diff options
Diffstat (limited to 're.c')
| -rw-r--r-- | re.c | 5610 |
1 files changed, 4416 insertions, 1194 deletions
@@ -5,18 +5,37 @@ $Author$ created at: Mon Aug 9 18:24:49 JST 1993 - Copyright (C) 1993-2003 Yukihiro Matsumoto + Copyright (C) 1993-2007 Yukihiro Matsumoto **********************************************************************/ -#include "ruby.h" -#include "re.h" -#include <ctype.h> +#include "ruby/internal/config.h" -static VALUE rb_eRegexpError; +#include <ctype.h> -#define BEG(no) regs->beg[no] -#define END(no) regs->end[no] +#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" +#include "internal/string.h" +#include "internal/object.h" +#include "internal/ractor.h" +#include "internal/variable.h" +#include "regint.h" +#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) #if 'a' == 97 /* it's ascii */ static const char casetable[] = { @@ -69,477 +88,810 @@ 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(p1, p2, len) - char *p1, *p2; - long len; +rb_memcicmp(const void *x, const void *y, long len) { + const unsigned char *p1 = x, *p2 = y; 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; } -int -rb_memcmp(p1, p2, len) - char *p1, *p2; - long len; +#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; + else + return -1; +} +#else +static inline long +rb_memsearch_ss(const unsigned char *xs, long m, const unsigned char *ys, long n) { - if (!ruby_ignorecase) { - return memcmp(p1, p2, len); + const unsigned char *x = xs, *xe = xs + m; + const unsigned char *y = ys, *ye = ys + n; +#define VALUE_MAX ((VALUE)~(VALUE)0) + VALUE hx, hy, mask = VALUE_MAX >> ((SIZEOF_VALUE - m) * CHAR_BIT); + + if (m > SIZEOF_VALUE) + rb_bug("!!too long pattern string!!"); + + if (!(y = memchr(y, *x, n - m + 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; + } + /* Searching */ + while (hx != hy) { + if (y == ye) + return -1; + hy <<= CHAR_BIT; + hy |= *y; + hy &= mask; + y++; } - return rb_memcicmp(p1, p2, len); + return y - ys - m; } +#endif -long -rb_memsearch(x0, m, y0, n) - char *x0, *y0; - long m, n; +static inline long +rb_memsearch_qs(const unsigned char *xs, long m, const unsigned char *ys, long n) { - unsigned char *x = (unsigned char *)x0, *y = (unsigned char *)y0; - unsigned char *s, *e; - long i; - int d; - unsigned long hx, hy; - -#define KR_REHASH(a, b, h) (((h) << 1) - ((a)<<d) + (b)) - - if (m > n) return -1; - s = y; e = s + n - m; + const unsigned char *x = xs, *xe = xs + m; + const unsigned char *y = ys; + VALUE i, qstable[256]; /* Preprocessing */ - /* computes d = 2^(m-1) with - the left-shift operator */ - d = sizeof(hx) * CHAR_BIT - 1; - if (d > m) d = m; - - if (ruby_ignorecase) { - if (n == m) { - return rb_memcicmp(x, s, m) == 0 ? 0 : -1; - } - /* Prepare hash value */ - for (hy = hx = i = 0; i < d; ++i) { - hx = KR_REHASH(0, casetable[x[i]], hx); - hy = KR_REHASH(0, casetable[s[i]], hy); - } - /* Searching */ - while (hx != hy || rb_memcicmp(x, s, m)) { - if (s >= e) return -1; - hy = KR_REHASH(casetable[*s], casetable[*(s+d)], hy); - s++; - } + for (i = 0; i < 256; ++i) + qstable[i] = m + 1; + for (; 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; + } + return -1; +} + +static inline unsigned int +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; + } + else if (h < 0xE0) { + h *= mix; + h += x[1]; + } + else if (h < 0xF0) { + 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]; } else { - if (n == m) { - return memcmp(x, s, m) == 0 ? 0 : -1; - } - /* Prepare hash value */ - for (hy = hx = i = 0; i < d; ++i) { - hx = KR_REHASH(0, x[i], hx); - hy = KR_REHASH(0, s[i], hy); - } - /* Searching */ - while (hx != hy || memcmp(x, s, m)) { - if (s >= e) return -1; - hy = KR_REHASH(*s, *(s+d), hy); - s++; - } - } - return s-y; -} - -#define REG_CASESTATE FL_USER0 -#define KCODE_NONE 0 -#define KCODE_EUC FL_USER1 -#define KCODE_SJIS FL_USER2 -#define KCODE_UTF8 FL_USER3 -#define KCODE_FIXED FL_USER4 -#define KCODE_MASK (KCODE_EUC|KCODE_SJIS|KCODE_UTF8) + return h + 256; + } + return (unsigned char)h; +} -static int reg_kcode = DEFAULT_KCODE; +static inline long +rb_memsearch_qs_utf8(const unsigned char *xs, long m, const unsigned char *ys, long n) +{ + const unsigned char *x = xs, *xe = xs + m; + const unsigned char *y = ys; + VALUE i, qstable[512]; -static void -kcode_euc(re) - struct RRegexp *re; + /* Preprocessing */ + for (i = 0; i < 512; ++i) { + qstable[i] = m + 1; + } + for (; 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; + } + return -1; +} + +static inline long +rb_memsearch_with_char_size(const unsigned char *xs, long m, const unsigned char *ys, long n, int char_size) { - FL_UNSET(re, KCODE_MASK); - FL_SET(re, KCODE_EUC); - FL_SET(re, KCODE_FIXED); + const unsigned char *x = xs, x0 = *xs, *y = ys; + + 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 void -kcode_sjis(re) - struct RRegexp *re; +static inline long +rb_memsearch_wchar(const unsigned char *xs, long m, const unsigned char *ys, long n) { - FL_UNSET(re, KCODE_MASK); - FL_SET(re, KCODE_SJIS); - FL_SET(re, KCODE_FIXED); + return rb_memsearch_with_char_size(xs, m, ys, n, 2); } -static void -kcode_utf8(re) - struct RRegexp *re; +static inline long +rb_memsearch_qchar(const unsigned char *xs, long m, const unsigned char *ys, long n) { - FL_UNSET(re, KCODE_MASK); - FL_SET(re, KCODE_UTF8); - FL_SET(re, KCODE_FIXED); + return rb_memsearch_with_char_size(xs, m, ys, n, 4); } -static void -kcode_none(re) - struct RRegexp *re; +long +rb_memsearch(const void *x0, long m, const void *y0, long n, rb_encoding *enc) { - FL_UNSET(re, KCODE_MASK); - FL_SET(re, KCODE_FIXED); + const unsigned char *x = x0, *y = y0; + + if (m > n) return -1; + else if (m == n) { + return memcmp(x0, y0, m) == 0 ? 0 : -1; + } + else if (m < 1) { + return 0; + } + else if (m == 1) { + const unsigned char *ys = memchr(y, *x, n); + + 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); + } + } + else if (LIKELY(rb_enc_mbminlen(enc) == 2)) { + 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_qs(x0, m, y0, n); } -static int curr_kcode; +#define REG_ENCODING_NONE FL_USER6 -static void -kcode_set_option(re) - VALUE re; -{ - if (!FL_TEST(re, KCODE_FIXED)) return; - - curr_kcode = RBASIC(re)->flags & KCODE_MASK; - if (reg_kcode == curr_kcode) return; - switch (curr_kcode) { - case KCODE_NONE: - re_mbcinit(MBCTYPE_ASCII); - break; - case KCODE_EUC: - re_mbcinit(MBCTYPE_EUC); - break; - case KCODE_SJIS: - re_mbcinit(MBCTYPE_SJIS); - break; - case KCODE_UTF8: - re_mbcinit(MBCTYPE_UTF8); - break; - } -} +#define KCODE_FIXED FL_USER4 -static void -kcode_reset_option() +static int +char_to_option(int c) { - if (reg_kcode == curr_kcode) return; - switch (reg_kcode) { - case KCODE_NONE: - re_mbcinit(MBCTYPE_ASCII); - break; - case KCODE_EUC: - re_mbcinit(MBCTYPE_EUC); - break; - case KCODE_SJIS: - re_mbcinit(MBCTYPE_SJIS); - break; - case KCODE_UTF8: - re_mbcinit(MBCTYPE_UTF8); - break; + int val; + + switch (c) { + case 'i': + val = ONIG_OPTION_IGNORECASE; + break; + case 'x': + val = ONIG_OPTION_EXTEND; + break; + case 'm': + val = ONIG_OPTION_MULTILINE; + break; + default: + val = 0; + break; } + return val; } -int -rb_reg_mbclen2(c, re) - unsigned int c; - VALUE re; +enum { OPTBUF_SIZE = 4 }; + +static char * +option_to_str(char str[OPTBUF_SIZE], int options) +{ + char *p = str; + if (options & ONIG_OPTION_MULTILINE) *p++ = 'm'; + if (options & ONIG_OPTION_IGNORECASE) *p++ = 'i'; + if (options & ONIG_OPTION_EXTEND) *p++ = 'x'; + *p = 0; + return str; +} + +extern int +rb_char_to_option_kcode(int c, int *option, int *kcode) { - int len; + *option = 0; - if (!FL_TEST(re, KCODE_FIXED)) - return mbclen(c); - kcode_set_option(re); - len = mbclen(c); - kcode_reset_option(); - return len; + switch (c) { + case 'n': + *kcode = rb_ascii8bit_encindex(); + return (*option = ARG_ENCODING_NONE); + case 'e': + *kcode = ENCINDEX_EUC_JP; + break; + case 's': + *kcode = ENCINDEX_Windows_31J; + break; + case 'u': + *kcode = rb_utf8_encindex(); + break; + default: + *kcode = -1; + return (*option = char_to_option(c)); + } + *option = ARG_ENCODING_FIXED; + return 1; } static void -rb_reg_check(re) - VALUE re; +rb_reg_check(VALUE re) { - if (!RREGEXP(re)->ptr || !RREGEXP(re)->str) { - rb_raise(rb_eTypeError, "uninitialized Regexp"); + if (!RREGEXP_PTR(re) || !RREGEXP_SRC(re) || !RREGEXP_SRC_PTR(re)) { + rb_raise(rb_eTypeError, "uninitialized Regexp"); } } -extern int ruby_in_compile; - static void -rb_reg_expr_str(str, s, len) - VALUE str; - const char *s; - long len; +rb_reg_expr_str(VALUE str, const char *s, long len, + rb_encoding *enc, rb_encoding *resenc, int term) { const char *p, *pend; + int cr = ENC_CODERANGE_UNKNOWN; int need_escape = 0; + int c, clen; p = s; pend = p + len; - while (p<pend) { - if (*p == '/' || (!ISPRINT(*p) && !ismbchar(*p))) { - need_escape = 1; - break; - } - p += mbclen(*p); + 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; + } + } + } + else { + need_escape = 1; } + if (!need_escape) { - rb_str_buf_cat(str, s, len); + rb_str_buf_cat(str, s, len); } else { - p = s; - while (p<pend) { - if (*p == '\\') { - int n = mbclen(p[1]) + 1; - rb_str_buf_cat(str, p, n); - p += n; - continue; - } - else if (*p == '/') { - char c = '\\'; - rb_str_buf_cat(str, &c, 1); - rb_str_buf_cat(str, p, 1); - } - else if (ismbchar(*p)) { - rb_str_buf_cat(str, p, mbclen(*p)); - p += mbclen(*p); - continue; - } - else if (ISPRINT(*p)) { - rb_str_buf_cat(str, p, 1); - } - else if (!ISSPACE(*p)) { - char b[8]; - - sprintf(b, "\\%03o", *p & 0377); - rb_str_buf_cat(str, b, 4); - } - else { - rb_str_buf_cat(str, p, 1); - } - p++; - } - } -} - -static VALUE -rb_reg_desc(s, len, re) - const char *s; - long len; - VALUE re; + 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; + } + } +} + +static VALUE +rb_reg_desc(VALUE re) { + rb_encoding *enc = rb_enc_get(re); VALUE str = rb_str_buf_new2("/"); + rb_encoding *resenc = rb_default_internal_encoding(); + if (resenc == NULL) resenc = rb_default_external_encoding(); + + if (re && rb_enc_asciicompat(enc)) { + rb_enc_copy(str, re); + } + else { + rb_enc_associate(str, rb_usascii_encoding()); + } + + 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_reg_expr_str(str, s, len); rb_str_buf_cat2(str, "/"); if (re) { - rb_reg_check(re); - if (RREGEXP(re)->ptr->options & RE_OPTION_MULTILINE) - rb_str_buf_cat2(str, "m"); - if (RREGEXP(re)->ptr->options & RE_OPTION_IGNORECASE) - rb_str_buf_cat2(str, "i"); - if (RREGEXP(re)->ptr->options & RE_OPTION_EXTENDED) - rb_str_buf_cat2(str, "x"); - - if (FL_TEST(re, KCODE_FIXED)) { - switch ((RBASIC(re)->flags & KCODE_MASK)) { - case KCODE_NONE: - rb_str_buf_cat2(str, "n"); - break; - case KCODE_EUC: - rb_str_buf_cat2(str, "e"); - break; - case KCODE_SJIS: - rb_str_buf_cat2(str, "s"); - break; - case KCODE_UTF8: - rb_str_buf_cat2(str, "u"); - break; - } - } - } - OBJ_INFECT(str, 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"); + } return str; } + +/* + * call-seq: + * source -> string + * + * Returns the original string of +self+: + * + * /ab+c/ix.source # => "ab+c" + * + * Regexp escape sequences are retained: + * + * /\x20\+/.source # => "\\x20\\+" + * + * Lexer escape characters are not retained: + * + * /\//.source # => "/" + * + */ + static VALUE -rb_reg_source(re) - VALUE re; +rb_reg_source(VALUE re) { VALUE str; rb_reg_check(re); - str = rb_str_new(RREGEXP(re)->str,RREGEXP(re)->len); - if (OBJ_TAINTED(re)) OBJ_TAINT(str); + str = rb_str_dup(RREGEXP_SRC(re)); return str; } +/* + * call-seq: + * inspect -> string + * + * Returns a nicely-formatted string representation of +self+: + * + * /ab+c/ix.inspect # => "/ab+c/ix" + * + * Related: Regexp#to_s. + */ + static VALUE -rb_reg_inspect(re) - VALUE re; +rb_reg_inspect(VALUE re) { - rb_reg_check(re); - return rb_reg_desc(RREGEXP(re)->str, RREGEXP(re)->len, re); + if (!RREGEXP_PTR(re) || !RREGEXP_SRC(re) || !RREGEXP_SRC_PTR(re)) { + return rb_any_to_s(re); + } + return rb_reg_desc(re); } +static VALUE rb_reg_str_with_term(VALUE re, int term); + +/* + * call-seq: + * to_s -> string + * + * Returns a string showing the options and string of +self+: + * + * r0 = /ab+c/ix + * s0 = r0.to_s # => "(?ix-m:ab+c)" + * + * The returned string may be used as an argument to Regexp.new, + * or as interpolated text for a + * {Regexp interpolation}[rdoc-ref:Regexp@Interpolation+Mode]: + * + * r1 = Regexp.new(s0) # => /(?ix-m:ab+c)/ + * r2 = /#{s0}/ # => /(?ix-m:ab+c)/ + * + * Note that +r1+ and +r2+ are not equal to +r0+ + * because their original strings are different: + * + * r0 == r1 # => false + * r0.source # => "ab+c" + * r1.source # => "(?ix-m:ab+c)" + * + * Related: Regexp#inspect. + * + */ + static VALUE -rb_reg_to_s(re) - VALUE re; +rb_reg_to_s(VALUE re) { - int options; - const int embeddable = RE_OPTION_MULTILINE|RE_OPTION_IGNORECASE|RE_OPTION_EXTENDED; - long len; - const char* ptr; + return rb_reg_str_with_term(re, '/'); +} + +static VALUE +rb_reg_str_with_term(VALUE re, int term) +{ + int options, opt; + const int embeddable = ONIG_OPTION_MULTILINE|ONIG_OPTION_IGNORECASE|ONIG_OPTION_EXTEND; VALUE str = rb_str_buf_new2("(?"); + char optbuf[OPTBUF_SIZE + 1]; /* for '-' */ + rb_encoding *enc = rb_enc_get(re); rb_reg_check(re); - options = RREGEXP(re)->ptr->options; - ptr = RREGEXP(re)->str; - len = RREGEXP(re)->len; + rb_enc_copy(str, re); + options = RREGEXP_PTR(re)->options; + 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 { - if (*ptr == 'm') { - options |= RE_OPTION_MULTILINE; - } - else if (*ptr == 'i') { - options |= RE_OPTION_IGNORECASE; - } - else if (*ptr == 'x') { - options |= RE_OPTION_EXTENDED; - } - else break; - ++ptr; - } while (--len > 0); - } - if (len > 1 && *ptr == '-') { - ++ptr; - --len; - do { - if (*ptr == 'm') { - options &= ~RE_OPTION_MULTILINE; - } - else if (*ptr == 'i') { - options &= ~RE_OPTION_IGNORECASE; - } - else if (*ptr == 'x') { - options &= ~RE_OPTION_EXTENDED; - } - else break; - ++ptr; - } while (--len > 0); - } - if (*ptr == ')') { - --len; - ++ptr; - goto again; - } - if (*ptr == ':' && ptr[len-1] == ')') { - Regexp *rp; - kcode_set_option(re); - rp = ALLOC(Regexp); - MEMZERO((char *)rp, Regexp, 1); - err = re_compile_pattern(++ptr, len -= 2, rp) != 0; - kcode_reset_option(); - re_free_pattern(rp); - } - if (err) { - options = RREGEXP(re)->ptr->options; - ptr = RREGEXP(re)->str; - len = RREGEXP(re)->len; - } - } - - if (options & RE_OPTION_MULTILINE) rb_str_buf_cat2(str, "m"); - if (options & RE_OPTION_IGNORECASE) rb_str_buf_cat2(str, "i"); - if (options & RE_OPTION_EXTENDED) rb_str_buf_cat2(str, "x"); + int err = 1; + ptr += 2; + if ((len -= 2) > 0) { + do { + opt = char_to_option((int )*ptr); + if (opt != 0) { + options |= opt; + } + else { + break; + } + ++ptr; + } while (--len > 0); + } + if (len > 1 && *ptr == '-') { + ++ptr; + --len; + do { + opt = char_to_option((int )*ptr); + if (opt != 0) { + options &= ~opt; + } + 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; + 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); + } + } + + if (*option_to_str(optbuf, options)) rb_str_buf_cat2(str, optbuf); if ((options & embeddable) != embeddable) { - rb_str_buf_cat2(str, "-"); - if (!