?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
* io.c (rb_open_file): encoding in mode string was ignored if perm is
[ruby-svn.git] / re.c
blobde7ebeb41bb2094c9bff521530a93adf54265cd8
1 /**********************************************************************
2
3 re.c -
4
5 $Author$
6 created at: Mon Aug 9 18:24:49 JST 1993
7
8 Copyright (C) 1993-2007 Yukihiro Matsumoto
9
10 **********************************************************************/
11
12 #include "ruby/ruby.h"
13 #include "ruby/re.h"
14 #include "ruby/encoding.h"
15 #include "ruby/util.h"
16 #include "regint.h"
17 #include <ctype.h>
18
19 VALUE rb_eRegexpError;
20
21 typedef char onig_errmsg_buffer[ONIG_MAX_ERROR_MESSAGE_LEN];
22
23 #define BEG(no) regs->beg[no]
24 #define END(no) regs->end[no]
25
26 #if 'a' == 97 /* it's ascii */
27 static const char casetable[] = {
28 '\000', '\001', '\002', '\003', '\004', '\005', '\006', '\007',
29 '\010', '\011', '\012', '\013', '\014', '\015', '\016', '\017',
30 '\020', '\021', '\022', '\023', '\024', '\025', '\026', '\027',
31 '\030', '\031', '\032', '\033', '\034', '\035', '\036', '\037',
32 /* ' ' '!' '"' '#' '$' '%' '&' ''' */
33 '\040', '\041', '\042', '\043', '\044', '\045', '\046', '\047',
34 /* '(' ')' '*' '+' ',' '-' '.' '/' */
35 '\050', '\051', '\052', '\053', '\054', '\055', '\056', '\057',
36 /* '0' '1' '2' '3' '4' '5' '6' '7' */
37 '\060', '\061', '\062', '\063', '\064', '\065', '\066', '\067',
38 /* '8' '9' ':' ';' '<' '=' '>' '?' */
39 '\070', '\071', '\072', '\073', '\074', '\075', '\076', '\077',
40 /* '@' 'A' 'B' 'C' 'D' 'E' 'F' 'G' */
41 '\100', '\141', '\142', '\143', '\144', '\145', '\146', '\147',
42 /* 'H' 'I' 'J' 'K' 'L' 'M' 'N' 'O' */
43 '\150', '\151', '\152', '\153', '\154', '\155', '\156', '\157',
44 /* 'P' 'Q' 'R' 'S' 'T' 'U' 'V' 'W' */
45 '\160', '\161', '\162', '\163', '\164', '\165', '\166', '\167',
46 /* 'X' 'Y' 'Z' '[' '\' ']' '^' '_' */
47 '\170', '\171', '\172', '\133', '\134', '\135', '\136', '\137',
48 /* '`' 'a' 'b' 'c' 'd' 'e' 'f' 'g' */
49 '\140', '\141', '\142', '\143', '\144', '\145', '\146', '\147',
50 /* 'h' 'i' 'j' 'k' 'l' 'm' 'n' 'o' */
51 '\150', '\151', '\152', '\153', '\154', '\155', '\156', '\157',
52 /* 'p' 'q' 'r' 's' 't' 'u' 'v' 'w' */
53 '\160', '\161', '\162', '\163', '\164', '\165', '\166', '\167',
54 /* 'x' 'y' 'z' '{' '|' '}' '~' */
55 '\170', '\171', '\172', '\173', '\174', '\175', '\176', '\177',
56 '\200', '\201', '\202', '\203', '\204', '\205', '\206', '\207',
57 '\210', '\211', '\212', '\213', '\214', '\215', '\216', '\217',
58 '\220', '\221', '\222', '\223', '\224', '\225', '\226', '\227',
59 '\230', '\231', '\232', '\233', '\234', '\235', '\236', '\237',
60 '\240', '\241', '\242', '\243', '\244', '\245', '\246', '\247',
61 '\250', '\251', '\252', '\253', '\254', '\255', '\256', '\257',
62 '\260', '\261', '\262', '\263', '\264', '\265', '\266', '\267',
63 '\270', '\271', '\272', '\273', '\274', '\275', '\276', '\277',
64 '\300', '\301', '\302', '\303', '\304', '\305', '\306', '\307',
65 '\310', '\311', '\312', '\313', '\314', '\315', '\316', '\317',
66 '\320', '\321', '\322', '\323', '\324', '\325', '\326', '\327',
67 '\330', '\331', '\332', '\333', '\334', '\335', '\336', '\337',
68 '\340', '\341', '\342', '\343', '\344', '\345', '\346', '\347',
69 '\350', '\351', '\352', '\353', '\354', '\355', '\356', '\357',
70 '\360', '\361', '\362', '\363', '\364', '\365', '\366', '\367',
71 '\370', '\371', '\372', '\373', '\374', '\375', '\376', '\377',
72 };
73 #else
74 # error >>> "You lose. You will need a translation table for your character set." <<<
75 #endif
76
77 int
78 rb_memcicmp(const void *x, const void *y, long len)
79 {
80 const unsigned char *p1 = x, *p2 = y;
81 int tmp;
82
83 while (len--) {
84 if ((tmp = casetable[(unsigned)*p1++] - casetable[(unsigned)*p2++]))
85 return tmp;
86 }
87 return 0;
88 }
89
90 #undef rb_memcmp
91
92 int
93 rb_memcmp(const void *p1, const void *p2, long len)
94 {
95 return memcmp(p1, p2, len);
96 }
97
98 static inline long
99 rb_memsearch_ss(const unsigned char *xs, long m, const unsigned char *ys, long n)
100 {
101 const unsigned char *x = xs, *xe = xs + m;
102 const unsigned char *y = ys, *ye = ys + n;
103 #ifndef VALUE_MAX
104 # if SIZEOF_VALUE == 8
105 # define VALUE_MAX 0xFFFFFFFFFFFFFFFFULL
106 # elif SIZEOF_VALUE == 4
107 # define VALUE_MAX 0xFFFFFFFFUL
108 # endif
109 #endif
110 VALUE hx, hy, mask = VALUE_MAX >> ((SIZEOF_VALUE - m) * CHAR_BIT);
111
112 if (m > SIZEOF_VALUE)
113 rb_bug("!!too long pattern string!!");
114
115 /* Prepare hash value */
116 for (hx = *x++, hy = *y++; x < xe; ++x, ++y) {
117 hx <<= CHAR_BIT;
118 hy <<= CHAR_BIT;
119 hx |= *x;
120 hy |= *y;
121 }
122 /* Searching */
123 while (hx != hy) {
124 if (y == ye)
125 return -1;
126 hy <<= CHAR_BIT;
127 hy |= *y;
128 hy &= mask;
129 y++;
130 }
131 return y - ys - m;
132 }
133
134 static inline long
135 rb_memsearch_qs(const unsigned char *xs, long m, const unsigned char *ys, long n)
136 {
137 const unsigned char *x = xs, *xe = xs + m;
138 const unsigned char *y = ys;
139 VALUE i, qstable[256];
140
141 /* Preprocessing */
142 for (i = 0; i < 256; ++i)
143 qstable[i] = m + 1;
144 for (; x < xe; ++x)
145 qstable[*x] = xe - x;
146 /* Searching */
147 for (; y + m <= ys + n; y += *(qstable + y[m])) {
148 if (*xs == *y && memcmp(xs, y, m) == 0)
149 return y - ys;
150 }
151 return -1;
152 }
153
154 static inline unsigned int
155 rb_memsearch_qs_utf8_hash(const unsigned char *x)
156 {
157 register const unsigned int mix = 8353;
158 register unsigned int h = *x;
159 if (h < 0xC0) {
160 return h + 256;
161 }
162 else if (h < 0xE0) {
163 h *= mix;
164 h += x[1];
165 }
166 else if (h < 0xF0) {
167 h *= mix;
168 h += x[1];
169 h *= mix;
170 h += x[2];
171 }
172 else if (h < 0xF5) {
173 h *= mix;
174 h += x[1];
175 h *= mix;
176 h += x[2];
177 h *= mix;
178 h += x[3];
179 }
180 else {
181 return h + 256;
182 }
183 return (unsigned char)h;
184 }
185
186 static inline long
187 rb_memsearch_qs_utf8(const unsigned char *xs, long m, const unsigned char *ys, long n)
188 {
189 const unsigned char *x = xs, *xe = xs + m;
190 const unsigned char *y = ys;
191 VALUE i, qstable[512];
192
193 /* Preprocessing */
194 for (i = 0; i < 512; ++i) {
195 qstable[i] = m + 1;
196 }
197 for (; x < xe; ++x) {
198 qstable[rb_memsearch_qs_utf8_hash(x)] = xe - x;
199 }
200 /* Searching */
201 for (; y + m <= ys + n; y += qstable[rb_memsearch_qs_utf8_hash(y+m)]) {
202 if (*xs == *y && memcmp(xs, y, m) == 0)
203 return y - ys;
204 }
205 return -1;
206 }
207
208 long
209 rb_memsearch(const void *x0, long m, const void *y0, long n, rb_encoding *enc)
210 {
211 const unsigned char *x = x0, *y = y0;
212
213 if (m > n) return -1;
214 else if (m == n) {
215 return memcmp(x0, y0, m) == 0 ? 0 : -1;
216 }
217 else if (m < 1) {
218 return 0;
219 }
220 else if (m == 1) {
221 const unsigned char *ys = y, *ye = ys + n;
222 for (; y < ye; ++y) {
223 if (*x == *y)
224 return y - ys;
225 }
226 return -1;
227 }
228 else if (m <= SIZEOF_VALUE) {
229 return rb_memsearch_ss(x0, m, y0, n);
230 }
231 else if (enc == rb_utf8_encoding()){
232 return rb_memsearch_qs_utf8(x0, m, y0, n);
233 }
234 else {
235 return rb_memsearch_qs(x0, m, y0, n);
236 }
237 }
238
239 #define REG_LITERAL FL_USER5
240 #define REG_ENCODING_NONE FL_USER6
241
242 #define KCODE_FIXED FL_USER4
243
244 #define ARG_REG_OPTION_MASK \
245 (ONIG_OPTION_IGNORECASE|ONIG_OPTION_MULTILINE|ONIG_OPTION_EXTEND)
246 #define ARG_ENCODING_FIXED 16
247 #define ARG_ENCODING_NONE 32
248
249 static int
250 char_to_option(int c)
251 {
252 int val;
253
254 switch (c) {
255 case 'i':
256 val = ONIG_OPTION_IGNORECASE;
257 break;
258 case 'x':
259 val = ONIG_OPTION_EXTEND;
260 break;
261 case 'm':
262 val = ONIG_OPTION_MULTILINE;
263 break;
264 default:
265 val = 0;
266 break;
267 }
268 return val;
269 }
270
271 static char *
272 option_to_str(char str[4], int options)
273 {
274 char *p = str;
275 if (options & ONIG_OPTION_MULTILINE) *p++ = 'm';
276 if (options & ONIG_OPTION_IGNORECASE) *p++ = 'i';
277 if (options & ONIG_OPTION_EXTEND) *p++ = 'x';
278 *p = 0;
279 return str;
280 }
281
282 extern int
283 rb_char_to_option_kcode(int c, int *option, int *kcode)
284 {
285 *option = 0;
286
287 switch (c) {
288 case 'n':
289 *kcode = -1;
290 return (*option = ARG_ENCODING_NONE);
291 case 'e':
292 *kcode = rb_enc_find_index("EUC-JP");
293 break;
294 case 's':
295 *kcode = rb_enc_find_index("Windows-31J");
296 break;
297 case 'u':
298 *kcode = rb_enc_find_index("UTF-8");
299 break;
300 default:
301 *kcode = -1;
302 return (*option = char_to_option(c));
303 }
304 *option = ARG_ENCODING_FIXED;
305 return 1;
306 }
307
308 static void
309 rb_reg_check(VALUE re)
310 {
311 if (!RREGEXP(re)->ptr || !RREGEXP_SRC(re) || !RREGEXP_SRC_PTR(re)) {
312 rb_raise(rb_eTypeError, "uninitialized Regexp");
313 }
314 }
315
316 static void
317 rb_reg_expr_str(VALUE str, const char *s, long len)
318 {
319 rb_encoding *enc = rb_enc_get(str);
320 const char *p, *pend;
321 int need_escape = 0;
322 int c, clen;
323
324 p = s; pend = p + len;
325 while (p<pend) {
326 c = rb_enc_ascget(p, pend, &clen, enc);
327 if (c == -1) {
328 p += mbclen(p, pend, enc);
329 }
330 else if (c != '/' && rb_enc_isprint(c, enc)) {
331 p += clen;
332 }
333 else {
334 need_escape = 1;
335 break;
336 }
337 }
338 if (!need_escape) {
339 rb_str_buf_cat(str, s, len);
340 }
341 else {
342 p = s;
343 while (p<pend) {
344 c = rb_enc_ascget(p, pend, &clen, enc);
345 if (c == '\\' && p+clen < pend) {
346 int n = clen + mbclen(p+clen, pend, enc);
347 rb_str_buf_cat(str, p, n);
348 p += n;
349 continue;
350 }
351 else if (c == '/') {
352 char c = '\\';
353 rb_str_buf_cat(str, &c, 1);
354 rb_str_buf_cat(str, p, clen);
355 }
356 else if (c == -1) {
357 int l = mbclen(p, pend, enc);
358 rb_str_buf_cat(str, p, l);
359 p += l;
360 continue;
361 }
362 else if (rb_enc_isprint(c, enc)) {
363 rb_str_buf_cat(str, p, clen);
364 }
365 else if (!rb_enc_isspace(c, enc)) {
366 char b[8];
367
368 sprintf(b, "\\x%02X", c);
369 rb_str_buf_cat(str, b, 4);
370 }
371 else {
372 rb_str_buf_cat(str, p, clen);
373 }
374 p += clen;
375 }
376 }
377 }
378
379 static VALUE
380 rb_reg_desc(const char *s, long len, VALUE re)
381 {
382 VALUE str = rb_str_buf_new2("/");
383
384 rb_enc_copy(str, re);
385 rb_reg_expr_str(str, s, len);
386 rb_str_buf_cat2(str, "/");
387 if (re) {
388 char opts[4];
389 rb_reg_check(re);
390 if (*option_to_str(opts, RREGEXP(re)->ptr->options))
391 rb_str_buf_cat2(str, opts);
392 }
393 OBJ_INFECT(str, re);
394 return str;
395 }
396
397
398 /*
399 * call-seq:
400 * rxp.source => str
401 *
402 * Returns the original string of the pattern.
403 *
404 * /ab+c/ix.source #=> "ab+c"
405 *
406 * Note that escape sequences are retained as is.
407 *
408 * /\x20\+/.source #=> "\\x20\\+"
409 *
410 */
411
412 static VALUE
413 rb_reg_source(VALUE re)
414 {
415 VALUE str;
416
417 rb_reg_check(re);
418 str = rb_enc_str_new(RREGEXP_SRC_PTR(re),RREGEXP_SRC_LEN(re), rb_enc_get(re));
419 if (OBJ_TAINTED(re)) OBJ_TAINT(str);
420 return str;
421 }
422
423 /*
424 * call-seq:
425 * rxp.inspect => string
426 *
427 * Produce a nicely formatted string-version of _rxp_. Perhaps surprisingly,
428 * <code>#inspect</code> actually produces the more natural version of
429 * the string than <code>#to_s</code>.
