: * * /\x20\+/.source # => "\\x20\\+" * + * Lexer escape characters are not retained: + * + * /\//.source # => "/" + * */ static VALUE @@ -520,7 +524,7 @@ rb_reg_inspect(VALUE re) if (!RREGEXP_PTR(re) || !RREGEXP_SRC(re) || !RREGEXP_SRC_PTR(re)) { return rb_any_to_s(re); } - return rb_reg_desc(RREGEXP_SRC_PTR(re), RREGEXP_SRC_LEN(re), re); + return rb_reg_desc(re); } static VALUE rb_reg_str_with_term(VALUE re, int term); @@ -536,7 +540,7 @@ static VALUE rb_reg_str_with_term(VALUE re, int term); * * The returned string may be used as an argument to Regexp.new, * or as interpolated text for a - * {Regexp literal}[rdoc-ref:regexp.rdoc@Regexp+Literal]: + * {Regexp interpolation}[rdoc-ref:Regexp@Interpolation+Mode]: * * r1 = Regexp.new(s0) # => /(?ix-m:ab+c)/ * r2 = /#{s0}/ # => /(?ix-m:ab+c)/ @@ -563,8 +567,6 @@ rb_reg_str_with_term(VALUE re, int term) { int options, opt; const int embeddable = ONIG_OPTION_MULTILINE|ONIG_OPTION_IGNORECASE|ONIG_OPTION_EXTEND; - long len; - const UChar* ptr; VALUE str = rb_str_buf_new2("(?"); char optbuf[OPTBUF_SIZE + 1]; /* for '-' */ rb_encoding *enc = rb_enc_get(re); @@ -573,14 +575,15 @@ rb_reg_str_with_term(VALUE re, int term) rb_enc_copy(str, re); options = RREGEXP_PTR(re)->options; - ptr = (UChar*)RREGEXP_SRC_PTR(re); - len = RREGEXP_SRC_LEN(re); + VALUE src_str = RREGEXP_SRC(re); + const UChar *ptr = (UChar *)RSTRING_PTR(src_str); + long len = RSTRING_LEN(src_str); again: if (len >= 4 && ptr[0] == '(' && ptr[1] == '?') { - int err = 1; - ptr += 2; - if ((len -= 2) > 0) { - do { + int err = 1; + ptr += 2; + if ((len -= 2) > 0) { + do { opt = char_to_option((int )*ptr); if (opt != 0) { options |= opt; @@ -588,13 +591,13 @@ rb_reg_str_with_term(VALUE re, int term) else { break; } - ++ptr; - } while (--len > 0); - } - if (len > 1 && *ptr == '-') { - ++ptr; - --len; - do { + ++ptr; + } while (--len > 0); + } + if (len > 1 && *ptr == '-') { + ++ptr; + --len; + do { opt = char_to_option((int )*ptr); if (opt != 0) { options &= ~opt; @@ -602,77 +605,79 @@ rb_reg_str_with_term(VALUE re, int term) else { break; } - ++ptr; - } while (--len > 0); - } - if (*ptr == ')') { - --len; - ++ptr; - goto again; - } - if (*ptr == ':' && ptr[len-1] == ')') { - Regexp *rp; - VALUE verbose = ruby_verbose; - ruby_verbose = Qfalse; - - ++ptr; - len -= 2; + ++ptr; + } while (--len > 0); + } + if (*ptr == ')') { + --len; + ++ptr; + goto again; + } + if (*ptr == ':' && ptr[len-1] == ')') { + Regexp *rp; + VALUE verbose = ruby_verbose; + ruby_verbose = Qfalse; + + ++ptr; + len -= 2; err = onig_new(&rp, ptr, ptr + len, options, - enc, OnigDefaultSyntax, NULL); - onig_free(rp); - ruby_verbose = verbose; - } - if (err) { - options = RREGEXP_PTR(re)->options; - ptr = (UChar*)RREGEXP_SRC_PTR(re); - len = RREGEXP_SRC_LEN(re); - } + enc, OnigDefaultSyntax, NULL); + onig_free(rp); + ruby_verbose = verbose; + } + if (err) { + options = RREGEXP_PTR(re)->options; + ptr = (UChar*)RREGEXP_SRC_PTR(re); + len = RREGEXP_SRC_LEN(re); + } } if (*option_to_str(optbuf, options)) rb_str_buf_cat2(str, optbuf); if ((options & embeddable) != embeddable) { - optbuf[0] = '-'; - option_to_str(optbuf + 1, ~options); - rb_str_buf_cat2(str, optbuf); + optbuf[0] = '-'; + option_to_str(optbuf + 1, ~options); + rb_str_buf_cat2(str, optbuf); } rb_str_buf_cat2(str, ":"); if (rb_enc_asciicompat(enc)) { - rb_reg_expr_str(str, (char*)ptr, len, enc, NULL, term); - rb_str_buf_cat2(str, ")"); + rb_reg_expr_str(str, (char*)ptr, len, enc, NULL, term); + rb_str_buf_cat2(str, ")"); } else { - const char *s, *e; - char *paren; - ptrdiff_t n; - rb_str_buf_cat2(str, ")"); - rb_enc_associate(str, rb_usascii_encoding()); - str = rb_str_encode(str, rb_enc_from_encoding(enc), 0, Qnil); - - /* backup encoded ")" to paren */ - s = RSTRING_PTR(str); - e = RSTRING_END(str); - s = rb_enc_left_char_head(s, e-1, e, enc); - n = e - s; - paren = ALLOCA_N(char, n); - memcpy(paren, s, n); - rb_str_resize(str, RSTRING_LEN(str) - n); - - rb_reg_expr_str(str, (char*)ptr, len, enc, NULL, term); - rb_str_buf_cat(str, paren, n); + const char *s, *e; + char *paren; + ptrdiff_t n; + rb_str_buf_cat2(str, ")"); + rb_enc_associate(str, rb_usascii_encoding()); + str = rb_str_encode(str, rb_enc_from_encoding(enc), 0, Qnil); + + /* backup encoded ")" to paren */ + s = RSTRING_PTR(str); + e = RSTRING_END(str); + s = rb_enc_left_char_head(s, e-1, e, enc); + n = e - s; + paren = ALLOCA_N(char, n); + memcpy(paren, s, n); + rb_str_resize(str, RSTRING_LEN(str) - n); + + rb_reg_expr_str(str, (char*)ptr, len, enc, NULL, term); + rb_str_buf_cat(str, paren, n); } rb_enc_copy(str, re); + RB_GC_GUARD(src_str); + return str; } -NORETURN(static void rb_reg_raise(const char *s, long len, const char *err, VALUE re)); +NORETURN(static void rb_reg_raise(const char *err, VALUE re)); static void -rb_reg_raise(const char *s, long len, const char *err, VALUE re) +rb_reg_raise(const char *err, VALUE re) { - VALUE desc = rb_reg_desc(s, len, re); + VALUE desc = rb_reg_desc(re); rb_raise(rb_eRegexpError, "%s: %"PRIsVALUE, err, desc); } @@ -706,7 +711,7 @@ static VALUE rb_reg_error_desc(VALUE str, int options, const char *err) { return rb_enc_reg_error_desc(RSTRING_PTR(str), RSTRING_LEN(str), - rb_enc_get(str), options, err); + rb_enc_get(str), options, err); } NORETURN(static void rb_reg_raise_str(VALUE str, int options, const char *err)); @@ -760,8 +765,8 @@ rb_reg_casefold_p(VALUE re) * /foo/mix.options # => 7 * * Note that additional bits may be set in the returned integer; - * these are maintained internally internally in +self+, - * are ignored if passed to Regexp.new, and may be ignored by the caller: + * these are maintained internally in +self+, are ignored if passed + * to Regexp.new, and may be ignored by the caller: * * Returns the set of bits corresponding to the options used when * creating this regexp (see Regexp::new for details). Note that @@ -860,8 +865,8 @@ rb_reg_named_captures(VALUE re) static int onig_new_with_source(regex_t** reg, const UChar* pattern, const UChar* pattern_end, - OnigOptionType option, OnigEncoding enc, const OnigSyntaxType* syntax, - OnigErrorInfo* einfo, const char *sourcefile, int sourceline) + OnigOptionType option, OnigEncoding enc, const OnigSyntaxType* syntax, + OnigErrorInfo* einfo, const char *sourcefile, int sourceline) { int r; @@ -874,15 +879,15 @@ onig_new_with_source(regex_t** reg, const UChar* pattern, const UChar* pattern_e r = onig_compile_ruby(*reg, pattern, pattern_end, einfo, sourcefile, sourceline); if (r) { err: - onig_free(*reg); - *reg = NULL; + onig_free(*reg); + *reg = NULL; } return r; } static Regexp* make_regexp(const char *s, long len, rb_encoding *enc, int flags, onig_errmsg_buffer err, - const char *sourcefile, int sourceline) + const char *sourcefile, int sourceline) { Regexp *rp; int r; @@ -896,10 +901,10 @@ make_regexp(const char *s, long len, rb_encoding *enc, int flags, onig_errmsg_bu */ r = onig_new_with_source(&rp, (UChar*)s, (UChar*)(s + len), flags, - enc, OnigDefaultSyntax, &einfo, sourcefile, sourceline); + enc, OnigDefaultSyntax, &einfo, sourcefile, sourceline); if (r) { - onig_error_code_to_str((UChar*)err, r, &einfo); - return 0; + onig_error_code_to_str((UChar*)err, r, &einfo); + return 0; } return rp; } @@ -951,24 +956,39 @@ make_regexp(const char *s, long len, rb_encoding *enc, int flags, onig_errmsg_bu * * <code>$'</code> is Regexp.last_match<code>.post_match</code>; * * <code>$+</code> is Regexp.last_match<code>[ -1 ]</code> (the last capture). * - * See also "Special global variables" section in Regexp documentation. + * See also Regexp@Global+Variables. */ VALUE rb_cMatch; static VALUE -match_alloc(VALUE klass) +match_alloc_n(VALUE klass, int num_regs) { - NEWOBJ_OF(match, struct RMatch, klass, T_MATCH); + int capa = num_regs * 2; + size_t alloc_size = offsetof(struct RMatch, as) + sizeof(OnigPosition) * capa; + if (alloc_size < sizeof(struct RMatch)) { + alloc_size = sizeof(struct RMatch); + } - match->str = 0; - match->rmatch = 0; - match->regexp = 0; - match->rmatch = ZALLOC(struct rmatch); + VALUE flags = T_MATCH; + if (!rb_gc_size_allocatable_p(alloc_size)) { + alloc_size = sizeof(struct RMatch); + flags |= RMATCH_ONIG; + capa = 0; + } + NEWOBJ_OF(match, struct RMatch, klass, flags, alloc_size); + memset(((char *)match) + sizeof(struct RBasic), 0, alloc_size - sizeof(struct RBasic)); + match->capa = capa; return (VALUE)match; } +static VALUE +match_alloc(VALUE klass) +{ + return match_alloc_n(klass, 0); +} + int rb_reg_region_copy(struct re_registers *to, const struct re_registers *from) { @@ -980,6 +1000,54 @@ rb_reg_region_copy(struct re_registers *to, const struct re_registers *from) return ONIGERR_MEMORY; } +static void +match_to_onig(VALUE match, int num_regs, const OnigPosition *src_beg, const OnigPosition *src_end) +{ + struct RMatch *rm = RMATCH(match); + struct re_registers tmp = {0}; + if (onig_region_resize(&tmp, num_regs)) { + rb_memerror(); + } + memcpy(tmp.beg, src_beg, num_regs * sizeof(OnigPosition)); + memcpy(tmp.end, src_end, num_regs * sizeof(OnigPosition)); + rm->as.onig = tmp; + FL_SET_RAW(match, RMATCH_ONIG); +} + +void +rb_match_ensure_onig(VALUE match) +{ + if (FL_TEST_RAW(match, RMATCH_ONIG)) return; + struct RMatch *rm = RMATCH(match); + int n = rm->num_regs; + match_to_onig(match, n, &rm->as.embed[0], &rm->as.embed[n]); +} + +/* Replace `match`'s registers with a copy of (num_regs, beg, end). If the + * data does not fit in the embed form, the match is evicted to onig form. + * Raises on OOM. */ +static void +match_set_regs(VALUE match, int num_regs, const OnigPosition *beg, const OnigPosition *end) +{ + struct RMatch *rm = RMATCH(match); + + if (FL_TEST_RAW(match, RMATCH_ONIG)) { + if (onig_region_resize(&rm->as.onig, num_regs)) { + rb_memerror(); + } + memcpy(rm->as.onig.beg, beg, num_regs * sizeof(OnigPosition)); + memcpy(rm->as.onig.end, end, num_regs * sizeof(OnigPosition)); + } + else if (num_regs * 2 <= rm->capa) { + memcpy(&rm->as.embed[0], beg, num_regs * sizeof(OnigPosition)); + memcpy(&rm->as.embed[num_regs], end, num_regs * sizeof(OnigPosition)); + } + else { + match_to_onig(match, num_regs, beg, end); + } + rm->num_regs = num_regs; +} + typedef struct { long byte_pos; long char_pos; @@ -999,41 +1067,41 @@ pair_byte_cmp(const void *pair1, const void *pair2) static void update_char_offset(VALUE match) { - struct rmatch *rm = RMATCH(match)->rmatch; - struct re_registers *regs; + struct RMatch *rm = RMATCH(match); int i, num_regs, num_pos; long c; char *s, *p, *q; rb_encoding *enc; pair_t *pairs; + VALUE pairs_obj = Qnil; if (rm->char_offset_num_allocated) return; - regs = &rm->regs; - num_regs = rm->regs.num_regs; + num_regs = RMATCH_NREGS(match); if (rm->char_offset_num_allocated < num_regs) { - REALLOC_N(rm->char_offset, struct rmatch_offset, num_regs); + SIZED_REALLOC_N(rm->char_offset, struct rmatch_offset, num_regs, rm->char_offset_num_allocated); rm->char_offset_num_allocated = num_regs; + FL_SET_RAW(match, RMATCH_OFFSETS_EXTERNAL); } enc = rb_enc_get(RMATCH(match)->str); if (rb_enc_mbmaxlen(enc) == 1) { for (i = 0; i < num_regs; i++) { - rm->char_offset[i].beg = BEG(i); - rm->char_offset[i].end = END(i); + rm->char_offset[i].beg = RMATCH_BEG(match, i); + rm->char_offset[i].end = RMATCH_END(match, i); } return; } - pairs = ALLOCA_N(pair_t, num_regs*2); + pairs = RB_ALLOCV_N(pair_t, pairs_obj, num_regs * 2); num_pos = 0; for (i = 0; i < num_regs; i++) { - if (BEG(i) < 0) + if (RMATCH_BEG(match, i) < 0) continue; - pairs[num_pos++].byte_pos = BEG(i); - pairs[num_pos++].byte_pos = END(i); + pairs[num_pos++].byte_pos = RMATCH_BEG(match, i); + pairs[num_pos++].byte_pos = RMATCH_END(match, i); } qsort(pairs, num_pos, sizeof(pair_t), pair_byte_cmp); @@ -1048,53 +1116,55 @@ update_char_offset(VALUE match) for (i = 0; i < num_regs; i++) { pair_t key, *found; - if (BEG(i) < 0) { + if (RMATCH_BEG(match, i) < 0) { rm->char_offset[i].beg = -1; rm->char_offset[i].end = -1; continue; } - key.byte_pos = BEG(i); + key.byte_pos = RMATCH_BEG(match, i); found = bsearch(&key, pairs, num_pos, sizeof(pair_t), pair_byte_cmp); rm->char_offset[i].beg = found->char_pos; - key.byte_pos = END(i); + key.byte_pos = RMATCH_END(match, i); found = bsearch(&key, pairs, num_pos, sizeof(pair_t), pair_byte_cmp); rm->char_offset[i].end = found->char_pos; } + + RB_ALLOCV_END(pairs_obj); } -static void +static VALUE match_check(VALUE match) { if (!RMATCH(match)->regexp) { - rb_raise(rb_eTypeError, "uninitialized MatchData"); + rb_raise(rb_eTypeError, "uninitialized MatchData"); } + return match; } /* :nodoc: */ static VALUE match_init_copy(VALUE obj, VALUE orig) { - struct rmatch *rm; + struct RMatch *rm = RMATCH(obj); if (!OBJ_INIT_COPY(obj, orig)) return obj; - RMATCH(obj)->str = RMATCH(orig)->str; - RMATCH(obj)->regexp = RMATCH(orig)->regexp; + RB_OBJ_WRITE(obj, &rm->str, RMATCH(orig)->str); + RB_OBJ_WRITE(obj, &rm->regexp, RMATCH(orig)->regexp); - rm = RMATCH(obj)->rmatch; - if (rb_reg_region_copy(&rm->regs, RMATCH_REGS(orig))) - rb_memerror(); + match_set_regs(obj, RMATCH_NREGS(orig), RMATCH_BEG_PTR(orig), RMATCH_END_PTR(orig)); - if (RMATCH(orig)->rmatch->char_offset_num_allocated) { - if (rm->char_offset_num_allocated < rm->regs.num_regs) { - REALLOC_N(rm->char_offset, struct rmatch_offset, rm->regs.num_regs); - rm->char_offset_num_allocated = rm->regs.num_regs; + if (RMATCH(orig)->char_offset_num_allocated) { + if (rm->char_offset_num_allocated < rm->num_regs) { + SIZED_REALLOC_N(rm->char_offset, struct rmatch_offset, rm->num_regs, rm->char_offset_num_allocated); + rm->char_offset_num_allocated = rm->num_regs; + FL_SET_RAW(obj, RMATCH_OFFSETS_EXTERNAL); } - MEMCPY(rm->char_offset, RMATCH(orig)->rmatch->char_offset, - struct rmatch_offset, rm->regs.num_regs); - RB_GC_GUARD(orig); + MEMCPY(rm->char_offset, RMATCH(orig)->char_offset, + struct rmatch_offset, rm->num_regs); + RB_GC_GUARD(orig); } return obj; @@ -1119,9 +1189,9 @@ match_regexp(VALUE match) match_check(match); regexp = RMATCH(match)->regexp; if (NIL_P(regexp)) { - VALUE str = rb_reg_nth_match(0, match); - regexp = rb_reg_regcomp(rb_reg_quote(str)); - RMATCH(match)->regexp = regexp; + VALUE str = rb_reg_nth_match(0, match); + regexp = rb_reg_regcomp(rb_reg_quote(str)); + RB_OBJ_WRITE(match, &RMATCH(match)->regexp, regexp); } return regexp; } @@ -1152,7 +1222,7 @@ match_names(VALUE match) { match_check(match); if (NIL_P(RMATCH(match)->regexp)) - return rb_ary_new_capa(0); + return rb_ary_new_capa(0); return rb_reg_names(RMATCH(match)->regexp); } @@ -1166,53 +1236,46 @@ match_names(VALUE match) * # => #<MatchData "HX1138" 1:"H" 2:"X" 3:"113" 4:"8"> * m.size # => 5 * - * MatchData#length is an alias for MatchData.size. - * */ static VALUE match_size(VALUE match) { match_check(match); - return INT2FIX(RMATCH_REGS(match)->num_regs); + return INT2FIX(RMATCH_NREGS(match)); } -static int name_to_backref_number(struct re_registers *, VALUE, const char*, const char*); +static int match_name_to_backref_number(VALUE match, VALUE name); NORETURN(static void name_to_backref_error(VALUE name)); static void name_to_backref_error(VALUE name) { rb_raise(rb_eIndexError, "undefined group name reference: % "PRIsVALUE, - name); + name); } static void -backref_number_check(struct re_registers *regs, int i) +backref_number_check(VALUE match, int i) { - if (i < 0 || regs->num_regs <= i) - rb_raise(rb_eIndexError, "index %d out of matches", i); + if (i < 0 || RMATCH_NREGS(match) <= i) + rb_raise(rb_eIndexError, "index %d out of matches", i); } static int match_backref_number(VALUE match, VALUE backref) { - const char *name; int num; - struct re_registers *regs = RMATCH_REGS(match); - VALUE regexp = RMATCH(match)->regexp; - match_check(match); if (SYMBOL_P(backref)) { - backref = rb_sym2str(backref); + backref = rb_sym2str(backref); } else if (!RB_TYPE_P(backref, T_STRING)) { - return NUM2INT(backref); + return NUM2INT(backref); } - name = StringValueCStr(backref); - num = name_to_backref_number(regs, regexp, name, name + RSTRING_LEN(backref)); + num = match_name_to_backref_number(match, backref); if (num < 1) { name_to_backref_error(backref); @@ -1240,17 +1303,16 @@ static VALUE match_offset(VALUE match, VALUE n) { int i = match_backref_number(match, n); - struct re_registers *regs = RMATCH_REGS(match); match_check(match); - backref_number_check(regs, i); + backref_number_check(match, i); - if (BEG(i) < 0) - return rb_assoc_new(Qnil, Qnil); + if (RMATCH_BEG(match, i) < 0) + return rb_assoc_new(Qnil, Qnil); update_char_offset(match); - return rb_assoc_new(LONG2NUM(RMATCH(match)->rmatch->char_offset[i].beg), - LONG2NUM(RMATCH(match)->rmatch->char_offset[i].end)); + return rb_assoc_new(LONG2NUM(RMATCH(match)->char_offset[i].beg), + LONG2NUM(RMATCH(match)->char_offset[i].end)); } /* @@ -1275,14 +1337,59 @@ static VALUE match_byteoffset(VALUE match, VALUE n) { int i = match_backref_number(match, n); - struct re_registers *regs = RMATCH_REGS(match); match_check(match); - backref_number_check(regs, i); + backref_number_check(match, i); - if (BEG(i) < 0) + if (RMATCH_BEG(match, i) < 0) return rb_assoc_new(Qnil, Qnil); - return rb_assoc_new(LONG2NUM(BEG(i)), LONG2NUM(END(i))); + return rb_assoc_new(LONG2NUM(RMATCH_BEG(match, i)), LONG2NUM(RMATCH_END(match, i))); +} + + +/* + * call-seq: + * bytebegin(n) -> integer + * bytebegin(name) -> integer + * + * :include: doc/matchdata/bytebegin.rdoc + * + */ + +static VALUE +match_bytebegin(VALUE match, VALUE n) +{ + int i = match_backref_number(match, n); + + match_check(match); + backref_number_check(match, i); + + if (RMATCH_BEG(match, i) < 0) + return Qnil; + return LONG2NUM(RMATCH_BEG(match, i)); +} + + +/* + * call-seq: + * byteend(n) -> integer + * byteend(name) -> integer + * + * :include: doc/matchdata/byteend.rdoc + * + */ + +static VALUE +match_byteend(VALUE match, VALUE n) +{ + int i = match_backref_number(match, n); + + match_check(match); + backref_number_check(match, i); + + if (RMATCH_BEG(match, i) < 0) + return Qnil; + return LONG2NUM(RMATCH_END(match, i)); } @@ -1299,16 +1406,15 @@ static VALUE match_begin(VALUE match, VALUE n) { int i = match_backref_number(match, n); - struct re_registers *regs = RMATCH_REGS(match); match_check(match); - backref_number_check(regs, i); + backref_number_check(match, i); - if (BEG(i) < 0) - return Qnil; + if (RMATCH_BEG(match, i) < 0) + return Qnil; update_char_offset(match); - return LONG2NUM(RMATCH(match)->rmatch->char_offset[i].beg); + return LONG2NUM(RMATCH(match)->char_offset[i].beg); } @@ -1325,16 +1431,15 @@ static VALUE match_end(VALUE match, VALUE n) { int i = match_backref_number(match, n); - struct re_registers *regs = RMATCH_REGS(match); match_check(match); - backref_number_check(regs, i); + backref_number_check(match, i); - if (BEG(i) < 0) - return Qnil; + if (RMATCH_BEG(match, i) < 0) + return Qnil; update_char_offset(match); - return LONG2NUM(RMATCH(match)->rmatch->char_offset[i].end); + return LONG2NUM(RMATCH(match)->char_offset[i].end); } /* @@ -1367,13 +1472,12 @@ static VALUE match_nth(VALUE match, VALUE n) { int i = match_backref_number(match, n); - struct re_registers *regs = RMATCH_REGS(match); - backref_number_check(regs, i); + backref_number_check(match, i); - long start = BEG(i), end = END(i); + long start = RMATCH_BEG(match, i), end = RMATCH_END(match, i); if (start < 0) - return Qnil; + return Qnil; return rb_str_subseq(RMATCH(match)->str, start, end - start); } @@ -1411,17 +1515,16 @@ static VALUE match_nth_length(VALUE match, VALUE n) { int i = match_backref_number(match, n); - struct re_registers *regs = RMATCH_REGS(match); match_check(match); - backref_number_check(regs, i); + backref_number_check(match, i); - if (BEG(i) < 0) - return Qnil; + if (RMATCH_BEG(match, i) < 0) + return Qnil; update_char_offset(match); const struct rmatch_offset *const ofs = - &RMATCH(match)->rmatch->char_offset[i]; + &RMATCH(match)->char_offset[i]; return LONG2NUM(ofs->end - ofs->beg); } @@ -1442,53 +1545,39 @@ rb_match_unbusy(VALUE match) int rb_match_count(VALUE match) { - struct re_registers *regs; if (NIL_P(match)) return -1; - regs = RMATCH_REGS(match); - if (!regs) return -1; - return regs->num_regs; + return RMATCH_NREGS(match); } -int -rb_match_nth_defined(int nth, VALUE match) +static VALUE +match_alloc_or_reuse(VALUE existing, int num_regs) { - struct re_registers *regs; - if (NIL_P(match)) return FALSE; - regs = RMATCH_REGS(match); - if (!regs) return FALSE; - if (nth >= regs->num_regs) { - return FALSE; + if (!NIL_P(existing) && + !FL_TEST(existing, MATCH_BUSY) && + RMATCH(existing)->capa >= num_regs * 2) { + return existing; } - if (nth < 0) { - nth += regs->num_regs; - if (nth <= 0) return FALSE; - } - return (BEG(nth) != -1); + return match_alloc_n(rb_cMatch, num_regs); } static void match_set_string(VALUE m, VALUE string, long pos, long len) { struct RMatch *match = (struct RMatch *)m; - struct rmatch *rmatch = match->rmatch; - match->str = string; - match->regexp = Qnil; - int err = onig_region_resize(&rmatch->regs, 1); - if (err) rb_memerror(); - rmatch->regs.beg[0] = pos; - rmatch->regs.end[0] = pos + len; + RB_OBJ_WRITE(match, &match->str, string); + RB_OBJ_WRITE(match, &match->regexp, Qnil); + OnigPosition beg = pos, end = pos + len; + match_set_regs(m, 1, &beg, &end); } -void +VALUE rb_backref_set_string(VALUE string, long pos, long len) { - VALUE match = rb_backref_get(); - if (NIL_P(match) || FL_TEST(match, MATCH_BUSY)) { - match = match_alloc(rb_cMatch); - } + VALUE match = match_alloc_or_reuse(rb_backref_get(), 1); match_set_string(match, string, pos, len); rb_backref_set(match); + return match; } /* @@ -1536,26 +1625,16 @@ static void reg_enc_error(VALUE re, VALUE str) { rb_raise(rb_eEncCompatError, - "incompatible encoding regexp match (%s regexp with %s string)", - rb_enc_name(rb_enc_get(re)), - rb_enc_name(rb_enc_get(str))); -} - -static inline int -str_coderange(VALUE str) -{ - int cr = ENC_CODERANGE(str); - if (cr == ENC_CODERANGE_UNKNOWN) { - cr = rb_enc_str_coderange(str); - } - return cr; + "incompatible encoding regexp match (%s regexp with %s string)", + rb_enc_inspect_name(rb_enc_get(re)), + rb_enc_inspect_name(rb_enc_get(str))); } static rb_encoding* rb_reg_prepare_enc(VALUE re, VALUE str, int warn) { rb_encoding *enc = 0; - int cr = str_coderange(str); + int cr = rb_enc_str_coderange(str); if (cr == ENC_CODERANGE_BROKEN) { rb_raise(rb_eArgError, @@ -1568,51 +1647,52 @@ rb_reg_prepare_enc(VALUE re, VALUE str, int warn) if (RREGEXP_PTR(re)->enc == enc) { } else if (cr == ENC_CODERANGE_7BIT && - RREGEXP_PTR(re)->enc == rb_usascii_encoding()) { - enc = RREGEXP_PTR(re)->enc; + RREGEXP_PTR(re)->enc == rb_usascii_encoding()) { + enc = RREGEXP_PTR(re)->enc; } else if (!rb_enc_asciicompat(enc)) { - reg_enc_error(re, str); + reg_enc_error(re, str); } else if (rb_reg_fixed_encoding_p(re)) { if ((!rb_enc_asciicompat(RREGEXP_PTR(re)->enc) || - cr != ENC_CODERANGE_7BIT)) { - reg_enc_error(re, str); - } - enc = RREGEXP_PTR(re)->enc; + cr != ENC_CODERANGE_7BIT)) { + reg_enc_error(re, str); + } + enc = RREGEXP_PTR(re)->enc; } else if (warn && (RBASIC(re)->flags & REG_ENCODING_NONE) && - enc != rb_ascii8bit_encoding() && - cr != ENC_CODERANGE_7BIT) { - rb_warn("historical binary regexp match /.../n against %s string", - rb_enc_name(enc)); + enc != rb_ascii8bit_encoding() && + cr != ENC_CODERANGE_7BIT) { + rb_warn("historical binary regexp match /.../n against %s string", + rb_enc_name(enc)); } return enc; } regex_t * -rb_reg_prepare_re0(VALUE re, VALUE str, onig_errmsg_buffer err) +rb_reg_prepare_re(VALUE re, VALUE str) { - regex_t *reg = RREGEXP_PTR(re); int r; OnigErrorInfo einfo; - const char *pattern; VALUE unescaped; rb_encoding *fixed_enc = 0; rb_encoding *enc = rb_reg_prepare_enc(re, str, 1); + regex_t *reg = RREGEXP_PTR(re); if (reg->enc == enc) return reg; rb_reg_check(re); - reg = RREGEXP_PTR(re); - pattern = RREGEXP_SRC_PTR(re); + VALUE src_str = RREGEXP_SRC(re); + const