/[pcre]/code/trunk/pcretest.c
ViewVC logotype

Contents of /code/trunk/pcretest.c

Parent Directory Parent Directory | Revision Log Revision Log


Revision 442 - (show annotations)
Fri Sep 11 10:21:02 2009 UTC (5 years, 10 months ago) by ph10
File MIME type: text/plain
File size: 73859 byte(s)
Error occurred while calculating annotation data.
Added PCRE_NOTEMPTY_ATSTART to fix /g bug when \K is present.
1 /*************************************************
2 * PCRE testing program *
3 *************************************************/
4
5 /* This program was hacked up as a tester for PCRE. I really should have
6 written it more tidily in the first place. Will I ever learn? It has grown and
7 been extended and consequently is now rather, er, *very* untidy in places.
8
9 -----------------------------------------------------------------------------
10 Redistribution and use in source and binary forms, with or without
11 modification, are permitted provided that the following conditions are met:
12
13 * Redistributions of source code must retain the above copyright notice,
14 this list of conditions and the following disclaimer.
15
16 * Redistributions in binary form must reproduce the above copyright
17 notice, this list of conditions and the following disclaimer in the
18 documentation and/or other materials provided with the distribution.
19
20 * Neither the name of the University of Cambridge nor the names of its
21 contributors may be used to endorse or promote products derived from
22 this software without specific prior written permission.
23
24 THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
25 AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
26 IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
27 ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE
28 LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
29 CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
30 SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
31 INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
32 CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
33 ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
34 POSSIBILITY OF SUCH DAMAGE.
35 -----------------------------------------------------------------------------
36 */
37
38
39 #ifdef HAVE_CONFIG_H
40 #include "config.h"
41 #endif
42
43 #include <ctype.h>
44 #include <stdio.h>
45 #include <string.h>
46 #include <stdlib.h>
47 #include <time.h>
48 #include <locale.h>
49 #include <errno.h>
50
51 #ifdef SUPPORT_LIBREADLINE
52 #ifdef HAVE_UNISTD_H
53 #include <unistd.h>
54 #endif
55 #include <readline/readline.h>
56 #include <readline/history.h>
57 #endif
58
59
60 /* A number of things vary for Windows builds. Originally, pcretest opened its
61 input and output without "b"; then I was told that "b" was needed in some
62 environments, so it was added for release 5.0 to both the input and output. (It
63 makes no difference on Unix-like systems.) Later I was told that it is wrong
64 for the input on Windows. I've now abstracted the modes into two macros that
65 are set here, to make it easier to fiddle with them, and removed "b" from the
66 input mode under Windows. */
67
68 #if defined(_WIN32) || defined(WIN32)
69 #include <io.h> /* For _setmode() */
70 #include <fcntl.h> /* For _O_BINARY */
71 #define INPUT_MODE "r"
72 #define OUTPUT_MODE "wb"
73
74 #ifndef isatty
75 #define isatty _isatty /* This is what Windows calls them, I'm told, */
76 #endif /* though in some environments they seem to */
77 /* be already defined, hence the #ifndefs. */
78 #ifndef fileno
79 #define fileno _fileno
80 #endif
81
82 #else
83 #include <sys/time.h> /* These two includes are needed */
84 #include <sys/resource.h> /* for setrlimit(). */
85 #define INPUT_MODE "rb"
86 #define OUTPUT_MODE "wb"
87 #endif
88
89
90 /* We have to include pcre_internal.h because we need the internal info for
91 displaying the results of pcre_study() and we also need to know about the
92 internal macros, structures, and other internal data values; pcretest has
93 "inside information" compared to a program that strictly follows the PCRE API.
94
95 Although pcre_internal.h does itself include pcre.h, we explicitly include it
96 here before pcre_internal.h so that the PCRE_EXP_xxx macros get set
97 appropriately for an application, not for building PCRE. */
98
99 #include "pcre.h"
100 #include "pcre_internal.h"
101
102 /* We need access to some of the data tables that PCRE uses. So as not to have
103 to keep two copies, we include the source file here, changing the names of the
104 external symbols to prevent clashes. */
105
106 #define _pcre_ucp_gentype ucp_gentype
107 #define _pcre_utf8_table1 utf8_table1
108 #define _pcre_utf8_table1_size utf8_table1_size
109 #define _pcre_utf8_table2 utf8_table2
110 #define _pcre_utf8_table3 utf8_table3
111 #define _pcre_utf8_table4 utf8_table4
112 #define _pcre_utt utt
113 #define _pcre_utt_size utt_size
114 #define _pcre_utt_names utt_names
115 #define _pcre_OP_lengths OP_lengths
116
117 #include "pcre_tables.c"
118
119 /* We also need the pcre_printint() function for printing out compiled
120 patterns. This function is in a separate file so that it can be included in
121 pcre_compile.c when that module is compiled with debugging enabled.