430 *
431 * /ab+c/ix.inspect #=> "/ab+c/ix"
432 *
433 */
434
435 static VALUE
436 rb_reg_inspect(VALUE re)
437 {
438 if (!RREGEXP(re)->ptr || !RREGEXP_SRC(re) || !RREGEXP_SRC_PTR(re)) {
439 return rb_any_to_s(re);
440 }
441 return rb_reg_desc(RREGEXP_SRC_PTR(re), RREGEXP_SRC_LEN(re), re);
442 }
443
444
445 /*
446 * call-seq:
447 * rxp.to_s => str
448 *
449 * Returns a string containing the regular expression and its options (using the
450 * <code>(?opts:source)</code> notation. This string can be fed back in to
451 * <code>Regexp::new</code> to a regular expression with the same semantics as
452 * the original. (However, <code>Regexp#==</code> may not return true when
453 * comparing the two, as the source of the regular expression itself may
454 * differ, as the example shows). <code>Regexp#inspect</code> produces a
455 * generally more readable version of <i>rxp</i>.
456 *
457 * r1 = /ab+c/ix #=> /ab+c/ix
458 * s1 = r1.to_s #=> "(?ix-m:ab+c)"
459 * r2 = Regexp.new(s1) #=> /(?ix-m:ab+c)/
460 * r1 == r2 #=> false
461 * r1.source #=> "ab+c"
462 * r2.source #=> "(?ix-m:ab+c)"
463 */
464
465 static VALUE
466 rb_reg_to_s(VALUE re)
467 {
468 int options, opt;
469 const int embeddable = ONIG_OPTION_MULTILINE|ONIG_OPTION_IGNORECASE|ONIG_OPTION_EXTEND;
470 long len;
471 const UChar* ptr;
472 VALUE str = rb_str_buf_new2("(?");
473 char optbuf[5];
474
475 rb_reg_check(re);
476
477 rb_enc_copy(str, re);
478 options = RREGEXP(re)->ptr->options;
479 ptr = (UChar*)RREGEXP_SRC_PTR(re);
480 len = RREGEXP_SRC_LEN(re);
481 again:
482 if (len >= 4 && ptr[0] == '(' && ptr[1] == '?') {
483 int err = 1;
484 ptr += 2;
485 if ((len -= 2) > 0) {
486 do {
487 opt = char_to_option((int )*ptr);
488 if (opt != 0) {
489 options |= opt;
490 }
491 else {
492 break;
493 }
494 ++ptr;
495 } while (--len > 0);
496 }
497 if (len > 1 && *ptr == '-') {
498 ++ptr;
499 --len;
500 do {
501 opt = char_to_option((int )*ptr);
502 if (opt != 0) {
503 options &= ~opt;
504 }
505 else {
506 break;
507 }
508 ++ptr;
509 } while (--len > 0);
510 }
511 if (*ptr == ')') {
512 --len;
513 ++ptr;
514 goto again;
515 }
516 if (*ptr == ':' && ptr[len-1] == ')') {
517 int r;
518 Regexp *rp;
519 r = onig_alloc_init(&rp, ONIG_OPTION_DEFAULT,
520 ONIGENC_CASE_FOLD_DEFAULT,
521 rb_enc_get(re),
522 OnigDefaultSyntax);
523 if (r == 0) {
524 ++ptr;
525 len -= 2;
526 err = (onig_compile(rp, ptr, ptr + len, NULL) != 0);
527 }
528 onig_free(rp);
529 }
530 if (err) {
531 options = RREGEXP(re)->ptr->options;
532 ptr = (UChar*)RREGEXP_SRC_PTR(re);
533 len = RREGEXP_SRC_LEN(re);
534 }
535 }
536
537 if (*option_to_str(optbuf, options)) rb_str_buf_cat2(str, optbuf);
538
539 if ((options & embeddable) != embeddable) {
540 optbuf[0] = '-';
541 option_to_str(optbuf + 1, ~options);
542 rb_str_buf_cat2(str, optbuf);
543 }
544
545 rb_str_buf_cat2(str, ":");
546 rb_reg_expr_str(str, (char*)ptr, len);
547 rb_str_buf_cat2(str, ")");
548 rb_enc_copy(str, re);
549
550 OBJ_INFECT(str, re);
551 return str;
552 }
553
554 static void
555 rb_reg_raise(const char *s, long len, const char *err, VALUE re)
556 {
557 VALUE desc = rb_reg_desc(s, len, re);
558
559 rb_raise(rb_eRegexpError, "%s: %s", err, RSTRING_PTR(desc));
560 }
561
562 static VALUE
563 rb_enc_reg_error_desc(const char *s, long len, rb_encoding *enc, int options, const char *err)
564 {
565 char opts[6];
566 VALUE desc = rb_str_buf_new2(err);
567
568 rb_enc_associate(desc, enc);
569 rb_str_buf_cat2(desc, ": /");
570 rb_reg_expr_str(desc, s, len);
571 opts[0] = '/';
572 option_to_str(opts + 1, options);
573 rb_str_buf_cat2(desc, opts);
574 return rb_exc_new3(rb_eRegexpError, desc);
575 }
576
577 static void
578 rb_enc_reg_raise(const char *s, long len, rb_encoding *enc, int options, const char *err)
579 {
580 rb_exc_raise(rb_enc_reg_error_desc(s, len, enc, options, err));
581 }
582
583 static VALUE
584 rb_reg_error_desc(VALUE str, int options, const char *err)
585 {
586 return rb_enc_reg_error_desc(RSTRING_PTR(str), RSTRING_LEN(str),
587 rb_enc_get(str), options, err);
588 }
589
590 static void
591 rb_reg_raise_str(VALUE str, int options, const char *err)
592 {
593 rb_exc_raise(rb_reg_error_desc(str, options, err));
594 }
595
596
597 /*
598 * call-seq:
599 * rxp.casefold? => true or false
600 *
601 * Returns the value of the case-insensitive flag.
602 *
603 * /a/.casefold? #=> false
604 * /a/i.casefold? #=> true
605 * /(?i:a)/.casefold? #=> false
606 */
607
608 static VALUE
609 rb_reg_casefold_p(VALUE re)
610 {
611 rb_reg_check(re);
612 if (RREGEXP(re)->ptr->options & ONIG_OPTION_IGNORECASE) return Qtrue;
613 return Qfalse;
614 }
615
616
617 /*
618 * call-seq:
619 * rxp.options => fixnum
620 *
621 * Returns the set of bits corresponding to the options used when creating this
622 * Regexp (see <code>Regexp::new</code> for details. Note that additional bits
623 * may be set in the returned options: these are used internally by the regular
624 * expression code. These extra bits are ignored if the options are passed to
625 * <code>Regexp::new</code>.
626 *
627 * Regexp::IGNORECASE #=> 1
628 * Regexp::EXTENDED #=> 2
629 * Regexp::MULTILINE #=> 4
630 *
631 * /cat/.options #=> 0
632 * /cat/ix.options #=> 3
633 * Regexp.new('cat', true).options #=> 1
634 * /\xa1\xa2/e.options #=> 16
635 *
636 * r = /cat/ix
637 * Regexp.new(r.source, r.options) #=> /cat/ix
638 */
639
640 static VALUE
641 rb_reg_options_m(VALUE re)
642 {
643 int options = rb_reg_options(re);
644 return INT2NUM(options);
645 }
646
647 static int
648 reg_names_iter(const OnigUChar *name, const OnigUChar *name_end,
649 int back_num, int *back_refs, OnigRegex regex, void *arg)
650 {
651 VALUE ary = (VALUE)arg;
652 rb_ary_push(ary, rb_str_new((const char *)name, name_end-name));
653 return 0;
654 }
655
656 /*
657 * call-seq:
658 * rxp.names => [name1, name2, ...]
659 *
660 * Returns a list of names of captures as an array of strings.
661 *
662 * /(?<foo>.)(?<bar>.)(?<baz>.)/.names
663 * #=> ["foo", "bar", "baz"]
664 *
665 * /(?<foo>.)(?<foo>.)/.names
666 * #=> ["foo"]
667 *
668 * /(.)(.)/.names
669 * #=> []
670 */
671
672 static VALUE
673 rb_reg_names(VALUE re)
674 {
675 VALUE ary = rb_ary_new();
676 rb_reg_check(re);
677 onig_foreach_name(RREGEXP(re)->ptr, reg_names_iter, (void*)ary);
678 return ary;
679 }
680
681 static int
682 reg_named_captures_iter(const OnigUChar *name, const OnigUChar *name_end,
683 int back_num, int *back_refs, OnigRegex regex, void *arg)
684 {
685 VALUE hash = (VALUE)arg;
686 VALUE ary = rb_ary_new2(back_num);
687 int i;
688
689 for(i = 0; i < back_num; i++)
690 rb_ary_store(ary, i, INT2NUM(back_refs[i]));
691
692 rb_hash_aset(hash, rb_str_new((const char*)name, name_end-name),ary);
693
694 return 0;
695 }
696
697 /*
698 * call-seq:
699 * rxp.named_captures => hash
700 *
701 * Returns a hash representing information about named captures of <i>rxp</i>.
702 *
703 * A key of the hash is a name of the named captures.
704 * A value of the hash is an array which is list of indexes of corresponding
705 * named captures.
706 *
707 * /(?<foo>.)(?<bar>.)/.named_captures
708 * #=> {"foo"=>[1], "bar"=>[2]}
709 *
710 * /(?<foo>.)(?<foo>.)/.named_captures
711 * #=> {"foo"=>[1, 2]}
712 *
713 * If there are no named captures, an empty hash is returned.
714 *
715 * /(.)(.)/.named_captures
716 * #=> {}
717 */
718
719 static VALUE
720 rb_reg_named_captures(VALUE re)
721 {
722 VALUE hash = rb_hash_new();
723 rb_reg_check(re);
724 onig_foreach_name(RREGEXP(re)->ptr, reg_named_captures_iter, (void*)hash);
725 return hash;
726 }
727
728 static Regexp*
729 make_regexp(const char *s, long len, rb_encoding *enc, int flags, onig_errmsg_buffer err)
730 {
731 Regexp *rp;
732 int r;
733 OnigErrorInfo einfo;
734
735 /* Handle escaped characters first. */
736
737 /* Build a copy of the string (in dest) with the
738 escaped characters translated, and generate the regex
739 from that.
740 */
741
742 r = onig_alloc_init(&rp, flags, ONIGENC_CASE_FOLD_DEFAULT,
743 enc, OnigDefaultSyntax);
744 if (r) {
745 onig_error_code_to_str((UChar*)err, r);
746 return 0;
747 }
748
749 r = onig_compile(rp, (UChar*)s, (UChar*)(s + len), &einfo);
750
751 if (r != 0) {
752 onig_free(rp);
753 (void )onig_error_code_to_str((UChar*)err, r, &einfo);
754 return 0;
755 }
756 return rp;
757 }
758
759
760 /*
761 * Document-class: MatchData
762 *
763 * <code>MatchData</code> is the type of the special variable <code>$~</code>,
764 * and is the type of the object returned by <code>Regexp#match</code> and
765 * <code>Regexp.last_match</code>. It encapsulates all the results of a pattern
766 * match, results normally accessed through the special variables
767 * <code>$&</code>, <code>$'</code>, <code>$`</code>, <code>$1</code>,
768 * <code>$2</code>, and so on.
769 *
770 */
771
772 VALUE rb_cMatch;
773
774 static VALUE
775 match_alloc(VALUE klass)
776 {
777 NEWOBJ(match, struct RMatch);
778 OBJSETUP(match, klass, T_MATCH);
779
780 match->str = 0;
781 match->rmatch = 0;
782 match->regexp = 0;
783 match->rmatch = ALLOC(struct rmatch);
784 MEMZERO(match->rmatch, struct rmatch, 1);
785
786 return (VALUE)match;
787 }
788
789 typedef struct {
790 int byte_pos;
791 int char_pos;
792 } pair_t;
793
794 static int
795 pair_byte_cmp(const void *pair1, const void *pair2)
796 {
797 return ((pair_t*)pair1)->byte_pos - ((pair_t*)pair2)->byte_pos;
798 }
799
800 static void
801 update_char_offset(VALUE match)
802 {
803 struct rmatch *rm = RMATCH(match)->rmatch;
804 struct re_registers *regs;
805 int num_regs;
806 int i, num_pos, c;
807 char *s, *p, *q, *e;
808 rb_encoding *enc;
809 pair_t *pairs;
810
811 if (rm->char_offset_updated)
812 return;
813
814 regs = &rm->regs;
815 num_regs = rm->regs.num_regs;
816
817 if (rm->char_offset_num_allocated < num_regs) {
818 REALLOC_N(rm->char_offset, struct rmatch_offset, num_regs);
819 rm->char_offset_num_allocated = num_regs;
820 }
821
822 enc = rb_enc_get(RMATCH(match)->str);
823 if (rb_enc_mbmaxlen(enc) == 1) {
824 for (i = 0; i < num_regs; i++) {
825 rm->char_offset[i].beg = BEG(i);
826 rm->char_offset[i].end = END(i);
827 }
828 rm->char_offset_updated = 1;
829 return;
830 }
831
832 pairs = ALLOCA_N(pair_t, num_regs*2);
833 num_pos = 0;
834 for (i = 0; i < num_regs; i++) {
835 if (BEG(i) < 0)
836 continue;
837 pairs[num_pos++].byte_pos = BEG(i);
838 pairs[num_pos++].byte_pos = END(i);
839 }
840 qsort(pairs, num_pos, sizeof(pair_t), pair_byte_cmp);
841
842 s = p = RSTRING_PTR(RMATCH(match)->str);
843 e = s + RSTRING_LEN(RMATCH(match)->str);
844 c = 0;
845 for (i = 0; i < num_pos; i++) {
846 q = s + pairs[i].byte_pos;
847 c += rb_enc_strlen(p, q, enc);
848 pairs[i].char_pos = c;
849 p = q;
850 }
851
852 for (i = 0; i < num_regs; i++) {
853 pair_t key, *found;
854 if (BEG(i) < 0) {
855 rm->char_offset[i].beg = -1;
856 rm->char_offset[i].end = -1;
857 continue;
858 }
859
860 key.byte_pos = BEG(i);
861 found = bsearch(&key, pairs, num_pos, sizeof(pair_t), pair_byte_cmp);
862 rm->char_offset[i].beg = found->char_pos;
863
864 key.byte_pos = END(i);
865 found = bsearch(&key, pairs, num_pos, sizeof(pair_t), pair_byte_cmp);
866 rm->char_offset[i].end = found->char_pos;
867 }
868
869 rm->char_offset_updated = 1;
870 }
871
872 static void
873 match_check(VALUE match)
874 {
875 if (!RMATCH(match)->regexp) {
876 rb_raise(rb_eTypeError, "uninitialized Match");
877 }
878 }
879
880 /* :nodoc: */
881 static VALUE
882 match_init_copy(VALUE obj, VALUE orig)
883 {
884 struct rmatch *rm;
885
886 if (obj == orig) return obj;
887
888 if (!rb_obj_is_instance_of(orig, rb_obj_class(obj))) {
889 rb_raise(rb_eTypeError, "wrong argument class");
890 }
891 RMATCH(obj)->str = RMATCH(orig)->str;
892 RMATCH(obj)->regexp = RMATCH(orig)->regexp;
893
894 rm = RMATCH(obj)->rmatch;
895 onig_region_copy(&rm->regs, RMATCH_REGS(orig));
896
897 if (!RMATCH(orig)->rmatch->char_offset_updated) {
898 rm->char_offset_updated = 0;
899 }
900 else {
901 if (rm->char_offset_num_allocated < rm->regs.num_regs) {
902 REALLOC_N(rm->char_offset, struct rmatch_offset, rm->regs.num_regs);
903 rm->char_offset_num_allocated = rm->regs.num_regs;
904 }
905 MEMCPY(rm->char_offset, RMATCH(orig)->rmatch->char_offset,
906 struct rmatch_offset, rm->regs.num_regs);
907 rm->char_offset_updated = 1;
908 }
909
910 return obj;
911 }
912
913
914 /*
915 * call-seq:
916 * mtch.regexp => regexp
917 *
918 * Returns the regexp.