(options & RE_OPTION_MULTILINE)) rb_str_buf_cat2(str, "m"); - if (!(options & RE_OPTION_IGNORECASE)) rb_str_buf_cat2(str, "i"); - if (!(options & RE_OPTION_EXTENDED)) rb_str_buf_cat2(str, "x"); + optbuf[0] = '-'; + option_to_str(optbuf + 1, ~options); + rb_str_buf_cat2(str, optbuf); } rb_str_buf_cat2(str, ":"); - rb_reg_expr_str(str, ptr, len); - 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, ")"); + } + 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); + } + rb_enc_copy(str, re); + + RB_GC_GUARD(src_str); - OBJ_INFECT(str, re); return str; } +NORETURN(static void rb_reg_raise(const char *err, VALUE re)); + static void -rb_reg_raise(s, len, err, re) - 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); - if (ruby_in_compile) - rb_compile_error("%s: %s", err, RSTRING(desc)->ptr); - else - rb_raise(rb_eRegexpError, "%s: %s", err, RSTRING(desc)->ptr); + rb_raise(rb_eRegexpError, "%s: %"PRIsVALUE, err, desc); } static VALUE -rb_reg_casefold_p(re) - VALUE re; +rb_enc_reg_error_desc(const char *s, long len, rb_encoding *enc, int options, const char *err) +{ + char opts[OPTBUF_SIZE + 1]; /* for '/' */ + VALUE desc = rb_str_buf_new2(err); + rb_encoding *resenc = rb_default_internal_encoding(); + if (resenc == NULL) resenc = rb_default_external_encoding(); + + rb_enc_associate(desc, enc); + rb_str_buf_cat2(desc, ": /"); + rb_reg_expr_str(desc, s, len, enc, resenc, '/'); + opts[0] = '/'; + option_to_str(opts + 1, options); + rb_str_buf_cat2(desc, opts); + return rb_exc_new3(rb_eRegexpError, desc); +} + +NORETURN(static void rb_enc_reg_raise(const char *s, long len, rb_encoding *enc, int options, const char *err)); + +static void +rb_enc_reg_raise(const char *s, long len, rb_encoding *enc, int options, const char *err) +{ + rb_exc_raise(rb_enc_reg_error_desc(s, len, enc, options, err)); +} + +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); +} + +NORETURN(static void rb_reg_raise_str(VALUE str, int options, const char *err)); + +static void +rb_reg_raise_str(VALUE str, int options, const char *err) +{ + rb_exc_raise(rb_reg_error_desc(str, options, err)); +} + + +/* + * call-seq: + * casefold?-> true or false + * + * Returns +true+ if the case-insensitivity flag in +self+ is set, + * +false+ otherwise: + * + * /a/.casefold? # => false + * /a/i.casefold? # => true + * /(?i:a)/.casefold? # => false + * + */ + +static VALUE +rb_reg_casefold_p(VALUE re) { rb_reg_check(re); - if (RREGEXP(re)->ptr->options & RE_OPTION_IGNORECASE) return Qtrue; - return Qfalse; + return RBOOL(RREGEXP_PTR(re)->options & ONIG_OPTION_IGNORECASE); } + +/* + * call-seq: + * options -> integer + * + * Returns an integer whose bits show the options set in +self+. + * + * The option bits are: + * + * Regexp::IGNORECASE # => 1 + * Regexp::EXTENDED # => 2 + * Regexp::MULTILINE # => 4 + * + * Examples: + * + * /foo/.options # => 0 + * /foo/i.options # => 1 + * /foo/x.options # => 2 + * /foo/m.options # => 4 + * /foo/mix.options # => 7 + * + * Note that additional bits may be set in the returned integer; + * 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 + * additional bits may be set in the returned options: these are used + * internally by the regular expression code. These extra bits are + * ignored if the options are passed to Regexp::new: + * + * r = /\xa1\xa2/e # => /\xa1\xa2/ + * r.source # => "\\xa1\\xa2" + * r.options # => 16 + * Regexp.new(r.source, r.options) # => /\xa1\xa2/ + * + */ + static VALUE -rb_reg_options_m(re) - VALUE re; +rb_reg_options_m(VALUE re) { int options = rb_reg_options(re); return INT2NUM(options); } +static int +reg_names_iter(const OnigUChar *name, const OnigUChar *name_end, + int back_num, int *back_refs, OnigRegex regex, void *arg) +{ + VALUE ary = (VALUE)arg; + rb_ary_push(ary, rb_enc_str_new((const char *)name, name_end-name, regex->enc)); + return 0; +} + +/* + * call-seq: + * names -> array_of_names + * + * Returns an array of names of captures + * (see {Named Captures}[rdoc-ref:Regexp@Named+Captures]): + * + * /(?<foo>.)(?<bar>.)(?<baz>.)/.names # => ["foo", "bar", "baz"] + * /(?<foo>.)(?<foo>.)/.names # => ["foo"] + * /(.)(.)/.names # => [] + * + */ + static VALUE -rb_reg_kcode_m(re) - VALUE re; +rb_reg_names(VALUE re) +{ + VALUE ary; + rb_reg_check(re); + ary = rb_ary_new_capa(onig_number_of_names(RREGEXP_PTR(re))); + onig_foreach_name(RREGEXP_PTR(re), reg_names_iter, (void*)ary); + return ary; +} + +static int +reg_named_captures_iter(const OnigUChar *name, const OnigUChar *name_end, + int back_num, int *back_refs, OnigRegex regex, void *arg) { - char *kcode; + VALUE hash = (VALUE)arg; + VALUE ary = rb_ary_new2(back_num); + int i; + + for (i = 0; i < back_num; i++) + rb_ary_store(ary, i, INT2NUM(back_refs[i])); - if (FL_TEST(re, KCODE_FIXED)) { - switch (RBASIC(re)->flags & KCODE_MASK) { - case KCODE_NONE: - kcode = "none"; break; - case KCODE_EUC: - kcode = "euc"; break; - case KCODE_SJIS: - kcode = "sjis"; break; - case KCODE_UTF8: - kcode = "utf8"; break; - default: - rb_bug("unknown kcode - should not happen"); - break; - } - return rb_str_new2(kcode); + rb_hash_aset(hash, rb_str_new((const char*)name, name_end-name),ary); + + return 0; +} + +/* + * call-seq: + * named_captures -> hash + * + * Returns a hash representing named captures of +self+ + * (see {Named Captures}[rdoc-ref:Regexp@Named+Captures]): + * + * - Each key is the name of a named capture. + * - Each value is an array of integer indexes for that named capture. + * + * Examples: + * + * /(?<foo>.)(?<bar>.)/.named_captures # => {"foo"=>[1], "bar"=>[2]} + * /(?<foo>.)(?<foo>.)/.named_captures # => {"foo"=>[1, 2]} + * /(.)(.)/.named_captures # => {} + * + */ + +static VALUE +rb_reg_named_captures(VALUE re) +{ + regex_t *reg = (rb_reg_check(re), RREGEXP_PTR(re)); + VALUE hash = rb_hash_new_with_size(onig_number_of_names(reg)); + onig_foreach_name(reg, reg_named_captures_iter, (void*)hash); + return hash; +} + +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) +{ + int r; + + *reg = (regex_t* )malloc(sizeof(regex_t)); + if (IS_NULL(*reg)) return ONIGERR_MEMORY; + + r = onig_reg_init(*reg, option, ONIGENC_CASE_FOLD_DEFAULT, enc, syntax); + if (r) goto err; + + r = onig_compile_ruby(*reg, pattern, pattern_end, einfo, sourcefile, sourceline); + if (r) { + err: + onig_free(*reg); + *reg = NULL; } - return Qnil; + return r; } static Regexp* -make_regexp(s, len, flags) - const char *s; - long len; - int flags; +make_regexp(const char *s, long len, rb_encoding *enc, int flags, onig_errmsg_buffer err, + const char *sourcefile, int sourceline) { Regexp *rp; - char *err; + int r; + OnigErrorInfo einfo; /* Handle escaped characters first. */ @@ -548,1196 +900,3871 @@ make_regexp(s, len, flags) from that. */ - rp = ALLOC(Regexp); - MEMZERO((char *)rp, Regexp, 1); - rp->buffer = ALLOC_N(char, 16); - rp->allocated = 16; - rp->fastmap = ALLOC_N(char, 256); - if (flags) { - rp->options = flags; - } - err = re_compile_pattern(s, len, rp); - - if (err != NULL) { - rb_reg_raise(s, len, err, 0); + r = onig_new_with_source(&rp, (UChar*)s, (UChar*)(s + len), flags, + enc, OnigDefaultSyntax, &einfo, sourcefile, sourceline); + if (r) { + onig_error_code_to_str((UChar*)err, r, &einfo); + return 0; } return rp; } -static VALUE rb_cMatch; -static VALUE match_alloc _((VALUE)); +/* + * Document-class: MatchData + * + * MatchData encapsulates the result of matching a Regexp against + * string. It is returned by Regexp#match and String#match, and also + * stored in a global variable returned by Regexp.last_match. + * + * Usage: + * + * url = 'https://docs.ruby-lang.org/en/2.5.0/MatchData.html' + * m = url.match(/(\d\.?)+/) # => #<MatchData "2.5.0" 1:"0"> + * m.string # => "https://docs.ruby-lang.org/en/2.5.0/MatchData.html" + * m.regexp # => /(\d\.?)+/ + * # entire matched substring: + * m[0] # => "2.5.0" + * + * # Working with unnamed captures + * m = url.match(%r{([^/]+)/([^/]+)\.html$}) + * m.captures # => ["2.5.0", "MatchData"] + * m[1] # => "2.5.0" + * m.values_at(1, 2) # => ["2.5.0", "MatchData"] + * + * # Working with named captures + * m = url.match(%r{(?<version>[^/]+)/(?<module>[^/]+)\.html$}) + * m.captures # => ["2.5.0", "MatchData"] + * m.named_captures # => {"version"=>"2.5.0", "module"=>"MatchData"} + * m[:version] # => "2.5.0" + * m.values_at(:version, :module) + * # => ["2.5.0", "MatchData"] + * # Numerical indexes are working, too + * m[1] # => "2.5.0" + * m.values_at(1, 2) # => ["2.5.0", "MatchData"] + * + * == Global variables equivalence + * + * Parts of last MatchData (returned by Regexp.last_match) are also + * aliased as global variables: + * + * * <code>$~</code> is Regexp.last_match; + * * <code>$&</code> is Regexp.last_match<code>[ 0 ]</code>; + * * <code>$1</code>, <code>$2</code>, and so on are + * Regexp.last_match<code>[ i ]</code> (captures by number); + * * <code>$`</code> is Regexp.last_match<code>.pre_match</code>; + * * <code>$'</code> is Regexp.last_match<code>.post_match</code>; + * * <code>$+</code> is Regexp.last_match<code>[ -1 ]</code> (the last capture). + * + * See also Regexp@Global+Variables. + */ + +VALUE rb_cMatch; + static VALUE -match_alloc(klass) - VALUE klass; +match_alloc_n(VALUE klass, int num_regs) { - NEWOBJ(match, struct RMatch); - OBJSETUP(match, 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->regs = 0; - match->regs = ALLOC(struct re_registers); - MEMZERO(match->regs, struct re_registers, 1); + 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_init_copy(obj, orig) - VALUE obj, orig; +match_alloc(VALUE klass) +{ + return match_alloc_n(klass, 0); +} + +int +rb_reg_region_copy(struct re_registers *to, const struct re_registers *from) +{ + onig_region_copy(to, (OnigRegion *)from); + if (to->allocated) return 0; + rb_gc(); + onig_region_copy(to, (OnigRegion *)from); + if (to->allocated) return 0; + 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) { - if (obj == orig) return obj; + struct RMatch *rm = RMATCH(match); - if (!rb_obj_is_instance_of(orig, rb_obj_class(obj))) { - rb_raise(rb_eTypeError, "wrong argument class"); + 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; +} pair_t; + +static int +pair_byte_cmp(const void *pair1, const void *pair2) +{ + long diff = ((pair_t*)pair1)->byte_pos - ((pair_t*)pair2)->byte_pos; +#if SIZEOF_LONG > SIZEOF_INT + return diff ? diff > 0 ? 1 : -1 : 0; +#else + return (int)diff; +#endif +} + +static void +update_char_offset(VALUE match) +{ + 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; + + num_regs = RMATCH_NREGS(match); + + if (rm->char_offset_num_allocated < 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 = RMATCH_BEG(match, i); + rm->char_offset[i].end = RMATCH_END(match, i); + } + return; + } + + pairs = RB_ALLOCV_N(pair_t, pairs_obj, num_regs * 2); + num_pos = 0; + for (i = 0; i < num_regs; i++) { + if (RMATCH_BEG(match, i) < 0) + continue; + 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); + + s = p = RSTRING_PTR(RMATCH(match)->str); + c = 0; + for (i = 0; i < num_pos; i++) { + q = s + pairs[i].byte_pos; + c += rb_enc_strlen(p, q, enc); + pairs[i].char_pos = c; + p = q; + } + + for (i = 0; i < num_regs; i++) { + pair_t key, *found; + if (RMATCH_BEG(match, i) < 0) { + rm->char_offset[i].beg = -1; + rm->char_offset[i].end = -1; + continue; + } + + 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 = 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 VALUE +match_check(VALUE match) +{ + if (!RMATCH(match)->regexp) { + rb_raise(rb_eTypeError, "uninitialized MatchData"); + } + return match; +} + +/* :nodoc: */ +static VALUE +match_init_copy(VALUE obj, VALUE orig) +{ + struct RMatch *rm = RMATCH(obj); + + if (!OBJ_INIT_COPY(obj, orig)) return obj; + + RB_OBJ_WRITE(obj, &rm->str, RMATCH(orig)->str); + RB_OBJ_WRITE(obj, &rm->regexp, RMATCH(orig)->regexp); + + match_set_regs(obj, RMATCH_NREGS(orig), RMATCH_BEG_PTR(orig), RMATCH_END_PTR(orig)); + + 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)->char_offset, + struct rmatch_offset, rm->num_regs); + RB_GC_GUARD(orig); } - RMATCH(obj)->str = RMATCH(orig)->str; - re_free_registers(RMATCH(obj)->regs); - RMATCH(obj)->regs->allocated = 0; - re_copy_registers(RMATCH(obj)->regs, RMATCH(orig)->regs); return obj; } + +/* + * call-seq: + * regexp -> regexp + * + * Returns the regexp that produced the match: + * + * m = /a.