char *pattern = RSTRING_PTR(src_str); + + onig_errmsg_buffer err = ""; unescaped = rb_reg_preprocess( - pattern, pattern + RREGEXP_SRC_LEN(re), enc, + pattern, pattern + RSTRING_LEN(src_str), enc, &fixed_enc, err, 0); if (NIL_P(unescaped)) { - rb_raise(rb_eArgError, "regexp preprocess failed: %s", err); + rb_raise(rb_eArgError, "regexp preprocess failed: %s", err); } // inherit the timeout settings @@ -1621,25 +1701,74 @@ rb_reg_prepare_re0(VALUE re, VALUE str, onig_errmsg_buffer err) const char *ptr; long len; RSTRING_GETMEM(unescaped, ptr, len); - r = onig_new(®, (UChar *)ptr, (UChar *)(ptr + len), - reg->options, enc, - OnigDefaultSyntax, &einfo); + + /* If there are no other users of this regex, then we can directly overwrite it. */ + if (ruby_single_main_ractor && RREGEXP(re)->usecnt == 0) { + regex_t tmp_reg; + r = onig_new_without_alloc(&tmp_reg, (UChar *)ptr, (UChar *)(ptr + len), + reg->options, enc, + OnigDefaultSyntax, &einfo); + + if (r) { + /* There was an error so perform cleanups. */ + onig_free_body(&tmp_reg); + } + else { + onig_free_body(reg); + /* There are no errors so set reg to tmp_reg. */ + *reg = tmp_reg; + } + } + else { + r = onig_new(®, (UChar *)ptr, (UChar *)(ptr + len), + reg->options, enc, + OnigDefaultSyntax, &einfo); + } + if (r) { - onig_error_code_to_str((UChar*)err, r, &einfo); - rb_reg_raise(pattern, RREGEXP_SRC_LEN(re), err, re); + onig_error_code_to_str((UChar*)err, r, &einfo); + rb_reg_raise(err, re); } reg->timelimit = timelimit; RB_GC_GUARD(unescaped); + RB_GC_GUARD(src_str); return reg; } -regex_t * -rb_reg_prepare_re(VALUE re, VALUE str) +OnigPosition +rb_reg_onig_match(VALUE re, VALUE str, + OnigPosition (*match)(regex_t *reg, VALUE str, struct re_registers *regs, void *args), + void *args, struct re_registers *regs) { - onig_errmsg_buffer err = ""; - return rb_reg_prepare_re0(re, str, err); + regex_t *reg = rb_reg_prepare_re(re, str); + + bool tmpreg = reg != RREGEXP_PTR(re); + if (!tmpreg) RREGEXP(re)->usecnt++; + + OnigPosition result = match(reg, str, regs, args); + + if (!tmpreg) RREGEXP(re)->usecnt--; + if (tmpreg) { + onig_free(reg); + } + + if (result < 0) { + switch (result) { + case ONIG_MISMATCH: + break; + case ONIGERR_TIMEOUT: + rb_raise(rb_eRegexpTimeoutError, "regexp match timeout"); + default: { + onig_errmsg_buffer err = ""; + onig_error_code_to_str((UChar*)err, (int)result); + rb_reg_raise(err, re); + } + } + } + + return result; } long @@ -1652,94 +1781,107 @@ rb_reg_adjust_startpos(VALUE re, VALUE str, long pos, int reverse) enc = rb_reg_prepare_enc(re, str, 0); if (reverse) { - range = -pos; + range = -pos; } else { - range = RSTRING_LEN(str) - pos; + range = RSTRING_LEN(str) - pos; } if (pos > 0 && ONIGENC_MBC_MAXLEN(enc) != 1 && pos < RSTRING_LEN(str)) { - string = (UChar*)RSTRING_PTR(str); + string = (UChar*)RSTRING_PTR(str); - if (range > 0) { - p = onigenc_get_right_adjust_char_head(enc, string, string + pos, string + RSTRING_LEN(str)); - } - else { - p = ONIGENC_LEFT_ADJUST_CHAR_HEAD(enc, string, string + pos, string + RSTRING_LEN(str)); - } - return p - string; + if (range > 0) { + p = onigenc_get_right_adjust_char_head(enc, string, string + pos, string + RSTRING_LEN(str)); + } + else { + p = ONIGENC_LEFT_ADJUST_CHAR_HEAD(enc, string, string + pos, string + RSTRING_LEN(str)); + } + return p - string; } return pos; } +struct reg_onig_search_args { + long pos; + long range; +}; + +static OnigPosition +reg_onig_search(regex_t *reg, VALUE str, struct re_registers *regs, void *args_ptr) +{ + struct reg_onig_search_args *args = (struct reg_onig_search_args *)args_ptr; + const char *ptr; + long len; + RSTRING_GETMEM(str, ptr, len); + + return onig_search( + reg, + (UChar *)ptr, + (UChar *)(ptr + len), + (UChar *)(ptr + args->pos), + (UChar *)(ptr + args->range), + regs, + ONIG_OPTION_NONE); +} + /* returns byte offset */ static long rb_reg_search_set_match(VALUE re, VALUE str, long pos, int reverse, int set_backref_str, VALUE *set_match) { - long result; - VALUE match; - struct re_registers regi, *regs = ®i; - char *start, *range; - long len; - regex_t *reg; - int tmpreg; - onig_errmsg_buffer err = ""; - - RSTRING_GETMEM(str, start, len); - range = start; + long len = RSTRING_LEN(str); if (pos > len || pos < 0) { - rb_backref_set(Qnil); - return -1; + rb_backref_set(Qnil); + return -1; } - reg = rb_reg_prepare_re0(re, str, err); - tmpreg = reg != RREGEXP_PTR(re); - if (!tmpreg) RREGEXP(re)->usecnt++; + struct reg_onig_search_args args = { + .pos = pos, + .range = reverse ? 0 : len, + }; - MEMZERO(regs, struct re_registers, 1); - if (!reverse) { - range += len; - } - result = onig_search(reg, - (UChar*)start, - ((UChar*)(start + len)), - ((UChar*)(start + pos)), - ((UChar*)range), - regs, ONIG_OPTION_NONE); - if (!tmpreg) RREGEXP(re)->usecnt--; - if (tmpreg) { - if (RREGEXP(re)->usecnt) { - onig_free(reg); - } - else { - onig_free(RREGEXP_PTR(re)); - RREGEXP_PTR(re) = reg; - } + rb_reg_check(re); + + /* Stack-backed regs sized to max(num_mem+1, ONIG_NREGION) so + * onig_region_resize_clear takes its no-op branch. */ + int n = RREGEXP_PTR(re)->num_mem + 1; + int cap = n < ONIG_NREGION ? ONIG_NREGION : n; + VALUE regs_buf; + OnigPosition *buf = ALLOCV_N(OnigPosition, regs_buf, (size_t)cap * 2); + struct re_registers regs = { + .allocated = cap, + .num_regs = 0, + .beg = buf, + .end = buf + cap, + }; + + OnigPosition result = rb_reg_onig_match(re, str, reg_onig_search, &args, ®s); + + if (result == ONIG_MISMATCH) { + ALLOCV_END(regs_buf); + rb_backref_set(Qnil); + return ONIG_MISMATCH; } - if (result < 0) { - if (regs == ®i) - onig_region_free(regs, 0); - if (result == ONIG_MISMATCH) { - rb_backref_set(Qnil); - return result; - } - else { - onig_error_code_to_str((UChar*)err, (int)result); - rb_reg_raise(RREGEXP_SRC_PTR(re), RREGEXP_SRC_LEN(re), err, re); - } - } - - match = match_alloc(rb_cMatch); - int copy_err = rb_reg_region_copy(RMATCH_REGS(match), regs); - onig_region_free(regs, 0); - if (copy_err) rb_memerror(); + + VALUE existing = (set_match && !NIL_P(*set_match)) ? *set_match : rb_backref_get(); + VALUE match = match_alloc_or_reuse(existing, regs.num_regs); + + match_set_regs(match, regs.num_regs, regs.beg, regs.end); + ALLOCV_END(regs_buf); if (set_backref_str) { - RMATCH(match)->str = rb_str_new4(str); + RB_OBJ_WRITE(match, &RMATCH(match)->str, rb_str_new4(str)); + rb_obj_reveal(match, rb_cMatch); + } + else { + /* Note that a MatchData object with RMATCH(match)->str == 0 is incomplete! + * We need to hide the object from ObjectSpace.each_object. + * https://bugs.ruby-lang.org/issues/19159 + */ + rb_obj_hide(match); } - RMATCH(match)->regexp = re; + RB_OBJ_WRITE(match, &RMATCH(match)->regexp, re); rb_backref_set(match); if (set_match) *set_match = match; @@ -1747,85 +1889,61 @@ rb_reg_search_set_match(VALUE re, VALUE str, long pos, int reverse, int set_back } long -rb_reg_search0(VALUE re, VALUE str, long pos, int reverse, int set_backref_str) +rb_reg_search0(VALUE re, VALUE str, long pos, int reverse, int set_backref_str, VALUE *match) { - return rb_reg_search_set_match(re, str, pos, reverse, set_backref_str, NULL); + return rb_reg_search_set_match(re, str, pos, reverse, set_backref_str, match); } long rb_reg_search(VALUE re, VALUE str, long pos, int reverse) { - return rb_reg_search0(re, str, pos, reverse, 1); + return rb_reg_search_set_match(re, str, pos, reverse, 1, NULL); } -bool -rb_reg_start_with_p(VALUE re, VALUE str) +static OnigPosition +reg_onig_match(regex_t *reg, VALUE str, struct re_registers *regs, void *_) { - long result; - VALUE match; - struct re_registers regi, *regs = ®i; - regex_t *reg; - int tmpreg; - onig_errmsg_buffer err = ""; - - reg = rb_reg_prepare_re0(re, str, err); - tmpreg = reg != RREGEXP_PTR(re); - if (!tmpreg) RREGEXP(re)->usecnt++; - - match = rb_backref_get(); - if (!NIL_P(match)) { - if (FL_TEST(match, MATCH_BUSY)) { - match = Qnil; - } - else { - regs = RMATCH_REGS(match); - } - } - if (NIL_P(match)) { - MEMZERO(regs, struct re_registers, 1); - } const char *ptr; long len; RSTRING_GETMEM(str, ptr, len); - result = onig_match(reg, - (UChar*)(ptr), - ((UChar*)(ptr + len)), - (UChar*)(ptr), - regs, ONIG_OPTION_NONE); - if (!tmpreg) RREGEXP(re)->usecnt--; - if (tmpreg) { - if (RREGEXP(re)->usecnt) { - onig_free(reg); - } - else { - onig_free(RREGEXP_PTR(re)); - RREGEXP_PTR(re) = reg; - } - } - if (result < 0) { - if (regs == ®i) - onig_region_free(regs, 0); - if (result == ONIG_MISMATCH) { - rb_backref_set(Qnil); - return false; - } - else { - onig_error_code_to_str((UChar*)err, (int)result); - rb_reg_raise(RREGEXP_SRC_PTR(re), RREGEXP_SRC_LEN(re), err, re); - } - } - if (NIL_P(match)) { - int err; - match = match_alloc(rb_cMatch); - err = rb_reg_region_copy(RMATCH_REGS(match), regs); - onig_region_free(regs, 0); - if (err) rb_memerror(); - } + return onig_match( + reg, + (UChar *)ptr, + (UChar *)(ptr + len), + (UChar *)ptr, + regs, + ONIG_OPTION_NONE); +} - RMATCH(match)->str = rb_str_new4(str); +bool +rb_reg_start_with_p(VALUE re, VALUE str) +{ + rb_reg_check(re); - RMATCH(match)->regexp = re; + int n = RREGEXP_PTR(re)->num_mem + 1; + int cap = n < ONIG_NREGION ? ONIG_NREGION : n; + VALUE regs_buf; + OnigPosition *buf = ALLOCV_N(OnigPosition, regs_buf, (size_t)cap * 2); + struct re_registers regs = { + .allocated = cap, + .num_regs = 0, + .beg = buf, + .end = buf + cap, + }; + + if (rb_reg_onig_match(re, str, reg_onig_match, NULL, ®s) == ONIG_MISMATCH) { + ALLOCV_END(regs_buf); + rb_backref_set(Qnil); + return false; + } + + VALUE match = match_alloc_or_reuse(rb_backref_get(), regs.num_regs); + match_set_regs(match, regs.num_regs, regs.beg, regs.end); + ALLOCV_END(regs_buf); + + RB_OBJ_WRITE(match, &RMATCH(match)->str, rb_str_new4(str)); + RB_OBJ_WRITE(match, &RMATCH(match)->regexp, re); rb_backref_set(match); return true; @@ -1834,18 +1952,17 @@ rb_reg_start_with_p(VALUE re, VALUE str) VALUE rb_reg_nth_defined(int nth, VALUE match) { - struct re_registers *regs; if (NIL_P(match)) return Qnil; match_check(match); - regs = RMATCH_REGS(match); - if (nth >= regs->num_regs) { - return Qnil; + int num_regs = RMATCH_NREGS(match); + if (nth >= num_regs) { + return Qnil; } if (nth < 0) { - nth += regs->num_regs; - if (nth <= 0) return Qnil; + nth += num_regs; + if (nth <= 0) return Qnil; } - return RBOOL(BEG(nth) != -1); + return RBOOL(RMATCH_BEG(match, nth) != -1); } VALUE @@ -1853,21 +1970,20 @@ rb_reg_nth_match(int nth, VALUE match) { VALUE str; long start, end, len; - struct re_registers *regs; if (NIL_P(match)) return Qnil; match_check(match); - regs = RMATCH_REGS(match); - if (nth >= regs->num_regs) { - return Qnil; + int num_regs = RMATCH_NREGS(match); + if (nth >= num_regs) { + return Qnil; } if (nth < 0) { - nth += regs->num_regs; - if (nth <= 0) return Qnil; + nth += num_regs; + if (nth <= 0) return Qnil; } - start = BEG(nth); + start = RMATCH_BEG(match, nth); if (start == -1) return Qnil; - end = END(nth); + end = RMATCH_END(match, nth); len = end - start; str = rb_str_subseq(RMATCH(match)->str, start, len); return str; @@ -1901,13 +2017,11 @@ VALUE rb_reg_match_pre(VALUE match) { VALUE str; - struct re_registers *regs; if (NIL_P(match)) return Qnil; match_check(match); - regs = RMATCH_REGS(match); - if (BEG(0) == -1) return Qnil; - str = rb_str_subseq(RMATCH(match)->str, 0, BEG(0)); + if (RMATCH_BEG(match, 0) == -1) return Qnil; + str = rb_str_subseq(RMATCH(match)->str, 0, RMATCH_BEG(match, 0)); return str; } @@ -1935,33 +2049,45 @@ rb_reg_match_post(VALUE match) { VALUE str; long pos; - struct re_registers *regs; if (NIL_P(match)) return Qnil; match_check(match); - regs = RMATCH_REGS(match); - if (BEG(0) == -1) return Qnil; + if (RMATCH_BEG(match, 0) == -1) return Qnil; str = RMATCH(match)->str; - pos = END(0); + pos = RMATCH_END(match, 0); str = rb_str_subseq(str, pos, RSTRING_LEN(str) - pos); return str; } -VALUE -rb_reg_match_last(VALUE match) +static int +match_last_index(VALUE match) { int i; - struct