919 *
920 * m = /a.*b/.match("abc")
921 * m.regexp #=> /a.*b/
922 */
923
924 static VALUE
925 match_regexp(VALUE match)
926 {
927 match_check(match);
928 return RMATCH(match)->regexp;
929 }
930
931 /*
932 * call-seq:
933 * mtch.names => [name1, name2, ...]
934 *
935 * Returns a list of names of captures as an array of strings.
936 * It is same as mtch.regexp.names.
937 *
938 * /(?<foo>.)(?<bar>.)(?<baz>.)/.match("hoge").names
939 * #=> ["foo", "bar", "baz"]
940 *
941 * m = /(?<x>.)(?<y>.)?/.match("a") #=> #<MatchData "a" x:"a" y:nil>
942 * m.names #=> ["x", "y"]
943 */
944
945 static VALUE
946 match_names(VALUE match)
947 {
948 match_check(match);
949 return rb_reg_names(RMATCH(match)->regexp);
950 }
951
952 /*
953 * call-seq:
954 * mtch.length => integer
955 * mtch.size => integer
956 *
957 * Returns the number of elements in the match array.
958 *
959 * m = /(.)(.)(\d+)(\d)/.match("THX1138.")
960 * m.length #=> 5
961 * m.size #=> 5
962 */
963
964 static VALUE
965 match_size(VALUE match)
966 {
967 match_check(match);
968 return INT2FIX(RMATCH_REGS(match)->num_regs);
969 }
970
971 static int
972 match_backref_number(VALUE match, VALUE backref)
973 {
974 const char *name;
975 int num;
976
977 struct re_registers *regs = RMATCH_REGS(match);
978 VALUE regexp = RMATCH(match)->regexp;
979
980 match_check(match);
981 switch(TYPE(backref)) {
982 default:
983 return NUM2INT(backref);
984
985 case T_SYMBOL:
986 name = rb_id2name(SYM2ID(backref));
987 break;
988
989 case T_STRING:
990 name = StringValueCStr(backref);
991 break;
992 }
993
994 num = onig_name_to_backref_number(RREGEXP(regexp)->ptr,
995 (const unsigned char*)name,
996 (const unsigned char*)name + strlen(name),
997 regs);
998
999 if (num < 1) {
1000 rb_raise(rb_eIndexError, "undefined group name reference: %s", name);
1001 }
1002
1003 return num;
1004 }
1005
1006
1007 /*
1008 * call-seq:
1009 * mtch.offset(n) => array
1010 *
1011 * Returns a two-element array containing the beginning and ending offsets of
1012 * the <em>n</em>th match.
1013 * <em>n</em> can be a string or symbol to reference a named capture.
1014 *
1015 * m = /(.)(.)(\d+)(\d)/.match("THX1138.")
1016 * m.offset(0) #=> [1, 7]
1017 * m.offset(4) #=> [6, 7]
1018 *
1019 * m = /(?<foo>.)(.)(?<bar>.)/.match("hoge")
1020 * p m.offset(:foo) #=> [0, 1]
1021 * p m.offset(:bar) #=> [2, 3]
1022 *
1023 */
1024
1025 static VALUE
1026 match_offset(VALUE match, VALUE n)
1027 {
1028 int i = match_backref_number(match, n);
1029 struct re_registers *regs = RMATCH_REGS(match);
1030
1031 match_check(match);
1032 if (i < 0 || regs->num_regs <= i)
1033 rb_raise(rb_eIndexError, "index %d out of matches", i);
1034
1035 if (BEG(i) < 0)
1036 return rb_assoc_new(Qnil, Qnil);
1037
1038 update_char_offset(match);
1039 return rb_assoc_new(INT2FIX(RMATCH(match)->rmatch->char_offset[i].beg),
1040 INT2FIX(RMATCH(match)->rmatch->char_offset[i].end));
1041 }
1042
1043
1044 /*
1045 * call-seq:
1046 * mtch.begin(n) => integer
1047 *
1048 * Returns the offset of the start of the <em>n</em>th element of the match
1049 * array in the string.
1050 * <em>n</em> can be a string or symbol to reference a named capture.
1051 *
1052 * m = /(.)(.)(\d+)(\d)/.match("THX1138.")
1053 * m.begin(0) #=> 1
1054 * m.begin(2) #=> 2
1055 *
1056 * m = /(?<foo>.)(.)(?<bar>.)/.match("hoge")
1057 * p m.begin(:foo) #=> 0
1058 * p m.begin(:bar) #=> 2
1059 */
1060
1061 static VALUE
1062 match_begin(VALUE match, VALUE n)
1063 {
1064 int i = match_backref_number(match, n);
1065 struct re_registers *regs = RMATCH_REGS(match);
1066
1067 match_check(match);
1068 if (i < 0 || regs->num_regs <= i)
1069 rb_raise(rb_eIndexError, "index %d out of matches", i);
1070
1071 if (BEG(i) < 0)
1072 return Qnil;
1073
1074 update_char_offset(match);
1075 return INT2FIX(RMATCH(match)->rmatch->char_offset[i].beg);
1076 }
1077
1078
1079 /*
1080 * call-seq:
1081 * mtch.end(n) => integer
1082 *
1083 * Returns the offset of the character immediately following the end of the
1084 * <em>n</em>th element of the match array in the string.
1085 * <em>n</em> can be a string or symbol to reference a named capture.
1086 *
1087 * m = /(.)(.)(\d+)(\d)/.match("THX1138.")
1088 * m.end(0) #=> 7
1089 * m.end(2) #=> 3
1090 *
1091 * m = /(?<foo>.)(.)(?<bar>.)/.match("hoge")
1092 * p m.end(:foo) #=> 1
1093 * p m.end(:bar) #=> 3
1094 */
1095
1096 static VALUE
1097 match_end(VALUE match, VALUE n)
1098 {
1099 int i = match_backref_number(match, n);
1100 struct re_registers *regs = RMATCH_REGS(match);
1101
1102 match_check(match);
1103 if (i < 0 || regs->num_regs <= i)
1104 rb_raise(rb_eIndexError, "index %d out of matches", i);
1105
1106 if (BEG(i) < 0)
1107 return Qnil;
1108
1109 update_char_offset(match);
1110 return INT2FIX(RMATCH(match)->rmatch->char_offset[i].end);
1111 }
1112
1113 #define MATCH_BUSY FL_USER2
1114
1115 void
1116 rb_match_busy(VALUE match)
1117 {
1118 FL_SET(match, MATCH_BUSY);
1119 }
1120
1121 /*
1122 * call-seq:
1123 * rxp.fixed_encoding? => true or false
1124 *
1125 * Returns false if rxp is applicable to
1126 * a string with any ASCII compatible encoding.
1127 * Returns true otherwise.
1128 *
1129 * r = /a/
1130 * r.fixed_encoding? #=> false
1131 * r =~ "\u{6666} a" #=> 2
1132 * r =~ "\xa1\xa2 a".force_encoding("euc-jp") #=> 2
1133 * r =~ "abc".force_encoding("euc-jp") #=> 0
1134 *
1135 * r = /a/u
1136 * r.fixed_encoding? #=> true
1137 * r.encoding #=> #<Encoding:UTF-8>
1138 * r =~ "\u{6666} a" #=> 2
1139 * r =~ "\xa1\xa2".force_encoding("euc-jp") #=> ArgumentError
1140 * r =~ "abc".force_encoding("euc-jp") #=> 0
1141 *
1142 * r = /\u{6666}/
1143 * r.fixed_encoding? #=> true
1144 * r.encoding #=> #<Encoding:UTF-8>
1145 * r =~ "\u{6666} a" #=> 0
1146 * r =~ "\xa1\xa2".force_encoding("euc-jp") #=> ArgumentError
1147 * r =~ "abc".force_encoding("euc-jp") #=> nil
1148 */
1149
1150 static VALUE
1151 rb_reg_fixed_encoding_p(VALUE re)
1152 {
1153 if (FL_TEST(re, KCODE_FIXED))
1154 return Qtrue;
1155 else
1156 return Qfalse;
1157 }
1158
1159 static VALUE
1160 rb_reg_preprocess(const char *p, const char *end, rb_encoding *enc,
1161 rb_encoding **fixed_enc, onig_errmsg_buffer err);
1162
1163
1164 static void
1165 reg_enc_error(VALUE re, VALUE str)
1166 {
1167 rb_raise(rb_eArgError,
1168 "incompatible encoding regexp match (%s regexp with %s string)",
1169 rb_enc_name(RREGEXP(re)->ptr->enc),
1170 rb_enc_name(rb_enc_get(str)));
1171 }
1172
1173 static rb_encoding*
1174 rb_reg_prepare_enc(VALUE re, VALUE str, int warn)
1175 {
1176 rb_encoding *enc = 0;
1177
1178 if (rb_enc_str_coderange(str) == ENC_CODERANGE_BROKEN) {
1179 rb_raise(rb_eArgError,
1180 "broken %s string",
1181 rb_enc_name(rb_enc_get(str)));
1182 }
1183
1184 rb_reg_check(re);
1185 enc = rb_enc_get(str);
1186 if (!rb_enc_str_asciicompat_p(str)) {
1187 if (RREGEXP(re)->ptr->enc != enc) {
1188 reg_enc_error(re, str);
1189 }
1190 }
1191 else if (rb_reg_fixed_encoding_p(re)) {
1192 if (RREGEXP(re)->ptr->enc != enc &&
1193 (!rb_enc_asciicompat(RREGEXP(re)->ptr->enc) ||
1194 rb_enc_str_coderange(str) != ENC_CODERANGE_7BIT)) {
1195 reg_enc_error(re, str);
1196 }
1197 enc = RREGEXP(re)->ptr->enc;
1198 }
1199 if (warn && (RBASIC(re)->flags & REG_ENCODING_NONE) &&
1200 enc != rb_ascii8bit_encoding() &&
1201 rb_enc_str_coderange(str) != ENC_CODERANGE_7BIT) {
1202 rb_warn("regexp match /.../n against to %s string",
1203 rb_enc_name(enc));
1204 }
1205 return enc;
1206 }
1207
1208 regex_t *
1209 rb_reg_prepare_re(VALUE re, VALUE str)
1210 {
1211 regex_t *reg = RREGEXP(re)->ptr;
1212 onig_errmsg_buffer err = "";
1213 int r;
1214 OnigErrorInfo einfo;
1215 const char *pattern;
1216 VALUE unescaped;
1217 rb_encoding *fixed_enc = 0;
1218 rb_encoding *enc = rb_reg_prepare_enc(re, str, 1);
1219
1220 if (reg->enc == enc) return reg;
1221
1222 rb_reg_check(re);
1223 reg = RREGEXP(re)->ptr;
1224 pattern = RREGEXP_SRC_PTR(re);
1225
1226 unescaped = rb_reg_preprocess(
1227 pattern, pattern + RREGEXP_SRC_LEN(re), enc,
1228 &fixed_enc, err);
1229
1230 if (unescaped == Qnil) {
1231 rb_raise(rb_eArgError, "regexp preprocess failed: %s", err);
1232 }
1233
1234 r = onig_new(&reg, (UChar* )RSTRING_PTR(unescaped),
1235 (UChar* )(RSTRING_PTR(unescaped) + RSTRING_LEN(unescaped)),
1236 reg->options, enc,
1237 OnigDefaultSyntax, &einfo);
1238 if (r) {
1239 onig_error_code_to_str((UChar*)err, r, &einfo);
1240 rb_reg_raise(pattern, RREGEXP_SRC_LEN(re), err, re);
1241 }
1242
1243 RB_GC_GUARD(unescaped);
1244 return reg;
1245 }
1246
1247 int
1248 rb_reg_adjust_startpos(VALUE re, VALUE str, int pos, int reverse)
1249 {
1250 int range;
1251 rb_encoding *enc;
1252 UChar *p, *string;
1253
1254 enc = rb_reg_prepare_enc(re, str, 0);
1255
1256 if (reverse) {
1257 range = -pos;
1258 }
1259 else {
1260 range = RSTRING_LEN(str) - pos;
1261 }
1262
1263 if (pos > 0 && ONIGENC_MBC_MAXLEN(enc) != 1 && pos < RSTRING_LEN(str)) {
1264 string = (UChar*)RSTRING_PTR(str);
1265
1266 if (range > 0) {
1267 p = onigenc_get_right_adjust_char_head(enc, string, string + pos);
1268 }
1269 else {
1270 p = ONIGENC_LEFT_ADJUST_CHAR_HEAD(enc, string, string + pos);
1271 }
1272 return p - string;
1273 }
1274
1275 return pos;
1276 }
1277
1278 int
1279 rb_reg_search(VALUE re, VALUE str, int pos, int reverse)
1280 {
1281 int result;
1282 VALUE match;
1283 struct re_registers regi, *regs = &regi;
1284 char *range = RSTRING_PTR(str);
1285 regex_t *reg;
1286 int tmpreg;
1287
1288 if (pos > RSTRING_LEN(str) || pos < 0) {
1289 rb_backref_set(Qnil);
1290 return -1;
1291 }
1292
1293 reg = rb_reg_prepare_re(re, str);
1294 tmpreg = reg != RREGEXP(re)->ptr;
1295 if (!tmpreg) RREGEXP(re)->usecnt++;
1296
1297 match = rb_backref_get();
1298 if (!NIL_P(match)) {
1299 if (FL_TEST(match, MATCH_BUSY)) {
1300 match = Qnil;
1301 }
1302 else {
1303 regs = RMATCH_REGS(match);
1304 }
1305 }
1306 if (NIL_P(match)) {
1307 MEMZERO(regs, struct re_registers, 1);
1308 }
1309 if (!reverse) {
1310 range += RSTRING_LEN(str);
1311 }
1312 result = onig_search(reg,
1313 (UChar*)(RSTRING_PTR(str)),
1314 ((UChar*)(RSTRING_PTR(str)) + RSTRING_LEN(str)),
1315 ((UChar*)(RSTRING_PTR(str)) + pos),
1316 ((UChar*)range),
1317 regs, ONIG_OPTION_NONE);
1318 if (!tmpreg) RREGEXP(re)->usecnt--;
1319 if (tmpreg) {
1320 if (RREGEXP(re)->usecnt) {
1321 onig_free(reg);
1322 }
1323 else {
1324 onig_free(RREGEXP(re)->ptr);
1325 RREGEXP(re)->ptr = reg;
1326 }
1327 }
1328 if (result < 0) {
1329 if (regs == &regi)
1330 onig_region_free(regs, 0);
1331 if (result == ONIG_MISMATCH) {
1332 rb_backref_set(Qnil);
1333 return result;
1334 }
1335 else {
1336 onig_errmsg_buffer err = "";
1337 onig_error_code_to_str((UChar*)err, result);
1338 rb_reg_raise(RREGEXP_SRC_PTR(re), RREGEXP_SRC_LEN(re), err, 0);
1339 }
1340 }
1341
1342 if (NIL_P(match)) {
1343 match = match_alloc(rb_cMatch);
1344 onig_region_copy(RMATCH_REGS(match), regs);
1345 onig_region_free(regs, 0);
1346 }
1347 else {
1348 if (rb_safe_level() >= 3)
1349 OBJ_TAINT(match);
1350 else
1351 FL_UNSET(match, FL_TAINT);
1352 }
1353
1354 RMATCH(match)->str = rb_str_new4(str);
1355 RMATCH(match)->regexp = re;
1356 RMATCH(match)->rmatch->char_offset_updated = 0;
1357 rb_backref_set(match);
1358
1359 OBJ_INFECT(match, re);
1360 OBJ_INFECT(match, str);
1361
1362 return result;
1363 }
1364
1365 VALUE
1366 rb_reg_nth_defined(int nth, VALUE match)
1367 {
1368 struct re_registers *regs;
1369 if (NIL_P(match)) return Qnil;
1370 match_check(match);
1371 regs = RMATCH_REGS(match);
1372 if (nth >= regs->num_regs) {
1373 return Qnil;
1374 }
1375 if (nth < 0) {
1376 nth += regs->num_regs;
1377 if (nth <= 0) return Qnil;
1378 }
1379 if (BEG(nth) == -1) return Qfalse;
1380 return Qtrue;
1381 }
1382
1383 VALUE
1384 rb_reg_nth_match(int nth, VALUE match)
1385 {
1386 VALUE str;
1387 long start, end, len;
1388 struct re_registers *regs;
1389
1390 if (NIL_P(match)) return Qnil;
1391 match_check(match);
1392 regs = RMATCH_REGS(match);
1393 if (nth >= regs->num_regs) {
1394 return Qnil;
1395 }
1396 if (nth < 0) {
1397 nth += regs->num_regs;
1398 if (nth <= 0) return Qnil;
1399 }
1400 start = BEG(nth);
1401 if (start == -1) return Qnil;
1402 end = END(nth);
1403 len = end - start;
1404 str = rb_str_subseq(RMATCH(match)->str, start, len);
1405 OBJ_INFECT(str, match);
1406 return str;
1407 }
1408
1409 VALUE
1410 rb_reg_last_match(VALUE match)
1411 {
1412 return rb_reg_nth_match(0, match);
1413 }
1414
1415
1416 /*
1417 * call-seq:
1418 * mtch.pre_match => str
1419 *
1420 * Returns the portion of the original string before the current match.