*b/.match("abc") # => #<MatchData "ab"> + * m.regexp # => /a.*b/ + * + */ + static VALUE -match_size(match) - VALUE match; +match_regexp(VALUE match) +{ + VALUE regexp; + 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)); + RB_OBJ_WRITE(match, &RMATCH(match)->regexp, regexp); + } + return regexp; +} + +/* + * call-seq: + * names -> array_of_names + * + * Returns an array of the capture names + * (see {Named Captures}[rdoc-ref:Regexp@Named+Captures]): + * + * m = /(?<foo>.)(?<bar>.)(?<baz>.)/.match("hoge") + * # => #<MatchData "hog" foo:"h" bar:"o" baz:"g"> + * m.names # => ["foo", "bar", "baz"] + * + * m = /foo/.match('foo') # => #<MatchData "foo"> + * m.names # => [] # No named captures. + * + * Equivalent to: + * + * m = /(?<foo>.)(?<bar>.)(?<baz>.)/.match("hoge") + * m.regexp.names # => ["foo", "bar", "baz"] + * + */ + +static VALUE +match_names(VALUE match) +{ + match_check(match); + if (NIL_P(RMATCH(match)->regexp)) + return rb_ary_new_capa(0); + return rb_reg_names(RMATCH(match)->regexp); +} + +/* + * call-seq: + * size -> integer + * + * Returns size of the match array: + * + * m = /(.)(.)(\d+)(\d)/.match("THX1138.") + * # => #<MatchData "HX1138" 1:"H" 2:"X" 3:"113" 4:"8"> + * m.size # => 5 + * + */ + +static VALUE +match_size(VALUE match) +{ + match_check(match); + return INT2FIX(RMATCH_NREGS(match)); +} + +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); +} + +static void +backref_number_check(VALUE match, int 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) +{ + int num; + + match_check(match); + if (SYMBOL_P(backref)) { + backref = rb_sym2str(backref); + } + else if (!RB_TYPE_P(backref, T_STRING)) { + return NUM2INT(backref); + } + + num = match_name_to_backref_number(match, backref); + + if (num < 1) { + name_to_backref_error(backref); + } + + return num; +} + +int +rb_reg_backref_number(VALUE match, VALUE backref) +{ + return match_backref_number(match, backref); +} + +/* + * call-seq: + * offset(n) -> [start_offset, end_offset] + * offset(name) -> [start_offset, end_offset] + * + * :include: doc/matchdata/offset.rdoc + * + */ + +static VALUE +match_offset(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 rb_assoc_new(Qnil, Qnil); + + update_char_offset(match); + return rb_assoc_new(LONG2NUM(RMATCH(match)->char_offset[i].beg), + LONG2NUM(RMATCH(match)->char_offset[i].end)); +} + +/* + * call-seq: + * mtch.byteoffset(n) -> array + * + * Returns a two-element array containing the beginning and ending byte-based offsets of + * the <em>n</em>th match. + * <em>n</em> can be a string or symbol to reference a named capture. + * + * m = /(.)(.)(\d+)(\d)/.match("THX1138.") + * m.byteoffset(0) #=> [1, 7] + * m.byteoffset(4) #=> [6, 7] + * + * m = /(?<foo>.)(.)(?<bar>.)/.match("hoge") + * p m.byteoffset(:foo) #=> [0, 1] + * p m.byteoffset(:bar) #=> [2, 3] + * + */ + +static VALUE +match_byteoffset(VALUE match, VALUE n) { - return INT2FIX(RMATCH(match)->regs->num_regs); + int i = match_backref_number(match, n); + + match_check(match); + backref_number_check(match, i); + + if (RMATCH_BEG(match, i) < 0) + return rb_assoc_new(Qnil, Qnil); + 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_offset(match, n) - VALUE match, n; +match_bytebegin(VALUE match, VALUE n) { - int i = NUM2INT(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 + * + */ - if (i < 0 || RMATCH(match)->regs->num_regs <= i) - rb_raise(rb_eIndexError, "index %d out of matches", i); +static VALUE +match_byteend(VALUE match, VALUE n) +{ + int i = match_backref_number(match, n); - if (RMATCH(match)->regs->beg[i] < 0) - return rb_assoc_new(Qnil, Qnil); + match_check(match); + backref_number_check(match, i); - return rb_assoc_new(INT2FIX(RMATCH(match)->regs->beg[i]), - INT2FIX(RMATCH(match)->regs->end[i])); + if (RMATCH_BEG(match, i) < 0) + return Qnil; + return LONG2NUM(RMATCH_END(match, i)); } + +/* + * call-seq: + * begin(n) -> integer + * begin(name) -> integer + * + * :include: doc/matchdata/begin.rdoc + * + */ + static VALUE -match_begin(match, n) - VALUE match, n; +match_begin(VALUE match, VALUE n) { - int i = NUM2INT(n); + int i = match_backref_number(match, n); - if (i < 0 || RMATCH(match)->regs->num_regs <= i) - rb_raise(rb_eIndexError, "index %d out of matches", i); + match_check(match); + backref_number_check(match, i); - if (RMATCH(match)->regs->beg[i] < 0) - return Qnil; + if (RMATCH_BEG(match, i) < 0) + return Qnil; - return INT2FIX(RMATCH(match)->regs->beg[i]); + update_char_offset(match); + return LONG2NUM(RMATCH(match)->char_offset[i].beg); } + +/* + * call-seq: + * end(n) -> integer + * end(name) -> integer + * + * :include: doc/matchdata/end.rdoc + * + */ + static VALUE -match_end(match, n) - VALUE match, n; +match_end(VALUE match, VALUE n) { - int i = NUM2INT(n); + int i = match_backref_number(match, n); - if (i < 0 || RMATCH(match)->regs->num_regs <= i) - rb_raise(rb_eIndexError, "index %d out of matches", i); + match_check(match); + backref_number_check(match, i); - if (RMATCH(match)->regs->beg[i] < 0) - return Qnil; + if (RMATCH_BEG(match, i) < 0) + return Qnil; - return INT2FIX(RMATCH(match)->regs->end[i]); + update_char_offset(match); + return LONG2NUM(RMATCH(match)->char_offset[i].end); +} + +/* + * call-seq: + * match(n) -> string or nil + * match(name) -> string or nil + * + * Returns the matched substring corresponding to the given argument. + * + * When non-negative argument +n+ is given, + * returns the matched substring for the <tt>n</tt>th match: + * + * m = /(.)(.)(\d+)(\d)(\w)?/.match("THX1138.") + * # => #<MatchData "HX1138" 1:"H" 2:"X" 3:"113" 4:"8" 5:nil> + * m.match(0) # => "HX1138" + * m.match(4) # => "8" + * m.match(5) # => nil + * + * When string or symbol argument +name+ is given, + * returns the matched substring for the given name: + * + * m = /(?<foo>.)(.)(?<bar>.+)/.match("hoge") + * # => #<MatchData "hoge" foo:"h" bar:"ge"> + * m.match('foo') # => "h" + * m.match(:bar) # => "ge" + * + */ + +static VALUE +match_nth(VALUE match, VALUE n) +{ + int i = match_backref_number(match, n); + + backref_number_check(match, i); + + long start = RMATCH_BEG(match, i), end = RMATCH_END(match, i); + if (start < 0) + return Qnil; + + return rb_str_subseq(RMATCH(match)->str, start, end - start); +} + +/* + * call-seq: + * match_length(n) -> integer or nil + * match_length(name) -> integer or nil + * + * Returns the length (in characters) of the matched substring + * corresponding to the given argument. + * + * When non-negative argument +n+ is given, + * returns the length of the matched substring + * for the <tt>n</tt>th match: + * + * m = /(.)(.)(\d+)(\d)(\w)?/.match("THX1138.") + * # => #<MatchData "HX1138" 1:"H" 2:"X" 3:"113" 4:"8" 5:nil> + * m.match_length(0) # => 6 + * m.match_length(4) # => 1 + * m.match_length(5) # => nil + * + * When string or symbol argument +name+ is given, + * returns the length of the matched substring + * for the named match: + * + * m = /(?<foo>.)(.)(?<bar>.+)/.match("hoge") + * # => #<MatchData "hoge" foo:"h" bar:"ge"> + * m.match_length('foo') # => 1 + * m.match_length(:bar) # => 2 + * + */ + +static VALUE +match_nth_length(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; + + update_char_offset(match); + const struct rmatch_offset *const ofs = + &RMATCH(match)->char_offset[i]; + return LONG2NUM(ofs->end - ofs->beg); } #define MATCH_BUSY FL_USER2 void -rb_match_busy(match) - VALUE match; +rb_match_busy(VALUE match) { FL_SET(match, MATCH_BUSY); } -int ruby_ignorecase; -static int may_need_recompile; +void +rb_match_unbusy(VALUE match) +{ + FL_UNSET(match, MATCH_BUSY); +} + +int +rb_match_count(VALUE match) +{ + if (NIL_P(match)) return -1; + return RMATCH_NREGS(match); +} + +static VALUE +match_alloc_or_reuse(VALUE existing, int num_regs) +{ + if (!NIL_P(existing) && + !FL_TEST(existing, MATCH_BUSY) && + RMATCH(existing)->capa >= num_regs * 2) { + return existing; + } + 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; + + 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); +} + +VALUE +rb_backref_set_string(VALUE string, long pos, long len) +{ + VALUE match = match_alloc_or_reuse(rb_backref_get(), 1); + match_set_string(match, string, pos, len); + rb_backref_set(match); + return match; +} + +/* + * call-seq: + * fixed_encoding? -> true or false + * + * Returns +false+ if +self+ is applicable to + * a string with any ASCII-compatible encoding; + * otherwise returns +true+: + * + * r = /a/ # => /a/ + * r.fixed_encoding? # => false + * r.match?("\u{6666} a") # => true + * r.match?("\xa1\xa2 a".force_encoding("euc-jp")) # => true + * r.match?("abc".force_encoding("euc-jp")) # => true + * + * r = /a/u # => /a/ + * r.fixed_encoding? # => true + * r.match?("\u{6666} a") # => true + * r.match?("\xa1\xa2".force_encoding("euc-jp")) # Raises exception. + * r.match?("abc".force_encoding("euc-jp")) # => true + * + * r = /\u{6666}/ # => /\u{6666}/ + * r.fixed_encoding? # => true + * r.encoding # => #<Encoding:UTF-8> + * r.match?("\u{6666} a") # => true + * r.match?("\xa1\xa2".force_encoding("euc-jp")) # Raises exception. + * r.match?("abc".force_encoding("euc-jp")) # => false + * + */ + +static VALUE +rb_reg_fixed_encoding_p(VALUE re) +{ + return RBOOL(FL_TEST(re, KCODE_FIXED)); +} + +static VALUE +rb_reg_preprocess(const char *p, const char *end, rb_encoding *enc, + rb_encoding **fixed_enc, onig_errmsg_buffer err, int options); + +NORETURN(static void reg_enc_error(VALUE re, VALUE str)); static void -rb_reg_prepare_re(re) - VALUE re; +reg_enc_error(VALUE re, VALUE str) { - int need_recompile = 0; - int state; + rb_raise(rb_eEncCompatError, + "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 = rb_enc_str_coderange(str); + + if (cr == ENC_CODERANGE_BROKEN) { + rb_raise(rb_eArgError, + "invalid byte sequence in %s", + rb_enc_name(rb_enc_get(str))); + } rb_reg_check(re); - state = FL_TEST(re, REG_CASESTATE); - /* ignorecase status */ - if (ruby_ignorecase && !state) { - FL_SET(re, REG_CASESTATE); - RREGEXP(re)->ptr->options |= RE_OPTION_IGNORECASE; - need_recompile = 1; + enc = rb_enc_get(str); + if (RREGEXP_PTR(re)->enc == enc) { } - if (!ruby_ignorecase && state) { - FL_UNSET(re, REG_CASESTATE); - RREGEXP(re)->ptr->options &= ~RE_OPTION_IGNORECASE; - need_recompile = 1; + else if (cr == ENC_CODERANGE_7BIT && + RREGEXP_PTR(re)->enc == rb_usascii_encoding()) { + enc = RREGEXP_PTR(re)->enc; } - - if (!FL_TEST(re, KCODE_FIXED) && - (RBASIC(re)->flags & KCODE_MASK) != reg_kcode) { - need_recompile = 1; - RBASIC(re)->flags &= ~KCODE_MASK; - RBASIC(re)->flags |= reg_kcode; + else if (!rb_enc_asciicompat(enc)) { + reg_enc_error(re, str); } - - if (need_recompile) { - char *err; - - if (FL_TEST(re, KCODE_FIXED)) - kcode_set_option(re); - rb_reg_check(re); - RREGEXP(re)->ptr->fastmap_accurate = 0; - err = re_compile_pattern(RREGEXP(re)->str, RREGEXP(re)->len, RREGEXP(re)->ptr); - if (err != NULL) { - rb_reg_raise(RREGEXP(re)->str, RREGEXP(re)->len, err, re); - } + 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; } + 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)); + } + return enc; } -long -rb_reg_adjust_startpos(re, str, pos, reverse) - VALUE re, str; - long pos, reverse; +regex_t * +rb_reg_prepare_re(VALUE re, VALUE str) { - long range; + int r; + OnigErrorInfo einfo; + 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); - if (may_need_recompile) rb_reg_prepare_re(re); - if (FL_TEST(re, KCODE_FIXED)) - kcode_set_option(re); - else if (reg_kcode != curr_kcode) - kcode_reset_option(); + VALUE src_str = RREGEXP_SRC(re); + const char *pattern = RSTRING_PTR(src_str); - if (reverse) { - range = -pos; + onig_errmsg_buffer err = ""; + unescaped = rb_reg_preprocess( + pattern, pattern + RSTRING_LEN(src_str), enc, + &fixed_enc, err, 0); + + if (NIL_P(unescaped)) { + rb_raise(rb_eArgError, "regexp preprocess failed: %s", err); + } + + // inherit the timeout settings + rb_hrtime_t timelimit = reg->timelimit; + + const char *ptr; + long len; + RSTRING_GETMEM(unescaped, ptr, len); + + /* 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 { - range = RSTRING(str)->len - pos; + 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(err, re); } - return re_adjust_startpos(RREGEXP(re)->ptr, - RSTRING(str)->ptr, RSTRING(str)->len, - pos, range); + + reg->timelimit = timelimit; + + RB_GC_GUARD(unescaped); + RB_GC_GUARD(src_str); + return reg; } -long -rb_reg_search(re, str, pos, reverse) - VALUE re, str; - long pos, reverse; +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) { - long result; - VALUE match; - static struct re_registers regs; - long range; + 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 (pos > RSTRING(str)->len || pos < 0) { - rb_backref_set(Qnil); - return -1; + if (!tmpreg) RREGEXP(re)->usecnt--; + if (tmpreg) { + onig_free(reg); } - rb_reg_check(re); - if (may_need_recompile) rb_reg_prepare_re(re); + 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 +rb_reg_adjust_startpos(VALUE re, VALUE str, long pos, int reverse) +{ + long range; + rb_encoding *enc; + UChar *p, *string; - if (FL_TEST(re, KCODE_FIXED)) - kcode_set_option(re); - else if (reg_kcode != curr_kcode) - kcode_reset_option(); + enc = rb_reg_prepare_enc(re, str, 0); if (reverse) { - range = -pos; + range = -pos; } else { - range = RSTRING(str)->len - pos; + range = RSTRING_LEN(str) - pos; } - result = re_search(RREGEXP(re)->ptr,RSTRING(str)->ptr,RSTRING(str)->len, - pos, range, ®s); - if (FL_TEST(re, KCODE_FIXED)) - kcode_reset_option(); + if (pos > 0 && ONIGENC_MBC_MAXLEN(enc) != 1 && pos < RSTRING_LEN(str)) { + string = (UChar*)RSTRING_PTR(str); - if (result == -2) { - rb_reg_raise(RREGEXP(re)->str, RREGEXP(re)->len, - "Stack overflow in regexp matcher", re); + 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 (result < 0) { - rb_backref_set(Qnil); - return result; + 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 len = RSTRING_LEN(str); + if (pos > len || pos < 0) { + rb_backref_set(Qnil); + return -1; + } + + struct reg_onig_search_args args = { + .pos = pos, + .range = reverse ? 