re_registers *regs; - if (NIL_P(match)) return Qnil; + if (NIL_P(match)) return -1; match_check(match); - regs = RMATCH_REGS(match); - if (BEG(0) == -1) return Qnil; + if (RMATCH_BEG(match, 0) == -1) return -1; + + for (i = RMATCH_NREGS(match) - 1; RMATCH_BEG(match, i) == -1 && i > 0; i--) + ; + return i; +} - for (i=regs->num_regs-1; BEG(i) == -1 && i > 0; i--) - ; - if (i == 0) return Qnil; - return rb_reg_nth_match(i, match); +VALUE +rb_reg_match_last(VALUE match) +{ + int i = match_last_index(match); + if (i <= 0) return Qnil; + long start = RMATCH_BEG(match, i); + return rb_str_subseq(RMATCH(match)->str, start, RMATCH_END(match, i) - start); +} + +VALUE +rb_reg_last_defined(VALUE match) +{ + int i = match_last_index(match); + if (i < 0) return Qnil; + return RBOOL(i); } static VALUE @@ -1991,24 +2117,24 @@ last_paren_match_getter(ID _x, VALUE *_y) static VALUE match_array(VALUE match, int start) { - struct re_registers *regs; VALUE ary; VALUE target; int i; match_check(match); - regs = RMATCH_REGS(match); - ary = rb_ary_new2(regs->num_regs); + int num_regs = RMATCH_NREGS(match); + ary = rb_ary_new2(num_regs); target = RMATCH(match)->str; - for (i=start; i<regs->num_regs; i++) { - if (regs->beg[i] == -1) { - rb_ary_push(ary, Qnil); - } - else { - VALUE str = rb_str_subseq(target, regs->beg[i], regs->end[i]-regs->beg[i]); - rb_ary_push(ary, str); - } + for (i = start; i < num_regs; i++) { + long beg = RMATCH_BEG(match, i); + if (beg == -1) { + rb_ary_push(ary, Qnil); + } + else { + VALUE str = rb_str_subseq(target, beg, RMATCH_END(match, i) - beg); + rb_ary_push(ary, str); + } } return ary; } @@ -2057,11 +2183,11 @@ match_captures(VALUE match) } static int -name_to_backref_number(struct re_registers *regs, VALUE regexp, const char* name, const char* name_end) +name_to_backref_number(const struct re_registers *regs, VALUE regexp, const char* name, const char* name_end) { if (NIL_P(regexp)) return -1; return onig_name_to_backref_number(RREGEXP_PTR(regexp), - (const unsigned char *)name, (const unsigned char *)name_end, regs); + (const unsigned char *)name, (const unsigned char *)name_end, regs); } #define NAME_TO_NUMBER(regs, re, name, name_ptr, name_end) \ @@ -2070,20 +2196,46 @@ name_to_backref_number(struct re_registers *regs, VALUE regexp, const char* name name_to_backref_number((regs), (re), (name_ptr), (name_end))) static int -namev_to_backref_number(struct re_registers *regs, VALUE re, VALUE name) +match_name_to_backref_number(VALUE match, VALUE name) +{ + VALUE regexp = RMATCH(match)->regexp; + if (NIL_P(regexp)) return -1; + + int *nums; + int n = onig_name_to_group_numbers(RREGEXP_PTR(regexp), + (const unsigned char *)RSTRING_PTR(name), + (const unsigned char *)RSTRING_END(name), &nums); + if (n < 0) return n; + if (n == 0) return ONIGERR_PARSER_BUG; + if (n == 1) return nums[0]; + for (int i = n - 1; i >= 0; i--) { + if (RMATCH_BEG(match, nums[i]) != ONIG_REGION_NOTPOS) + return nums[i]; + } + return nums[n - 1]; +} + +static int +namev_to_backref_number(VALUE match, VALUE name) { int num; if (SYMBOL_P(name)) { - name = rb_sym2str(name); + name = rb_sym2str(name); } else if (!RB_TYPE_P(name, T_STRING)) { - return -1; + return -1; + } + + VALUE re = RMATCH(match)->regexp; + if (NIL_P(re) || !rb_enc_compatible(RREGEXP_SRC(re), name)) { + num = 0; + } + else { + num = match_name_to_backref_number(match, name); } - num = NAME_TO_NUMBER(regs, re, name, - RSTRING_PTR(name), RSTRING_END(name)); if (num < 1) { - name_to_backref_error(name); + name_to_backref_error(name); } return num; } @@ -2091,16 +2243,16 @@ namev_to_backref_number(struct re_registers *regs, VALUE re, VALUE name) static VALUE match_ary_subseq(VALUE match, long beg, long len, VALUE result) { - long olen = RMATCH_REGS(match)->num_regs; + long olen = RMATCH_NREGS(match); long j, end = olen < beg+len ? olen : beg+len; if (NIL_P(result)) result = rb_ary_new_capa(len); if (len == 0) return result; for (j = beg; j < end; j++) { - rb_ary_push(result, rb_reg_nth_match((int)j, match)); + rb_ary_push(result, rb_reg_nth_match((int)j, match)); } if (beg + len > j) { - rb_ary_resize(result, RARRAY_LEN(result) + (beg + len) - j); + rb_ary_resize(result, RARRAY_LEN(result) + (beg + len) - j); } return result; } @@ -2109,29 +2261,29 @@ static VALUE match_ary_aref(VALUE match, VALUE idx, VALUE result) { long beg, len; - int num_regs = RMATCH_REGS(match)->num_regs; + int num_regs = RMATCH_NREGS(match); /* check if idx is Range */ switch (rb_range_beg_len(idx, &beg, &len, (long)num_regs, !NIL_P(result))) { case Qfalse: - if (NIL_P(result)) return rb_reg_nth_match(NUM2INT(idx), match); - rb_ary_push(result, rb_reg_nth_match(NUM2INT(idx), match)); - return result; + if (NIL_P(result)) return rb_reg_nth_match(NUM2INT(idx), match); + rb_ary_push(result, rb_reg_nth_match(NUM2INT(idx), match)); + return result; case Qnil: - return Qnil; + return Qnil; default: - return match_ary_subseq(match, beg, len, result); + return match_ary_subseq(match, beg, len, result); } } /* * call-seq: - * matchdata[index] -> string or nil - * matchdata[start, length] -> array - * matchdata[range] -> array - * matchdata[name] -> string or nil + * self[offset] -> string or nil + * self[offset, size] -> array + * self[range] -> array + * self[name] -> string or nil * - * When arguments +index+, +start and +length+, or +range+ are given, + * When arguments +offset+, +offset+ and +size+, or +range+ are given, * returns match and captures in the style of Array#[]: * * m = /(.)(.)(\d+)(\d)/.match("THX1138.") @@ -2149,6 +2301,17 @@ match_ary_aref(VALUE match, VALUE idx, VALUE result) * m['foo'] # => "h" * m[:bar] # => "ge" * + * If multiple captures have the same name, returns the last matched + * substring. + * + * m = /(?<foo>.)(?<foo>.+)/.match("hoge") + * # => #<MatchData "hoge" foo:"h" foo:"oge"> + * m[:foo] #=> "oge" + * + * m = /\W(?<foo>.+)|\w(?<foo>.+)|(?<foo>.+)/.match("hoge") + * #<MatchData "hoge" foo:nil foo:"oge" foo:nil> + * m[:foo] #=> "oge" + * */ static VALUE @@ -2160,37 +2323,37 @@ match_aref(int argc, VALUE *argv, VALUE match) rb_scan_args(argc, argv, "11", &idx, &length); if (NIL_P(length)) { - if (FIXNUM_P(idx)) { - return rb_reg_nth_match(FIX2INT(idx), match); - } - else { - int num = namev_to_backref_number(RMATCH_REGS(match), RMATCH(match)->regexp, idx); - if (num >= 0) { - return rb_reg_nth_match(num, match); - } - else { - return match_ary_aref(match, idx, Qnil); - } - } + if (FIXNUM_P(idx)) { + return rb_reg_nth_match(FIX2INT(idx), match); + } + else { + int num = namev_to_backref_number(match, idx); + if (num >= 0) { + return rb_reg_nth_match(num, match); + } + else { + return match_ary_aref(match, idx, Qnil); + } + } } else { - long beg = NUM2LONG(idx); - long len = NUM2LONG(length); - long num_regs = RMATCH_REGS(match)->num_regs; - if (len < 0) { - return Qnil; - } - if (beg < 0) { - beg += num_regs; - if (beg < 0) return Qnil; - } - else if (beg > num_regs) { - return Qnil; - } - if (beg+len > num_regs) { - len = num_regs - beg; - } - return match_ary_subseq(match, beg, len, Qnil); + long beg = NUM2LONG(idx); + long len = NUM2LONG(length); + long num_regs = RMATCH_NREGS(match); + if (len < 0) { + return Qnil; + } + if (beg < 0) { + beg += num_regs; + if (beg < 0) return Qnil; + } + else if (beg > num_regs) { + return Qnil; + } + if (beg+len > num_regs) { + len = num_regs - beg; + } + return match_ary_subseq(match, beg, len, Qnil); } } @@ -2230,18 +2393,18 @@ match_values_at(int argc, VALUE *argv, VALUE match) result = rb_ary_new2(argc); for (i=0; i<argc; i++) { - if (FIXNUM_P(argv[i])) { - rb_ary_push(result, rb_reg_nth_match(FIX2INT(argv[i]), match)); - } - else { - int num = namev_to_backref_number(RMATCH_REGS(match), RMATCH(match)->regexp, argv[i]); - if (num >= 0) { - rb_ary_push(result, rb_reg_nth_match(num, match)); - } - else { - match_ary_aref(match, argv[i], result); - } - } + if (FIXNUM_P(argv[i])) { + rb_ary_push(result, rb_reg_nth_match(FIX2INT(argv[i]), match)); + } + else { + int num = namev_to_backref_number(match, argv[i]); + if (num >= 0) { + rb_ary_push(result, rb_reg_nth_match(num, match)); + } + else { + match_ary_aref(match, argv[i], result); + } + } } return result; } @@ -2268,36 +2431,47 @@ match_values_at(int argc, VALUE *argv, VALUE match) static VALUE match_to_s(VALUE match) { - VALUE str = rb_reg_last_match(match); + VALUE str = rb_reg_last_match(match_check(match)); - match_check(match); if (NIL_P(str)) str = rb_str_new(0,0); return str; } +struct named_captures_data { + VALUE hash; + VALUE match; + int symbolize; +}; + static int match_named_captures_iter(const OnigUChar *name, const OnigUChar *name_end, - int back_num, int *back_refs, OnigRegex regex, void *arg) { - struct MEMO *memo = MEMO_CAST(arg); - VALUE hash = memo->v1; - VALUE match = memo->v2; + int back_num, int *back_refs, OnigRegex regex, void *arg) +{ + struct named_captures_data *data = arg; + VALUE hash = data->hash; + VALUE match = data->match; VALUE key = rb_enc_str_new((const char *)name, name_end-name, regex->enc); + + if (data->symbolize) { + key = rb_str_intern(key); + } + VALUE value; int i; int found = 0; for (i = 0; i < back_num; i++) { - value = rb_reg_nth_match(back_refs[i], match); - if (RTEST(value)) { - rb_hash_aset(hash, key, value); - found = 1; - } + value = rb_reg_nth_match(back_refs[i], match); + if (RTEST(value)) { + rb_hash_aset(hash, key, value); + found = 1; + } } if (found == 0) { - rb_hash_aset(hash, key, Qnil); + rb_hash_aset(hash, key, Qnil); } return 0; @@ -2305,7 +2479,7 @@ match_named_captures_iter(const OnigUChar *name, const OnigUChar *name_end, /* * call-seq: - * named_captures -> hash + * named_captures(symbolize_names: false) -> hash * * Returns a hash of the named captures; * each key is a capture name; each value is its captured string or +nil+: @@ -2326,28 +2500,120 @@ match_named_captures_iter(const OnigUChar *name, const OnigUChar *name_end, * # => #<MatchData "01" a:"0" a:"1"> * m.named_captures #=> {"a" => "1"} * + * If keyword argument +symbolize_names+ is given + * a true value, the keys in the resulting hash are Symbols: + * + * m = /(?<a>.)(?<a>.)/.match("01") + * # => #<MatchData "01" a:"0" a:"1"> + * m.named_captures(symbolize_names: true) #=> {:a => "1"} + * */ static VALUE -match_named_captures(VALUE match) +match_named_captures(int argc, VALUE *argv, VALUE match) { VALUE hash; - struct MEMO *memo; match_check(match); if (NIL_P(RMATCH(match)->regexp)) - return rb_hash_new(); + return rb_hash_new(); + + VALUE opt; + int symbolize_names = 0; + + rb_scan_args(argc, argv, "0:", &opt); + + if (!NIL_P(opt)) { + static ID keyword_ids[1]; + + VALUE symbolize_names_val; + + if (!keyword_ids[0]) { + keyword_ids[0] = rb_intern_const("symbolize_names"); + } + rb_get_kwargs(opt, keyword_ids, 0, 1, &symbolize_names_val); + if (!UNDEF_P(symbolize_names_val) && RTEST(symbolize_names_val)) { + symbolize_names = 1; + } + } hash = rb_hash_new(); - memo = MEMO_NEW(hash, match, 0); + struct named_captures_data data = { hash, match, symbolize_names }; - onig_foreach_name(RREGEXP(RMATCH(match)->regexp)->ptr, match_named_captures_iter, (void*)memo); + onig_foreach_name(RREGEXP(RMATCH(match)->regexp)->ptr, match_named_captures_iter, &data); return hash; } /* * call-seq: + * deconstruct_keys(array_of_names) -> hash + * + * Returns a hash of the named captures for the given names. + * + * m = /(?<hours>\d{2}):(?<minutes>\d{2}):(?