1421 * Equivalent to the special variable <code>$`</code>.
1422 *
1423 * m = /(.)(.)(\d+)(\d)/.match("THX1138.")
1424 * m.pre_match #=> "T"
1425 */
1426
1427 VALUE
1428 rb_reg_match_pre(VALUE match)
1429 {
1430 VALUE str;
1431 struct re_registers *regs;
1432
1433 if (NIL_P(match)) return Qnil;
1434 match_check(match);
1435 regs = RMATCH_REGS(match);
1436 if (BEG(0) == -1) return Qnil;
1437 str = rb_str_subseq(RMATCH(match)->str, 0, BEG(0));
1438 if (OBJ_TAINTED(match)) OBJ_TAINT(str);
1439 return str;
1440 }
1441
1442
1443 /*
1444 * call-seq:
1445 * mtch.post_match => str
1446 *
1447 * Returns the portion of the original string after the current match.
1448 * Equivalent to the special variable <code>$'</code>.
1449 *
1450 * m = /(.)(.)(\d+)(\d)/.match("THX1138: The Movie")
1451 * m.post_match #=> ": The Movie"
1452 */
1453
1454 VALUE
1455 rb_reg_match_post(VALUE match)
1456 {
1457 VALUE str;
1458 long pos;
1459 struct re_registers *regs;
1460
1461 if (NIL_P(match)) return Qnil;
1462 match_check(match);
1463 regs = RMATCH_REGS(match);
1464 if (BEG(0) == -1) return Qnil;
1465 str = RMATCH(match)->str;
1466 pos = END(0);
1467 str = rb_str_subseq(str, pos, RSTRING_LEN(str) - pos);
1468 if (OBJ_TAINTED(match)) OBJ_TAINT(str);
1469 return str;
1470 }
1471
1472 VALUE
1473 rb_reg_match_last(VALUE match)
1474 {
1475 int i;
1476 struct re_registers *regs;
1477
1478 if (NIL_P(match)) return Qnil;
1479 match_check(match);
1480 regs = RMATCH_REGS(match);
1481 if (BEG(0) == -1) return Qnil;
1482
1483 for (i=regs->num_regs-1; BEG(i) == -1 && i > 0; i--)
1484 ;
1485 if (i == 0) return Qnil;
1486 return rb_reg_nth_match(i, match);
1487 }
1488
1489 static VALUE
1490 last_match_getter(void)
1491 {
1492 return rb_reg_last_match(rb_backref_get());
1493 }
1494
1495 static VALUE
1496 prematch_getter(void)
1497 {
1498 return rb_reg_match_pre(rb_backref_get());
1499 }
1500
1501 static VALUE
1502 postmatch_getter(void)
1503 {
1504 return rb_reg_match_post(rb_backref_get());
1505 }
1506
1507 static VALUE
1508 last_paren_match_getter(void)
1509 {
1510 return rb_reg_match_last(rb_backref_get());
1511 }
1512
1513 static VALUE
1514 match_array(VALUE match, int start)
1515 {
1516 struct re_registers *regs = RMATCH_REGS(match);
1517 VALUE ary = rb_ary_new2(regs->num_regs);
1518 VALUE target = RMATCH(match)->str;
1519 int i;
1520 int taint = OBJ_TAINTED(match);
1521
1522 match_check(match);
1523 for (i=start; i<regs->num_regs; i++) {
1524 if (regs->beg[i] == -1) {
1525 rb_ary_push(ary, Qnil);
1526 }
1527 else {
1528 VALUE str = rb_str_subseq(target, regs->beg[i], regs->end[i]-regs->beg[i]);
1529 if (taint) OBJ_TAINT(str);
1530 rb_ary_push(ary, str);
1531 }
1532 }
1533 return ary;
1534 }
1535
1536
1537 /* [MG]:FIXME: I put parens around the /.../.match() in the first line of the
1538 second example to prevent the '*' followed by a '/' from ending the
1539 comment. */
1540
1541 /*
1542 * call-seq:
1543 * mtch.to_a => anArray
1544 *
1545 * Returns the array of matches.
1546 *
1547 * m = /(.)(.)(\d+)(\d)/.match("THX1138.")
1548 * m.to_a #=> ["HX1138", "H", "X", "113", "8"]
1549 *
1550 * Because <code>to_a</code> is called when expanding
1551 * <code>*</code><em>variable</em>, there's a useful assignment
1552 * shortcut for extracting matched fields. This is slightly slower than
1553 * accessing the fields directly (as an intermediate array is
1554 * generated).
1555 *
1556 * all,f1,f2,f3 = *(/(.)(.)(\d+)(\d)/.match("THX1138."))
1557 * all #=> "HX1138"
1558 * f1 #=> "H"
1559 * f2 #=> "X"
1560 * f3 #=> "113"
1561 */
1562
1563 static VALUE
1564 match_to_a(VALUE match)
1565 {
1566 return match_array(match, 0);
1567 }
1568
1569
1570 /*
1571 * call-seq:
1572 * mtch.captures => array
1573 *
1574 * Returns the array of captures; equivalent to <code>mtch.to_a[1..-1]</code>.
1575 *
1576 * f1,f2,f3,f4 = /(.)(.)(\d+)(\d)/.match("THX1138.").captures
1577 * f1 #=> "H"
1578 * f2 #=> "X"
1579 * f3 #=> "113"
1580 * f4 #=> "8"
1581 */
1582 static VALUE
1583 match_captures(VALUE match)
1584 {
1585 return match_array(match, 1);
1586 }
1587
1588 static int
1589 name_to_backref_number(struct re_registers *regs, VALUE regexp, const char* name, const char* name_end)
1590 {
1591 int num;
1592
1593 num = onig_name_to_backref_number(RREGEXP(regexp)->ptr,
1594 (const unsigned char* )name, (const unsigned char* )name_end, regs);
1595 if (num >= 1) {
1596 return num;
1597 }
1598 else {
1599 VALUE s = rb_str_new(name, (long )(name_end - name));
1600 rb_raise(rb_eIndexError, "undefined group name reference: %s",
1601 StringValuePtr(s));
1602 }
1603 }
1604
1605 /*
1606 * call-seq:
1607 * mtch[i] => str or nil
1608 * mtch[start, length] => array
1609 * mtch[range] => array
1610 * mtch[name] => str or nil
1611 *
1612 * Match Reference---<code>MatchData</code> acts as an array, and may be
1613 * accessed using the normal array indexing techniques. <i>mtch</i>[0] is
1614 * equivalent to the special variable <code>$&</code>, and returns the entire
1615 * matched string. <i>mtch</i>[1], <i>mtch</i>[2], and so on return the values
1616 * of the matched backreferences (portions of the pattern between parentheses).
1617 *
1618 * m = /(.)(.)(\d+)(\d)/.match("THX1138.")
1619 * m #=> #<MatchData "HX1138" 1:"H" 2:"X" 3:"113" 4:"8">
1620 * m[0] #=> "HX1138"
1621 * m[1, 2] #=> ["H", "X"]
1622 * m[1..3] #=> ["H", "X", "113"]
1623 * m[-3, 2] #=> ["X", "113"]
1624 *
1625 * m = /(?<foo>a+)b/.match("ccaaab")
1626 * m #=> #<MatchData "aaab" foo:"aaa">
1627 * m["foo"] #=> "aaa"
1628 * m[:foo] #=> "aaa"
1629 */
1630
1631 static VALUE
1632 match_aref(int argc, VALUE *argv, VALUE match)
1633 {
1634 VALUE idx, rest;
1635
1636 match_check(match);
1637 rb_scan_args(argc, argv, "11", &idx, &rest);
1638
1639 if (NIL_P(rest)) {
1640 if (FIXNUM_P(idx)) {
1641 if (FIX2INT(idx) >= 0) {
1642 return rb_reg_nth_match(FIX2INT(idx), match);
1643 }
1644 }
1645 else {
1646 const char *p;
1647 int num;
1648
1649 switch (TYPE(idx)) {
1650 case T_SYMBOL:
1651 p = rb_id2name(SYM2ID(idx));
1652 goto name_to_backref;
1653 break;
1654 case T_STRING:
1655 p = StringValuePtr(idx);
1656
1657 name_to_backref:
1658 num = name_to_backref_number(RMATCH_REGS(match),
1659 RMATCH(match)->regexp, p, p + strlen(p));
1660 return rb_reg_nth_match(num, match);
1661 break;
1662
1663 default:
1664 break;
1665 }
1666 }
1667 }
1668
1669 return rb_ary_aref(argc, argv, match_to_a(match));
1670 }
1671
1672 static VALUE
1673 match_entry(VALUE match, long n)
1674 {
1675 return rb_reg_nth_match(n, match);
1676 }
1677
1678
1679 /*
1680 * call-seq:
1681 *
1682 * mtch.values_at([index]*) => array
1683 *
1684 * Uses each <i>index</i> to access the matching values, returning an array of
1685 * the corresponding matches.
1686 *
1687 * m = /(.)(.)(\d+)(\d)/.match("THX1138: The Movie")
1688 * m.to_a #=> ["HX1138", "H", "X", "113", "8"]
1689 * m.values_at(0, 2, -2) #=> ["HX1138", "X", "113"]
1690 */
1691
1692 static VALUE
1693 match_values_at(int argc, VALUE *argv, VALUE match)
1694 {
1695 struct re_registers *regs = RMATCH_REGS(match);
1696 match_check(match);
1697 return rb_get_values_at(match, regs->num_regs, argc, argv, match_entry);
1698 }
1699
1700
1701 /*
1702 * call-seq:
1703 * mtch.to_s => str
1704 *
1705 * Returns the entire matched string.
1706 *
1707 * m = /(.)(.)(\d+)(\d)/.match("THX1138.")
1708 * m.to_s #=> "HX1138"
1709 */
1710
1711 static VALUE
1712 match_to_s(VALUE match)
1713 {
1714 VALUE str = rb_reg_last_match(match);
1715
1716 match_check(match);
1717 if (NIL_P(str)) str = rb_str_new(0,0);
1718 if (OBJ_TAINTED(match)) OBJ_TAINT(str);
1719 if (OBJ_TAINTED(RMATCH(match)->str)) OBJ_TAINT(str);
1720 return str;
1721 }
1722
1723
1724 /*
1725 * call-seq:
1726 * mtch.string => str
1727 *
1728 * Returns a frozen copy of the string passed in to <code>match</code>.
1729 *
1730 * m = /(.)(.)(\d+)(\d)/.match("THX1138.")
1731 * m.string #=> "THX1138."
1732 */
1733
1734 static VALUE
1735 match_string(VALUE match)
1736 {
1737 match_check(match);
1738 return RMATCH(match)->str; /* str is frozen */
1739 }
1740
1741 struct backref_name_tag {
1742 const UChar *name;
1743 long len;
1744 };
1745
1746 static int
1747 match_inspect_name_iter(const OnigUChar *name, const OnigUChar *name_end,
1748 int back_num, int *back_refs, OnigRegex regex, void *arg0)
1749 {
1750 struct backref_name_tag *arg = (struct backref_name_tag *)arg0;
1751 int i;
1752
1753 for (i = 0; i < back_num; i++) {
1754 arg[back_refs[i]].name = name;
1755 arg[back_refs[i]].len = name_end - name;
1756 }
1757 return 0;
1758 }
1759
1760 /*
1761 * call-seq:
1762 * mtch.inspect => str
1763 *
1764 * Returns a printable version of <i>mtch</i>.