122
123 The definition of the macro PRINTABLE, which determines whether to print an
124 output character as-is or as a hex value when showing compiled patterns, is
125 contained in this file. We uses it here also, in cases when the locale has not
126 been explicitly changed, so as to get consistent output from systems that
127 differ in their output from isprint() even in the "C" locale. */
128
129 #include "pcre_printint.src"
130
131 #define PRINTHEX(c) (locale_set? isprint(c) : PRINTABLE(c))
132
133
134 /* It is possible to compile this test program without including support for
135 testing the POSIX interface, though this is not available via the standard
136 Makefile. */
137
138 #if !defined NOPOSIX
139 #include "pcreposix.h"
140 #endif
141
142 /* It is also possible, for the benefit of the version currently imported into
143 Exim, to build pcretest without support for UTF8 (define NOUTF8), without the
144 interface to the DFA matcher (NODFA), and without the doublecheck of the old
145 "info" function (define NOINFOCHECK). In fact, we automatically cut out the
146 UTF8 support if PCRE is built without it. */
147
148 #ifndef SUPPORT_UTF8
149 #ifndef NOUTF8
150 #define NOUTF8
151 #endif
152 #endif
153
154
155 /* Other parameters */
156
157 #ifndef CLOCKS_PER_SEC
158 #ifdef CLK_TCK
159 #define CLOCKS_PER_SEC CLK_TCK
160 #else
161 #define CLOCKS_PER_SEC 100
162 #endif
163 #endif
164
165 /* This is the default loop count for timing. */
166
167 #define LOOPREPEAT 500000
168
169 /* Static variables */
170
171 static FILE *outfile;
172 static int log_store = 0;
173 static int callout_count;
174 static int callout_extra;
175 static int callout_fail_count;
176 static int callout_fail_id;
177 static int debug_lengths;
178 static int first_callout;
179 static int locale_set = 0;
180 static int show_malloc;
181 static int use_utf8;
182 static size_t gotten_store;
183
184 /* The buffers grow automatically if very long input lines are encountered. */
185
186 static int buffer_size = 50000;
187 static uschar *buffer = NULL;
188 static uschar *dbuffer = NULL;
189 static uschar *pbuffer = NULL;
190
191
192
193 /*************************************************
194 * Read or extend an input line *
195 *************************************************/
196
197 /* Input lines are read into buffer, but both patterns and data lines can be
198 continued over multiple input lines. In addition, if the buffer fills up, we
199 want to automatically expand it so as to be able to handle extremely large
200 lines that are needed for certain stress tests. When the input buffer is
201 expanded, the other two buffers must also be expanded likewise, and the
202 contents of pbuffer, which are a copy of the input for callouts, must be
203 preserved (for when expansion happens for a data line). This is not the most
204 optimal way of handling this, but hey, this is just a test program!