<seconds>\d{2})/.match("18:37:22") + * m.deconstruct_keys([:hours, :minutes]) # => {:hours => "18", :minutes => "37"} + * m.deconstruct_keys(nil) # => {:hours => "18", :minutes => "37", :seconds => "22"} + * + * Returns an empty hash if no named captures were defined: + * + * m = /(\d{2}):(\d{2}):(\d{2})/.match("18:37:22") + * m.deconstruct_keys(nil) # => {} + * + */ +static VALUE +match_deconstruct_keys(VALUE match, VALUE keys) +{ + VALUE h; + long i; + + match_check(match); + + if (NIL_P(RMATCH(match)->regexp)) { + return rb_hash_new_with_size(0); + } + + if (NIL_P(keys)) { + h = rb_hash_new_with_size(onig_number_of_names(RREGEXP_PTR(RMATCH(match)->regexp))); + + struct named_captures_data data = { h, match, 1 }; + + onig_foreach_name(RREGEXP_PTR(RMATCH(match)->regexp), match_named_captures_iter, &data); + + return h; + } + + Check_Type(keys, T_ARRAY); + + if (onig_number_of_names(RREGEXP_PTR(RMATCH(match)->regexp)) < RARRAY_LEN(keys)) { + return rb_hash_new_with_size(0); + } + + h = rb_hash_new_with_size(RARRAY_LEN(keys)); + + for (i=0; i<RARRAY_LEN(keys); i++) { + VALUE key = RARRAY_AREF(keys, i); + VALUE name; + + Check_Type(key, T_SYMBOL); + + name = rb_sym2str(key); + + int num = match_name_to_backref_number(match, name); + + if (num >= 0) { + rb_hash_aset(h, key, rb_reg_nth_match(num, match)); + } + else { + return h; + } + } + + return h; +} + +/* + * call-seq: * string -> string * * Returns the target string if it was frozen; @@ -2386,10 +2652,10 @@ match_inspect_name_iter(const OnigUChar *name, const OnigUChar *name_end, } /* - * call-seq: - * inspect -> string + * call-seq: + * inspect -> string * - * Returns a string representation of +self+: + * Returns a string representation of +self+: * * m = /.$/.match("foo") * # => #<MatchData "o"> @@ -2404,7 +2670,6 @@ match_inspect_name_iter(const OnigUChar *name, const OnigUChar *name_end, * m.inspect # => "#<MatchData \"fo\" 1:\"f\" 2:nil 3:\"o\">" * * Related: MatchData#to_s. - * */ static VALUE @@ -2413,9 +2678,9 @@ match_inspect(VALUE match) VALUE cname = rb_class_path(rb_obj_class(match)); VALUE str; int i; - struct re_registers *regs = RMATCH_REGS(match); - int num_regs = regs->num_regs; + int num_regs = RMATCH_NREGS(match); struct backref_name_tag *names; + VALUE names_obj = Qnil; VALUE regexp = RMATCH(match)->regexp; if (regexp == 0) { @@ -2423,10 +2688,10 @@ match_inspect(VALUE match) } else if (NIL_P(regexp)) { return rb_sprintf("#<%"PRIsVALUE": %"PRIsVALUE">", - cname, rb_reg_nth_match(0, match)); + cname, rb_reg_nth_match(0, match)); } - names = ALLOCA_N(struct backref_name_tag, num_regs); + names = RB_ALLOCV_N(struct backref_name_tag, names_obj, num_regs); MEMZERO(names, struct backref_name_tag, num_regs); onig_foreach_name(RREGEXP_PTR(regexp), @@ -2454,6 +2719,7 @@ match_inspect(VALUE match) } rb_str_buf_cat2(str, ">"); + RB_ALLOCV_END(names_obj); return str; } @@ -2717,14 +2983,18 @@ unescape_unicode_bmp(const char **pp, const char *end, } static int -unescape_nonascii(const char *p, const char *end, rb_encoding *enc, +unescape_nonascii0(const char **pp, const char *end, rb_encoding *enc, VALUE buf, rb_encoding **encp, int *has_property, - onig_errmsg_buffer err, int options) + onig_errmsg_buffer err, int options, int recurse) { + const char *p = *pp; unsigned char c; char smallbuf[2]; int in_char_class = 0; + int parens = 1; /* ignored unless recurse is true */ + int extended_mode = options & ONIG_OPTION_EXTEND; +begin_scan: while (p < end) { int chlen = rb_enc_precise_mbclen(p, end, enc); if (!MBCLEN_CHARFOUND_P(chlen)) { @@ -2757,7 +3027,7 @@ unescape_nonascii(const char *p, const char *end, rb_encoding *enc, goto invalid_multibyte; } if ((chlen = MBCLEN_CHARFOUND_LEN(chlen)) > 1) { - /* include the previous backslash */ + /* include the previous backslash */ --p; ++chlen; goto multibyte; @@ -2783,17 +3053,17 @@ unescape_nonascii(const char *p, const char *end, rb_encoding *enc, case 'C': /* \C-X, \C-\M-X */ case 'M': /* \M-X, \M-\C-X, \M-\cX */ p = p-2; - if (enc == rb_usascii_encoding()) { - const char *pbeg = p; + if (rb_is_usascii_enc(enc)) { + const char *pbeg = p; int byte = read_escaped_byte(&p, end, err); if (byte == -1) return -1; c = byte; - rb_str_buf_cat(buf, pbeg, p-pbeg); - } - else { - if (unescape_escaped_nonascii(&p, end, enc, buf, encp, err) != 0) - return -1; - } + rb_str_buf_cat(buf, pbeg, p-pbeg); + } + else { + if (unescape_escaped_nonascii(&p, end, enc, buf, encp, err) != 0) + return -1; + } break; case 'u': @@ -2836,9 +3106,13 @@ escape_asis: break; case '#': - if ((options & ONIG_OPTION_EXTEND) && !in_char_class) { + if (extended_mode && !in_char_class) { /* consume and ignore comment in extended regexp */ - while ((p < end) && ((c = *p++) != '\n')); + while ((p < end) && ((c = *p++) != '\n')) { + if ((c & 0x80) && !*encp && enc == rb_utf8_encoding()) { + *encp = enc; + } + } break; } rb_str_buf_cat(buf, (char *)&c, 1); @@ -2853,51 +3127,141 @@ escape_asis: } rb_str_buf_cat(buf, (char *)&c, 1); break; + case ')': + rb_str_buf_cat(buf, (char *)&c, 1); + if (!in_char_class && recurse) { + if (--parens == 0) { + *pp = p; + return 0; + } + } + break; case '(': - if (!in_char_class && p + 1 < end && *p == '?' && *(p+1) == '#') { - /* (?# is comment inside any regexp, and content inside should be ignored */ - const char *orig_p = p; - int cont = 1; - - while (cont && (p < end)) { - switch (c = *p++) { - default: - if (!(c & 0x80)) break; - --p; - /* fallthrough */ - case '\\': - chlen = rb_enc_precise_mbclen(p, end, enc); - if (!MBCLEN_CHARFOUND_P(chlen)) { - goto invalid_multibyte; + if (!in_char_class && p + 1 < end && *p == '?') { + if (*(p+1) == '#') { + /* (?# is comment inside any regexp, and content inside should be ignored */ + const char *orig_p = p; + int cont = 1; + + while (cont && (p < end)) { + switch (c = *p++) { + default: + if (!(c & 0x80)) break; + if (!*encp && enc == rb_utf8_encoding()) { + *encp = enc; + } + --p; + /* fallthrough */ + case '\\': + chlen = rb_enc_precise_mbclen(p, end, enc); + if (!MBCLEN_CHARFOUND_P(chlen)) { + goto invalid_multibyte; + } + p += MBCLEN_CHARFOUND_LEN(chlen); + break; + case ')': + cont = 0; + break; } - p += MBCLEN_CHARFOUND_LEN(chlen); - break; - case ')': - cont = 0; - break; } + + if (cont) { + /* unterminated (?#, rewind so it is syntax error */ + p = orig_p; + c = '('; + rb_str_buf_cat(buf, (char *)&c, 1); + } + break; } + else { + /* potential change of extended option */ + int invert = 0; + int local_extend = 0; + const char *s; - if (cont) { - /* unterminated (?#, rewind so it is syntax error */ - p = orig_p; - c = '('; - rb_str_buf_cat(buf, (char *)&c, 1); + if (recurse) { + parens++; + } + + for (s = p+1; s < end; s++) { + switch(*s) { + case 'x': + local_extend = invert ? -1 : 1; + break; + case '-': + invert = 1; + break; + case ':': + case ')': + if (local_extend == 0 || + (local_extend == -1 && !extended_mode) || + (local_extend == 1 && extended_mode)) { + /* no changes to extended flag */ + goto fallthrough; + } + + if (*s == ':') { + /* change extended flag until ')' */ + int local_options = options; + if (local_extend == 1) { + local_options |= ONIG_OPTION_EXTEND; + } + else { + local_options &= ~ONIG_OPTION_EXTEND; + } + + rb_str_buf_cat(buf, (char *)&c, 1); + int ret = unescape_nonascii0(&p, end, enc, buf, encp, + has_property, err, + local_options, 1); + if (ret < 0) return ret; + goto begin_scan; + } + else { + /* change extended flag for rest of expression */ + extended_mode = local_extend == 1; + goto fallthrough; + } + case 'i': + case 'm': + case 'a': + case 'd': + case 'u': + /* other option flags, ignored during scanning */ + break; + default: + /* other character, no extended flag change*/ + goto fallthrough; + } + } } } - else { - rb_str_buf_cat(buf, (char *)&c, 1); + else if (!in_char_class && recurse) { + parens++; } - break; + /* FALLTHROUGH */ default: +fallthrough: rb_str_buf_cat(buf, (char *)&c, 1); break; } } + if (recurse) { + *pp = p; + } return 0; } +static int +unescape_nonascii(const char *p, const char *end, rb_encoding *enc, + VALUE buf, rb_encoding **encp, int *has_property, + onig_errmsg_buffer err, int options) +{ + return unescape_nonascii0(&p, end, enc, buf, encp, has_property, + err, options, 0); +} + static VALUE rb_reg_preprocess(const char *p, const char *end, rb_encoding *enc, rb_encoding **fixed_enc, onig_errmsg_buffer err, int options) @@ -2946,7 +3310,7 @@ rb_reg_check_preprocess(VALUE str) RB_GC_GUARD(str); if (NIL_P(buf)) { - return rb_reg_error_desc(str, 0, err); + return rb_reg_error_desc(str, 0, err); } return Qnil; } @@ -2971,14 +3335,14 @@ rb_reg_preprocess_dregexp(VALUE ary, int options) char *p, *end; rb_encoding *src_enc; - src_enc = rb_enc_get(str); - if (options & ARG_ENCODING_NONE && - src_enc != ascii8bit) { - if (str_coderange(str) != ENC_CODERANGE_7BIT) - rb_raise(rb_eRegexpError, "/.../n has a non escaped non ASCII character in non ASCII-8BIT script"); - else - src_enc = ascii8bit; - } + src_enc = rb_enc_get(str); + if (options & ARG_ENCODING_NONE && + src_enc != ascii8bit) { + if (rb_enc_str_coderange(str) != ENC_CODERANGE_7BIT) + rb_raise(rb_eRegexpError, "/.../n has a non escaped non ASCII character in non ASCII-8BIT script"); + else + src_enc = ascii8bit; + } StringValue(str); p = RSTRING_PTR(str); @@ -3009,26 +3373,30 @@ rb_reg_preprocess_dregexp(VALUE ary, int options) return result; } +static void +rb_reg_initialize_check(VALUE obj) +{ + rb_check_frozen(obj); + if (RREGEXP_PTR(obj)) { + rb_raise(rb_eTypeError, "already initialized regexp"); + } +} + static int rb_reg_initialize(VALUE obj, const char *s, long len, rb_encoding *enc, - int options, onig_errmsg_buffer err, - const char *sourcefile, int sourceline) + int options, onig_errmsg_buffer err, + const char *sourcefile, int sourceline) { struct RRegexp *re = RREGEXP(obj); VALUE unescaped; rb_encoding *fixed_enc = 0; rb_encoding *a_enc = rb_ascii8bit_encoding(); - rb_check_frozen(obj); - if (FL_TEST(obj, REG_LITERAL)) - rb_raise(rb_eSecurityError, "can't modify literal regexp"); - if (re->ptr) - rb_raise(rb_eTypeError, "already initialized regexp"); - re->ptr = 0; + rb_reg_initialize_check(obj); if (rb_enc_dummy_p(enc)) { - errcpy(err, "can't make regexp with dummy encoding"); - return -1; + errcpy(err, "can't make regexp with dummy encoding"); + return -1; } unescaped = rb_reg_preprocess(s, s+len, enc, &fixed_enc, err, options); @@ -3036,15 +3404,15 @@ rb_reg_initialize(VALUE obj, const char *s, long len, rb_encoding *enc, return -1; if (fixed_enc) { - if ((fixed_enc != enc && (options & ARG_ENCODING_FIXED)) || + if ((fixed_enc != enc && (options & ARG_ENCODING_FIXED)) || (fixed_enc != a_enc && (options & ARG_ENCODING_NONE))) { - errcpy(err, "incompatible character encoding"); - return -1; - } + errcpy(err, "incompatible character encoding"); + return -1; + } if (fixed_enc != a_enc) { - options |= ARG_ENCODING_FIXED; - enc = fixed_enc; - } + options |= ARG_ENCODING_FIXED; + enc = fixed_enc; + } } else if (!