0 : len, + }; + + 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; } - match = rb_backref_get(); - if (NIL_P(match) || FL_TEST(match, MATCH_BUSY)) { - match = match_alloc(rb_cMatch); + 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) { + RB_OBJ_WRITE(match, &RMATCH(match)->str, rb_str_new4(str)); + rb_obj_reveal(match, rb_cMatch); } else { - if (rb_safe_level() >= 3) - OBJ_TAINT(match); - else - FL_UNSET(match, FL_TAINT); + /* 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); } - re_copy_registers(RMATCH(match)->regs, ®s); - RMATCH(match)->str = rb_str_new4(str); + RB_OBJ_WRITE(match, &RMATCH(match)->regexp, re); rb_backref_set(match); + if (set_match) *set_match = match; - OBJ_INFECT(match, re); - OBJ_INFECT(match, str); return result; } +long +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, match); +} + +long +rb_reg_search(VALUE re, VALUE str, long pos, int reverse) +{ + return rb_reg_search_set_match(re, str, pos, reverse, 1, NULL); +} + +static OnigPosition +reg_onig_match(regex_t *reg, VALUE str, struct re_registers *regs, void *_) +{ + const char *ptr; + long len; + RSTRING_GETMEM(str, ptr, len); + + return onig_match( + reg, + (UChar *)ptr, + (UChar *)(ptr + len), + (UChar *)ptr, + regs, + ONIG_OPTION_NONE); +} + +bool +rb_reg_start_with_p(VALUE re, VALUE str) +{ + rb_reg_check(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; +} + VALUE -rb_reg_nth_defined(nth, match) - int nth; - VALUE match; +rb_reg_nth_defined(int nth, VALUE match) { if (NIL_P(match)) return Qnil; - if (nth >= RMATCH(match)->regs->num_regs) { - return Qnil; + match_check(match); + int num_regs = RMATCH_NREGS(match); + if (nth >= num_regs) { + return Qnil; } if (nth < 0) { - nth += RMATCH(match)->regs->num_regs; - if (nth <= 0) return Qnil; + nth += num_regs; + if (nth <= 0) return Qnil; } - if (RMATCH(match)->BEG(nth) == -1) return Qfalse; - return Qtrue; + return RBOOL(RMATCH_BEG(match, nth) != -1); } VALUE -rb_reg_nth_match(nth, match) - int nth; - VALUE match; +rb_reg_nth_match(int nth, VALUE match) { VALUE str; long start, end, len; if (NIL_P(match)) return Qnil; - if (nth >= RMATCH(match)->regs->num_regs) { - return Qnil; + match_check(match); + int num_regs = RMATCH_NREGS(match); + if (nth >= num_regs) { + return Qnil; } if (nth < 0) { - nth += RMATCH(match)->regs->num_regs; - if (nth <= 0) return Qnil; + nth += num_regs; + if (nth <= 0) return Qnil; } - start = RMATCH(match)->BEG(nth); + start = RMATCH_BEG(match, nth); if (start == -1) return Qnil; - end = RMATCH(match)->END(nth); + end = RMATCH_END(match, nth); len = end - start; - str = rb_str_substr(RMATCH(match)->str, start, len); - OBJ_INFECT(str, match); + str = rb_str_subseq(RMATCH(match)->str, start, len); return str; } VALUE -rb_reg_last_match(match) - VALUE match; +rb_reg_last_match(VALUE match) { return rb_reg_nth_match(0, match); } + +/* + * call-seq: + * pre_match -> string + * + * Returns the substring of the target string from its beginning + * up to the first match in +self+ (that is, <tt>self[0]</tt>); + * equivalent to regexp global variable <tt>$`</tt>: + * + * m = /(.)(.)(\d+)(\d)/.match("THX1138.") + * # => #<MatchData "HX1138" 1:"H" 2:"X" 3:"113" 4:"8"> + * m[0] # => "HX1138" + * m.pre_match # => "T" + * + * Related: MatchData#post_match. + * + */ + VALUE -rb_reg_match_pre(match) - VALUE match; +rb_reg_match_pre(VALUE match) { VALUE str; if (NIL_P(match)) return Qnil; - if (RMATCH(match)->BEG(0) == -1) return Qnil; - str = rb_str_substr(RMATCH(match)->str, 0, RMATCH(match)->BEG(0)); - if (OBJ_TAINTED(match)) OBJ_TAINT(str); + match_check(match); + if (RMATCH_BEG(match, 0) == -1) return Qnil; + str = rb_str_subseq(RMATCH(match)->str, 0, RMATCH_BEG(match, 0)); return str; } + +/* + * call-seq: + * post_match -> str + * + * Returns the substring of the target string from + * the end of the first match in +self+ (that is, <tt>self[0]</tt>) + * to the end of the string; + * equivalent to regexp global variable <tt>$'</tt>: + * + * m = /(.)(.)(\d+)(\d)/.match("THX1138: The Movie") + * # => #<MatchData "HX1138" 1:"H" 2:"X" 3:"113" 4:"8"> + * m[0] # => "HX1138" + * m.post_match # => ": The Movie"\ + * + * Related: MatchData.pre_match. + * + */ + VALUE -rb_reg_match_post(match) - VALUE match; +rb_reg_match_post(VALUE match) { VALUE str; long pos; if (NIL_P(match)) return Qnil; - if (RMATCH(match)->BEG(0) == -1) return Qnil; + match_check(match); + if (RMATCH_BEG(match, 0) == -1) return Qnil; str = RMATCH(match)->str; - pos = RMATCH(match)->END(0); - str = rb_str_substr(str, pos, RSTRING(str)->len - pos); - if (OBJ_TAINTED(match)) OBJ_TAINT(str); + pos = RMATCH_END(match, 0); + str = rb_str_subseq(str, pos, RSTRING_LEN(str) - pos); return str; } -VALUE -rb_reg_match_last(match) - VALUE match; +static int +match_last_index(VALUE match) { int i; - if (NIL_P(match)) return Qnil; - if (RMATCH(match)->BEG(0) == -1) return Qnil; + if (NIL_P(match)) return -1; + match_check(match); + if (RMATCH_BEG(match, 0) == -1) return -1; - for (i=RMATCH(match)->regs->num_regs-1; RMATCH(match)->BEG(i) == -1 && i > 0; i--) - ; - if (i == 0) return Qnil; - return rb_reg_nth_match(i, match); + for (i = RMATCH_NREGS(match) - 1; RMATCH_BEG(match, i) == -1 && i > 0; i--) + ; + return i; +} + +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 -last_match_getter() +last_match_getter(ID _x, VALUE *_y) { return rb_reg_last_match(rb_backref_get()); } static VALUE -prematch_getter() +prematch_getter(ID _x, VALUE *_y) { return rb_reg_match_pre(rb_backref_get()); } static VALUE -postmatch_getter() +postmatch_getter(ID _x, VALUE *_y) { return rb_reg_match_post(rb_backref_get()); } static VALUE -last_paren_match_getter() +last_paren_match_getter(ID _x, VALUE *_y) { return rb_reg_match_last(rb_backref_get()); } static VALUE -match_array(match, start) - VALUE match; - int start; +match_array(VALUE match, int start) { - struct re_registers *regs = RMATCH(match)->regs; - VALUE ary = rb_ary_new2(regs->num_regs); - VALUE target = RMATCH(match)->str; + VALUE ary; + VALUE target; int i; - int taint = OBJ_TAINTED(match); - - for (i=start; i<regs->num_regs; i++) { - if (regs->beg[i] == -1) { - rb_ary_push(ary, Qnil); - } - else { - VALUE str = rb_str_substr(target, regs->beg[i], regs->end[i]-regs->beg[i]); - if (taint) OBJ_TAINT(str); - rb_ary_push(ary, str); - } + + match_check(match); + int num_regs = RMATCH_NREGS(match); + ary = rb_ary_new2(num_regs); + target = RMATCH(match)->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; } + +/* + * call-seq: + * to_a -> array + * + * Returns the array of matches: + * + * m = /(.)(.)(\d+)(\d)/.match("THX1138.") + * # => #<MatchData "HX1138" 1:"H" 2:"X" 3:"113" 4:"8"> + * m.to_a # => ["HX1138", "H", "X", "113", "8"] + * + * Related: MatchData#captures. + * + */ + static VALUE -match_to_a(match) - VALUE match; +match_to_a(VALUE match) { return match_array(match, 0); } + +/* + * call-seq: + * captures -> array + * + * Returns the array of captures, + * which are all matches except <tt>m[0]</tt>: + * + * m = /(.)(.)(\d+)(\d)/.match("THX1138.") + * # => #<MatchData "HX1138" 1:"H" 2:"X" 3:"113" 4:"8"> + * m[0] # => "HX1138" + * m.captures # => ["H", "X", "113", "8"] + * + * Related: MatchData.to_a. + * + */ static VALUE -match_captures(match) - VALUE match; +match_captures(VALUE match) { return match_array(match, 1); } -static VALUE -match_aref(argc, argv, match) - int argc; - VALUE *argv; - VALUE match; +static int +name_to_backref_number(const struct re_registers *regs, VALUE regexp, const char* name, const char* name_end) { - VALUE idx, rest; + 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); +} - rb_scan_args(argc, argv, "11", &idx, &rest); +#define NAME_TO_NUMBER(regs, re, name, name_ptr, name_end) \ + (NIL_P(re) ? 0 : \ + !rb_enc_compatible(RREGEXP_SRC(re), (name)) ? 0 : \ + name_to_backref_number((regs), (re), (name_ptr), (name_end))) - if (!NIL_P(rest) || !FIXNUM_P(idx) || FIX2INT(idx) < 0) { - return rb_ary_aref(argc, argv, match_to_a(match)); +static int +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 rb_reg_nth_match(FIX2INT(idx), match); + return nums[n - 1]; +} + +static int +namev_to_backref_number(VALUE match, VALUE name) +{ + int num; + + if (SYMBOL_P(name)) { + name = rb_sym2str(name); + } + else if (!RB_TYPE_P(name, T_STRING)) { + 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); + } + if (num < 1) { + name_to_backref_error(name); + } + return num; } -static VALUE match_entry _((VALUE, long)); static VALUE -match_entry(match, n) - VALUE match; - long n; +match_ary_subseq(VALUE match, long beg, long len, VALUE result) { - return rb_reg_nth_match(n, match); + 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)); + } + if (beg + len > j) { + rb_ary_resize(result, RARRAY_LEN(result) + (beg + len) - j); + } + return result; } static VALUE -match_values_at(argc, argv, match) - int argc; - VALUE *argv; - VALUE match; +match_ary_aref(VALUE match, VALUE idx, VALUE result) { - return rb_values_at(match, RMATCH(match)->regs->num_regs, argc, argv, match_entry); + long beg, len; + 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; + case Qnil: + return Qnil; + default: + return match_ary_subseq(match, beg, len, result); + } } +/* + * call-seq: + * self[offset] -> string or nil + * self[offset, size] -> array + * self[range] -> array + * self[name] -> string or nil + * + * When arguments +offset+, +offset+ and +size+, or +range+ are given, + * returns match and captures in the style of Array#[]: + * + * m = /(.)(.)(\d+)(\d)/.match("THX1138.") + * # => #<MatchData "HX1138" 1:"H" 2:"X" 3:"113" 4:"8"> + * m[0] # => "HX1138" + * m[1, 2] # => ["H", "X"] + * m[1..3] # => ["H", "X", "113"] + * m[-3, 2] # => ["X", "113"] + * + * When string or symbol argument +name+ is given, + * returns the matched substring for the given name: + * + * m = /(?<foo>.)(.)(?<bar>.+)/.match("hoge") + * # => #<MatchData "hoge" foo:"h" bar:"ge"> + * 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 -match_select(argc, argv, match) - int argc; - VALUE *argv; - VALUE match; +match_aref(int argc, VALUE *argv, VALUE match) { - if (argc > 0) { - rb_raise(rb_eArgError, "wrong number arguments(%d for 0)", argc); + VALUE idx, length; + + match_check(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(match, idx); + if (num >= 0) { + return rb_reg_nth_match(num, match); + } + else { + return match_ary_aref(match, idx, Qnil); + } + } } else { - struct re_registers *regs = RMATCH(match)->regs; - VALUE target = RMATCH(match)->str; - VALUE result = rb_ary_new(); - int i; - int taint = OBJ_TAINTED(match); + 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); + } +} + +/* + * call-seq: + * values_at(*indexes) -> array + * + * Returns match and captures at the given +indexes+, + * which may include any mixture of: + * + * - Integers. + * - Ranges. + * - Names (strings and symbols). + * + * + * Examples: + * + * m = /(.)(.)(\d+)(\d)/.match("THX1138: The Movie") + * # => #<MatchData "HX1138" 1:"H" 2:"X" 3:"113" 4:"8"> + * m.values_at(0, 2, -2) # => ["HX1138", "X", "113"] + * m.values_at(1..2, -1) # => ["H", "X", "8"] + * + * m = /(?<a>\d+) *(?<op>[+\-*\/]) *(?<b>\d+)/.match("1 + 2") + * # => #<MatchData "1 + 2" a:"1" op:"+" b:"2"> + * m.values_at(0, 1..2, :a, :b, :op) + * # => ["1 + 2", "1", "+", "1", "2", "+"] + * + */ + +static VALUE +match_values_at(int argc, VALUE *argv, VALUE match) +{ + VALUE result; + int i; - for (i=0; i<regs->num_regs; i++) { - VALUE str = rb_str_substr(target, regs->beg[i], regs->end[i]-regs->beg[i]); - if (taint) OBJ_TAINT(str); - if (RTEST(rb_yield(str))) { - rb_ary_push(result, str); - } - } - return result; + match_check(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(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; } + +/* + * call-seq: + * to_s -> string + * + * Returns the matched string: + * + * m = /(.)(.)(\d+)(\d)/.match("THX1138.") + * # => #<MatchData "HX1138" 1:"H" 2:"X" 3:"113" 4:"8"> + * m.to_s # => "HX1138" + * + * m = /(?<foo>.)(.)(?<bar>.+)/.match("hoge") + * # => #<MatchData "hoge" foo:"h" bar:"ge"> + * m.to_s # => "hoge" + * + * Related: MatchData.inspect. + * + */ + static VALUE -match_to_s(match) - VALUE match; +match_to_s(VALUE match) { - VALUE str = rb_reg_last_match(match); + VALUE str = rb_reg_last_match(match_check(match)); if (NIL_P(str)) str = rb_str_new(0,0); - if (OBJ_TAINTED(match)) OBJ_TAINT(str); - if (OBJ_TAINTED(RMATCH(match)->str)) OBJ_TAINT(str); return str; } -static VALUE -match_string(match) +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 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; + } + } + + if (found == 0) { + rb_hash_aset(hash, key, Qnil); + } + + return 0; +} + +/* + * call-seq: + * 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+: + * + * m = /(?<foo>.)(.)(?<bar>.+)/.match("hoge") + * # => #<MatchData "hoge" foo:"h" bar:"ge"> + * m.named_captures # => {"foo"=>"h", "bar"=>"ge"} + * + * m = /(?<a>.)(?<b>.)/.match("01") + * # => #<MatchData "01" a:"0" b:"1"> + * m.named_captures #=> {"a" => "0", "b" => "1"} + * + * m = /(?<a>.)(?<b>.)?