205
206 Arguments:
207 f the file to read
208 start where in buffer to start (this *must* be within buffer)
209 prompt for stdin or readline()
210
211 Returns: pointer to the start of new data
212 could be a copy of start, or could be moved
213 NULL if no data read and EOF reached
214 */
215
216 static uschar *
217 extend_inputline(FILE *f, uschar *start, const char *prompt)
218 {
219 uschar *here = start;
220
221 for (;;)
222 {
223 int rlen = buffer_size - (here - buffer);
224
225 if (rlen > 1000)
226 {
227 int dlen;
228
229 /* If libreadline support is required, use readline() to read a line if the
230 input is a terminal. Note that readline() removes the trailing newline, so
231 we must put it back again, to be compatible with fgets(). */
232
233 #ifdef SUPPORT_LIBREADLINE
234 if (isatty(fileno(f)))
235 {
236 size_t len;
237 char *s = readline(prompt);
238 if (s == NULL) return (here == start)? NULL : start;
239 len = strlen(s);
240 if (len > 0) add_history(s);
241 if (len > rlen - 1) len = rlen - 1;
242 memcpy(here, s, len);
243 here[len] = '\n';
244 here[len+1] = 0;
245 free(s);
246 }
247 else
248 #endif
249
250 /* Read the next line by normal means, prompting if the file is stdin. */
251
252 {
253 if (f == stdin) printf(prompt);
254 if (fgets((char *)here, rlen, f) == NULL)
255 return (here == start)? NULL : start;
256 }
257
258 dlen = (int)strlen((char *)here);
259 if (dlen > 0 && here[dlen - 1] == '\n') return start;
260 here += dlen;
261 }
262
263 else
264 {
265 int new_buffer_size = 2*buffer_size;
266 uschar *new_buffer = (unsigned char *)malloc(new_buffer_size);
267 uschar *new_dbuffer = (unsigned char *)malloc(new_buffer_size);
268 uschar *new_pbuffer = (unsigned char *)malloc(new_buffer_size);
269
270 if (new_buffer == NULL || new_dbuffer == NULL || new_pbuffer == NULL)
271 {
272 fprintf(stderr, "pcretest: malloc(%d) failed\n", new_buffer_size);
273 exit(1);
274 }
275
276 memcpy(new_buffer, buffer, buffer_size);
277 memcpy(new_pbuffer, pbuffer, buffer_size);
278
279 buffer_size = new_buffer_size;
280
281 start = new_buffer + (start - buffer);
282 here = new_buffer + (here - buffer);
283
284 free(buffer);
285 free(dbuffer);
286 free(pbuffer);
287
288 buffer = new_buffer;
289 dbuffer = new_dbuffer;
290 pbuffer = new_pbuffer;
291 }
292 }
293
294 return NULL; /* Control never gets here */
295 }
296
297
298
299
300
301
302
303 /*************************************************
304 * Read number from string *
305 *************************************************/
306
307 /* We don't use strtoul() because SunOS4 doesn't have it. Rather than mess
308 around with conditional compilation, just do the job by hand. It is only used
309 for unpicking arguments, so just keep it simple.
310
311 Arguments:
312 str string to be converted
313 endptr where to put the end pointer
314
315 Returns: the unsigned long
316 */
317
318 static int
319 get_value(unsigned char *str, unsigned char **endptr)
320 {
321 int result = 0;
322 while(*str != 0 && isspace(*str)) str++;
323 while (isdigit(*str)) result = result * 10 + (int)(*str++ - '0');
324 *endptr = str;
325 return(result);
326 }
327
328
329
330
331 /*************************************************
332 * Convert UTF-8 string to value *
333 *************************************************/
334
335 /* This function takes one or more bytes that represents a UTF-8 character,
336 and returns the value of the character.
337
338 Argument:
339 utf8bytes a pointer to the byte vector
340 vptr a pointer to an int to receive the value
341
342 Returns: > 0 => the number of bytes consumed
343 -6 to 0 => malformed UTF-8 character at offset = (-return)
344 */
345
346 #if !defined NOUTF8
347
348 static int
349 utf82ord(unsigned char *utf8bytes, int *vptr)
350 {
351 int c = *utf8bytes++;
352 int d = c;
353 int i, j, s;
354
355 for (i = -1; i < 6; i++) /* i is number of additional bytes */
356 {
357 if ((d & 0x80) == 0) break;
358 d <<= 1;
359 }
360
361 if (i == -1) { *vptr = c; return 1; } /* ascii character */
362 if (i == 0 || i == 6) return 0; /* invalid UTF-8 */
363
364 /* i now has a value in the range 1-5 */
365
366 s = 6*i;
367 d = (c & utf8_table3[i]) << s;
368
369 for (j = 0; j < i; j++)
370 {
371 c = *utf8bytes++;
372 if ((c & 0xc0) != 0x80) return -(j+1);
373 s -= 6;
374 d |= (c & 0x3f) << s;
375 }
376
377 /* Check that encoding was the correct unique one */
378
379 for (j = 0; j < utf8_table1_size; j++)
380 if (d <= utf8_table1[j]) break;
381 if (j != i) return -(i+1);
382
383 /* Valid value */
384
385 *vptr = d;
386 return i+1;
387 }
388
389 #endif
390
391
392
393 /*************************************************
394 * Convert character value to UTF-8 *
395 *************************************************/
396
397 /* This function takes an integer value in the range 0 - 0x7fffffff
398 and encodes it as a UTF-8 character in 0 to 6 bytes.