(options & ARG_ENCODING_FIXED)) { enc = rb_usascii_encoding(); @@ -3052,16 +3420,19 @@ rb_reg_initialize(VALUE obj, const char *s, long len, rb_encoding *enc, rb_enc_associate((VALUE)re, enc); if ((options & ARG_ENCODING_FIXED) || fixed_enc) { - re->basic.flags |= KCODE_FIXED; + re->basic.flags |= KCODE_FIXED; } if (options & ARG_ENCODING_NONE) { re->basic.flags |= REG_ENCODING_NONE; } re->ptr = make_regexp(RSTRING_PTR(unescaped), RSTRING_LEN(unescaped), enc, - options & ARG_REG_OPTION_MASK, err, - sourcefile, sourceline); + options & ARG_REG_OPTION_MASK, err, + sourcefile, sourceline); if (!re->ptr) return -1; + if (RBASIC_CLASS(obj) == rb_cRegexp) { + OBJ_FREEZE(obj); + } RB_GC_GUARD(unescaped); return 0; } @@ -3070,22 +3441,25 @@ static void reg_set_source(VALUE reg, VALUE str, rb_encoding *enc) { rb_encoding *regenc = rb_enc_get(reg); + if (regenc != enc) { - str = rb_enc_associate(rb_str_dup(str), enc = regenc); + VALUE dup = rb_str_dup(str); + str = rb_enc_associate(dup, enc = regenc); } - RB_OBJ_WRITE(reg, &RREGEXP(reg)->src, rb_fstring(str)); + str = rb_fstring(str); + RB_OBJ_WRITE(reg, &RREGEXP(reg)->src, str); } static int rb_reg_initialize_str(VALUE obj, VALUE str, int options, onig_errmsg_buffer err, - const char *sourcefile, int sourceline) + const char *sourcefile, int sourceline) { int ret; rb_encoding *str_enc = rb_enc_get(str), *enc = str_enc; if (options & ARG_ENCODING_NONE) { rb_encoding *ascii8bit = rb_ascii8bit_encoding(); if (enc != ascii8bit) { - if (str_coderange(str) != ENC_CODERANGE_7BIT) { + if (rb_enc_str_coderange(str) != ENC_CODERANGE_7BIT) { errcpy(err, "/.../n has a non escaped non ASCII character in non ASCII-8BIT script"); return -1; } @@ -3093,15 +3467,15 @@ rb_reg_initialize_str(VALUE obj, VALUE str, int options, onig_errmsg_buffer err, } } ret = rb_reg_initialize(obj, RSTRING_PTR(str), RSTRING_LEN(str), enc, - options, err, sourcefile, sourceline); + options, err, sourcefile, sourceline); if (ret == 0) reg_set_source(obj, str, str_enc); return ret; } -static VALUE +VALUE rb_reg_s_alloc(VALUE klass) { - NEWOBJ_OF(re, struct RRegexp, klass, T_REGEXP | (RGENGC_WB_PROTECTED_REGEXP ? FL_WB_PROTECTED : 0)); + NEWOBJ_OF(re, struct RRegexp, klass, T_REGEXP, sizeof(struct RRegexp)); re->ptr = 0; RB_OBJ_WRITE(re, &re->src, 0); @@ -3128,7 +3502,7 @@ rb_reg_init_str(VALUE re, VALUE s, int options) onig_errmsg_buffer err = ""; if (rb_reg_initialize_str(re, s, options, err, NULL, 0) != 0) { - rb_reg_raise_str(s, options, err); + rb_reg_raise_str(s, options, err); } return re; @@ -3140,20 +3514,21 @@ rb_reg_init_str_enc(VALUE re, VALUE s, rb_encoding *enc, int options) onig_errmsg_buffer err = ""; if (rb_reg_initialize(re, RSTRING_PTR(s), RSTRING_LEN(s), - enc, options, err, NULL, 0) != 0) { - rb_reg_raise_str(s, options, err); + enc, options, err, NULL, 0) != 0) { + rb_reg_raise_str(s, options, err); } reg_set_source(re, s, enc); return re; } -MJIT_FUNC_EXPORTED VALUE -rb_reg_new_ary(VALUE ary, int opt) +VALUE +rb_reg_new_from_values(long cnt, const VALUE *elements, int opt) { - VALUE re = rb_reg_new_str(rb_reg_preprocess_dregexp(ary, opt), opt); - rb_obj_freeze(re); - return re; + const VALUE ary = rb_ary_tmp_new_from_values(0, cnt, elements); + VALUE val = rb_reg_new_str(rb_reg_preprocess_dregexp(ary, opt), opt); + rb_ary_clear(ary); + return val; } VALUE @@ -3163,7 +3538,7 @@ rb_enc_reg_new(const char *s, long len, rb_encoding *enc, int options) onig_errmsg_buffer err = ""; if (rb_reg_initialize(re, s, len, enc, options, err, NULL, 0) != 0) { - rb_enc_reg_raise(s, len, enc, options, err); + rb_enc_reg_raise(s, len, enc, options, err); } RB_OBJ_WRITE(re, &RREGEXP(re)->src, rb_fstring(rb_enc_str_new(s, len, enc))); @@ -3184,11 +3559,9 @@ rb_reg_compile(VALUE str, int options, const char *sourcefile, int sourceline) if (!str) str = rb_str_new(0,0); if (rb_reg_initialize_str(re, str, options, err, sourcefile, sourceline) != 0) { - rb_set_errinfo(rb_reg_error_desc(str, options, err)); - return Qnil; + rb_set_errinfo(rb_reg_error_desc(str, options, err)); + return Qnil; } - FL_SET(re, REG_LITERAL); - rb_obj_freeze(re); return re; } @@ -3197,12 +3570,17 @@ static VALUE reg_cache; VALUE rb_reg_regcomp(VALUE str) { - if (reg_cache && RREGEXP_SRC_LEN(reg_cache) == RSTRING_LEN(str) - && ENCODING_GET(reg_cache) == ENCODING_GET(str) - && memcmp(RREGEXP_SRC_PTR(reg_cache), RSTRING_PTR(str), RSTRING_LEN(str)) == 0) - return reg_cache; + if (rb_ractor_main_p()) { + if (reg_cache && RREGEXP_SRC_LEN(reg_cache) == RSTRING_LEN(str) + && ENCODING_GET(reg_cache) == ENCODING_GET(str) + && memcmp(RREGEXP_SRC_PTR(reg_cache), RSTRING_PTR(str), RSTRING_LEN(str)) == 0) + return reg_cache; - return reg_cache = rb_reg_new_str(str, 0); + return reg_cache = rb_reg_new_str(str, 0); + } + else { + return rb_reg_new_str(str, 0); + } } static st_index_t reg_hash(VALUE re); @@ -3237,18 +3615,16 @@ reg_hash(VALUE re) /* * call-seq: - * regexp == object -> true or false + * self == other -> true or false * - * Returns +true+ if +object+ is another \Regexp whose pattern, - * flags, and encoding are the same as +self+, +false+ otherwise: + * Returns whether +other+ is another \Regexp whose pattern, + * flags, and encoding are the same as +self+: * * /foo/ == Regexp.new('foo') # => true * /foo/ == /foo/i # => false * /foo/ == Regexp.new('food') # => false * /foo/ == Regexp.new("abc".force_encoding("euc-jp")) # => false * - * Regexp#eql? is an alias for Regexp#==. - * */ VALUE @@ -3278,55 +3654,118 @@ rb_reg_equal(VALUE re1, VALUE re2) static VALUE match_hash(VALUE match) { - const struct re_registers *regs; st_index_t hashval; match_check(match); hashval = rb_hash_start(rb_str_hash(RMATCH(match)->str)); hashval = rb_hash_uint(hashval, reg_hash(match_regexp(match))); - regs = RMATCH_REGS(match); - hashval = rb_hash_uint(hashval, regs->num_regs); - hashval = rb_hash_uint(hashval, rb_memhash(regs->beg, regs->num_regs * sizeof(*regs->beg))); - hashval = rb_hash_uint(hashval, rb_memhash(regs->end, regs->num_regs * sizeof(*regs->end))); + int num_regs = RMATCH_NREGS(match); + hashval = rb_hash_uint(hashval, num_regs); + hashval = rb_hash_uint(hashval, rb_memhash(RMATCH_BEG_PTR(match), num_regs * sizeof(OnigPosition))); + hashval = rb_hash_uint(hashval, rb_memhash(RMATCH_END_PTR(match), num_regs * sizeof(OnigPosition))); hashval = rb_hash_end(hashval); return ST2FIX(hashval); } /* * call-seq: - * matchdata == object -> true or false + * self == other -> true or false * - * Returns +true+ if +object+ is another \MatchData object + * Returns whether +other+ is another \MatchData object * whose target string, regexp, match, and captures - * are the same as +self+, +false+ otherwise. - * - * MatchData#eql? is an alias for MatchData#==. - * + * are the same as +self+. */ static VALUE match_equal(VALUE match1, VALUE match2) { - const struct re_registers *regs1, *regs2; - if (match1 == match2) return Qtrue; if (!RB_TYPE_P(match2, T_MATCH)) return Qfalse; if (!RMATCH(match1)->regexp || !RMATCH(match2)->regexp) return Qfalse; if (!rb_str_equal(RMATCH(match1)->str, RMATCH(match2)->str)) return Qfalse; if (!rb_reg_equal(match_regexp(match1), match_regexp(match2))) return Qfalse; - regs1 = RMATCH_REGS(match1); - regs2 = RMATCH_REGS(match2); - if (regs1->num_regs != regs2->num_regs) return Qfalse; - if (memcmp(regs1->beg, regs2->beg, regs1->num_regs * sizeof(*regs1->beg))) return Qfalse; - if (memcmp(regs1->end, regs2->end, regs1->num_regs * sizeof(*regs1->end))) return Qfalse; + int num_regs = RMATCH_NREGS(match1); + if (num_regs != RMATCH_NREGS(match2)) return Qfalse; + if (memcmp(RMATCH_BEG_PTR(match1), RMATCH_BEG_PTR(match2), num_regs * sizeof(OnigPosition))) return Qfalse; + if (memcmp(RMATCH_END_PTR(match1), RMATCH_END_PTR(match2), num_regs * sizeof(OnigPosition))) return Qfalse; return Qtrue; } +/* + * call-seq: + * integer_at(index, base = 10) -> integer or nil + * integer_at(name, base = 10) -> integer or nil + * + * Converts the matched substring to integer and return the result. + * +$~.integer_at(N)+ is equivalent to +$N&.to_i+. + * + * m = /(\d+{4})(\d+{2})(\d+{2})/.match("20260308") + * # => #<MatchData "20260308" 1:"2026" 2:"03" 3:"08"> + * m.integer_at(0) # => 20260308 + * m.integer_at(1) # => 2026 + * m.integer_at(2) # => 3 + * m.integer_at(3) # => 8 + * + * m = /(?<y>\d+{4})(?<m>\d+{2})(?<d>\d+{2})/.match("20260308") + * m.integer_at("y") # => 2026 + * m.integer_at("m") # => 3 + * m.integer_at("d") # => 8 + * + * If the substring does not match, returns +nil+. + * + * re = /(\d+)?/ + * re.match("123").integer_at(1) #=> 123 + * re.match("abc").integer_at(1) #=> nil + * + * The string is converted in decimal by default. + * + * /\d+/.match("011").integer_at(0) #=> 10 + * /\d+/.match("011").integer_at(0, 12) #=> 13 + * /\d+/.match("011").integer_at(0, 0) #=> 9 + * + * See also MatchData#[], String#to_i. + */ +static VALUE +match_integer_at(int argc, VALUE *argv, VALUE match) +{ + match_check(match); + + int base = 10; + VALUE idx; + int nth; + + argc = rb_check_arity(argc, 1, 2); + if (FIXNUM_P(idx = argv[0])) { + nth = NUM2INT(idx); + } + else if ((nth = namev_to_backref_number(match, idx)) < 0) { + name_to_backref_error(idx); + } + + if (argc > 1 && (base = NUM2INT(argv[1])) < 0) { + rb_raise(rb_eArgError, "invalid radix %d", base); + } + + if (nth >= RMATCH_NREGS(match)) return Qnil; + if (nth < 0 && (nth += RMATCH_NREGS(match)) <= 0) return Qnil; + + long start = RMATCH_BEG(match, nth), end = RMATCH_END(match, nth); + if (start < 0) return Qnil; + RUBY_ASSERT(start <= end, "%ld > %ld", start, end); + + VALUE str = RMATCH(match)->str; + RUBY_ASSERT(end <= RSTRING_LEN(str), "%ld > %ld", end, RSTRING_LEN(str)); + + char *endp; + return rb_int_parse_cstr(RSTRING_PTR(str) + start, end - start, &endp, NULL, + base, RB_INT_PARSE_DEFAULT); +} + static VALUE reg_operand(VALUE s, int check) { if (SYMBOL_P(s)) { - return rb_sym2str(s); + return rb_sym2str(s); } else if (RB_TYPE_P(s, T_STRING)) { return s; @@ -3342,31 +3781,30 @@ reg_match_pos(VALUE re, VALUE *strp, long pos, VALUE* set_match) VALUE str = *strp; if (NIL_P(str)) { - rb_backref_set(Qnil); - return -1; + rb_backref_set(Qnil); + return -1; } *strp = str = reg_operand(str, TRUE); if (pos != 0) { - if (pos < 0) { - VALUE l = rb_str_length(str); - pos += NUM2INT(l); - if (pos < 0) { - return pos; - } - } - pos = rb_str_offset(str, pos); + if (pos < 0) { + VALUE l = rb_str_length(str); + pos += NUM2INT(l); + if (pos < 0) { + return pos; + } + } + pos = rb_str_offset(str, pos); } return rb_reg_search_set_match(re, str, pos, 0, 1, set_match); } /* * call-seq: - * regexp =~ string -> integer or nil + * self =~ other -> integer or nil * * Returns the integer index (in characters) of the first match - * for +self+ and +string+, or +nil+ if none; - * also sets the - * {rdoc-ref:Regexp Global Variables}[rdoc-ref:Regexp@Regexp+Global+Variables]: + * for +self+ and +other+, or +nil+ if none; + * updates {Regexp-related global variables}[rdoc-ref:Regexp@Global+Variables]. * * /at/ =~ 'input data' # => 7 * $~ # => #<MatchData "at"> @@ -3377,9 +3815,9 @@ reg_match_pos(VALUE re, VALUE *strp, long pos, VALUE* set_match) * if and only if +self+: * * - Is a regexp literal; - * see {Regexp Literals}[rdoc-ref:literals.rdoc@Regexp+Literals]. + * see {Regexp Literals}[rdoc-ref:syntax/literals.rdoc@Regexp+Literals]. * - Does not contain interpolations; - * see {Regexp Interpolation}[rdoc-ref:Regexp@Regexp+Interpolation]. + * see {Regexp interpolation}[rdoc-ref:Regexp@Interpolation+Mode]. * - Is at the left of the expression. * * Example: @@ -3426,9 +3864,9 @@ rb_reg_match(VALUE re, VALUE str) /* * call-seq: - * regexp === string -> true or false + * self === other -> true or false * - * Returns +true+ if +self+ finds a match in +string+: + * Returns whether +self+ finds a match in +other+: * * /^[a-z]*$/ === 'HELLO' # => false * /^[A-Z]*$/ === 'HELLO' # => true @@ -3451,8 +3889,8 @@ rb_reg_eqq(VALUE re, VALUE str) str = reg_operand(str, FALSE); if (NIL_P(str)) { - rb_backref_set(Qnil); - return Qfalse; + rb_backref_set(Qnil); + return Qfalse; } start = rb_reg_search(re, str, 0, 0); return RBOOL(start >= 0); @@ -3477,13 +3915,13 @@ rb_reg_match2(VALUE re) VALUE line = rb_lastline_get(); if (!RB_TYPE_P(line, T_STRING)) { - rb_backref_set(Qnil); - return Qnil; + rb_backref_set(Qnil); + return Qnil; } start = rb_reg_search(re, line, 0, 0); if (start < 0) { - return Qnil; + return Qnil; } start = rb_str_sublen(line, start); return LONG2FIX(start); @@ -3497,13 +3935,18 @@ rb_reg_match2(VALUE re) * * With no block given, returns the MatchData object * that describes the match, if any, or +nil+ if none; - * the search begins at the given byte +offset+ in +self+: + * the search begins at the given character +offset+ in +string+: * * /abra/.match('abracadabra') # => #<MatchData "abra"> * /abra/.match('abracadabra', 4) # => #<MatchData "abra"> * /abra/.match('abracadabra', 8) # => nil * /abra/.match('abracadabra', 800) # => nil * + * string = "\u{5d0 5d1 5e8 5d0}cadabra" + * /abra/.match(string, 7) #=> #<MatchData "abra"> + * /abra/.match(string, 8) #=> nil + * /abra/.match(string.b, 8) #=> #<MatchData "abra"> + * * With a block given, calls the block if and only if a match is found; * returns the block's value: * @@ -3533,20 +3976,20 @@ rb_reg_match_m(int argc, VALUE *argv, VALUE re) long pos; if (rb_scan_args(argc, argv, "11", &str, &initpos) == 2) { - pos = NUM2LONG(initpos); + pos = NUM2LONG(initpos); } else { - pos = 0; + pos = 0; } pos = reg_match_pos(re, &str, pos, &result); if (pos < 0) { - rb_backref_set(Qnil); - return Qnil; + rb_backref_set(Qnil); + return Qnil; } rb_match_busy(result); if (!NIL_P(result) && rb_block_given_p()) { - return rb_yield(result); + return rb_yield(result); } return result; } @@ -3577,53 +4020,27 @@ rb_reg_match_m_p(int argc, VALUE *argv, VALUE re) VALUE rb_reg_match_p(VALUE re, VALUE str, long pos) { - regex_t *reg; - onig_errmsg_buffer err = ""; - OnigPosition result; - const UChar *start, *end; - int tmpreg; - if (NIL_P(str)) return Qfalse; str = SYMBOL_P(str) ? rb_sym2str(str) : StringValue(str); if (pos) { - if (pos < 0) { - pos += NUM2LONG(rb_str_length(str)); - if (pos < 0) return Qfalse; - } - if (pos > 0) { - long len = 1; - const char *beg = rb_str_subpos(str, pos, &len); - if (!beg) return Qfalse; - pos = beg - RSTRING_PTR(str); - } - } - reg = rb_reg_prepare_re0(re, str, err); - tmpreg = reg != RREGEXP_PTR(re); - if (!tmpreg) RREGEXP(re)->usecnt++; - start = ((UChar*)RSTRING_PTR(str)); - end = start + RSTRING_LEN(str); - result = onig_search(reg, start, end, start + pos, end, - NULL, ONIG_OPTION_NONE); - if (!tmpreg) RREGEXP(re)->usecnt--; - if (tmpreg) { - if (RREGEXP(re)->usecnt) { - onig_free(reg); - } - else { - onig_free(RREGEXP_PTR(re)); - RREGEXP_PTR(re) = reg; - } - } - if (result < 0) { - if (result == ONIG_MISMATCH) { - return Qfalse; - } - else { - onig_error_code_to_str((UChar*)err, (int)result); - rb_reg_raise(RREGEXP_SRC_PTR(re), RREGEXP_SRC_LEN(re), err, re); - } + if (pos < 0) { + pos += NUM2LONG(rb_str_length(str)); + if (pos < 0) return Qfalse; + } + if (pos > 0) { + long len = 1; + const char *beg = rb_str_subpos(str, pos, &len); + if (!beg) return Qfalse; + pos = beg - RSTRING_PTR(str); + } } - return Qtrue; + + struct reg_onig_search_args args = { + .pos = pos, + .range = RSTRING_LEN(str), + }; + + return rb_reg_onig_match(re, str, reg_onig_search, &args, NULL) == ONIG_MISMATCH ? Qfalse : Qtrue; } /* @@ -3632,9 +4049,71 @@ rb_reg_match_p(VALUE re, VALUE str, long pos) * Alias for Regexp.new */ +static int +str_to_option(VALUE str) +{ + int flag = 0; + const char *ptr; + long len; + str = rb_check_string_type(str); + if (NIL_P(str)) return -1; + RSTRING_GETMEM(str, ptr, len); + for (long i = 0; i < len; ++i) { + int f = char_to_option(ptr[i]); + if (!f) { + rb_raise(rb_eArgError, "unknown regexp option: %"PRIsVALUE, str); + } + flag |= f; + } + return flag; +} + +static void +set_timeout(rb_hrtime_t *hrt, VALUE timeout) +{ + double timeout_d = NIL_P(timeout) ? 0.0 : NUM2DBL(timeout); + if (!NIL_P(timeout) && !(timeout_d > 0)) { + rb_raise(rb_eArgError, "invalid timeout: %"PRIsVALUE, timeout); + } + double2hrtime(hrt, timeout_d); +} + +static VALUE +reg_copy(VALUE copy, VALUE orig) +{ + int r; + regex_t *re; + + rb_reg_initialize_check(copy); + if ((r = onig_reg_copy(&re, RREGEXP_PTR(orig))) != 0) { + /* ONIGERR_MEMORY only */ + rb_raise(rb_eRegexpError, "%s", onig_error_code_to_format(r)); + } + RREGEXP_PTR(copy) = re; + RB_OBJ_WRITE(copy, &RREGEXP(copy)->src, RREGEXP(orig)->src); + RREGEXP_PTR(copy)->timelimit = RREGEXP_PTR(orig)->timelimit; + rb_enc_copy(copy, orig); + FL_SET_RAW(copy, FL_TEST_RAW(orig, KCODE_FIXED|REG_ENCODING_NONE)); + if (RBASIC_CLASS(copy) == rb_cRegexp) { + OBJ_FREEZE(copy); + } + + return copy; +} + +struct reg_init_args { + VALUE str; + VALUE timeout; + rb_encoding *enc; + int flags; +}; + +static VALUE reg_extract_args(int argc, VALUE *argv, struct reg_init_args *args); +static VALUE reg_init_args(VALUE self, VALUE str, rb_encoding *enc, int flags); + /* * call-seq: - * Regexp.new(string, options = 0, n_flag = nil, timeout: nil) -> regexp + * Regexp.new(string, options = 0, timeout: nil) -> regexp * Regexp.new(regexp, timeout: nil) -> regexp * * With argument +string+ given, returns a new regexp with the given string @@ -3646,28 +4125,31 @@ rb_reg_match_p(VALUE re, VALUE str, long pos) * * Optional argument +options+ is one of the following: * - * - The logical OR of one or more of the constants - * Regexp::EXTENDED, Regexp::IGNORECASE, and Regexp::MULTILINE: + * - A String of options: + * + * Regexp.new('foo', 'i') # => /foo/i + * Regexp.new('foo', 'im') # => /foo/im + * + * - The bit-wise OR of one or more of the constants + * Regexp::EXTENDED, Regexp::IGNORECASE, Regexp::MULTILINE, and + * Regexp::NOENCODING: * * Regexp.new('foo', Regexp::IGNORECASE) # => /foo/i * Regexp.new('foo', Regexp::EXTENDED) # => /foo/x * Regexp.new('foo', Regexp::MULTILINE) # => /foo/m + * Regexp.new('foo', Regexp::NOENCODING) # => /foo/n * flags = Regexp::IGNORECASE | Regexp::EXTENDED | Regexp::MULTILINE * Regexp.new('foo', flags) # => /foo/mix * * - +nil+ or +false+, which is ignored. - * - * If optional argument +n_flag+ if it is a string starts with - * <code>'n'</code> or <code>'N'</code>, the encoding of +string+ is - * ignored and the new regexp encoding is fixed to +ASCII-8BIT+ or - * +US-ASCII+, by its content. - * - * Regexp.new('foo', nil, 'n') # => /foo/n - * Regexp.new("\u3042", nil, 'n') # => /\xE3\x81\x82/n + * - Any other truthy value, in which case the regexp will be + * case-insensitive. * * If optional keyword argument +timeout+ is given, * its float value overrides the timeout interval for the class, * Regexp.timeout. + * If +nil+ is passed as +timeout, it uses the timeout interval + * for the class, Regexp.timeout. * * With argument +regexp+ given, returns a new regexp. The source, * options, timeout are the same as +regexp+. +options+ and +n_flag+ @@ -3675,76 +4157,87 @@ rb_reg_match_p(VALUE re, VALUE str, long pos) * +timeout+ keyword. * * options = Regexp::MULTILINE - * r = Regexp.new('foo', optinos, timeout: 1.1) # => /foo/m + * r = Regexp.new('foo', options, timeout: 1.1) # => /foo/m * r2 = Regexp.new(r) # => /foo/m * r2.timeout # => 1.1 * r3 = Regexp.new(r, timeout: 3.14) # => /foo/m * r3.timeout # => 3.14 * - * Regexp.compile is an alias for Regexp.new. - * */ static VALUE rb_reg_initialize_m(int argc, VALUE *argv, VALUE self) { + struct reg_init_args args; + VALUE re = reg_extract_args(argc, argv, &args); + + if (NIL_P(re)) { + reg_init_args(self, args.str, args.enc, args.flags); + } + else { + reg_copy(self, re); + } + + set_timeout(&RREGEXP_PTR(self)->timelimit, args.timeout); + if (RBASIC_CLASS(self) == rb_cRegexp) { + OBJ_FREEZE(self); + } + + return self; +} + +static VALUE +reg_extract_args(int argc, VALUE *argv, struct reg_init_args *args) +{ int flags = 0; - VALUE str; rb_encoding *enc = 0; + VALUE str, src, opts = Qundef, kwargs; + VALUE re = Qnil; - VALUE src, opts = Qundef, n_flag = Qundef, kwargs, timeout = Qnil; - - rb_scan_args(argc, argv, "12:", &src, &opts, &n_flag, &kwargs); + rb_scan_args(argc, argv, "11:", &src, &opts, &kwargs); + args->timeout = Qnil; if (!NIL_P(kwargs)) { - static ID keywords[1]; - if (!keywords[0]) { - keywords[0] = rb_intern_const("timeout"); - } - rb_get_kwargs(kwargs, keywords, 0, 1, &timeout); + static ID keywords[1]; + if (!keywords[0]) { + keywords[0] = rb_intern_const("timeout"); + } + rb_get_kwargs(kwargs, keywords, 0, 1, &args->timeout); } if (RB_TYPE_P(src, T_REGEXP)) { - VALUE re = src; + re = src; - if (opts != Qnil) { - rb_warn("flags ignored"); - } - rb_reg_check(re); - flags = rb_reg_options(re); - str = RREGEXP_SRC(re); + if (!NIL_P(opts)) { + rb_warn("flags ignored"); + } + rb_reg_check(re); + flags = rb_reg_options(re); + str = RREGEXP_SRC(re); } else { - if (opts != Qundef) { - if (FIXNUM_P(opts)) flags = FIX2INT(opts); - else if (!NIL_P(opts) && rb_bool_expected(opts, "ignorecase", FALSE)) - flags = ONIG_OPTION_IGNORECASE; - } - if (n_flag != Qundef && !NIL_P(n_flag)) { - char *kcode = StringValuePtr(n_flag); - if (kcode[0] == 'n' || kcode[0] == 'N') { - enc = rb_ascii8bit_encoding(); - flags |= ARG_ENCODING_NONE; - } - else { - rb_category_warn(RB_WARN_CATEGORY_DEPRECATED, "encoding option is ignored - %s", kcode); - } - } - str = StringValue(src); + if (!NIL_P(opts)) { + int f; + if (FIXNUM_P(opts)) flags = FIX2INT(opts); + else if ((f = str_to_option(opts)) >= 0) flags = f; + else if (rb_bool_expected(opts, "ignorecase", FALSE)) + flags = ONIG_OPTION_IGNORECASE; + } + str = StringValue(src); } + args->str = str; + args->enc = enc; + args->flags = flags; + return re; +} + +static VALUE +reg_init_args(VALUE self, VALUE str, rb_encoding *enc, int flags) +{ if (enc && rb_enc_get(str) != enc) - rb_reg_init_str_enc(self, str, enc, flags); + rb_reg_init_str_enc(self, str, enc, flags); else - rb_reg_init_str(self, str, flags); - - regex_t *reg = RREGEXP_PTR(self); - - { - double limit = NIL_P(timeout) ? 0.0 : NUM2DBL(timeout); - if (limit < 0) limit = 0; - double2hrtime(®->timelimit, limit); - } - + rb_reg_init_str(self, str, flags); return self; } @@ -3761,19 +4254,19 @@ rb_reg_quote(VALUE str) send = s + RSTRING_LEN(str); while (s < send) { c = rb_enc_ascget(s, send, &clen, enc); - if (c == -1) { + if (c == -1) { s += mbclen(s, send, enc); - continue; - } - switch (c) { - case '[': case ']': case '{': case '}': - case '(': case ')': case '|': case '-': - case '*': case '.': case '\\': - case '?': case '+': case '^': case '$': - case ' ': case '#': - case '\t': case '\f': case '\v': case '\n': case '\r': - goto meta_found; - } + continue; + } + switch (c) { + case '[': case ']': case '{': case '}': + case '(': case ')': case '|': case '-': + case '*': case '.': case '\\': + case '?': case '+': case '^': case '$': + case ' ': case '#': + case '\t': case '\f': case '\v': case '\n': case '\r': + goto meta_found; + } s += clen; } tmp = rb_str_new3(str); @@ -3798,47 +4291,47 @@ rb_reg_quote(VALUE str) while (s < send) { c = rb_enc_ascget(s, send, &clen, enc); - if (c == -1) { - int n = mbclen(s, send, enc); + if (c == -1) { + int n = mbclen(s, send, enc); - while (n--) - *t++ = *s++; - continue; - } + while (n--) + *t++ = *s++; + continue; + } s += clen; - switch (c) { - case '[': case ']': case '{': case '}': - case '(': case ')': case '|': case '-': - case '*': case '.': case '\\': - case '?': case '+': case '^': case '$': - case '#': + switch (c) { + case '[': case ']': case '{': case '}': + case '(': case ')': case '|': case '-': + case '*': case '.': case '\\': + case '?': case '+': case '^': case '$': + case '#': t += rb_enc_mbcput('\\', t, enc); - break; - case ' ': + break; + case ' ': t += rb_enc_mbcput('\\', t, enc); t += rb_enc_mbcput(' ', t, enc); - continue; - case '\t': + continue; + case '\t': t += rb_enc_mbcput('\\', t, enc); t += rb_enc_mbcput('t', t, enc); - continue; - case '\n': + continue; + case '\n': t += rb_enc_mbcput('\\', t, enc); t += rb_enc_mbcput('n', t, enc); - continue; - case '\r': + continue; + case '\r': t += rb_enc_mbcput('\\', t, enc); t += rb_enc_mbcput('r', t, enc); - continue; - case '\f': + continue; + case '\f': t += rb_enc_mbcput('\\', t, enc); t += rb_enc_mbcput('f', t, enc); - continue; - case '\v': + continue; + case '\v': t += rb_enc_mbcput('\\', t, enc); t += rb_enc_mbcput('v', t, enc); - continue; - } + continue; + } t += rb_enc_mbcput(c, t, enc); } rb_str_resize(tmp, t - RSTRING_PTR(tmp)); @@ -3860,8 +4353,6 @@ rb_reg_quote(VALUE str) * r = Regexp.new(Regexp.escape(s)) # => /\\\\\\\*\\\?\\\{\\\}\\\./ * r.match(s) # => #<MatchData "\\\\\\*\\?\\{\\}\\."