/.match("0") + * # => #<MatchData "0" a:"0" b:nil> + * m.named_captures #=> {"a" => "0", "b" => nil} + * + * m = /(?<a>.)(?<a>.)/.match("01") + * # => #<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(int argc, VALUE *argv, VALUE match) +{ + VALUE hash; + + match_check(match); + if (NIL_P(RMATCH(match)->regexp)) + 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(); + struct named_captures_data data = { hash, match, symbolize_names }; + + 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; + * otherwise, returns a frozen copy of the target string: + * + * m = /(.)(.)(\d+)(\d)/.match("THX1138.") + * # => #<MatchData "HX1138" 1:"H" 2:"X" 3:"113" 4:"8"> + * m.string # => "THX1138." + * + */ + +static VALUE +match_string(VALUE match) { + match_check(match); return RMATCH(match)->str; /* str is frozen */ } +struct backref_name_tag { + const UChar *name; + long len; +}; + +static int +match_inspect_name_iter(const OnigUChar *name, const OnigUChar *name_end, + int back_num, int *back_refs, OnigRegex regex, void *arg0) +{ + struct backref_name_tag *arg = (struct backref_name_tag *)arg0; + int i; + + for (i = 0; i < back_num; i++) { + arg[back_refs[i]].name = name; + arg[back_refs[i]].len = name_end - name; + } + return 0; +} + +/* + * call-seq: + * inspect -> string + * + * Returns a string representation of +self+: + * + * m = /.$/.match("foo") + * # => #<MatchData "o"> + * m.inspect # => "#<MatchData \"o\">" + * + * m = /(.)(.)(.)/.match("foo") + * # => #<MatchData "foo" 1:"f" 2:"o" 3:"o"> + * m.inspect # => "#<MatchData \"foo\" 1:\"f\" 2:\"o\ + * + * m = /(.)(.)?(.)/.match("fo") + * # => #<MatchData "fo" 1:"f" 2:nil 3:"o"> + * m.inspect # => "#<MatchData \"fo\" 1:\"f\" 2:nil 3:\"o\">" + * + * Related: MatchData#to_s. + */ + +static VALUE +match_inspect(VALUE match) +{ + VALUE cname = rb_class_path(rb_obj_class(match)); + VALUE str; + int i; + int num_regs = RMATCH_NREGS(match); + struct backref_name_tag *names; + VALUE names_obj = Qnil; + VALUE regexp = RMATCH(match)->regexp; + + if (regexp == 0) { + return rb_sprintf("#<%"PRIsVALUE":%p>", cname, (void*)match); + } + else if (NIL_P(regexp)) { + return rb_sprintf("#<%"PRIsVALUE": %"PRIsVALUE">", + cname, rb_reg_nth_match(0, match)); + } + + 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), + match_inspect_name_iter, names); + + str = rb_str_buf_new2("#<"); + rb_str_append(str, cname); + + for (i = 0; i < num_regs; i++) { + VALUE v; + rb_str_buf_cat2(str, " "); + if (0 < i) { + if (names[i].name) + rb_str_buf_cat(str, (const char *)names[i].name, names[i].len); + else { + rb_str_catf(str, "%d", i); + } + rb_str_buf_cat2(str, ":"); + } + v = rb_reg_nth_match(i, match); + if (NIL_P(v)) + rb_str_buf_cat2(str, "nil"); + else + rb_str_buf_append(str, rb_str_inspect(v)); + } + rb_str_buf_cat2(str, ">"); + + RB_ALLOCV_END(names_obj); + return str; +} + VALUE rb_cRegexp; -static void -rb_reg_initialize(obj, s, len, options) - VALUE obj; - const char *s; - long len; - int options; /* CASEFOLD = 1 */ - /* EXTENDED = 2 */ - /* MULTILINE = 4 */ - /* CODE_NONE = 16 */ - /* CODE_EUC = 32 */ - /* CODE_SJIS = 48 */ - /* CODE_UTF8 = 64 */ +static int +read_escaped_byte(const char **pp, const char *end, onig_errmsg_buffer err) { - struct RRegexp *re = RREGEXP(obj); + const char *p = *pp; + int code; + int meta_prefix = 0, ctrl_prefix = 0; + size_t len; + + if (p == end || *p++ != '\\') { + errcpy(err, "too short escaped multibyte character"); + return -1; + } - if (re->ptr) re_free_pattern(re->ptr); - if (re->str) free(re->str); - re->ptr = 0; - re->str = 0; +again: + if (p == end) { + errcpy(err, "too short escape sequence"); + return -1; + } + switch (*p++) { + case '\\': code = '\\'; break; + case 'n': code = '\n'; break; + case 't': code = '\t'; break; + case 'r': code = '\r'; break; + case 'f': code = '\f'; break; + case 'v': code = '\013'; break; + case 'a': code = '\007'; break; + case 'e': code = '\033'; break; + + /* \OOO */ + case '0': case '1': case '2': case '3': + case '4': case '5': case '6': case '7': + p--; + code = scan_oct(p, end < p+3 ? end-p : 3, &len); + p += len; + break; + + case 'x': /* \xHH */ + code = scan_hex(p, end < p+2 ? end-p : 2, &len); + if (len < 1) { + errcpy(err, "invalid hex escape"); + return -1; + } + p += len; + break; + + case 'M': /* \M-X, \M-\C-X, \M-\cX */ + if (meta_prefix) { + errcpy(err, "duplicate meta escape"); + return -1; + } + meta_prefix = 1; + if (p+1 < end && *p++ == '-' && (*p & 0x80) == 0) { + if (*p == '\\') { + p++; + goto again; + } + else { + code = *p++; + break; + } + } + errcpy(err, "too short meta escape"); + return -1; + + case 'C': /* \C-X, \C-\M-X */ + if (p == end || *p++ != '-') { + errcpy(err, "too short control escape"); + return -1; + } + case 'c': /* \cX, \c\M-X */ + if (ctrl_prefix) { + errcpy(err, "duplicate control escape"); + return -1; + } + ctrl_prefix = 1; + if (p < end && (*p & 0x80) == 0) { + if (*p == '\\') { + p++; + goto again; + } + else { + code = *p++; + break; + } + } + errcpy(err, "too short control escape"); + return -1; - switch (options & ~0xf) { - case 0: default: - FL_SET(re, reg_kcode); - break; - case 16: - kcode_none(re); - break; - case 32: - kcode_euc(re); - break; - case 48: - kcode_sjis(re); - break; - case 64: - kcode_utf8(re); - break; + errcpy(err, "unexpected escape sequence"); + return -1; + } + if (code < 0 || 0xff < code) { + errcpy(err, "invalid escape code"); + return -1; + } + + if (ctrl_prefix) + code &= 0x1f; + if (meta_prefix) + code |= 0x80; + + *pp = p; + return code; +} + +static int +unescape_escaped_nonascii(const char **pp, const char *end, rb_encoding *enc, + VALUE buf, rb_encoding **encp, onig_errmsg_buffer err) +{ + const char *p = *pp; + int chmaxlen = rb_enc_mbmaxlen(enc); + unsigned char *area = ALLOCA_N(unsigned char, chmaxlen); + char *chbuf = (char *)area; + int chlen = 0; + int byte; + int l; + + memset(chbuf, 0, chmaxlen); + + byte = read_escaped_byte(&p, end, err); + if (byte == -1) { + return -1; + } + + area[chlen++] = byte; + while (chlen < chmaxlen && + MBCLEN_NEEDMORE_P(rb_enc_precise_mbclen(chbuf, chbuf+chlen, enc))) { + byte = read_escaped_byte(&p, end, err); + if (byte == -1) { + return -1; + } + area[chlen++] = byte; + } + + l = rb_enc_precise_mbclen(chbuf, chbuf+chlen, enc); + if (MBCLEN_INVALID_P(l)) { + errcpy(err, "invalid multibyte escape"); + return -1; + } + if (1 < chlen || (area[0] & 0x80)) { + rb_str_buf_cat(buf, chbuf, chlen); + + if (*encp == 0) + *encp = enc; + else if (*encp != enc) { + errcpy(err, "escaped non ASCII character in UTF-8 regexp"); + return -1; + } + } + else { + char escbuf[5]; + snprintf(escbuf, sizeof(escbuf), "\\x%02X", area[0]&0xff); + rb_str_buf_cat(buf, escbuf, 4); + } + *pp = p; + return 0; +} + +static int +check_unicode_range(unsigned long code, onig_errmsg_buffer err) +{ + if ((0xd800 <= code && code <= 0xdfff) || /* Surrogates */ + 0x10ffff < code) { + errcpy(err, "invalid Unicode range"); + return -1; } + return 0; +} - if (options & ~0xf) { - kcode_set_option((VALUE)re); +static int +append_utf8(unsigned long uv, + VALUE buf, rb_encoding **encp, onig_errmsg_buffer err) +{ + if (check_unicode_range(uv, err) != 0) + return -1; + if (uv < 0x80) { + char escbuf[5]; + snprintf(escbuf, sizeof(escbuf), "\\x%02X", (int)uv); + rb_str_buf_cat(buf, escbuf, 4); + } + else { + int len; + char utf8buf[6]; + len = rb_uv_to_utf8(utf8buf, uv); + rb_str_buf_cat(buf, utf8buf, len); + + if (*encp == 0) + *encp = rb_utf8_encoding(); + else if (*encp != rb_utf8_encoding()) { + errcpy(err, "UTF-8 character in non UTF-8 regexp"); + return -1; + } } - if (ruby_ignorecase) { - options |= RE_OPTION_IGNORECASE; - FL_SET(re, REG_CASESTATE); + return 0; +} + +static int +unescape_unicode_list(const char **pp, const char *end, + VALUE buf, rb_encoding **encp, onig_errmsg_buffer err) +{ + const char *p = *pp; + int has_unicode = 0; + unsigned long code; + size_t len; + + while (p < end && ISSPACE(*p)) p++; + + while (1) { + code = ruby_scan_hex(p, end-p, &len); + if (len == 0) + break; + if (6 < len) { /* max 10FFFF */ + errcpy(err, "invalid Unicode range"); + return -1; + } + p += len; + if (append_utf8(code, buf, encp, err) != 0) + return -1; + has_unicode = 1; + + while (p < end && ISSPACE(*p)) p++; } - re->ptr = make_regexp(s, len, options & 0xf); - re->str = ALLOC_N(char, len+1); - memcpy(re->str, s, len); - re->str[len] = '\0'; - re->len = len; - if (options & ~0xf) { - kcode_reset_option(); + + if (has_unicode == 0) { + errcpy(err, "invalid Unicode list"); + return -1; } + + *pp = p; + + return 0; +} + +static int +unescape_unicode_bmp(const char **pp, const char *end, + VALUE buf, rb_encoding **encp, onig_errmsg_buffer err) +{ + const char *p = *pp; + size_t len; + unsigned long code; + + if (end < p+4) { + errcpy(err, "invalid Unicode escape"); + return -1; + } + code = ruby_scan_hex(p, 4, &len); + if (len != 4) { + errcpy(err, "invalid Unicode escape"); + return -1; + } + if (append_utf8(code, buf, encp, err) != 0) + return -1; + *pp = p + 4; + return 0; +} + +static int +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, 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)) { + invalid_multibyte: + errcpy(err, "invalid multibyte character"); + return -1; + } + chlen = MBCLEN_CHARFOUND_LEN(chlen); + if (1 < chlen || (*p & 0x80)) { + multibyte: + rb_str_buf_cat(buf, p, chlen); + p += chlen; + if (*encp == 0) + *encp = enc; + else if (*encp != enc) { + errcpy(err, "non ASCII character in UTF-8 regexp"); + return -1; + } + continue; + } + + switch (c = *p++) { + case '\\': + if (p == end) { + errcpy(err, "too short escape sequence"); + return -1; + } + chlen = rb_enc_precise_mbclen(p, end, enc); + if (!MBCLEN_CHARFOUND_P(chlen)) { + goto invalid_multibyte; + } + if ((chlen = MBCLEN_CHARFOUND_LEN(chlen)) > 1) { + /* include the previous backslash */ + --p; + ++chlen; + goto multibyte; + } + switch (c = *p++) { + case '1': case '2': case '3': + case '4': case '5': case '6': case '7': /* \O, \OO, \OOO or backref */ + { + size_t len = end-(p-1), octlen; + if (ruby_scan_oct(p-1, len < 3 ? len : 3, &octlen) <= 0177) { + /* backref or 7bit octal. + no need to unescape anyway. + re-escaping may break backref */ + goto escape_asis; + } + } + /* xxx: How about more than 199 subexpressions? */ + + case '0': /* \0, \0O, \0OO */ + + case 'x': /* \xHH */ + case 'c': /* \cX, \c\M-X */ + case 'C': /* \C-X, \C-\M-X */ + case 'M': /* \M-X, \M-\C-X, \M-\cX */ + p = p-2; + 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; + } + break; + + case 'u': + if (p == end) { + errcpy(err, "too short escape sequence"); + return -1; + } + if (*p == '{') { + /* \u{H HH HHH HHHH HHHHH HHHHHH ...} */ + p++; + if (unescape_unicode_list(&p, end, buf, encp, err) != 0) + return -1; + if (p == end || *p++ != '}') { + errcpy(err, "invalid Unicode list"); + return -1; + } + break; + } + else { + /* \uHHHH */ + if (unescape_unicode_bmp(&p, end, buf, encp, err) != 0) + return -1; + break; + } + + case 'p': /* \p{Hiragana} */ + case 'P': + if (!*encp) { + *has_property = 1; + } + goto escape_asis; + + default: /* \n, \\, \d, \9, etc. */ +escape_asis: + smallbuf[0] = '\\'; + smallbuf[1] = c; + rb_str_buf_cat(buf, smallbuf, 2); + break; + } + break; + + case '#': + if (extended_mode && !in_char_class) { + /* consume and ignore comment in extended regexp */ + 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); + break; + case '[': + in_char_class++; + rb_str_buf_cat(buf, (char *)&c, 1); + break; + case ']': + if (in_char_class) { + in_char_class--; + } + 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 == '?') { + 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; + } + } + + 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 (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 if (!in_char_class && recurse) { + parens++; + } + /* 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_s_alloc _((VALUE)); static VALUE -rb_reg_s_alloc(klass) - VALUE klass; +rb_reg_preprocess(const char *p, const char *end, rb_encoding *enc, + rb_encoding **fixed_enc, onig_errmsg_buffer err, int options) { - NEWOBJ(re, struct RRegexp); - OBJSETUP(re, klass, T_REGEXP); + VALUE buf; + int has_property = 0; + + buf = rb_str_buf_new(0); + + if (rb_enc_asciicompat(enc)) + *fixed_enc = 0; + else { + *fixed_enc = enc; + rb_enc_associate(buf, enc); + } + + if (unescape_nonascii(p, end, enc, buf, fixed_enc, &has_property, err, options) != 0) + return Qnil; + + if (has_property && !