399
400 Arguments:
401 cvalue the character value
402 utf8bytes pointer to buffer for result - at least 6 bytes long
403
404 Returns: number of characters placed in the buffer
405 */
406
407 #if !defined NOUTF8
408
409 static int
410 ord2utf8(int cvalue, uschar *utf8bytes)
411 {
412 register int i, j;
413 for (i = 0; i < utf8_table1_size; i++)
414 if (cvalue <= utf8_table1[i]) break;
415 utf8bytes += i;
416 for (j = i; j > 0; j--)
417 {
418 *utf8bytes-- = 0x80 | (cvalue & 0x3f);
419 cvalue >>= 6;
420 }
421 *utf8bytes = utf8_table2[i] | cvalue;
422 return i + 1;
423 }
424
425 #endif
426
427
428
429 /*************************************************
430 * Print character string *
431 *************************************************/
432
433 /* Character string printing function. Must handle UTF-8 strings in utf8
434 mode. Yields number of characters printed. If handed a NULL file, just counts
435 chars without printing. */
436
437 static int pchars(unsigned char *p, int length, FILE *f)
438 {
439 int c = 0;
440 int yield = 0;
441
442 while (length-- > 0)
443 {
444 #if !defined NOUTF8
445 if (use_utf8)
446 {
447 int rc = utf82ord(p, &c);
448
449 if (rc > 0 && rc <= length + 1) /* Mustn't run over the end */
450 {
451 length -= rc - 1;
452 p += rc;
453 if (PRINTHEX(c))
454 {
455 if (f != NULL) fprintf(f, "%c", c);
456 yield++;
457 }
458 else
459 {
460 int n = 4;
461 if (f != NULL) fprintf(f, "\\x{%02x}", c);
462 yield += (n <= 0x000000ff)? 2 :
463 (n <= 0x00000fff)? 3 :
464 (n <= 0x0000ffff)? 4 :
465 (n <= 0x000fffff)? 5 : 6;
466 }
467 continue;
468 }
469 }
470 #endif
471
472 /* Not UTF-8, or malformed UTF-8 */
473
474 c = *p++;
475 if (PRINTHEX(c))
476 {
477 if (f != NULL) fprintf(f, "%c", c);
478 yield++;
479 }
480 else
481 {
482 if (f != NULL) fprintf(f, "\\x%02x", c);
483 yield += 4;
484 }
485 }
486
487 return yield;
488 }
489
490
491
492 /*************************************************
493 * Callout function *
494 *************************************************/
495
496 /* Called from PCRE as a result of the (?C) item. We print out where we are in
497 the match. Yield zero unless more callouts than the fail count, or the callout
498 data is not zero. */
499
500 static int callout(pcre_callout_block *cb)
501 {
502 FILE *f = (first_callout | callout_extra)? outfile : NULL;
503 int i, pre_start, post_start, subject_length;
504
505 if (callout_extra)
506 {
507 fprintf(f, "Callout %d: last capture = %d\n",
508 cb->callout_number, cb->capture_last);
509
510 for (i = 0; i < cb->capture_top * 2; i += 2)
511 {
512 if (cb->offset_vector[i] < 0)
513 fprintf(f, "%2d: <unset>\n", i/2);
514 else
515 {
516 fprintf(f, "%2d: ", i/2);
517 (void)pchars((unsigned char *)cb->subject + cb->offset_vector[i],
518 cb->offset_vector[i+1] - cb->offset_vector[i], f);
519 fprintf(f, "\n");
520 }
521 }
522 }
523
524 /* Re-print the subject in canonical form, the first time or if giving full
525 datails. On subsequent calls in the same match, we use pchars just to find the
526 printed lengths of the substrings. */
527
528 if (f != NULL) fprintf(f, "--->");
529
530 pre_start = pchars((unsigned char *)cb->subject, cb->start_match, f);
531 post_start = pchars((unsigned char *)(cb->subject + cb->start_match),
532 cb->current_position - cb->start_match, f);
533
534 subject_length = pchars((unsigned char *)cb->subject, cb->subject_length, NULL);
535
536 (void)pchars((unsigned char *)(cb->subject + cb->current_position),
537 cb->subject_length - cb->current_position, f);
538
539 if (f != NULL) fprintf(f, "\n");
540
541 /* Always print appropriate indicators, with callout number if not already