1765 *
1766 * puts /.$/.match("foo").inspect
1767 * #=> #<MatchData "o">
1768 *
1769 * puts /(.)(.)(.)/.match("foo").inspect
1770 * #=> #<MatchData "foo" 1:"f" 2:"o" 3:"o">
1771 *
1772 * puts /(.)(.)?(.)/.match("fo").inspect
1773 * #=> #<MatchData "fo" 1:"f" 2:nil 3:"o">
1774 *
1775 * puts /(?<foo>.)(?<bar>.)(?<baz>.)/.match("hoge").inspect
1776 * #=> #<MatchData "hog" foo:"h" bar:"o" baz:"g">
1777 *
1778 */
1779
1780 static VALUE
1781 match_inspect(VALUE match)
1782 {
1783 const char *cname = rb_obj_classname(match);
1784 VALUE str;
1785 int i;
1786 struct re_registers *regs = RMATCH_REGS(match);
1787 int num_regs = regs->num_regs;
1788 struct backref_name_tag *names;
1789 VALUE regexp = RMATCH(match)->regexp;
1790
1791 if (regexp == 0) {
1792 return rb_sprintf("#<%s:%p>", cname, (void*)match);
1793 }
1794
1795 names = ALLOCA_N(struct backref_name_tag, num_regs);
1796 MEMZERO(names, struct backref_name_tag, num_regs);
1797
1798 onig_foreach_name(RREGEXP(regexp)->ptr,
1799 match_inspect_name_iter, names);
1800
1801 str = rb_str_buf_new2("#<");
1802 rb_str_buf_cat2(str, cname);
1803
1804 for (i = 0; i < num_regs; i++) {
1805 VALUE v;
1806 rb_str_buf_cat2(str, " ");
1807 if (0 < i) {
1808 if (names[i].name)
1809 rb_str_buf_cat(str, (const char *)names[i].name, names[i].len);
1810 else {
1811 rb_str_catf(str, "%d", i);
1812 }
1813 rb_str_buf_cat2(str, ":");
1814 }
1815 v = rb_reg_nth_match(i, match);
1816 if (v == Qnil)
1817 rb_str_buf_cat2(str, "nil");
1818 else
1819 rb_str_buf_append(str, rb_str_inspect(v));
1820 }
1821 rb_str_buf_cat2(str, ">");
1822
1823 return str;
1824 }
1825
1826 VALUE rb_cRegexp;
1827
1828 static int
1829 read_escaped_byte(const char **pp, const char *end, onig_errmsg_buffer err)
1830 {
1831 const char *p = *pp;
1832 int code;
1833 int meta_prefix = 0, ctrl_prefix = 0;
1834 int len;
1835 int retbyte;
1836
1837 retbyte = -1;
1838 if (p == end || *p++ != '\\') {
1839 strcpy(err, "too short escaped multibyte character");
1840 return -1;
1841 }
1842
1843 again:
1844 if (p == end) {
1845 strcpy(err, "too short escape sequence");
1846 return -1;
1847 }
1848 switch (*p++) {
1849 case '\\': code = '\\'; break;
1850 case 'n': code = '\n'; break;
1851 case 't': code = '\t'; break;
1852 case 'r': code = '\r'; break;
1853 case 'f': code = '\f'; break;
1854 case 'v': code = '\013'; break;
1855 case 'a': code = '\007'; break;
1856 case 'e': code = '\033'; break;
1857
1858 /* \OOO */
1859 case '0': case '1': case '2': case '3':
1860 case '4': case '5': case '6': case '7':
1861 p--;
1862 code = ruby_scan_oct(p, end < p+3 ? end-p : 3, &len);
1863 p += len;
1864 break;
1865
1866 case 'x': /* \xHH */
1867 code = ruby_scan_hex(p, end < p+2 ? end-p : 2, &len);
1868 if (len < 1) {
1869 strcpy(err, "invalid hex escape");
1870 return -1;
1871 }
1872 p += len;
1873 break;
1874
1875 case 'M': /* \M-X, \M-\C-X, \M-\cX */
1876 if (meta_prefix) {
1877 strcpy(err, "duplicate meta escape");
1878 return -1;
1879 }
1880 meta_prefix = 1;
1881 if (p+1 < end && *p++ == '-' && (*p & 0x80) == 0) {
1882 if (*p == '\\') {
1883 p++;
1884 goto again;
1885 }
1886 else {
1887 code = *p++;
1888 break;
1889 }
1890 }
1891 strcpy(err, "too short meta escape");
1892 return -1;
1893
1894 case 'C': /* \C-X, \C-\M-X */
1895 if (p == end || *p++ != '-') {
1896 strcpy(err, "too short control escape");
1897 return -1;
1898 }
1899 case 'c': /* \cX, \c\M-X */
1900 if (ctrl_prefix) {
1901 strcpy(err, "duplicate control escape");
1902 return -1;
1903 }
1904 ctrl_prefix = 1;
1905 if (p < end && (*p & 0x80) == 0) {
1906 if (*p == '\\') {
1907 p++;
1908 goto again;
1909 }
1910 else {
1911 code = *p++;
1912 break;
1913 }
1914 }
1915 strcpy(err, "too short control escape");
1916 return -1;
1917
1918 default:
1919 strcpy(err, "unexpected escape sequence");
1920 return -1;
1921 }
1922 if (code < 0 || 0xff < code) {
1923 strcpy(err, "invalid escape code");
1924 return -1;
1925 }
1926
1927 if (ctrl_prefix)
1928 code &= 0x1f;
1929 if (meta_prefix)
1930 code |= 0x80;
1931
1932 *pp = p;
1933 return code;
1934 }
1935
1936 static int
1937 unescape_escaped_nonascii(const char **pp, const char *end, rb_encoding *enc,
1938 VALUE buf, rb_encoding **encp, onig_errmsg_buffer err)
1939 {
1940 const char *p = *pp;
1941 int chmaxlen = rb_enc_mbmaxlen(enc);
1942 char *chbuf = ALLOCA_N(char, chmaxlen);
1943 int chlen = 0;
1944 int byte;
1945 int l;
1946
1947 memset(chbuf, 0, chmaxlen);
1948
1949 byte = read_escaped_byte(&p, end, err);
1950 if (byte == -1) {
1951 return -1;
1952 }
1953
1954 chbuf[chlen++] = byte;
1955 while (chlen < chmaxlen &&
1956 MBCLEN_NEEDMORE_P(rb_enc_precise_mbclen(chbuf, chbuf+chlen, enc))) {
1957 byte = read_escaped_byte(&p, end, err);
1958 if (byte == -1) {
1959 return -1;
1960 }
1961 chbuf[chlen++] = byte;
1962 }
1963
1964 l = rb_enc_precise_mbclen(chbuf, chbuf+chlen, enc);
1965 if (MBCLEN_INVALID_P(l)) {
1966 strcpy(err, "invalid multibyte escape");
1967 return -1;
1968 }
1969 if (1 < chlen || (chbuf[0] & 0x80)) {
1970 rb_str_buf_cat(buf, chbuf, chlen);
1971
1972 if (*encp == 0)
1973 *encp = enc;
1974 else if (*encp != enc) {
1975 strcpy(err, "escaped non ASCII character in UTF-8 regexp");
1976 return -1;
1977 }
1978 }
1979 else {
1980 char escbuf[5];
1981 snprintf(escbuf, sizeof(escbuf), "\\x%02X", chbuf[0]&0xff);
1982 rb_str_buf_cat(buf, escbuf, 4);
1983 }
1984 *pp = p;
1985 return 0;
1986 }
1987
1988 static int
1989 check_unicode_range(unsigned long code, onig_errmsg_buffer err)
1990 {
1991 if ((0xd800 <= code && code <= 0xdfff) || /* Surrogates */
1992 0x10ffff < code) {
1993 strcpy(err, "invalid Unicode range");
1994 return -1;
1995 }
1996 return 0;
1997 }
1998
1999 static int
2000 append_utf8(unsigned long uv,
2001 VALUE buf, rb_encoding **encp, onig_errmsg_buffer err)
2002 {
2003 if (check_unicode_range(uv, err) != 0)
2004 return -1;
2005 if (uv < 0x80) {
2006 char escbuf[5];
2007 snprintf(escbuf, sizeof(escbuf), "\\x%02X", (int)uv);
2008 rb_str_buf_cat(buf, escbuf, 4);
2009 }
2010 else {
2011 int len;
2012 char utf8buf[6];
2013 len = rb_uv_to_utf8(utf8buf, uv);
2014 rb_str_buf_cat(buf, utf8buf, len);
2015
2016 if (*encp == 0)
2017 *encp = rb_utf8_encoding();
2018 else if (*encp != rb_utf8_encoding()) {
2019 strcpy(err, "UTF-8 character in non UTF-8 regexp");
2020 return -1;
2021 }
2022 }
2023 return 0;
2024 }
2025
2026 static int
2027 unescape_unicode_list(const char **pp, const char *end,
2028 VALUE buf, rb_encoding **encp, onig_errmsg_buffer err)
2029 {
2030 const char *p = *pp;
2031 int has_unicode = 0;
2032 unsigned long code;
2033 int len;
2034
2035 while (p < end && ISSPACE(*p)) p++;
2036
2037 while (1) {
2038 code = ruby_scan_hex(p, end-p, &len);
2039 if (len == 0)
2040 break;
2041 if (6 < len) { /* max 10FFFF */
2042 strcpy(err, "invalid Unicode range");
2043 return -1;
2044 }
2045 p += len;
2046 if (append_utf8(code, buf, encp, err) != 0)
2047 return -1;
2048 has_unicode = 1;
2049
2050 while (p < end && ISSPACE(*p)) p++;
2051 }
2052
2053 if (has_unicode == 0) {
2054 strcpy(err, "invalid Unicode list");
2055 return -1;
2056 }
2057
2058 *pp = p;
2059
2060 return 0;
2061 }
2062
2063 static int
2064 unescape_unicode_bmp(const char **pp, const char *end,
2065 VALUE buf, rb_encoding **encp, onig_errmsg_buffer err)
2066 {
2067 const char *p = *pp;
2068 int len;
2069 unsigned long code;
2070
2071 if (end < p+4) {
2072 strcpy(err, "invalid Unicode escape");
2073 return -1;
2074 }
2075 code = ruby_scan_hex(p, 4, &len);
2076 if (len != 4) {
2077 strcpy(err, "invalid Unicode escape");
2078 return -1;
2079 }
2080 if (append_utf8(code, buf, encp, err) != 0)
2081 return -1;
2082 *pp = p + 4;
2083 return 0;
2084 }
2085
2086 static int
2087 unescape_nonascii(const char *p, const char *end, rb_encoding *enc,
2088 VALUE buf, rb_encoding **encp, int *has_property,
2089 onig_errmsg_buffer err)
2090 {
2091 char c;
2092 char smallbuf[2];
2093
2094 while (p < end) {
2095 int chlen = rb_enc_precise_mbclen(p, end, enc);
2096 if (!MBCLEN_CHARFOUND_P(chlen)) {
2097 strcpy(err, "invalid multibyte character");
2098 return -1;
2099 }
2100 chlen = MBCLEN_CHARFOUND_LEN(chlen);
2101 if (1 < chlen || (*p & 0x80)) {
2102 rb_str_buf_cat(buf, p, chlen);
2103 p += chlen;
2104 if (*encp == 0)
2105 *encp = enc;
2106 else if (*encp != enc) {
2107 strcpy(err, "non ASCII character in UTF-8 regexp");
2108 return -1;
2109 }
2110 continue;
2111 }
2112
2113 switch (c = *p++) {
2114 case '\\':
2115 if (p == end) {
2116 strcpy(err, "too short escape sequence");
2117 return -1;
2118 }
2119 switch (c = *p++) {
2120 case '1': case '2': case '3':
2121 case '4': case '5': case '6': case '7': /* \O, \OO, \OOO or backref */
2122 {
2123 int octlen;
2124 if (ruby_scan_oct(p-1, end-(p-1), &octlen) <= 0177) {
2125 /* backref or 7bit octal.
2126 no need to unescape anyway.