*fixed_enc) { + *fixed_enc = enc; + } + + if (*fixed_enc) { + rb_enc_associate(buf, *fixed_enc); + } + + return buf; +} + +VALUE +rb_reg_check_preprocess(VALUE str) +{ + rb_encoding *fixed_enc = 0; + onig_errmsg_buffer err = ""; + VALUE buf; + char *p, *end; + rb_encoding *enc; + + StringValue(str); + p = RSTRING_PTR(str); + end = p + RSTRING_LEN(str); + enc = rb_enc_get(str); + + buf = rb_reg_preprocess(p, end, enc, &fixed_enc, err, 0); + RB_GC_GUARD(str); + + if (NIL_P(buf)) { + return rb_reg_error_desc(str, 0, err); + } + return Qnil; +} + +static VALUE +rb_reg_preprocess_dregexp(VALUE ary, int options) +{ + rb_encoding *fixed_enc = 0; + rb_encoding *regexp_enc = 0; + onig_errmsg_buffer err = ""; + int i; + VALUE result = 0; + rb_encoding *ascii8bit = rb_ascii8bit_encoding(); + + if (RARRAY_LEN(ary) == 0) { + rb_raise(rb_eArgError, "no arguments given"); + } + + for (i = 0; i < RARRAY_LEN(ary); i++) { + VALUE str = RARRAY_AREF(ary, i); + VALUE buf; + char *p, *end; + rb_encoding *src_enc; + + 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); + end = p + RSTRING_LEN(str); + + buf = rb_reg_preprocess(p, end, src_enc, &fixed_enc, err, options); + + if (NIL_P(buf)) + rb_raise(rb_eArgError, "%s", err); + + if (fixed_enc != 0) { + if (regexp_enc != 0 && regexp_enc != fixed_enc) { + rb_raise(rb_eRegexpError, "encoding mismatch in dynamic regexp : %s and %s", + rb_enc_name(regexp_enc), rb_enc_name(fixed_enc)); + } + regexp_enc = fixed_enc; + } + + if (!result) + result = rb_str_new3(str); + else + rb_str_buf_append(result, str); + } + if (regexp_enc) { + rb_enc_associate(result, regexp_enc); + } + + 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) +{ + struct RRegexp *re = RREGEXP(obj); + VALUE unescaped; + rb_encoding *fixed_enc = 0; + rb_encoding *a_enc = rb_ascii8bit_encoding(); + + rb_reg_initialize_check(obj); + + if (rb_enc_dummy_p(enc)) { + errcpy(err, "can't make regexp with dummy encoding"); + return -1; + } + + unescaped = rb_reg_preprocess(s, s+len, enc, &fixed_enc, err, options); + if (NIL_P(unescaped)) + return -1; + + if (fixed_enc) { + if ((fixed_enc != enc && (options & ARG_ENCODING_FIXED)) || + (fixed_enc != a_enc && (options & ARG_ENCODING_NONE))) { + errcpy(err, "incompatible character encoding"); + return -1; + } + if (fixed_enc != a_enc) { + options |= ARG_ENCODING_FIXED; + enc = fixed_enc; + } + } + else if (!(options & ARG_ENCODING_FIXED)) { + enc = rb_usascii_encoding(); + } + + rb_enc_associate((VALUE)re, enc); + if ((options & ARG_ENCODING_FIXED) || fixed_enc) { + 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); + if (!re->ptr) return -1; + if (RBASIC_CLASS(obj) == rb_cRegexp) { + OBJ_FREEZE(obj); + } + RB_GC_GUARD(unescaped); + return 0; +} + +static void +reg_set_source(VALUE reg, VALUE str, rb_encoding *enc) +{ + rb_encoding *regenc = rb_enc_get(reg); + + if (regenc != enc) { + VALUE dup = rb_str_dup(str); + str = rb_enc_associate(dup, enc = regenc); + } + 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) +{ + 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 (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; + } + enc = ascii8bit; + } + } + ret = rb_reg_initialize(obj, RSTRING_PTR(str), RSTRING_LEN(str), enc, + options, err, sourcefile, sourceline); + if (ret == 0) reg_set_source(obj, str, str_enc); + return ret; +} + +VALUE +rb_reg_s_alloc(VALUE klass) +{ + NEWOBJ_OF(re, struct RRegexp, klass, T_REGEXP, sizeof(struct RRegexp)); re->ptr = 0; - re->len = 0; - re->str = 0; + RB_OBJ_WRITE(re, &re->src, 0); + re->usecnt = 0; return (VALUE)re; } VALUE -rb_reg_new(s, len, options) - const char *s; - long len; - int options; +rb_reg_alloc(void) { - VALUE re = rb_reg_s_alloc(rb_cRegexp); + return rb_reg_s_alloc(rb_cRegexp); +} - rb_reg_initialize(re, s, len, options); - return (VALUE)re; +VALUE +rb_reg_new_str(VALUE s, int options) +{ + return rb_reg_init_str(rb_reg_alloc(), s, options); } -static int case_cache; -static int kcode_cache; -static VALUE reg_cache; +VALUE +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); + } + + return re; +} + +static VALUE +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); + } + reg_set_source(re, s, enc); + + return re; +} VALUE -rb_reg_regcomp(str) - VALUE str; +rb_reg_new_from_values(long cnt, const VALUE *elements, int opt) +{ + 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 +rb_enc_reg_new(const char *s, long len, rb_encoding *enc, int options) { - if (reg_cache && RREGEXP(reg_cache)->len == RSTRING(str)->len - && case_cache == ruby_ignorecase - && kcode_cache == reg_kcode - && memcmp(RREGEXP(reg_cache)->str, RSTRING(str)->ptr, RSTRING(str)->len) == 0) - return reg_cache; + VALUE re = rb_reg_alloc(); + onig_errmsg_buffer err = ""; - case_cache = ruby_ignorecase; - kcode_cache = reg_kcode; - return reg_cache = rb_reg_new(RSTRING(str)->ptr, RSTRING(str)->len, - ruby_ignorecase); + if (rb_reg_initialize(re, s, len, enc, options, err, NULL, 0) != 0) { + 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))); + + return re; } -static int -rb_reg_cur_kcode(re) - VALUE re; +VALUE +rb_reg_new(const char *s, long len, int options) +{ + return rb_enc_reg_new(s, len, rb_ascii8bit_encoding(), options); +} + +VALUE +rb_reg_compile(VALUE str, int options, const char *sourcefile, int sourceline) { - if (FL_TEST(re, KCODE_FIXED)) { - return RBASIC(re)->flags & KCODE_MASK; + VALUE re = rb_reg_alloc(); + onig_errmsg_buffer err = ""; + + 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; } - return 0; + return re; } -static VALUE -rb_reg_hash(re) - VALUE re; +static VALUE reg_cache; + +VALUE +rb_reg_regcomp(VALUE str) { - int hashval, len; - char *p; + 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; - rb_reg_check(re); - hashval = RREGEXP(re)->ptr->options; - len = RREGEXP(re)->len; - p = RREGEXP(re)->str; - while (len--) { - hashval = hashval * 33 + *p++; + return reg_cache = rb_reg_new_str(str, 0); + } + else { + return rb_reg_new_str(str, 0); } - hashval = hashval + (hashval>>5); - - return INT2FIX(hashval); } -static VALUE -rb_reg_equal(re1, re2) - VALUE re1, re2; +static st_index_t reg_hash(VALUE re); +/* + * call-seq: + * hash -> integer + * + * Returns the integer hash value for +self+. + * + * Related: Object#hash. + * + */ + +VALUE +rb_reg_hash(VALUE re) +{ + st_index_t hashval = reg_hash(re); + return ST2FIX(hashval); +} + +static st_index_t +reg_hash(VALUE re) +{ + st_index_t hashval; + + rb_reg_check(re); + hashval = RREGEXP_PTR(re)->options; + hashval = rb_hash_uint(hashval, rb_memhash(RREGEXP_SRC_PTR(re), RREGEXP_SRC_LEN(re))); + return rb_hash_end(hashval); +} + + +/* + * call-seq: + * self == other -> true or false + * + * 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 + * + */ + +VALUE +rb_reg_equal(VALUE re1, VALUE re2) { if (re1 == re2) return Qtrue; - if (TYPE(re2) != T_REGEXP) return Qfalse; + if (!RB_TYPE_P(re2, T_REGEXP)) return Qfalse; rb_reg_check(re1); rb_reg_check(re2); - if (RREGEXP(re1)->len != RREGEXP(re2)->len) return Qfalse; - if (memcmp(RREGEXP(re1)->str, RREGEXP(re2)->str, RREGEXP(re1)->len) == 0 && - rb_reg_cur_kcode(re1) == rb_reg_cur_kcode(re2) && - RREGEXP(re1)->ptr->options == RREGEXP(re2)->ptr->options) { - return Qtrue; + if (FL_TEST(re1, KCODE_FIXED) != FL_TEST(re2, KCODE_FIXED)) return Qfalse; + if (RREGEXP_PTR(re1)->options != RREGEXP_PTR(re2)->options) return Qfalse; + if (RREGEXP_SRC_LEN(re1) != RREGEXP_SRC_LEN(re2)) return Qfalse; + if (ENCODING_GET(re1) != ENCODING_GET(re2)) return Qfalse; + return RBOOL(memcmp(RREGEXP_SRC_PTR(re1), RREGEXP_SRC_PTR(re2), RREGEXP_SRC_LEN(re1)) == 0); +} + +/* + * call-seq: + * hash -> integer + * + * Returns the integer hash value for +self+, + * based on the target string, regexp, match, and captures. + * + * See also Object#hash. + * + */ + +static VALUE +match_hash(VALUE match) +{ + 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))); + 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: + * self == other -> true or false + * + * Returns whether +other+ is another \MatchData object + * whose target string, regexp, match, and captures + * are the same as +self+. + */ + +static VALUE +match_equal(VALUE match1, VALUE match2) +{ + 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; + 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); } - return Qfalse; + 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); } -VALUE -rb_reg_match(re, str) - VALUE re, str; +static VALUE +reg_operand(VALUE s, int check) { - long start; + if (SYMBOL_P(s)) { + return rb_sym2str(s); + } + else if (RB_TYPE_P(s, T_STRING)) { + return s; + } + else { + return check ? rb_str_to_str(s) : rb_check_string_type(s); + } +} + +static long +reg_match_pos(VALUE re, VALUE *strp, long pos, VALUE* set_match) +{ + VALUE str = *strp; if (NIL_P(str)) { - rb_backref_set(Qnil); - return Qnil; + rb_backref_set(Qnil); + return -1; } - StringValue(str); - start = rb_reg_search(re, str, 0, 0); - if (start < 0) { - return Qnil; + *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); } - return LONG2FIX(start); + return rb_reg_search_set_match(re, str, pos, 0, 1, set_match); } +/* + * call-seq: + * self =~ other -> integer or nil + * + * Returns the integer index (in characters) of the first match + * for +self+ and +other+, or +nil+ if none; + * updates {Regexp-related global variables}[rdoc-ref:Regexp@Global+Variables]. + * + * /at/ =~ 'input data' # => 7 + * $~ # => #<MatchData "at"> + * /ax/ =~ 'input data' # => nil + * $~ # => nil + * + * Assigns named captures to local variables of the same names + * if and only if +self+: + * + * - Is a regexp literal; + * see {Regexp Literals}[rdoc-ref:syntax/literals.rdoc@Regexp+Literals]. + * - Does not contain interpolations; + * see {Regexp interpolation}[rdoc-ref:Regexp@Interpolation+Mode]. + * - Is at the left of the expression. + * + * Example: + * + * /(?<lhs>\w+)\s*=\s*(?<rhs>\w+)/ =~ ' x = y ' + * p lhs # => "x" + * p rhs # => "y" + * + * Assigns +nil+ if not matched: + * + * /(?<lhs>\w+)\s*=\s*(?<rhs>\w+)/ =~ ' x = ' + * p lhs # => nil + * p rhs # => nil + * + * Does not make local variable assignments if +self+ is not a regexp literal: + * + * r = /(?<foo>\w+)\s*=\s*(?<foo>\w+)/ + * r =~ ' x = y ' + * p foo # Undefined local variable + * p bar # Undefined local variable + * + * The assignment does not occur if the regexp is not at the left: + * + * ' x = y ' =~ /(?<foo>\w+)\s*=\s*(?<foo>\w+)/ + * p foo, foo # Undefined local variables + * + * A regexp interpolation, <tt>#{}</tt>, also disables + * the assignment: + * + * r = /(?<foo>\w+)/ + * /(?<foo>\w+)\s*=\s*#{r}/ =~ 'x = y' + * p foo # Undefined local variable + * + */ + VALUE -rb_reg_eqq(re, str) - VALUE re, str; +rb_reg_match(VALUE re, VALUE str) +{ + long pos = reg_match_pos(re, &str, 0, NULL); + if (pos < 0) return Qnil; + pos = rb_str_sublen(str, pos); + return LONG2FIX(pos); +} + +/* + * call-seq: + * self === other -> true or false + * + * Returns whether +self+ finds a match in +other+: + * + * /^[a-z]*$/ === 'HELLO' # => false + * /^[A-Z]*$/ === 'HELLO' # => true + * + * This method is called in case statements: + * + * s = 'HELLO' + * case s + * when /\A[a-z]*\z/; print "Lower case\n" + * when /\A[A-Z]*\z/; print "Upper case\n" + * else print "Mixed case\n" + * end # => "Upper case" + * + */ + +static VALUE +rb_reg_eqq(VALUE re, VALUE str) { long start; - if (TYPE(str) != T_STRING) { - str = rb_check_string_type(str); - if (NIL_P(str)) { - rb_backref_set(Qnil); - return Qfalse; - } + str = reg_operand(str, FALSE); + if (NIL_P(str)) { + rb_backref_set(Qnil); + return Qfalse; } - StringValue(str); start = rb_reg_search(re, str, 0, 0); - if (start < 0) { - return Qfalse; - } - return Qtrue; + return RBOOL(start >= 0); } + +/* + * call-seq: + * ~ rxp -> integer or nil + * + * Equivalent to <tt><i>rxp</i> =~ $_</tt>: + * + * $_ = "input data" + * ~ /at/ # => 7 + * + */ + VALUE -rb_reg_match2(re) - VALUE re; +rb_reg_match2(VALUE re) { long start; VALUE line = rb_lastline_get(); - if (TYPE(line) != T_STRING) { - rb_backref_set(Qnil); - return Qnil; + if (!RB_TYPE_P(line, T_STRING)) { + 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); } + +/* + * call-seq: + * match(string, offset = 0) -> matchdata or nil + * match(string, offset = 0) {|matchdata| ... } -> object + * + * With no block given, returns the MatchData object + * that describes the match, if any, or +nil+ if none; + * 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: + * + * /abra/.match('abracadabra') {|matchdata| p matchdata } + * # => #<MatchData "abra"> + * /abra/.match('abracadabra', 4) {|matchdata| p matchdata } + * # => #<MatchData "abra"> + * /abra/.match('abracadabra', 8) {|matchdata| p matchdata } + * # => nil + * /abra/.match('abracadabra', 8) {|marchdata| fail 'Cannot happen' } + * # => nil + * + * Output (from the first two blocks above): + * + * #<MatchData "abra"> + * #<MatchData "abra"> + * + * /(.)(.)(.)/.match("abc")[2] # => "b" + * /(.)(.)/.match("abc", 1)[2] # => "c" + * + */ + static VALUE -rb_reg_match_m(re, str) - VALUE re, str; +rb_reg_match_m(int argc, VALUE *argv, VALUE re) { - VALUE result = rb_reg_match(re, str); + VALUE result = Qnil, str, initpos; + long pos; - if (NIL_P(result)) return Qnil; - result = rb_backref_get(); + if (rb_scan_args(argc, argv, "11", &str, &initpos) == 2) { + pos = NUM2LONG(initpos); + } + else { + pos = 0; + } + + pos = reg_match_pos(re, &str, pos, &result); + if (pos < 0) { + rb_backref_set(Qnil); + return Qnil; + } rb_match_busy(result); + if (!NIL_P(result) && rb_block_given_p()) { + return rb_yield(result); + } return result; } +/* + * call-seq: + * match?(string) -> true or false + * match?(string, offset = 0) -> true or false + * + * Returns <code>true</code> or <code>false</code> to indicate whether the + * regexp is matched or not without updating $~ and other related variables. + * If the second parameter is present, it specifies the position in the string + * to begin the search. + * + * /R.../.match?("Ruby") # => true + * /R.../.match?("Ruby", 1) # => false + * /P.../.match?