542 shown. For automatic callouts, show the pattern offset. */
543
544 if (cb->callout_number == 255)
545 {
546 fprintf(outfile, "%+3d ", cb->pattern_position);
547 if (cb->pattern_position > 99) fprintf(outfile, "\n ");
548 }
549 else
550 {
551 if (callout_extra) fprintf(outfile, " ");
552 else fprintf(outfile, "%3d ", cb->callout_number);
553 }
554
555 for (i = 0; i < pre_start; i++) fprintf(outfile, " ");
556 fprintf(outfile, "^");
557
558 if (post_start > 0)
559 {
560 for (i = 0; i < post_start - 1; i++) fprintf(outfile, " ");
561 fprintf(outfile, "^");
562 }
563
564 for (i = 0; i < subject_length - pre_start - post_start + 4; i++)
565 fprintf(outfile, " ");
566
567 fprintf(outfile, "%.*s", (cb->next_item_length == 0)? 1 : cb->next_item_length,
568 pbuffer + cb->pattern_position);
569
570 fprintf(outfile, "\n");
571 first_callout = 0;
572
573 if (cb->callout_data != NULL)
574 {
575 int callout_data = *((int *)(cb->callout_data));
576 if (callout_data != 0)
577 {
578 fprintf(outfile, "Callout data = %d\n", callout_data);
579 return callout_data;
580 }
581 }
582
583 return (cb->callout_number != callout_fail_id)? 0 :
584 (++callout_count >= callout_fail_count)? 1 : 0;
585 }
586
587
588 /*************************************************
589 * Local malloc functions *
590 *************************************************/
591
592 /* Alternative malloc function, to test functionality and show the size of the
593 compiled re. */
594
595 static void *new_malloc(size_t size)
596 {
597 void *block = malloc(size);
598 gotten_store = size;
599 if (show_malloc)
600 fprintf(outfile, "malloc %3d %p\n", (int)size, block);
601 return block;
602 }
603
604 static void new_free(void *block)
605 {
606 if (show_malloc)
607 fprintf(outfile, "free %p\n", block);
608 free(block);
609 }
610
611
612 /* For recursion malloc/free, to test stacking calls */
613
614 static void *stack_malloc(size_t size)
615 {
616 void *block = malloc(size);
617 if (show_malloc)
618 fprintf(outfile, "stack_malloc %3d %p\n", (int)size, block);
619 return block;
620 }
621
622 static void stack_free(void *block)
623 {
624 if (show_malloc)
625 fprintf(outfile, "stack_free %p\n", block);
626 free(block);
627 }
628
629
630 /*************************************************
631 * Call pcre_fullinfo() *
632 *************************************************/
633
634 /* Get one piece of information from the pcre_fullinfo() function */
635
636 static void new_info(pcre *re, pcre_extra *study, int option, void *ptr)
637 {
638 int rc;
639 if ((rc = pcre_fullinfo(re, study, option, ptr)) < 0)
640 fprintf(outfile, "Error %d from pcre_fullinfo(%d)\n", rc, option);
641 }
642
643
644
645 /*************************************************
646 * Byte flipping function *
647 *************************************************/
648
649 static unsigned long int
650 byteflip(unsigned long int value, int n)
651 {
652 if (n == 2) return ((value & 0x00ff) << 8) | ((value & 0xff00) >> 8);
653 return ((value & 0x000000ff) << 24) |
654 ((value & 0x0000ff00) << 8) |
655 ((value & 0x00ff0000) >> 8) |
656 ((value & 0xff000000) >> 24);
657 }
658
659
660
661
662 /*************************************************
663 * Check match or recursion limit *
664 *************************************************/
665
666 static int
667 check_match_limit(pcre *re, pcre_extra *extra, uschar *bptr, int len,
668 int start_offset, int options, int *use_offsets, int use_size_offsets,
669 int flag, unsigned long int *limit, int errnumber, const char *msg)
670 {
671 int count;
672 int min = 0;
673 int mid = 64;
674 int max = -1;
675
676 extra->flags |= flag;