2127 re-escaping may break backref */
2128 goto escape_asis;
2129 }
2130 }
2131 /* xxx: How about more than 199 subexpressions? */
2132
2133 case '0': /* \0, \0O, \0OO */
2134
2135 case 'x': /* \xHH */
2136 case 'c': /* \cX, \c\M-X */
2137 case 'C': /* \C-X, \C-\M-X */
2138 case 'M': /* \M-X, \M-\C-X, \M-\cX */
2139 p = p-2;
2140 if (unescape_escaped_nonascii(&p, end, enc, buf, encp, err) != 0)
2141 return -1;
2142 break;
2143
2144 case 'u':
2145 if (p == end) {
2146 strcpy(err, "too short escape sequence");
2147 return -1;
2148 }
2149 if (*p == '{') {
2150 /* \u{H HH HHH HHHH HHHHH HHHHHH ...} */
2151 p++;
2152 if (unescape_unicode_list(&p, end, buf, encp, err) != 0)
2153 return -1;
2154 if (p == end || *p++ != '}') {
2155 strcpy(err, "invalid Unicode list");
2156 return -1;
2157 }
2158 break;
2159 }
2160 else {
2161 /* \uHHHH */
2162 if (unescape_unicode_bmp(&p, end, buf, encp, err) != 0)
2163 return -1;
2164 break;
2165 }
2166
2167 case 'p': /* \p{Hiragana} */
2168 if (!*encp) {
2169 *has_property = 1;
2170 }
2171 goto escape_asis;
2172
2173 default: /* \n, \\, \d, \9, etc. */
2174 escape_asis:
2175 smallbuf[0] = '\\';
2176 smallbuf[1] = c;
2177 rb_str_buf_cat(buf, smallbuf, 2);
2178 break;
2179 }
2180 break;
2181
2182 default:
2183 rb_str_buf_cat(buf, &c, 1);
2184 break;
2185 }
2186 }
2187
2188 return 0;
2189 }
2190
2191 static VALUE
2192 rb_reg_preprocess(const char *p, const char *end, rb_encoding *enc,
2193 rb_encoding **fixed_enc, onig_errmsg_buffer err)
2194 {
2195 VALUE buf;
2196 int has_property = 0;
2197
2198 buf = rb_str_buf_new(0);
2199
2200 if (rb_enc_asciicompat(enc))
2201 *fixed_enc = 0;
2202 else {
2203 *fixed_enc = enc;
2204 rb_enc_associate(buf, enc);
2205 }
2206
2207 if (unescape_nonascii(p, end, enc, buf, fixed_enc, &has_property, err) != 0)
2208 return Qnil;
2209
2210 if (has_property && !*fixed_enc) {
2211 *fixed_enc = enc;
2212 }
2213
2214 if (*fixed_enc) {
2215 rb_enc_associate(buf, *fixed_enc);
2216 }
2217
2218 return buf;
2219 }
2220
2221 VALUE
2222 rb_reg_check_preprocess(VALUE str)
2223 {
2224 rb_encoding *fixed_enc = 0;
2225 onig_errmsg_buffer err = "";
2226 VALUE buf;
2227 char *p, *end;
2228 rb_encoding *enc;
2229
2230 StringValue(str);
2231 p = RSTRING_PTR(str);
2232 end = p + RSTRING_LEN(str);
2233 enc = rb_enc_get(str);
2234
2235 buf = rb_reg_preprocess(p, end, enc, &fixed_enc, err);
2236 RB_GC_GUARD(str);
2237
2238 if (buf == Qnil) {
2239 return rb_reg_error_desc(str, 0, err);
2240 }
2241 return Qnil;
2242 }
2243
2244 static VALUE
2245 rb_reg_preprocess_dregexp(VALUE ary)
2246 {
2247 rb_encoding *fixed_enc = 0;
2248 rb_encoding *regexp_enc = 0;
2249 onig_errmsg_buffer err = "";
2250 int i;
2251 VALUE result = 0;
2252 int argc = RARRAY_LEN(ary);
2253 VALUE *argv = RARRAY_PTR(ary);
2254
2255 if (argc == 0) {
2256 rb_raise(rb_eArgError, "no arguments given");
2257 }
2258
2259 for (i = 0; i < argc; i++) {
2260 VALUE str = argv[i];
2261 VALUE buf;
2262 char *p, *end;
2263 rb_encoding *src_enc;
2264
2265 StringValue(str);
2266 p = RSTRING_PTR(str);
2267 end = p + RSTRING_LEN(str);
2268 src_enc = rb_enc_get(str);
2269
2270 buf = rb_reg_preprocess(p, end, src_enc, &fixed_enc, err);
2271
2272 if (buf == Qnil)
2273 rb_raise(rb_eArgError, "%s", err);
2274
2275 if (fixed_enc != 0) {
2276 if (regexp_enc != 0 && regexp_enc != fixed_enc) {
2277 rb_raise(rb_eArgError, "encoding mismatch in dynamic regexp : %s and %s",
2278 rb_enc_name(regexp_enc), rb_enc_name(fixed_enc));
2279 }
2280 regexp_enc = fixed_enc;
2281 }
2282
2283 if (!result)
2284 result = rb_str_new3(str);
2285 else
2286 rb_str_buf_append(result, str);
2287 }
2288 if (regexp_enc) {
2289 rb_enc_associate(result, regexp_enc);
2290 }
2291
2292 return result;
2293 }
2294
2295 static int
2296 rb_reg_initialize(VALUE obj, const char *s, int len, rb_encoding *enc,
2297 int options, onig_errmsg_buffer err)
2298 {
2299 struct RRegexp *re = RREGEXP(obj);
2300 VALUE unescaped;
2301 rb_encoding *fixed_enc = 0;
2302 rb_encoding *a_enc = rb_ascii8bit_encoding();
2303
2304 if (!OBJ_UNTRUSTED(obj) && rb_safe_level() >= 4)
2305 rb_raise(rb_eSecurityError, "Insecure: can't modify regexp");
2306 rb_check_frozen(obj);
2307 if (FL_TEST(obj, REG_LITERAL))
2308 rb_raise(rb_eSecurityError, "can't modify literal regexp");
2309 if (re->ptr)
2310 rb_raise(rb_eTypeError, "already initialized regexp");
2311 re->ptr = 0;
2312
2313 unescaped = rb_reg_preprocess(s, s+len, enc, &fixed_enc, err);
2314 if (unescaped == Qnil)
2315 return -1;
2316
2317 if (fixed_enc) {
2318 if ((fixed_enc != enc && (options & ARG_ENCODING_FIXED)) ||
2319 (fixed_enc != a_enc && (options & ARG_ENCODING_NONE))) {
2320 strcpy(err, "incompatible character encoding");
2321 return -1;
2322 }
2323 if (fixed_enc != a_enc) {
2324 options |= ARG_ENCODING_FIXED;
2325 enc = fixed_enc;
2326 }
2327 }
2328 else if (!(options & ARG_ENCODING_FIXED)) {
2329 enc = rb_usascii_encoding();
2330 }
2331
2332 rb_enc_associate((VALUE)re, enc);
2333 if ((options & ARG_ENCODING_FIXED) || fixed_enc) {
2334 re->basic.flags |= KCODE_FIXED;
2335 }
2336 if (options & ARG_ENCODING_NONE) {
2337 re->basic.flags |= REG_ENCODING_NONE;
2338 }
2339
2340 re->ptr = make_regexp(RSTRING_PTR(unescaped), RSTRING_LEN(unescaped), enc,
2341 options & ARG_REG_OPTION_MASK, err);
2342 if (!re->ptr) return -1;
2343 re->src = rb_enc_str_new(s, len, enc);
2344 OBJ_FREEZE(re->src);
2345 RB_GC_GUARD(unescaped);
2346 return 0;
2347 }
2348
2349 static int
2350 rb_reg_initialize_str(VALUE obj, VALUE str, int options, onig_errmsg_buffer err)
2351 {
2352 int ret;
2353 rb_encoding *enc = rb_enc_get(str);
2354 if (options & ARG_ENCODING_NONE) {
2355 rb_encoding *ascii8bit = rb_ascii8bit_encoding();
2356 if (enc != ascii8bit) {
2357 if (rb_enc_str_coderange(str) != ENC_CODERANGE_7BIT) {
2358 strcpy(err, "/.../n has a non escaped non ASCII character in non ASCII-8BIT script");
2359 return -1;
2360 }
2361 enc = ascii8bit;
2362 }
2363 }
2364 ret = rb_reg_initialize(obj, RSTRING_PTR(str), RSTRING_LEN(str), enc,
2365 options, err);
2366 RB_GC_GUARD(str);
2367 return ret;
2368 }
2369
2370 static VALUE
2371 rb_reg_s_alloc(VALUE klass)
2372 {
2373 NEWOBJ(re, struct RRegexp);
2374 OBJSETUP(re, klass, T_REGEXP);
2375
2376 re->ptr = 0;
2377 re->src = 0;
2378 re->usecnt = 0;
2379
2380 return (VALUE)re;
2381 }
2382
2383 VALUE
2384 rb_reg_new_str(VALUE s, int options)
2385 {
2386 VALUE re = rb_reg_s_alloc(rb_cRegexp);
2387 onig_errmsg_buffer err = "";
2388
2389 if (rb_reg_initialize_str(re, s, options, err) != 0) {
2390 rb_reg_raise_str(s, options, err);
2391 }
2392
2393 return re;
2394 }
2395
2396 VALUE
2397 rb_reg_new_ary(VALUE ary, int opt)
2398 {
2399 return rb_reg_new_str(rb_reg_preprocess_dregexp(ary), opt);
2400 }
2401
2402 VALUE
2403 rb_enc_reg_new(const char *s, long len, rb_encoding *enc, int options)
2404 {
2405 VALUE re = rb_reg_s_alloc(rb_cRegexp);
2406 onig_errmsg_buffer err = "";
2407
2408 if (rb_reg_initialize(re, s, len, enc, options, err) != 0) {
2409 rb_enc_reg_raise(s, len, enc, options, err);
2410 }
2411
2412 return re;
2413 }
2414
2415 VALUE
2416 rb_reg_new(const char *s, long len, int options)
2417 {
2418 return rb_enc_reg_new(s, len, rb_ascii8bit_encoding(), options);
2419 }
2420
2421 VALUE
2422 rb_reg_compile(VALUE str, int options)
2423 {
2424 VALUE re = rb_reg_s_alloc(rb_cRegexp);
2425 onig_errmsg_buffer err = "";
2426
2427 if (!str) str = rb_str_new(0,0);
2428 if (rb_reg_initialize_str(re, str, options, err) != 0) {
2429 rb_set_errinfo(rb_reg_error_desc(str, options, err));
2430 return Qnil;
2431 }
2432 FL_SET(re, REG_LITERAL);
2433 return re;
2434 }
2435
2436 static VALUE reg_cache;
2437
2438 VALUE
2439 rb_reg_regcomp(VALUE str)
2440 {
2441 volatile VALUE save_str = str;
2442 if (reg_cache && RREGEXP_SRC_LEN(reg_cache) == RSTRING_LEN(str)
2443 && ENCODING_GET(reg_cache) == ENCODING_GET(str)
2444 && memcmp(RREGEXP_SRC_PTR(reg_cache), RSTRING_PTR(str), RSTRING_LEN(str)) == 0)
2445 return reg_cache;
2446
2447 return reg_cache = rb_reg_new_str(save_str, 0);
2448 }
2449
2450 /*
2451 * call-seq:
2452 * rxp.hash => fixnum
2453 *
2454 * Produce a hash based on the text and options of this regular expression.
2455 */
2456
2457 static VALUE
2458 rb_reg_hash(VALUE re)
2459 {
2460 int hashval, len;
2461 char *p;
2462
2463 rb_reg_check(re);
2464 hashval = RREGEXP(re)->ptr->options;
2465 len = RREGEXP_SRC_LEN(re);
2466 p = RREGEXP_SRC_PTR(re);
2467 while (len--) {
2468 hashval = hashval * 33 + *p++;
2469 }
2470 hashval = hashval + (hashval>>5);
2471
2472 return INT2FIX(hashval);
2473 }
2474
2475
2476 /*
2477 * call-seq:
2478 * rxp == other_rxp => true or false
2479 * rxp.eql?(other_rxp) => true or false
2480 *
2481 * Equality---Two regexps are equal if their patterns are identical, they have
2482 * the same character set code, and their <code>casefold?</code> values are the
2483 * same.
2484 *
2485 * /abc/ == /abc/x #=> false
2486 * /abc/ == /abc/i #=> false
2487 * /abc/ == /abc/n #=> false
2488 * /abc/u == /abc/n #=> false
2489 */
2490
2491 static VALUE
2492 rb_reg_equal(VALUE re1, VALUE re2)
2493 {
2494 if (re1 == re2) return Qtrue;
2495 if (TYPE(re2) != T_REGEXP) return Qfalse;
2496 rb_reg_check(re1); rb_reg_check(re2);
2497 if (FL_TEST(re1, KCODE_FIXED) != FL_TEST(re2, KCODE_FIXED)) return Qfalse;
2498 if (RREGEXP(re1)->ptr->options != RREGEXP(re2)->ptr->options) return Qfalse;
2499 if (RREGEXP_SRC_LEN(re1) != RREGEXP_SRC_LEN(re2)) return Qfalse;
2500 if (ENCODING_GET(re1) != ENCODING_GET(re2)) return Qfalse;
2501 if (memcmp(RREGEXP_SRC_PTR(re1), RREGEXP_SRC_PTR(re2), RREGEXP_SRC_LEN(re1)) == 0) {
2502 return Qtrue;
2503 }
2504 return Qfalse;
2505 }
2506
2507 static VALUE
2508 reg_operand(VALUE s, int check)
2509 {
2510 if (SYMBOL_P(s)) {
2511 return rb_sym_to_s(s);
2512 }
2513 else {
2514 VALUE tmp = rb_check_string_type(s);
2515 if (check && NIL_P(tmp)) {
2516 rb_raise(rb_eTypeError, "can't convert %s to String",
2517 rb_obj_classname(s));
2518 }
2519 return tmp;
2520 }
2521 }
2522
2523 static long
2524 reg_match_pos(VALUE re, VALUE *strp, long pos)
2525 {
2526 VALUE str = *strp;
2527
2528 if (NIL_P(str)) {
2529 rb_backref_set(Qnil);
2530 return -1;
2531 }
2532 *strp = str = reg_operand(str, Qtrue);
2533 if (pos != 0) {
2534 if (pos < 0) {
2535 VALUE l = rb_str_length(str);
2536 pos += NUM2INT(l);
2537 if (pos < 0) {
2538 return pos;
2539 }
2540 }
2541 pos = rb_reg_adjust_startpos(re, str, pos, 0);
2542 }
2543 return rb_reg_search(re, str, pos, 0);
2544 }
2545
2546 /*
2547 * call-seq:
2548 * rxp =~ str => integer or nil
2549 *
2550 * Match---Matches <i>rxp</i> against <i>str</i>.
2551 *
2552 * /at/ =~ "input data" #=> 7
2553 * /ax/ =~ "input data" #=> nil
2554 *
2555 * If <code>=~</code> is used with a regexp literal with named captures,
2556 * captured strings (or nil) is assigned to local variables named by
2557 * the capture names.
2558 *
2559 * /(?<lhs>\w+)\s*=\s*(?<rhs>\w+)/ =~ " x = y "
2560 * p lhs #=> "x"
2561 * p rhs #=> "y"
2562 *
2563 * If it is not matched, nil is assigned for the variables.
2564 *
2565 * /(?<lhs>\w+)\s*=\s*(?<rhs>\w+)/ =~ " x = "
2566 * p lhs #=> nil
2567 * p rhs #=> nil
2568 *
2569 * This assignment is implemented in the Ruby parser.
2570 * So a regexp literal is required for the assignment.
2571 * The assignment is not occur if the regexp is not a literal.
2572 *
2573 * re = /(?<lhs>\w+)\s*=\s*(?<rhs>\w+)/
2574 * re =~ " x = "
2575 * p lhs # undefined local variable
2576 * p rhs # undefined local variable
2577 *
2578 * A regexp interpolation, <code>#{}</code>, also disables
2579 * the assignment.
2580 *
2581 * rhs_pat = /(?<rhs>\w+)/
2582 * /(?<lhs>\w+)\s*=\s*#{rhs_pat}/ =~ "x = y"
2583 * p lhs # undefined local variable
2584 *
2585 */
2586
2587 VALUE
2588 rb_reg_match(VALUE re, VALUE str)
2589 {
2590 long pos = reg_match_pos(re, &str, 0);
2591 if (pos < 0) return Qnil;
2592 pos = rb_str_sublen(str, pos);
2593 return LONG2FIX(pos);
2594 }
2595
2596 /*
2597 * call-seq:
2598 * rxp === str => true or false
2599 *
2600 * Case Equality---Synonym for <code>Regexp#=~</code> used in case statements.
2601 *
2602 * a = "HELLO"
2603 * case a
2604 * when /^[a-z]*$/; print "Lower case\n"
2605 * when /^[A-Z]*$/; print "Upper case\n"
2606 * else; print "Mixed case\n"
2607 * end
2608 *
2609 * <em>produces:</em>
2610 *
2611 * Upper case
2612 */
2613
2614 VALUE
2615 rb_reg_eqq(VALUE re, VALUE str)
2616 {
2617 long start;
2618
2619 str = reg_operand(str, Qfalse);
2620 if (NIL_P(str)) {
2621 rb_backref_set(Qnil);
2622 return Qfalse;
2623 }
2624 start = rb_reg_search(re, str, 0, 0);
2625 if (start < 0) {
2626 return Qfalse;
2627 }
2628 return Qtrue;
2629 }
2630
2631
2632 /*
2633 * call-seq:
2634 * ~ rxp => integer or nil
2635 *
2636 * Match---Matches <i>rxp</i> against the contents of <code>$_</code>.
2637 * Equivalent to <code><i>rxp</i> =~ $_</code>.