("Ruby") # => false + * $& # => nil + */ + static VALUE -rb_reg_initialize_m(argc, argv, self) - int argc; - VALUE *argv; - VALUE self; +rb_reg_match_m_p(int argc, VALUE *argv, VALUE re) +{ + long pos = rb_check_arity(argc, 1, 2) > 1 ? NUM2LONG(argv[1]) : 0; + return rb_reg_match_p(re, argv[0], pos); +} + +VALUE +rb_reg_match_p(VALUE re, VALUE str, long pos) { - const char *s; + 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); + } + } + + 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; +} + +/* + * Document-method: compile + * + * 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, timeout: nil) -> regexp + * Regexp.new(regexp, timeout: nil) -> regexp + * + * With argument +string+ given, returns a new regexp with the given string + * and options: + * + * r = Regexp.new('foo') # => /foo/ + * r.source # => "foo" + * r.options # => 0 + * + * Optional argument +options+ is one of the following: + * + * - 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. + * - 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+ + * arguments are ineffective. The timeout can be overridden by + * +timeout+ keyword. + * + * options = Regexp::MULTILINE + * 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 + * + */ + +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; + rb_encoding *enc = 0; + VALUE str, src, opts = Qundef, kwargs; + VALUE re = Qnil; - rb_check_frozen(self); - if (argc == 0 || argc > 3) { - rb_raise(rb_eArgError, "wrong number of argument"); - } - if (TYPE(argv[0]) == T_REGEXP) { - if (argc > 1) { - rb_warn("flags%s ignored", (argc == 3) ? " and encoding": ""); - } - rb_reg_check(argv[0]); - flags = RREGEXP(argv[0])->ptr->options & 0xf; - if (FL_TEST(argv[0], KCODE_FIXED)) { - switch (RBASIC(argv[0])->flags & KCODE_MASK) { - case KCODE_NONE: - flags |= 16; - break; - case KCODE_EUC: - flags |= 32; - break; - case KCODE_SJIS: - flags |= 48; - break; - case KCODE_UTF8: - flags |= 64; - break; - default: - break; - } - } - s = RREGEXP(argv[0])->str; - len = RREGEXP(argv[0])->len; + 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, &args->timeout); + } + + if (RB_TYPE_P(src, T_REGEXP)) { + re = src; + + if (!NIL_P(opts)) { + rb_warn("flags ignored"); + } + rb_reg_check(re); + flags = rb_reg_options(re); + str = RREGEXP_SRC(re); } else { - s = StringValuePtr(argv[0]); - len = RSTRING(argv[0])->len; - if (argc >= 2) { - if (FIXNUM_P(argv[1])) flags = FIX2INT(argv[1]); - else if (RTEST(argv[1])) flags = RE_OPTION_IGNORECASE; - } - if (argc == 3 && !NIL_P(argv[2])) { - char *kcode = StringValuePtr(argv[2]); - - flags &= ~0x70; - switch (kcode[0]) { - case 'n': case 'N': - flags |= 16; - break; - case 'e': case 'E': - flags |= 32; - break; - case 's': case 'S': - flags |= 48; - break; - case 'u': case 'U': - flags |= 64; - break; - default: - break; - } - } - } - rb_reg_initialize(self, s, len, flags); + 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); + else + rb_reg_init_str(self, str, flags); return self; } VALUE -rb_reg_quote(str) - VALUE str; +rb_reg_quote(VALUE str) { + rb_encoding *enc = rb_enc_get(str); char *s, *send, *t; VALUE tmp; - int c; - - s = RSTRING(str)->ptr; - send = s + RSTRING(str)->len; - for (; s < send; s++) { - c = *s; - if (ismbchar(c)) { - int n = mbclen(c); - - while (n-- && s < send) - s++; - s--; - 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 '\n': case '\r': - goto meta_found; - } + int c, clen; + int ascii_only = rb_enc_str_asciionly_p(str); + + s = RSTRING_PTR(str); + send = s + RSTRING_LEN(str); + while (s < send) { + c = rb_enc_ascget(s, send, &clen, enc); + 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 '$': + case ' ': case '#': + case '\t': case '\f': case '\v': case '\n': case '\r': + goto meta_found; + } + s += clen; } - return str; + tmp = rb_str_new3(str); + if (ascii_only) { + rb_enc_associate(tmp, rb_usascii_encoding()); + } + return tmp; meta_found: - tmp = rb_str_new(0, RSTRING(str)->len*2); - t = RSTRING(tmp)->ptr; + tmp = rb_str_new(0, RSTRING_LEN(str)*2); + if (ascii_only) { + rb_enc_associate(tmp, rb_usascii_encoding()); + } + else { + rb_enc_copy(tmp, str); + } + t = RSTRING_PTR(tmp); /* copy upto metacharacter */ - memcpy(t, RSTRING(str)->ptr, s - RSTRING(str)->ptr); - t += s - RSTRING(str)->ptr; - - for (; s < send; s++) { - c = *s; - if (ismbchar(c)) { - int n = mbclen(c); - - while (n-- && s < send) - *t++ = *s++; - s--; - continue; - } - switch (c) { - case '[': case ']': case '{': case '}': - case '(': case ')': case '|': case '-': - case '*': case '.': case '\\': - case '?': case '+': case '^': case '$': - case '#': - *t++ = '\\'; - break; - case ' ': - *t++ = '\\'; - *t++ = ' '; - continue; - case '\t': - *t++ = '\\'; - *t++ = 't'; - continue; - case '\n': - *t++ = '\\'; - *t++ = 'n'; - continue; - case '\r': - *t++ = '\\'; - *t++ = 'r'; - continue; - case '\f': - *t++ = '\\'; - *t++ = 'f'; - continue; - } - *t++ = c; - } - rb_str_resize(tmp, t - RSTRING(tmp)->ptr); - OBJ_INFECT(tmp, str); + const char *p = RSTRING_PTR(str); + memcpy(t, p, s - p); + t += s - p; + + while (s < send) { + c = rb_enc_ascget(s, send, &clen, enc); + if (c == -1) { + int n = mbclen(s, send, enc); + + 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 '#': + t += rb_enc_mbcput('\\', t, enc); + break; + case ' ': + t += rb_enc_mbcput('\\', t, enc); + t += rb_enc_mbcput(' ', t, enc); + continue; + case '\t': + t += rb_enc_mbcput('\\', t, enc); + t += rb_enc_mbcput('t', t, enc); + continue; + case '\n': + t += rb_enc_mbcput('\\', t, enc); + t += rb_enc_mbcput('n', t, enc); + continue; + case '\r': + t += rb_enc_mbcput('\\', t, enc); + t += rb_enc_mbcput('r', t, enc); + continue; + case '\f': + t += rb_enc_mbcput('\\', t, enc); + t += rb_enc_mbcput('f', t, enc); + continue; + case '\v': + t += rb_enc_mbcput('\\', t, enc); + t += rb_enc_mbcput('v', t, enc); + continue; + } + t += rb_enc_mbcput(c, t, enc); + } + rb_str_resize(tmp, t - RSTRING_PTR(tmp)); return tmp; } -static VALUE -rb_reg_s_quote(argc, argv) - int argc; - VALUE *argv; -{ - VALUE str, kcode; - int kcode_saved = reg_kcode; - rb_scan_args(argc, argv, "11", &str, &kcode); - if (!NIL_P(kcode)) { - rb_set_kcode(StringValuePtr(kcode)); - curr_kcode = reg_kcode; - reg_kcode = kcode_saved; - } - StringValue(str); - str = rb_reg_quote(str); - kcode_reset_option(); - return str; -} +/* + * call-seq: + * Regexp.escape(string) -> new_string + * + * Returns a new string that escapes any characters + * that have special meaning in a regular expression: + * + * s = Regexp.escape('\*?{}.') # => "\\\\\\*\\?\\{\\}\\." + * + * For any string +s+, this call returns a MatchData object: + * + * r = Regexp.new(Regexp.escape(s)) # => /\\\\\\\*\\\?\\\{\\\}\\\./ + * r.match(s) # => #<MatchData "\\\\\\*\\?\\{\\}\\."> + * + */ -int -rb_kcode() +static VALUE +rb_reg_s_quote(VALUE c, VALUE str) { - switch (reg_kcode) { - case KCODE_EUC: - return MBCTYPE_EUC; - case KCODE_SJIS: - return MBCTYPE_SJIS; - case KCODE_UTF8: - return MBCTYPE_UTF8; - case KCODE_NONE: - return MBCTYPE_ASCII; - } - rb_bug("wrong reg_kcode value (0x%x)", reg_kcode); -} - -static int -rb_reg_get_kcode(re) - VALUE re; -{ - switch (RBASIC(re)->flags & KCODE_MASK) { - case KCODE_NONE: - return 16; - case KCODE_EUC: - return 32; - case KCODE_SJIS: - return 48; - case KCODE_UTF8: - return 64; - default: - return 0; - } + return rb_reg_quote(reg_operand(str, TRUE)); } int -rb_reg_options(re) - VALUE re; +rb_reg_options(VALUE re) { int options; rb_reg_check(re); - options = RREGEXP(re)->ptr->options & - (RE_OPTION_IGNORECASE|RE_OPTION_MULTILINE|RE_OPTION_EXTENDED); - if (FL_TEST(re, KCODE_FIXED)) { - options |= rb_reg_get_kcode(re); - } + options = RREGEXP_PTR(re)->options & ARG_REG_OPTION_MASK; + if (RBASIC(re)->flags & KCODE_FIXED) options |= ARG_ENCODING_FIXED; + if (RBASIC(re)->flags & REG_ENCODING_NONE) options |= ARG_ENCODING_NONE; return options; } static VALUE -rb_reg_s_union(argc, argv) - int argc; - VALUE *argv; +rb_check_regexp_type(VALUE re) { + return rb_check_convert_type(re, T_REGEXP, "Regexp", "to_regexp"); +} + +/* + * call-seq: + * Regexp.try_convert(object) -> regexp or nil + * + * Returns +object+ if it is a regexp: + * + * Regexp.try_convert(/re/) # => /re/ + * + * Otherwise if +object+ responds to <tt>:to_regexp</tt>, + * calls <tt>object.to_regexp</tt> and returns the result. + * + * Returns +nil+ if +object+ does not respond to <tt>:to_regexp</tt>. + * + * Regexp.try_convert('re') # => nil + * + * Raises an exception unless <tt>object.to_regexp</tt> returns a regexp. + * + */ +static VALUE +rb_reg_s_try_convert(VALUE dummy, VALUE re) +{ + return rb_check_regexp_type(re); +} + +static VALUE +rb_reg_s_union(VALUE self, VALUE args0) +{ + long argc = RARRAY_LEN(args0); + if (argc == 0) { VALUE args[1]; args[0] = rb_str_new2("(?!)"); return rb_class_new_instance(1, args, rb_cRegexp); } else if (argc == 1) { - VALUE v; - v = rb_check_convert_type(argv[0], T_REGEXP, "Regexp", "to_regexp"); - if (!NIL_P(v)) - return v; + VALUE arg = rb_ary_entry(args0, 0); + VALUE re = rb_check_regexp_type(arg); + if (!NIL_P(re)) + return re; else { - VALUE args[1]; - args[0] = rb_reg_s_quote(argc, argv); - return rb_class_new_instance(1, args, rb_cRegexp); + VALUE quoted; + quoted = rb_reg_s_quote(Qnil, arg); + return rb_reg_new_str(quoted, 0); } } else { - int i, kcode = -1; - VALUE kcode_re = Qnil; + int i; VALUE source = rb_str_buf_new(0); - VALUE args[3]; + 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); + if (0 < i) - rb_str_buf_cat2(source, "|"); - v = rb_check_convert_type(argv[i], T_REGEXP, "Regexp", "to_regexp"); + rb_str_buf_cat_ascii(source, "|"); + + v = rb_check_regexp_type(e); if (!NIL_P(v)) { - if (FL_TEST(v, KCODE_FIXED)) { - if (kcode == -1) { - kcode_re = v; - kcode = RBASIC(v)->flags & KCODE_MASK; - } - else if ((RBASIC(v)->flags & KCODE_MASK) != kcode) { - volatile VALUE str1, str2; - str1 = rb_inspect(kcode_re); - str2 = rb_inspect(v); - rb_raise(rb_eArgError, "mixed kcode: %s and %s", - RSTRING(str1)->ptr, RSTRING(str2)->ptr); - } + rb_encoding *enc = rb_enc_get(v); + if (!rb_enc_asciicompat(enc)) { + if (!has_ascii_incompat) + has_ascii_incompat = enc; + else if (has_ascii_incompat != enc) + rb_raise(rb_eArgError, "incompatible encodings: %s and %s", + rb_enc_name(has_ascii_incompat), rb_enc_name(enc)); + } + else if (rb_reg_fixed_encoding_p(v)) { + if (!has_ascii_compat_fixed) + has_ascii_compat_fixed = enc; + else if (has_ascii_compat_fixed != enc) + rb_raise(rb_eArgError, "incompatible encodings: %s and %s", + rb_enc_name(has_ascii_compat_fixed), rb_enc_name(enc)); + } + else { + has_asciionly = 1; } - v = rb_reg_to_s(v); + v = rb_reg_str_with_term(v, -1); } else { - args[0] = argv[i]; - v = rb_reg_s_quote(1, args); + rb_encoding *enc; + StringValue(e); + enc = rb_enc_get(e); + if (!rb_enc_asciicompat(enc)) { + if (!has_ascii_incompat) + has_ascii_incompat = enc; + else if (has_ascii_incompat != enc) + rb_raise(rb_eArgError, "incompatible encodings: %s and %s", + rb_enc_name(has_ascii_incompat), rb_enc_name(enc)); + } + else if (rb_enc_str_asciionly_p(e)) { + has_asciionly = 1; + } + else { + if (!has_ascii_compat_fixed) + has_ascii_compat_fixed = enc; + else if (has_ascii_compat_fixed != enc) + 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); } - rb_str_buf_append(source, v); + if (has_ascii_incompat) { + if (has_asciionly) { + rb_raise(rb_eArgError, "ASCII incompatible encoding: %s", + rb_enc_name(has_ascii_incompat)); + } + if (has_ascii_compat_fixed) { + rb_raise(rb_eArgError, "incompatible encodings: %s and %s", + rb_enc_name(has_ascii_incompat), rb_enc_name(has_ascii_compat_fixed)); + } + } + + if (i == 0) { + rb_enc_copy(source, v); + } + rb_str_append(source, v); } - args[0] = source; - args[1] = Qnil; - switch (kcode) { - case -1: - args[2] = Qnil; - break; - case KCODE_NONE: - args[2] = rb_str_new2("n"); - break; - case KCODE_EUC: - args[2] = rb_str_new2("e"); - break; - case KCODE_SJIS: - args[2] = rb_str_new2("s"); - break; - case KCODE_UTF8: - args[2] = rb_str_new2("u"); - break; + + if (has_ascii_incompat) { + result_enc = has_ascii_incompat; + } + else if (has_ascii_compat_fixed) { + result_enc = has_ascii_compat_fixed; } - return rb_class_new_instance(3, args, rb_cRegexp); + else { + result_enc = rb_ascii8bit_encoding(); + } + + rb_enc_associate(source, result_enc); + return rb_class_new_instance(1, &source, rb_cRegexp); } } +/* + * call-seq: + * Regexp.union(*patterns) -> regexp + * Regexp.union(array_of_patterns) -> regexp + * + * Returns a new regexp that is the union of the given patterns: + * + * r = Regexp.union(%w[cat dog]) # => /cat|dog/ + * r.match('cat') # => #<MatchData "cat"> + * r.match('dog') # => #<MatchData "dog"> + * r.match('cog') # => nil + * + * For each pattern that is a string, <tt>Regexp.new(pattern)</tt> is used: + * + * Regexp.union('penzance') # => /penzance/ + * Regexp.union('a+b*c') # => /a\+b\*c/ + * Regexp.union('skiing', 'sledding') # => /skiing|sledding/ + * Regexp.union(['skiing', 'sledding']) # => /skiing|sledding/ + * + * For each pattern that is a regexp, it is used as is, + * including its flags: + * + * Regexp.union(/foo/i, /bar/m, /baz/x) + * # => /(?i-mx:foo)|(?m-ix:bar)|(?x-mi:baz)/ + * Regexp.union([/foo/i, /bar/m, /baz/x]) + * # => /(?i-mx:foo)|(?m-ix:bar)|(?x-mi:baz)/ + * + * With no arguments, returns <tt>/(?!)/</tt>: + * + * Regexp.union # => /(?!)/ + * + * If any regexp pattern contains captures, the behavior is unspecified. + * + */ static VALUE -rb_reg_init_copy(copy, re) - VALUE copy, re; +rb_reg_s_union_m(VALUE self, VALUE args) { - if (copy == re) return copy; - rb_check_frozen(copy); - /* need better argument type check */ - if (!rb_obj_is_instance_of(re, rb_obj_class(copy))) { - rb_raise(rb_eTypeError, "wrong argument type"); + VALUE v; + if (RARRAY_LEN(args) == 1 && + !NIL_P(v = rb_check_array_type(rb_ary_entry(args, 0)))) { + return rb_reg_s_union(self, v); + } + 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); - rb_reg_initialize(copy, RREGEXP(re)->str, RREGEXP(re)->len, - rb_reg_options(re)); - return copy; + return reg_copy(copy, re); } -VALUE -rb_reg_regsub(str, src, regs) - VALUE str, src; - struct re_registers *regs; +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, c; - int no; + char *p, *s, *e; + int no, clen; + rb_encoding *str_enc = rb_enc_get(str); + rb_encoding *src_enc = rb_enc_get(src); + int acompat = rb_enc_asciicompat(str_enc); + long n; +#define ASCGET(s,e,cl) (acompat ? (*(cl)=1,ISASCII((s)[0])?