> * - * Regexp.quote is an alias for Regexp.escape. - * */ static VALUE @@ -3934,23 +4425,23 @@ rb_reg_s_union(VALUE self, VALUE args0) } } else { - int i; - VALUE source = rb_str_buf_new(0); - rb_encoding *result_enc; + int i; + VALUE source = rb_str_buf_new(0); + rb_encoding *result_enc; int has_asciionly = 0; rb_encoding *has_ascii_compat_fixed = 0; rb_encoding *has_ascii_incompat = 0; - for (i = 0; i < argc; i++) { - volatile VALUE v; - VALUE e = rb_ary_entry(args0, i); + for (i = 0; i < argc; i++) { + volatile VALUE v; + VALUE e = rb_ary_entry(args0, i); - if (0 < i) - rb_str_buf_cat_ascii(source, "|"); + if (0 < i) + rb_str_buf_cat_ascii(source, "|"); - v = rb_check_regexp_type(e); - if (!NIL_P(v)) { + v = rb_check_regexp_type(e); + if (!NIL_P(v)) { rb_encoding *enc = rb_enc_get(v); if (!rb_enc_asciicompat(enc)) { if (!has_ascii_incompat) @@ -3969,9 +4460,9 @@ rb_reg_s_union(VALUE self, VALUE args0) else { has_asciionly = 1; } - v = rb_reg_str_with_term(v, -1); - } - else { + v = rb_reg_str_with_term(v, -1); + } + else { rb_encoding *enc; StringValue(e); enc = rb_enc_get(e); @@ -3992,8 +4483,8 @@ rb_reg_s_union(VALUE self, VALUE args0) rb_raise(rb_eArgError, "incompatible encodings: %s and %s", rb_enc_name(has_ascii_compat_fixed), rb_enc_name(enc)); } - v = rb_reg_s_quote(Qnil, e); - } + v = rb_reg_s_quote(Qnil, e); + } if (has_ascii_incompat) { if (has_asciionly) { rb_raise(rb_eArgError, "ASCII incompatible encoding: %s", @@ -4008,8 +4499,8 @@ rb_reg_s_union(VALUE self, VALUE args0) if (i == 0) { rb_enc_copy(source, v); } - rb_str_append(source, v); - } + rb_str_append(source, v); + } if (has_ascii_incompat) { result_enc = has_ascii_incompat; @@ -4071,17 +4562,51 @@ rb_reg_s_union_m(VALUE self, VALUE args) return rb_reg_s_union(self, args); } +/* + * call-seq: + * Regexp.linear_time?(re) + * Regexp.linear_time?(string, options = 0) + * + * Returns +true+ if matching against <tt>re</tt> can be + * done in linear time to the input string. + * + * Regexp.linear_time?(/re/) # => true + * + * Note that this is a property of the ruby interpreter, not of the argument + * regular expression. Identical regexp can or cannot run in linear time + * depending on your ruby binary. Neither forward nor backward compatibility + * is guaranteed about the return value of this method. Our current algorithm + * is (*1) but this is subject to change in the future. Alternative + * implementations can also behave differently. They might always return + * false for everything. + * + * (*1): https://doi.org/10.1109/SP40001.2021.00032 + * + */ +static VALUE +rb_reg_s_linear_time_p(int argc, VALUE *argv, VALUE self) +{ + struct reg_init_args args; + VALUE re = reg_extract_args(argc, argv, &args); + + if (NIL_P(re)) { + re = reg_init_args(rb_reg_alloc(), args.str, args.enc, args.flags); + } + + return RBOOL(onig_check_linear_time(RREGEXP_PTR(re))); +} + /* :nodoc: */ static VALUE rb_reg_init_copy(VALUE copy, VALUE re) { if (!OBJ_INIT_COPY(copy, re)) return copy; rb_reg_check(re); - return rb_reg_init_str(copy, RREGEXP_SRC(re), rb_reg_options(re)); + return reg_copy(copy, re); } -VALUE -rb_reg_regsub(VALUE str, VALUE src, struct re_registers *regs, VALUE regexp) +static VALUE +do_regsub(VALUE str, VALUE src, VALUE regexp, int num_regs, const OnigPosition *beg, const OnigPosition *end) { VALUE val = 0; char *p, *s, *e; @@ -4098,42 +4623,42 @@ rb_reg_regsub(VALUE str, VALUE src, struct re_registers *regs, VALUE regexp) while (s < e) { int c = ASCGET(s, e, &clen); - char *ss; + char *ss; - if (c == -1) { - s += mbclen(s, e, str_enc); - continue; - } - ss = s; + if (c == -1) { + s += mbclen(s, e, str_enc); + continue; + } + ss = s; s += clen; - if (c != '\\' || s == e) continue; + if (c != '\\' || s == e) continue; - if (!val) { - val = rb_str_buf_new(ss-p); - } + if (!val) { + val = rb_str_buf_new(ss-p); + } rb_enc_str_buf_cat(val, p, ss-p, str_enc); c = ASCGET(s, e, &clen); if (c == -1) { s += mbclen(s, e, str_enc); - rb_enc_str_buf_cat(val, ss, s-ss, str_enc); + rb_enc_str_buf_cat(val, ss, s-ss, str_enc); p = s; - continue; + continue; } s += clen; - p = s; - switch (c) { - case '1': case '2': case '3': case '4': - case '5': case '6': case '7': case '8': case '9': + p = s; + switch (c) { + case '1': case '2': case '3': case '4': + case '5': case '6': case '7': case '8': case '9': if (!NIL_P(regexp) && onig_noname_group_capture_is_active(RREGEXP_PTR(regexp))) { no = c - '0'; } else { continue; } - break; + break; case 'k': if (s < e && ASCGET(s, e, &clen) == '<') { @@ -4146,11 +4671,17 @@ rb_reg_regsub(VALUE str, VALUE src, struct re_registers *regs, VALUE regexp) name_end += c == -1 ? mbclen(name_end, e, str_enc) : clen; } if (name_end < e) { - VALUE n = rb_str_subseq(str, (long)(name - RSTRING_PTR(str)), - (long)(name_end - name)); - if ((no = NAME_TO_NUMBER(regs, regexp, n, name, name_end)) < 1) { - name_to_backref_error(n); - } + VALUE n = rb_str_subseq(str, (long)(name - RSTRING_PTR(str)), + (long)(name_end - name)); + struct re_registers tmp = { + .allocated = num_regs, + .num_regs = num_regs, + .beg = (OnigPosition *)beg, + .end = (OnigPosition *)end, + }; + if ((no = NAME_TO_NUMBER(&tmp, regexp, n, name, name_end)) < 1) { + name_to_backref_error(n); + } p = s = name_end + clen; break; } @@ -4163,38 +4694,38 @@ rb_reg_regsub(VALUE str, VALUE src, struct re_registers *regs, VALUE regexp) continue; case '0': - case '&': - no = 0; - break; + case '&': + no = 0; + break; - case '`': - rb_enc_str_buf_cat(val, RSTRING_PTR(src), BEG(0), src_enc); - continue; + case '`': + rb_enc_str_buf_cat(val, RSTRING_PTR(src), beg[0], src_enc); + continue; - case '\'': - rb_enc_str_buf_cat(val, RSTRING_PTR(src)+END(0), RSTRING_LEN(src)-END(0), src_enc); - continue; + case '\'': + rb_enc_str_buf_cat(val, RSTRING_PTR(src)+end[0], RSTRING_LEN(src)-end[0], src_enc); + continue; - case '+': - no = regs->num_regs-1; - while (BEG(no) == -1 && no > 0) no--; - if (no == 0) continue; - break; + case '+': + no = num_regs-1; + while (beg[no] == -1 && no > 0) no--; + if (no == 0) continue; + break; - case '\\': - rb_enc_str_buf_cat(val, s-clen, clen, str_enc); - continue; + case '\\': + rb_enc_str_buf_cat(val, s-clen, clen, str_enc); + continue; - default: - rb_enc_str_buf_cat(val, ss, s-ss, str_enc); - continue; - } + default: + rb_enc_str_buf_cat(val, ss, s-ss, str_enc); + continue; + } - if (no >= 0) { - if (no >= regs->num_regs) continue; - if (BEG(no) == -1) continue; - rb_enc_str_buf_cat(val, RSTRING_PTR(src)+BEG(no), END(no)-BEG(no), src_enc); - } + if (no >= 0) { + if (no >= num_regs) continue; + if (beg[no] == -1) continue; + rb_enc_str_buf_cat(val, RSTRING_PTR(src)+beg[no], end[no]-beg[no], src_enc); + } } if (!val) return str; @@ -4203,6 +4734,20 @@ rb_reg_regsub(VALUE str, VALUE src, struct re_registers *regs, VALUE regexp) } return val; +#undef ASCGET +} + +VALUE +rb_reg_regsub(VALUE str, VALUE src, struct re_registers *regs, VALUE regexp) +{ + return do_regsub(str, src, regexp, regs->num_regs, regs->beg, regs->end); +} + +VALUE +rb_reg_regsub_match(VALUE str, VALUE src, VALUE match) +{ + return do_regsub(str, src, RMATCH(match)->regexp, + RMATCH_NREGS(match), RMATCH_BEG_PTR(match), RMATCH_END_PTR(match)); } static VALUE @@ -4238,7 +4783,7 @@ static void match_setter(VALUE val, ID _x, VALUE *_y) { if (!NIL_P(val)) { - Check_Type(val, T_MATCH); + Check_Type(val, T_MATCH); } rb_backref_set(val); } @@ -4249,9 +4794,9 @@ match_setter(VALUE val, ID _x, VALUE *_y) * Regexp.last_match(n) -> string or nil * Regexp.last_match(name) -> string or nil * - * With no argument, returns the value of <tt>$!</tt>, + * With no argument, returns the value of <tt>$~</tt>, * which is the result of the most recent pattern match - * (see {Regexp Global Variables}[rdoc-ref:Regexp@Regexp+Global+Variables]): + * (see {Regexp global variables}[rdoc-ref:Regexp@Global+Variables]): * * /c(.)t/ =~ 'cat' # => 0 * Regexp.last_match # => #<MatchData "cat" 1:"a"> @@ -4289,7 +4834,7 @@ rb_reg_s_last_match(int argc, VALUE *argv, VALUE _) int n; if (NIL_P(match)) return Qnil; n = match_backref_number(match, argv[0]); - return rb_reg_nth_match(n, match); + return rb_reg_nth_match(n, match); } return match_getter(); } @@ -4300,12 +4845,9 @@ re_warn(const char *s) rb_warn("%s", s); } -// The process-global timeout for regexp matching -rb_hrtime_t rb_reg_match_time_limit = 0; - // This function is periodically called during regexp matching -void -rb_reg_check_timeout(regex_t *reg, void *end_time_) +bool +rb_reg_timeout_p(regex_t *reg, void *end_time_) { rb_hrtime_t *end_time = (rb_hrtime_t *)end_time_; @@ -4330,10 +4872,12 @@ rb_reg_check_timeout(regex_t *reg, void *end_time_) } else { if (*end_time < rb_hrtime_now()) { - // timeout is exceeded - rb_raise(rb_eRegexpTimeoutError, "regexp match timeout"); + // Timeout has exceeded + return true; } } + + return false; } /* @@ -4366,16 +4910,13 @@ rb_reg_s_timeout_get(VALUE dummy) */ static VALUE -rb_reg_s_timeout_set(VALUE dummy, VALUE limit) +rb_reg_s_timeout_set(VALUE dummy, VALUE timeout) { - double timeout = NIL_P(limit) ? 0.0 : NUM2DBL(limit); - rb_ractor_ensure_main_ractor("can not access Regexp.timeout from non-main Ractors"); - if (timeout < 0) timeout = 0; - double2hrtime(&rb_reg_match_time_limit, timeout); + set_timeout(&rb_reg_match_time_limit, timeout); - return limit; + return timeout; } /* @@ -4417,7 +4958,7 @@ rb_reg_timeout_get(VALUE re) /* * Document-class: Regexp * - * :include: doc/regexp.rdoc + * :include: doc/_regexp.rdoc */ void @@ -4440,17 +4981,23 @@ Init_Regexp(void) rb_gvar_ractor_local("$`"); rb_gvar_ractor_local("$'"); rb_gvar_ractor_local("$+"); + rb_gvar_box_dynamic("$~"); + rb_gvar_box_ready("$&"); + rb_gvar_box_ready("$`"); + rb_gvar_box_ready("$'"); + rb_gvar_box_ready("$+"); rb_define_virtual_variable("$=", ignorecase_getter, ignorecase_setter); rb_cRegexp = rb_define_class("Regexp", rb_cObject); rb_define_alloc_func(rb_cRegexp, rb_reg_s_alloc); - rb_define_singleton_method(rb_cRegexp, "compile", rb_class_new_instance, -1); + rb_define_singleton_method(rb_cRegexp, "compile", rb_class_new_instance_pass_kw, -1); rb_define_singleton_method(rb_cRegexp, "quote", rb_reg_s_quote, 1); rb_define_singleton_method(rb_cRegexp, "escape", rb_reg_s_quote, 1); rb_define_singleton_method(rb_cRegexp, "union", rb_reg_s_union_m, -2); rb_define_singleton_method(rb_cRegexp, "last_match", rb_reg_s_last_match, -1); rb_define_singleton_method(rb_cRegexp, "try_convert", rb_reg_s_try_convert, 1); + rb_define_singleton_method(rb_cRegexp, "linear_time?", rb_reg_s_linear_time_p, -1); rb_define_method(rb_cRegexp, "initialize", rb_reg_initialize_m, -1); rb_define_method(rb_cRegexp, "initialize_copy", rb_reg_init_copy, 1); @@ -4473,6 +5020,7 @@ Init_Regexp(void) rb_define_method(rb_cRegexp, "named_captures", rb_reg_named_captures, 0); rb_define_method(rb_cRegexp, "timeout", rb_reg_timeout_get, 0); + /* Raised when regexp matching timed out. */ rb_eRegexpTimeoutError = rb_define_class_under(rb_cRegexp, "TimeoutError", rb_eRegexpError); rb_define_singleton_method(rb_cRegexp, "timeout", rb_reg_s_timeout_get, 0); rb_define_singleton_method(rb_cRegexp, "timeout=", rb_reg_s_timeout_set, 1); @@ -4502,6 +5050,8 @@ Init_Regexp(void) rb_define_method(rb_cMatch, "length", match_size, 0); rb_define_method(rb_cMatch, "offset", match_offset, 1); rb_define_method(rb_cMatch, "byteoffset", match_byteoffset, 1); + rb_define_method(rb_cMatch, "bytebegin", match_bytebegin, 1); + rb_define_method(rb_cMatch, "byteend", match_byteend, 1); rb_define_method(rb_cMatch, "begin", match_begin, 1); rb_define_method(rb_cMatch, "end", match_end, 1); rb_define_method(rb_cMatch, "match", match_nth, 1); @@ -4509,7 +5059,9 @@ Init_Regexp(void) rb_define_method(rb_cMatch, "to_a", match_to_a, 0); rb_define_method(rb_cMatch, "[]", match_aref, -1); rb_define_method(rb_cMatch, "captures", match_captures, 0); - rb_define_method(rb_cMatch, "named_captures", match_named_captures, 0); + rb_define_alias(rb_cMatch, "deconstruct", "captures"); + rb_define_method(rb_cMatch, "named_captures", match_named_captures, -1); + rb_define_method(rb_cMatch, "deconstruct_keys", match_deconstruct_keys, 1); rb_define_method(rb_cMatch, "values_at", match_values_at, -1); rb_define_method(rb_cMatch, "pre_match", rb_reg_match_pre, 0); rb_define_method(rb_cMatch, "post_match", rb_reg_match_post, 0); @@ -4519,4 +5071,5 @@ Init_Regexp(void) rb_define_method(rb_cMatch, "hash", match_hash, 0); rb_define_method(rb_cMatch, "eql?", match_equal, 1); rb_define_method(rb_cMatch, "==", match_equal, 1); + rb_define_method(rb_cMatch, "integer_at", match_integer_at, -1); } |