2638 *
2639 * $_ = "input data"
2640 * ~ /at/ #=> 7
2641 */
2642
2643 VALUE
2644 rb_reg_match2(VALUE re)
2645 {
2646 long start;
2647 VALUE line = rb_lastline_get();
2648
2649 if (TYPE(line) != T_STRING) {
2650 rb_backref_set(Qnil);
2651 return Qnil;
2652 }
2653
2654 start = rb_reg_search(re, line, 0, 0);
2655 if (start < 0) {
2656 return Qnil;
2657 }
2658 start = rb_str_sublen(line, start);
2659 return LONG2FIX(start);
2660 }
2661
2662
2663 /*
2664 * call-seq:
2665 * rxp.match(str) => matchdata or nil
2666 * rxp.match(str,pos) => matchdata or nil
2667 *
2668 * Returns a <code>MatchData</code> object describing the match, or
2669 * <code>nil</code> if there was no match. This is equivalent to retrieving the
2670 * value of the special variable <code>$~</code> following a normal match.
2671 * If the second parameter is present, it specifies the position in the string
2672 * to begin the search.
2673 *
2674 * /(.)(.)(.)/.match("abc")[2] #=> "b"
2675 * /(.)(.)/.match("abc", 1)[2] #=> "c"
2676 *
2677 * If a block is given, invoke the block with MatchData if match succeed, so
2678 * that you can write
2679 *
2680 * pat.match(str) {|m| ...}
2681 *
2682 * instead of
2683 *
2684 * if m = pat.match(str)
2685 * ...
2686 * end
2687 *
2688 * The return value is a value from block execution in this case.
2689 */
2690
2691 static VALUE
2692 rb_reg_match_m(int argc, VALUE *argv, VALUE re)
2693 {
2694 VALUE result, str, initpos;
2695 long pos;
2696
2697 if (rb_scan_args(argc, argv, "11", &str, &initpos) == 2) {
2698 pos = NUM2LONG(initpos);
2699 }
2700 else {
2701 pos = 0;
2702 }
2703
2704 pos = reg_match_pos(re, &str, pos);
2705 if (pos < 0) {
2706 rb_backref_set(Qnil);
2707 return Qnil;
2708 }
2709 result = rb_backref_get();
2710 rb_match_busy(result);
2711 if (!NIL_P(result) && rb_block_given_p()) {
2712 return rb_yield(result);
2713 }
2714 return result;
2715 }
2716
2717 /*
2718 * Document-method: compile
2719 *
2720 * Synonym for <code>Regexp.new</code>
2721 */
2722
2723
2724 /*
2725 * call-seq:
2726 * Regexp.new(string [, options]) => regexp
2727 * Regexp.new(regexp) => regexp
2728 * Regexp.compile(string [, options]) => regexp
2729 * Regexp.compile(regexp) => regexp
2730 *
2731 * Constructs a new regular expression from <i>pattern</i>, which can be either
2732 * a <code>String</code> or a <code>Regexp</code> (in which case that regexp's
2733 * options are propagated, and new options may not be specified (a change as of
2734 * Ruby 1.8). If <i>options</i> is a <code>Fixnum</code>, it should be one or
2735 * more of the constants <code>Regexp::EXTENDED</code>,
2736 * <code>Regexp::IGNORECASE</code>, and <code>Regexp::MULTILINE</code>,
2737 * <em>or</em>-ed together. Otherwise, if <i>options</i> is not
2738 * <code>nil</code>, the regexp will be case insensitive.
2739 *
2740 * r1 = Regexp.new('^a-z+:\\s+\w+') #=> /^a-z+:\s+\w+/
2741 * r2 = Regexp.new('cat', true) #=> /cat/i
2742 * r3 = Regexp.new('dog', Regexp::EXTENDED) #=> /dog/x
2743 * r4 = Regexp.new(r2) #=> /cat/i
2744 */
2745
2746 static VALUE
2747 rb_reg_initialize_m(int argc, VALUE *argv, VALUE self)
2748 {
2749 onig_errmsg_buffer err = "";
2750 int flags = 0;
2751 VALUE str;
2752 rb_encoding *enc;
2753 const char *ptr;
2754 long len;
2755
2756 if (argc == 0 || argc > 3) {
2757 rb_raise(rb_eArgError, "wrong number of arguments");
2758 }
2759 if (TYPE(argv[0]) == T_REGEXP) {
2760 VALUE re = argv[0];
2761
2762 if (argc > 1) {
2763 rb_warn("flags ignored");
2764 }
2765 rb_reg_check(re);
2766 flags = rb_reg_options(re);
2767 ptr = RREGEXP_SRC_PTR(re);
2768 len = RREGEXP_SRC_LEN(re);
2769 enc = rb_enc_get(re);
2770 if (rb_reg_initialize(self, ptr, len, enc, flags, err)) {
2771 str = rb_enc_str_new(ptr, len, enc);
2772 rb_reg_raise_str(str, flags, err);
2773 }
2774 }
2775 else {
2776 if (argc >= 2) {
2777 if (FIXNUM_P(argv[1])) flags = FIX2INT(argv[1]);
2778 else if (RTEST(argv[1])) flags = ONIG_OPTION_IGNORECASE;
2779 }
2780 enc = 0;
2781 if (argc == 3 && !NIL_P(argv[2])) {
2782 char *kcode = StringValuePtr(argv[2]);
2783 if (kcode[0] == 'n' || kcode[1] == 'N') {
2784 enc = rb_ascii8bit_encoding();
2785 flags |= ARG_ENCODING_FIXED;
2786 }
2787 else {
2788 rb_warning("encoding option is obsolete - %s", kcode);
2789 }
2790 }
2791 str = argv[0];
2792 ptr = StringValuePtr(str);
2793 if (enc
2794 ? rb_reg_initialize(self, ptr, RSTRING_LEN(str), enc, flags, err)
2795 : rb_reg_initialize_str(self, str, flags, err)) {
2796 rb_reg_raise_str(str, flags, err);
2797 }
2798 }
2799 return self;
2800 }
2801
2802 VALUE
2803 rb_reg_quote(VALUE str)
2804 {
2805 rb_encoding *enc = rb_enc_get(str);
2806 char *s, *send, *t;
2807 VALUE tmp;
2808 int c, clen;
2809 int ascii_only = rb_enc_str_asciionly_p(str);
2810
2811 s = RSTRING_PTR(str);
2812 send = s + RSTRING_LEN(str);
2813 while (s < send) {
2814 c = rb_enc_ascget(s, send, &clen, enc);
2815 if (c == -1) {
2816 s += mbclen(s, send, enc);
2817 continue;
2818 }
2819 switch (c) {
2820 case '[': case ']': case '{': case '}':
2821 case '(': case ')': case '|': case '-':
2822 case '*': case '.': case '\\':
2823 case '?': case '+': case '^': case '$':
2824 case ' ': case '#':
2825 case '\t': case '\f': case '\v': case '\n': case '\r':
2826 goto meta_found;
2827 }
2828 s += clen;
2829 }
2830 tmp = rb_str_new3(str);
2831 if (ascii_only) {
2832 rb_enc_associate(tmp, rb_usascii_encoding());
2833 }
2834 return tmp;
2835
2836 meta_found:
2837 tmp = rb_str_new(0, RSTRING_LEN(str)*2);
2838 if (ascii_only) {
2839 rb_enc_associate(tmp, rb_usascii_encoding());
2840 }
2841 else {
2842 rb_enc_copy(tmp, str);
2843 }
2844 t = RSTRING_PTR(tmp);
2845 /* copy upto metacharacter */
2846 memcpy(t, RSTRING_PTR(str), s - RSTRING_PTR(str));
2847 t += s - RSTRING_PTR(str);
2848
2849 while (s < send) {
2850 c = rb_enc_ascget(s, send, &clen, enc);
2851 if (c == -1) {
2852 int n = mbclen(s, send, enc);
2853
2854 while (n--)
2855 *t++ = *s++;
2856 continue;
2857 }
2858 s += clen;
2859 switch (c) {
2860 case '[': case ']': case '{': case '}':
2861 case '(': case ')': case '|': case '-':
2862 case '*': case '.': case '\\':
2863 case '?': case '+': case '^': case '$':
2864 case '#':
2865 *t++ = '\\';
2866 break;
2867 case ' ':
2868 *t++ = '\\';
2869 *t++ = ' ';
2870 continue;
2871 case '\t':
2872 *t++ = '\\';
2873 *t++ = 't';
2874 continue;
2875 case '\n':
2876 *t++ = '\\';
2877 *t++ = 'n';
2878 continue;
2879 case '\r':
2880 *t++ = '\\';
2881 *t++ = 'r';
2882 continue;
2883 case '\f':
2884 *t++ = '\\';
2885 *t++ = 'f';
2886 continue;
2887 case '\v':
2888 *t++ = '\\';
2889 *t++ = 'v';
2890 continue;
2891 }
2892 *t++ = c;
2893 }
2894 rb_str_resize(tmp, t - RSTRING_PTR(tmp));
2895 OBJ_INFECT(tmp, str);
2896 return tmp;
2897 }
2898
2899
2900 /*
2901 * call-seq:
2902 * Regexp.escape(str) => string
2903 * Regexp.quote(str) => string
2904 *
2905 * Escapes any characters that would have special meaning in a regular
2906 * expression. Returns a new escaped string, or self if no characters are
2907 * escaped. For any string,
2908 * <code>Regexp.new(Regexp.escape(<i>str</i>))=~<i>str</i></code> will be true.
2909 *
2910 * Regexp.escape('\*?{}.') #=> \\\*\?\{\}\.
2911 *
2912 */
2913
2914 static VALUE
2915 rb_reg_s_quote(VALUE c, VALUE str)
2916 {
2917 return rb_reg_quote(reg_operand(str, Qtrue));
2918 }
2919
2920 int
2921 rb_reg_options(VALUE re)
2922 {
2923 int options;
2924
2925 rb_reg_check(re);
2926 options = RREGEXP(re)->ptr->options & ARG_REG_OPTION_MASK;
2927 if (RBASIC(re)->flags & KCODE_FIXED) options |= ARG_ENCODING_FIXED;
2928 if (RBASIC(re)->flags & REG_ENCODING_NONE) options |= ARG_ENCODING_NONE;
2929 return options;
2930 }
2931
2932 VALUE
2933 rb_check_regexp_type(VALUE re)
2934 {
2935 return rb_check_convert_type(re, T_REGEXP, "Regexp", "to_regexp");
2936 }
2937
2938 /*
2939 * call-seq:
2940 * Regexp.try_convert(obj) -> re or nil
2941 *
2942 * Try to convert <i>obj</i> into a Regexp, using to_regexp method.
2943 * Returns converted regexp or nil if <i>obj</i> cannot be converted
2944 * for any reason.
2945 *
2946 * Regexp.try_convert(/re/) #=> /re/
2947 * Regexp.try_convert("re") #=> nil
2948 *
2949 * o = Object.new
2950 * Regexp.try_convert(o) #=> nil
2951 * def o.to_regexp() /foo/ end
2952 * Regexp.try_convert(o) #=> /foo/
2953 *
2954 */
2955 static VALUE
2956 rb_reg_s_try_convert(VALUE dummy, VALUE re)
2957 {
2958 return rb_check_regexp_type(re);
2959 }
2960
2961 static VALUE
2962 rb_reg_s_union(VALUE self, VALUE args0)
2963 {
2964 long argc = RARRAY_LEN(args0);
2965
2966 if (argc == 0) {
2967 VALUE args[1];
2968 args[0] = rb_str_new2("(?!)");
2969 return rb_class_new_instance(1, args, rb_cRegexp);
2970 }
2971 else if (argc == 1) {
2972 VALUE arg = rb_ary_entry(args0, 0);
2973 VALUE re = rb_check_regexp_type(arg);
2974 if (!NIL_P(re))
2975 return re;
2976 else {
2977 VALUE quoted;
2978 quoted = rb_reg_s_quote(Qnil, arg);
2979 return rb_reg_new_str(quoted, 0);
2980 }
2981 }
2982 else {
2983 int i;
2984 VALUE source = rb_str_buf_new(0);
2985 rb_encoding *result_enc;
2986
2987 int has_asciionly = 0;
2988 rb_encoding *has_ascii_compat_fixed = 0;
2989 rb_encoding *has_ascii_incompat = 0;
2990
2991 for (i = 0; i < argc; i++) {
2992 volatile VALUE v;
2993 VALUE e = rb_ary_entry(args0, i);
2994
2995 if (0 < i)
2996 rb_str_buf_cat_ascii(source, "|");
2997
2998 v = rb_check_regexp_type(e);
2999 if (!NIL_P(v)) {
3000 rb_encoding *enc = rb_enc_get(v);
3001 if (!rb_enc_asciicompat(enc)) {
3002 if (!has_ascii_incompat)
3003 has_ascii_incompat = enc;
3004 else if (has_ascii_incompat != enc)
3005 rb_raise(rb_eArgError, "incompatible encodings: %s and %s",
3006 rb_enc_name(has_ascii_incompat), rb_enc_name(enc));
3007 }
3008 else if (rb_reg_fixed_encoding_p(v)) {
3009 if (!has_ascii_compat_fixed)
3010 has_ascii_compat_fixed = enc;
3011 else if (has_ascii_compat_fixed != enc)
3012 rb_raise(rb_eArgError, "incompatible encodings: %s and %s",
3013 rb_enc_name(has_ascii_compat_fixed), rb_enc_name(enc));
3014 }
3015 else {
3016 has_asciionly = 1;
3017 }
3018 v = rb_reg_to_s(v);
3019 }
3020 else {
3021 rb_encoding *enc;
3022 StringValue(e);
3023 enc = rb_enc_get(e);
3024 if (!rb_enc_str_asciicompat_p(e)) {
3025 if (!has_ascii_incompat)
3026 has_ascii_incompat = enc;
3027 else if (has_ascii_incompat != enc)
3028 rb_raise(rb_eArgError, "incompatible encodings: %s and %s",
3029 rb_enc_name(has_ascii_incompat), rb_enc_name(enc));
3030 }
3031 else if (rb_enc_str_asciionly_p(e)) {
3032 has_asciionly = 1;
3033 }
3034 else {
3035 if (!has_ascii_compat_fixed)
3036 has_ascii_compat_fixed = enc;
3037 else if (has_ascii_compat_fixed != enc)
3038 rb_raise(rb_eArgError, "incompatible encodings: %s and %s",
3039 rb_enc_name(has_ascii_compat_fixed), rb_enc_name(enc));
3040 }
3041 v = rb_reg_s_quote(Qnil, e);
3042 }
3043 if (has_ascii_incompat) {
3044 if (has_asciionly) {
3045 rb_raise(rb_eArgError, "ASCII incompatible encoding: %s",
3046 rb_enc_name(has_ascii_incompat));
3047 }
3048 if (has_ascii_compat_fixed) {
3049 rb_raise(rb_eArgError, "incompatible encodings: %s and %s",
3050 rb_enc_name(has_ascii_incompat), rb_enc_name(has_ascii_compat_fixed));
3051 }
3052 }
3053
3054 if (i == 0) {
3055 rb_enc_copy(source, v);
3056 }
3057 rb_str_append(source, v);
3058 }
3059
3060 if (has_ascii_incompat) {
3061 result_enc = has_ascii_incompat;
3062 }
3063 else if (has_ascii_compat_fixed) {
3064 result_enc = has_ascii_compat_fixed;
3065 }
3066 else {
3067 result_enc = rb_ascii8bit_encoding();
3068 }
3069
3070 rb_enc_associate(source, result_enc);
3071 return rb_class_new_instance(1, &source, rb_cRegexp);
3072 }
3073 }
3074
3075 /*
3076 * call-seq:
3077 * Regexp.union(pat1, pat2, ...) => new_regexp
3078 * Regexp.union(pats_ary) => new_regexp
3079 *
3080 * Return a <code>Regexp</code> object that is the union of the given
3081 * <em>pattern</em>s, i.e., will match any of its parts. The <em>pattern</em>s
3082 * can be Regexp objects, in which case their options will be preserved, or
3083 * Strings. If no patterns are given, returns <code>/(?!)/</code>.