(s)[0]:-1) : rb_enc_ascget((s), (e), (cl), str_enc)) - p = s = RSTRING(str)->ptr; - e = s + RSTRING(str)->len; + RSTRING_GETMEM(str, s, n); + p = s; + e = s + n; while (s < e) { - char *ss = s; - - c = *s++; - if (ismbchar(c)) { - s += mbclen(c) - 1; - continue; - } - if (c != '\\' || s == e) continue; - - if (!val) { - val = rb_str_buf_new(ss-p); - rb_str_buf_cat(val, p, ss-p); - } - else { - rb_str_buf_cat(val, p, ss-p); - } - - c = *s++; - p = s; - switch (c) { - case '0': case '1': case '2': case '3': case '4': - case '5': case '6': case '7': case '8': case '9': - no = c - '0'; - break; - case '&': - no = 0; - break; - - case '`': - rb_str_buf_cat(val, RSTRING(src)->ptr, BEG(0)); - continue; - - case '\'': - rb_str_buf_cat(val, RSTRING(src)->ptr+END(0), RSTRING(src)->len-END(0)); - continue; - - case '+': - no = regs->num_regs-1; - while (BEG(no) == -1 && no > 0) no--; - if (no == 0) continue; - break; - - case '\\': - rb_str_buf_cat(val, s-1, 1); - continue; - - default: - rb_str_buf_cat(val, s-2, 2); - continue; - } - - if (no >= 0) { - if (no >= regs->num_regs) continue; - if (BEG(no) == -1) continue; - rb_str_buf_cat(val, RSTRING(src)->ptr+BEG(no), END(no)-BEG(no)); - } + int c = ASCGET(s, e, &clen); + char *ss; + + if (c == -1) { + s += mbclen(s, e, str_enc); + continue; + } + ss = s; + s += clen; + + if (c != '\\' || s == e) continue; + + 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); + p = s; + 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': + if (!NIL_P(regexp) && onig_noname_group_capture_is_active(RREGEXP_PTR(regexp))) { + no = c - '0'; + } + else { + continue; + } + break; + + case 'k': + if (s < e && ASCGET(s, e, &clen) == '<') { + char *name, *name_end; + + name_end = name = s + clen; + while (name_end < e) { + c = ASCGET(name_end, e, &clen); + if (c == '>') break; + 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)); + 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; + } + else { + rb_raise(rb_eRuntimeError, "invalid group name reference format"); + } + } + + rb_enc_str_buf_cat(val, ss, s-ss, str_enc); + continue; + + case '0': + 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)+end[0], RSTRING_LEN(src)-end[0], src_enc); + continue; + + 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; + + default: + rb_enc_str_buf_cat(val, ss, s-ss, str_enc); + continue; + } + + 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; if (p < e) { - if (!val) { - val = rb_str_buf_new(e-p); - rb_str_buf_cat(val, p, e-p); - } - else { - rb_str_buf_cat(val, p, e-p); - } + rb_enc_str_buf_cat(val, p, e-p, str_enc); } - if (!val) return str; return val; +#undef ASCGET } -const char* -rb_get_kcode() -{ - switch (reg_kcode) { - case KCODE_SJIS: - return "SJIS"; - case KCODE_EUC: - return "EUC"; - case KCODE_UTF8: - return "UTF8"; - default: - return "NONE"; - } -} - -static VALUE -kcode_getter() -{ - return rb_str_new2(rb_get_kcode()); -} - -void -rb_set_kcode(code) - const char *code; +VALUE +rb_reg_regsub(VALUE str, VALUE src, struct re_registers *regs, VALUE regexp) { - if (code == 0) goto set_no_conversion; - - switch (code[0]) { - case 'E': - case 'e': - reg_kcode = KCODE_EUC; - re_mbcinit(MBCTYPE_EUC); - break; - case 'S': - case 's': - reg_kcode = KCODE_SJIS; - re_mbcinit(MBCTYPE_SJIS); - break; - case 'U': - case 'u': - reg_kcode = KCODE_UTF8; - re_mbcinit(MBCTYPE_UTF8); - break; - default: - case 'N': - case 'n': - case 'A': - case 'a': - set_no_conversion: - reg_kcode = KCODE_NONE; - re_mbcinit(MBCTYPE_ASCII); - break; - } + return do_regsub(str, src, regexp, regs->num_regs, regs->beg, regs->end); } -static void -kcode_setter(val) - VALUE val; +VALUE +rb_reg_regsub_match(VALUE str, VALUE src, VALUE match) { - may_need_recompile = 1; - rb_set_kcode(StringValuePtr(val)); + return do_regsub(str, src, RMATCH(match)->regexp, + RMATCH_NREGS(match), RMATCH_BEG_PTR(match), RMATCH_END_PTR(match)); } static VALUE -ignorecase_getter() +ignorecase_getter(ID _x, VALUE *_y) { - return ruby_ignorecase?Qtrue:Qfalse; + rb_category_warn(RB_WARN_CATEGORY_DEPRECATED, "variable $= is no longer effective"); + return Qfalse; } static void -ignorecase_setter(val, id) - VALUE val; - ID id; +ignorecase_setter(VALUE val, ID id, VALUE *_) { - rb_warn("modifying %s is deprecated", rb_id2name(id)); - may_need_recompile = 1; - ruby_ignorecase = RTEST(val); + rb_category_warn(RB_WARN_CATEGORY_DEPRECATED, "variable $= is no longer effective; ignored"); } static VALUE -match_getter() +match_getter(void) { VALUE match = rb_backref_get(); @@ -1746,66 +4773,231 @@ match_getter() return match; } +static VALUE +get_LAST_MATCH_INFO(ID _x, VALUE *_y) +{ + return match_getter(); +} + static void -match_setter(val) - VALUE val; +match_setter(VALUE val, ID _x, VALUE *_y) { if (!NIL_P(val)) { - Check_Type(val, T_MATCH); + Check_Type(val, T_MATCH); } rb_backref_set(val); } +/* + * call-seq: + * Regexp.last_match -> matchdata or nil + * Regexp.last_match(n) -> string or nil + * Regexp.last_match(name) -> string or nil + * + * With no argument, returns the value of <tt>$~</tt>, + * which is the result of the most recent pattern match + * (see {Regexp global variables}[rdoc-ref:Regexp@Global+Variables]): + * + * /c(.)t/ =~ 'cat' # => 0 + * Regexp.last_match # => #<MatchData "cat" 1:"a"> + * /a/ =~ 'foo' # => nil + * Regexp.last_match # => nil + * + * With non-negative integer argument +n+, returns the _n_th field in the + * matchdata, if any, or nil if none: + * + * /c(.)t/ =~ 'cat' # => 0 + * Regexp.last_match(0) # => "cat" + * Regexp.last_match(1) # => "a" + * Regexp.last_match(2) # => nil + * + * With negative integer argument +n+, counts backwards from the last field: + * + * Regexp.last_match(-1) # => "a" + * + * With string or symbol argument +name+, + * returns the string value for the named capture, if any: + * + * /(?<lhs>\w+)\s*=\s*(?<rhs>\w+)/ =~ 'var = val' + * Regexp.last_match # => #<MatchData "var = val" lhs:"var"rhs:"val"> + * Regexp.last_match(:lhs) # => "var" + * Regexp.last_match('rhs') # => "val" + * Regexp.last_match('foo') # Raises IndexError. + * + */ + static VALUE -rb_reg_s_last_match(argc, argv) - int argc; - VALUE *argv; +rb_reg_s_last_match(int argc, VALUE *argv, VALUE _) { - VALUE nth; - - if (rb_scan_args(argc, argv, "01", &nth) == 1) { - return rb_reg_nth_match(NUM2INT(nth), rb_backref_get()); + if (rb_check_arity(argc, 0, 1) == 1) { + VALUE match = rb_backref_get(); + int n; + if (NIL_P(match)) return Qnil; + n = match_backref_number(match, argv[0]); + return rb_reg_nth_match(n, match); } return match_getter(); } +static void +re_warn(const char *s) +{ + rb_warn("%s", s); +} + +// This function is periodically called during regexp matching +bool +rb_reg_timeout_p(regex_t *reg, void *end_time_) +{ + rb_hrtime_t *end_time = (rb_hrtime_t *)end_time_; + + if (*end_time == 0) { + // This is the first time to check interrupts; + // just measure the current time and determine the end time + // if timeout is set. + rb_hrtime_t timelimit = reg->timelimit; + + if (!timelimit) { + // no per-object timeout. + timelimit = rb_reg_match_time_limit; + } + + if (timelimit) { + *end_time = rb_hrtime_add(timelimit, rb_hrtime_now()); + } + else { + // no timeout is set + *end_time = RB_HRTIME_MAX; + } + } + else { + if (*end_time < rb_hrtime_now()) { + // Timeout has exceeded + return true; + } + } + + return false; +} + +/* + * call-seq: + * Regexp.timeout -> float or nil + * + * It returns the current default timeout interval for Regexp matching in second. + * +nil+ means no default timeout configuration. + */ + +static VALUE +rb_reg_s_timeout_get(VALUE dummy) +{ + double d = hrtime2double(rb_reg_match_time_limit); + if (d == 0.0) return Qnil; + return DBL2NUM(d); +} + +/* + * call-seq: + * Regexp.timeout = float or nil + * + * It sets the default timeout interval for Regexp matching in second. + * +nil+ means no default timeout configuration. + * This configuration is process-global. If you want to set timeout for + * each Regexp, use +timeout+ keyword for <code>Regexp.new</code>. + * + * Regexp.timeout = 1 + * /^a*b?a*$/ =~ "a" * 100000 + "x" #=> regexp match timeout (RuntimeError) + */ + +static VALUE +rb_reg_s_timeout_set(VALUE dummy, VALUE timeout) +{ + rb_ractor_ensure_main_ractor("can not access Regexp.timeout from non-main Ractors"); + + set_timeout(&rb_reg_match_time_limit, timeout); + + return timeout; +} + +/* + * call-seq: + * rxp.timeout -> float or nil + * + * It returns the timeout interval for Regexp matching in second. + * +nil+ means no default timeout configuration. + * + * This configuration is per-object. The global configuration set by + * Regexp.timeout= is ignored if per-object configuration is set. + * + * re = Regexp.new("^a*b?a*$", timeout: 1) + * re.timeout #=> 1.0 + * re =~ "a" * 100000 + "x" #=> regexp match timeout (RuntimeError) + */ + +static VALUE +rb_reg_timeout_get(VALUE re) +{ + rb_reg_check(re); + double d = hrtime2double(RREGEXP_PTR(re)->timelimit); + if (d == 0.0) return Qnil; + return DBL2NUM(d); +} + +/* + * Document-class: RegexpError + * + * Raised when given an invalid regexp expression. + * + * Regexp.new("?") + * + * <em>raises the exception:</em> + * + * RegexpError: target of repeat operator is not specified: /?/ + */ + +/* + * Document-class: Regexp + * + * :include: doc/_regexp.rdoc + */ + void -Init_Regexp() +Init_Regexp(void) { rb_eRegexpError = rb_define_class("RegexpError", rb_eStandardError); - re_set_casetable(casetable); -#if DEFAULT_KCODE == KCODE_EUC - re_mbcinit(MBCTYPE_EUC); -#else -#if DEFAULT_KCODE == KCODE_SJIS - re_mbcinit(MBCTYPE_SJIS); -#else -#if DEFAULT_KCODE == KCODE_UTF8 - re_mbcinit(MBCTYPE_UTF8); -#else - re_mbcinit(MBCTYPE_ASCII); -#endif -#endif -#endif + onigenc_set_default_encoding(ONIG_ENCODING_ASCII); + onig_set_warn_func(re_warn); + onig_set_verb_warn_func(re_warn); - rb_define_virtual_variable("$~", match_getter, match_setter); + rb_define_virtual_variable("$~", get_LAST_MATCH_INFO, match_setter); rb_define_virtual_variable("$&", last_match_getter, 0); rb_define_virtual_variable("$`", prematch_getter, 0); rb_define_virtual_variable("$'", postmatch_getter, 0); rb_define_virtual_variable("$+", last_paren_match_getter, 0); + rb_gvar_ractor_local("$~"); + rb_gvar_ractor_local("$&"); + 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("$=", ignorecase_getter, ignorecase_setter); - rb_define_virtual_variable("$KCODE", kcode_getter, kcode_setter); - rb_define_virtual_variable("$-K", kcode_getter, kcode_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, "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, -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); @@ -1815,39 +5007,69 @@ Init_Regexp() rb_define_method(rb_cRegexp, "=~", rb_reg_match, 1); rb_define_method(rb_cRegexp, "===", rb_reg_eqq, 1); rb_define_method(rb_cRegexp, "~", rb_reg_match2, 0); - rb_define_method(rb_cRegexp, "match", rb_reg_match_m, 1); + rb_define_method(rb_cRegexp, "match", rb_reg_match_m, -1); + rb_define_method(rb_cRegexp, "match?", rb_reg_match_m_p, -1); rb_define_method(rb_cRegexp, "to_s", rb_reg_to_s, 0); rb_define_method(rb_cRegexp, "inspect", rb_reg_inspect, 0); rb_define_method(rb_cRegexp, "source", rb_reg_source, 0); rb_define_method(rb_cRegexp, "casefold?", rb_reg_casefold_p, 0); rb_define_method(rb_cRegexp, "options", rb_reg_options_m, 0); - rb_define_method(rb_cRegexp, "kcode", rb_reg_kcode_m, 0); - - rb_define_const(rb_cRegexp, "IGNORECASE", INT2FIX(RE_OPTION_IGNORECASE)); - rb_define_const(rb_cRegexp, "EXTENDED", INT2FIX(RE_OPTION_EXTENDED)); - rb_define_const(rb_cRegexp, "MULTILINE", INT2FIX(RE_OPTION_MULTILINE)); + rb_define_method(rb_cRegexp, "encoding", rb_obj_encoding, 0); /* in encoding.c */ + rb_define_method(rb_cRegexp, "fixed_encoding?", rb_reg_fixed_encoding_p, 0); + rb_define_method(rb_cRegexp, "names", rb_reg_names, 0); + 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); + + /* see Regexp.options and Regexp.new */ + rb_define_const(rb_cRegexp, "IGNORECASE", INT2FIX(ONIG_OPTION_IGNORECASE)); + /* see Regexp.options and Regexp.new */ + rb_define_const(rb_cRegexp, "EXTENDED", INT2FIX(ONIG_OPTION_EXTEND)); + /* see Regexp.options and Regexp.new */ + rb_define_const(rb_cRegexp, "MULTILINE", INT2FIX(ONIG_OPTION_MULTILINE)); + /* see Regexp.options and Regexp.new */ + rb_define_const(rb_cRegexp, "FIXEDENCODING", INT2FIX(ARG_ENCODING_FIXED)); + /* see Regexp.options and Regexp.new */ + rb_define_const(rb_cRegexp, "NOENCODING", INT2FIX(ARG_ENCODING_NONE)); rb_global_variable(®_cache); rb_cMatch = rb_define_class("MatchData", rb_cObject); - rb_define_global_const("MatchingData", rb_cMatch); rb_define_alloc_func(rb_cMatch, match_alloc); rb_undef_method(CLASS_OF(rb_cMatch), "new"); + rb_undef_method(CLASS_OF(rb_cMatch), "allocate"); rb_define_method(rb_cMatch, "initialize_copy", match_init_copy, 1); + rb_define_method(rb_cMatch, "regexp", match_regexp, 0); + rb_define_method(rb_cMatch, "names", match_names, 0); rb_define_method(rb_cMatch, "size", match_size, 0); 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); + rb_define_method(rb_cMatch, "match_length", match_nth_length, 1); 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, "select", match_select, -1); + 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); rb_define_method(rb_cMatch, "to_s", match_to_s, 0); - rb_define_method(rb_cMatch, "inspect", rb_any_to_s, 0); + rb_define_method(rb_cMatch, "inspect", match_inspect, 0); rb_define_method(rb_cMatch, "string", match_string, 0); + 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); } |