3084 *
3085 * Regexp.union #=> /(?!)/
3086 * Regexp.union("penzance") #=> /penzance/
3087 * Regexp.union("a+b*c") #=> /a\+b\*c/
3088 * Regexp.union("skiing", "sledding") #=> /skiing|sledding/
3089 * Regexp.union(["skiing", "sledding"]) #=> /skiing|sledding/
3090 * Regexp.union(/dogs/, /cats/i) #=> /(?-mix:dogs)|(?i-mx:cats)/
3091 */
3092 static VALUE
3093 rb_reg_s_union_m(VALUE self, VALUE args)
3094 {
3095 VALUE v;
3096 if (RARRAY_LEN(args) == 1 &&
3097 !NIL_P(v = rb_check_array_type(rb_ary_entry(args, 0)))) {
3098 return rb_reg_s_union(self, v);
3099 }
3100 return rb_reg_s_union(self, args);
3101 }
3102
3103 /* :nodoc: */
3104 static VALUE
3105 rb_reg_init_copy(VALUE copy, VALUE re)
3106 {
3107 onig_errmsg_buffer err = "";
3108 const char *s;
3109 long len;
3110
3111 if (copy == re) return copy;
3112 rb_check_frozen(copy);
3113 /* need better argument type check */
3114 if (!rb_obj_is_instance_of(re, rb_obj_class(copy))) {
3115 rb_raise(rb_eTypeError, "wrong argument type");
3116 }
3117 rb_reg_check(re);
3118 s = RREGEXP_SRC_PTR(re);
3119 len = RREGEXP_SRC_LEN(re);
3120 if (rb_reg_initialize(copy, s, len, rb_enc_get(re), rb_reg_options(re), err) != 0) {
3121 rb_reg_raise(s, len, err, re);
3122 }
3123 return copy;
3124 }
3125
3126 VALUE
3127 rb_reg_regsub(VALUE str, VALUE src, struct re_registers *regs, VALUE regexp)
3128 {
3129 VALUE val = 0;
3130 char *p, *s, *e;
3131 int no, clen;
3132 rb_encoding *str_enc = rb_enc_get(str);
3133 rb_encoding *src_enc = rb_enc_get(src);
3134 int acompat = rb_enc_asciicompat(str_enc);
3135 #define ASCGET(s,e,cl) (acompat ? (*cl=1,s[0]) : rb_enc_ascget(s, e, cl, str_enc))
3136
3137 p = s = RSTRING_PTR(str);
3138 e = s + RSTRING_LEN(str);
3139
3140 while (s < e) {
3141 int c = ASCGET(s, e, &clen);
3142 char *ss;
3143
3144 if (c == -1) {
3145 s += mbclen(s, e, str_enc);
3146 continue;
3147 }
3148 ss = s;
3149 s += clen;
3150
3151 if (c != '\\' || s == e) continue;
3152
3153 if (!val) {
3154 val = rb_str_buf_new(ss-p);
3155 }
3156 rb_enc_str_buf_cat(val, p, ss-p, str_enc);
3157
3158 c = ASCGET(s, e, &clen);
3159 if (c == -1) {
3160 s += mbclen(s, e, str_enc);
3161 rb_enc_str_buf_cat(val, ss, s-ss, str_enc);
3162 p = s;
3163 continue;
3164 }
3165 s += clen;
3166
3167 p = s;
3168 switch (c) {
3169 case '1': case '2': case '3': case '4':
3170 case '5': case '6': case '7': case '8': case '9':
3171 if (onig_noname_group_capture_is_active(RREGEXP(regexp)->ptr)) {
3172 no = c - '0';
3173 }
3174 else {
3175 continue;
3176 }
3177 break;
3178
3179 case 'k':
3180 if (s < e && ASCGET(s, e, &clen) == '<') {
3181 char *name, *name_end;
3182
3183 name_end = name = s + clen;
3184 while (name_end < e) {
3185 c = ASCGET(name_end, e, &clen);
3186 if (c == '>') break;
3187 name_end += c == -1 ? mbclen(name_end, e, str_enc) : clen;
3188 }
3189 if (name_end < e) {
3190 no = name_to_backref_number(regs, regexp, name, name_end);
3191 p = s = name_end + clen;
3192 break;
3193 }
3194 else {
3195 rb_raise(rb_eRuntimeError, "invalid group name reference format");
3196 }
3197 }
3198
3199 rb_enc_str_buf_cat(val, ss, s-ss, str_enc);
3200 continue;
3201
3202 case '0':
3203 case '&':
3204 no = 0;
3205 break;
3206
3207 case '`':
3208 rb_enc_str_buf_cat(val, RSTRING_PTR(src), BEG(0), src_enc);
3209 continue;
3210
3211 case '\'':
3212 rb_enc_str_buf_cat(val, RSTRING_PTR(src)+END(0), RSTRING_LEN(src)-END(0), src_enc);
3213 continue;
3214
3215 case '+':
3216 no = regs->num_regs-1;
3217 while (BEG(no) == -1 && no > 0) no--;
3218 if (no == 0) continue;
3219 break;
3220
3221 case '\\':
3222 rb_enc_str_buf_cat(val, s-clen, clen, str_enc);
3223 continue;
3224
3225 default:
3226 rb_enc_str_buf_cat(val, ss, s-ss, str_enc);
3227 continue;
3228 }
3229
3230 if (no >= 0) {
3231 if (no >= regs->num_regs) continue;
3232 if (BEG(no) == -1) continue;
3233 rb_enc_str_buf_cat(val, RSTRING_PTR(src)+BEG(no), END(no)-BEG(no), src_enc);
3234 }
3235 }
3236
3237 if (!val) return str;
3238 if (p < e) {
3239 rb_enc_str_buf_cat(val, p, e-p, str_enc);
3240 }
3241
3242 return val;
3243 }
3244
3245 static VALUE
3246 kcode_getter(void)
3247 {
3248 rb_warn("variable $KCODE is no longer effective");
3249 return Qnil;
3250 }
3251
3252 static void
3253 kcode_setter(VALUE val, ID id)
3254 {
3255 rb_warn("variable $KCODE is no longer effective; ignored");
3256 }
3257
3258 static VALUE
3259 ignorecase_getter(void)
3260 {
3261 rb_warn("variable $= is no longer effective");
3262 return Qfalse;
3263 }
3264
3265 static void
3266 ignorecase_setter(VALUE val, ID id)
3267 {
3268 rb_warn("variable $= is no longer effective; ignored");
3269 }
3270
3271 static VALUE
3272 match_getter(void)
3273 {
3274 VALUE match = rb_backref_get();
3275
3276 if (NIL_P(match)) return Qnil;
3277 rb_match_busy(match);
3278 return match;
3279 }
3280
3281 static void
3282 match_setter(VALUE val)
3283 {
3284 if (!NIL_P(val)) {
3285 Check_Type(val, T_MATCH);
3286 }
3287 rb_backref_set(val);
3288 }
3289
3290 /*
3291 * call-seq:
3292 * Regexp.last_match => matchdata
3293 * Regexp.last_match(n) => str
3294 *
3295 * The first form returns the <code>MatchData</code> object generated by the
3296 * last successful pattern match. Equivalent to reading the global variable
3297 * <code>$~</code>. The second form returns the <i>n</i>th field in this
3298 * <code>MatchData</code> object.
3299 * <em>n</em> can be a string or symbol to reference a named capture.
3300 *
3301 * /c(.)t/ =~ 'cat' #=> 0
3302 * Regexp.last_match #=> #<MatchData "cat" 1:"a">
3303 * Regexp.last_match(0) #=> "cat"
3304 * Regexp.last_match(1) #=> "a"
3305 * Regexp.last_match(2) #=> nil
3306 *
3307 * /(?<lhs>\w+)\s*=\s*(?<rhs>\w+)/ =~ "var = val"
3308 * Regexp.last_match #=> #<MatchData "var = val" lhs:"var" rhs:"val">
3309 * Regexp.last_match(:lhs) #=> "var"
3310 * Regexp.last_match(:rhs) #=> "val"
3311 */
3312
3313 static VALUE
3314 rb_reg_s_last_match(int argc, VALUE *argv)
3315 {
3316 VALUE nth;
3317
3318 if (argc > 0 && rb_scan_args(argc, argv, "01", &nth) == 1) {
3319 VALUE match = rb_backref_get();
3320 int n;
3321 if (NIL_P(match)) return Qnil;
3322 n = match_backref_number(match, nth);
3323 return rb_reg_nth_match(n, match);
3324 }
3325 return match_getter();
3326 }
3327
3328 static void
3329 re_warn(const char *s)
3330 {
3331 rb_warn("%s", s);
3332 }
3333
3334 /*
3335 * Document-class: Regexp
3336 *
3337 * A <code>Regexp</code> holds a regular expression, used to match a pattern
3338 * against strings. Regexps are created using the <code>/.../</code> and
3339 * <code>%r{...}</code> literals, and by the <code>Regexp::new</code>
3340 * constructor.
3341 *
3342 */
3343
3344 void
3345 Init_Regexp(void)
3346 {
3347 rb_eRegexpError = rb_define_class("RegexpError", rb_eStandardError);
3348
3349 onigenc_set_default_caseconv_table((UChar*)casetable);
3350 onigenc_set_default_encoding(ONIG_ENCODING_ASCII);
3351 onig_set_warn_func(re_warn);
3352 onig_set_verb_warn_func(re_warn);
3353
3354 rb_define_virtual_variable("$~", match_getter, match_setter);
3355 rb_define_virtual_variable("$&", last_match_getter, 0);
3356 rb_define_virtual_variable("$`", prematch_getter, 0);
3357 rb_define_virtual_variable("$'", postmatch_getter, 0);
3358 rb_define_virtual_variable("$+", last_paren_match_getter, 0);
3359
3360 rb_define_virtual_variable("$=", ignorecase_getter, ignorecase_setter);
3361 rb_define_virtual_variable("$KCODE", kcode_getter, kcode_setter);
3362 rb_define_virtual_variable("$-K", kcode_getter, kcode_setter);
3363
3364 rb_cRegexp = rb_define_class("Regexp", rb_cObject);
3365 rb_define_alloc_func(rb_cRegexp, rb_reg_s_alloc);
3366 rb_define_singleton_method(rb_cRegexp, "compile", rb_class_new_instance, -1);
3367 rb_define_singleton_method(rb_cRegexp, "quote", rb_reg_s_quote, 1);
3368 rb_define_singleton_method(rb_cRegexp, "escape", rb_reg_s_quote, 1);
3369 rb_define_singleton_method(rb_cRegexp, "union", rb_reg_s_union_m, -2);
3370 rb_define_singleton_method(rb_cRegexp, "last_match", rb_reg_s_last_match, -1);
3371 rb_define_singleton_method(rb_cRegexp, "try_convert", rb_reg_s_try_convert, 1);
3372
3373 rb_define_method(rb_cRegexp, "initialize", rb_reg_initialize_m, -1);
3374 rb_define_method(rb_cRegexp, "initialize_copy", rb_reg_init_copy, 1);
3375 rb_define_method(rb_cRegexp, "hash", rb_reg_hash, 0);
3376 rb_define_method(rb_cRegexp, "eql?", rb_reg_equal, 1);
3377 rb_define_method(rb_cRegexp, "==", rb_reg_equal, 1);
3378 rb_define_method(rb_cRegexp, "=~", rb_reg_match, 1);
3379 rb_define_method(rb_cRegexp, "===", rb_reg_eqq, 1);
3380 rb_define_method(rb_cRegexp, "~", rb_reg_match2, 0);
3381 rb_define_method(rb_cRegexp, "match", rb_reg_match_m, -1);
3382 rb_define_method(rb_cRegexp, "to_s", rb_reg_to_s, 0);
3383 rb_define_method(rb_cRegexp, "inspect", rb_reg_inspect, 0);
3384 rb_define_method(rb_cRegexp, "source", rb_reg_source, 0);
3385 rb_define_method(rb_cRegexp, "casefold?", rb_reg_casefold_p, 0);
3386 rb_define_method(rb_cRegexp, "options", rb_reg_options_m, 0);
3387 rb_define_method(rb_cRegexp, "encoding", rb_obj_encoding, 0); /* in encoding.c */
3388 rb_define_method(rb_cRegexp, "fixed_encoding?", rb_reg_fixed_encoding_p, 0);
3389 rb_define_method(rb_cRegexp, "names", rb_reg_names, 0);
3390 rb_define_method(rb_cRegexp, "named_captures", rb_reg_named_captures, 0);
3391
3392 rb_define_const(rb_cRegexp, "IGNORECASE", INT2FIX(ONIG_OPTION_IGNORECASE));
3393 rb_define_const(rb_cRegexp, "EXTENDED", INT2FIX(ONIG_OPTION_EXTEND));
3394 rb_define_const(rb_cRegexp, "MULTILINE", INT2FIX(ONIG_OPTION_MULTILINE));
3395
3396 rb_global_variable(&reg_cache);
3397
3398 rb_cMatch = rb_define_class("MatchData", rb_cObject);
3399 rb_define_alloc_func(rb_cMatch, match_alloc);
3400 rb_undef_method(CLASS_OF(rb_cMatch), "new");
3401
3402 rb_define_method(rb_cMatch, "initialize_copy", match_init_copy, 1);
3403 rb_define_method(rb_cMatch, "regexp", match_regexp, 0);
3404 rb_define_method(rb_cMatch, "names", match_names, 0);
3405 rb_define_method(rb_cMatch, "size", match_size, 0);
3406 rb_define_method(rb_cMatch, "length", match_size, 0);
3407 rb_define_method(rb_cMatch, "offset", match_offset, 1);
3408 rb_define_method(rb_cMatch, "begin", match_begin, 1);
3409 rb_define_method(rb_cMatch, "end", match_end, 1);
3410 rb_define_method(rb_cMatch, "to_a", match_to_a, 0);
3411 rb_define_method(rb_cMatch, "[]", match_aref, -1);
3412 rb_define_method(rb_cMatch, "captures", match_captures, 0);
3413 rb_define_method(rb_cMatch, "values_at", match_values_at, -1);
3414 rb_define_method(rb_cMatch, "pre_match", rb_reg_match_pre, 0);
3415 rb_define_method(rb_cMatch, "post_match", rb_reg_match_post, 0);
3416 rb_define_method(rb_cMatch, "to_s", match_to_s, 0);
3417 rb_define_method(rb_cMatch, "inspect", match_inspect, 0);
3418 rb_define_method(rb_cMatch, "string", match_string, 0);
3419 }
ruby svn mirror