677
678 for (;;)
679 {
680 *limit = mid;
681
682 count = pcre_exec(re, extra, (char *)bptr, len, start_offset, options,
683 use_offsets, use_size_offsets);
684
685 if (count == errnumber)
686 {
687 /* fprintf(outfile, "Testing %s limit = %d\n", msg, mid); */
688 min = mid;
689 mid = (mid == max - 1)? max : (max > 0)? (min + max)/2 : mid*2;
690 }
691
692 else if (count >= 0 || count == PCRE_ERROR_NOMATCH ||
693 count == PCRE_ERROR_PARTIAL)
694 {
695 if (mid == min + 1)
696 {
697 fprintf(outfile, "Minimum %s limit = %d\n", msg, mid);
698 break;
699 }
700 /* fprintf(outfile, "Testing %s limit = %d\n", msg, mid); */
701 max = mid;
702 mid = (min + mid)/2;
703 }
704 else break; /* Some other error */
705 }
706
707 extra->flags &= ~flag;
708 return count;
709 }
710
711
712
713 /*************************************************
714 * Case-independent strncmp() function *
715 *************************************************/
716
717 /*
718 Arguments:
719 s first string
720 t second string
721 n number of characters to compare
722
723 Returns: < 0, = 0, or > 0, according to the comparison
724 */
725
726 static int
727 strncmpic(uschar *s, uschar *t, int n)
728 {
729 while (n--)
730 {
731 int c = tolower(*s++) - tolower(*t++);
732 if (c) return c;
733 }
734 return 0;
735 }
736
737
738
739 /*************************************************
740 * Check newline indicator *
741 *************************************************/
742
743 /* This is used both at compile and run-time to check for <xxx> escapes, where
744 xxx is LF, CR, CRLF, ANYCRLF, or ANY. Print a message and return 0 if there is
745 no match.
746
747 Arguments:
748 p points after the leading '<'
749 f file for error message
750
751 Returns: appropriate PCRE_NEWLINE_xxx flags, or 0
752 */
753
754 static int
755 check_newline(uschar *p, FILE *f)
756 {
757 if (strncmpic(p, (uschar *)"cr>", 3) == 0) return PCRE_NEWLINE_CR;
758 if (strncmpic(p, (uschar *)"lf>", 3) == 0) return PCRE_NEWLINE_LF;
759 if (strncmpic(p, (uschar *)"crlf>", 5) == 0) return PCRE_NEWLINE_CRLF;
760 if (strncmpic(p, (uschar *)"anycrlf>", 8) == 0) return PCRE_NEWLINE_ANYCRLF;
761 if (strncmpic(p, (uschar *)"any>", 4) == 0) return PCRE_NEWLINE_ANY;
762 if (strncmpic(p, (uschar *)"bsr_anycrlf>", 12) == 0) return PCRE_BSR_ANYCRLF;
763 if (strncmpic(p, (uschar *)"bsr_unicode>", 12) == 0) return PCRE_BSR_UNICODE;
764 fprintf(f, "Unknown newline type at: <%s\n", p);
765 return 0;
766 }
767
768
769
770 /*************************************************
771 * Usage function *
772 *************************************************/
773
774 static void
775 usage(void)
776 {
777 printf("Usage: pcretest [options] [<input file> [<output file>]]\n\n");
778 printf("Input and output default to stdin and stdout.\n");
779 #ifdef SUPPORT_LIBREADLINE
780 printf("If input is a terminal, readline() is used to read from it.\n");
781 #else
782 printf("This version of pcretest is not linked with readline().\n");
783 #endif
784 printf("\nOptions:\n");
785 printf(" -b show compiled code (bytecode)\n");
786 printf(" -C show PCRE compile-time options and exit\n");
787 printf(" -d debug: show compiled code and information (-b and -i)\n");
788 #if !defined NODFA
789 printf(" -dfa force DFA matching for all subjects\n");
790 #endif
791 printf(" -help show usage information\n");
792 printf(" -i show information about compiled patterns\n"
793 " -M find MATCH_LIMIT minimum for each subject\n"
794 " -m output memory used information\n"
795 " -o <n> set size of offsets vector to <n>\n");
796 #if !defined NOPOSIX
797 printf(" -p use POSIX interface\n");
798 #endif
799 printf(" -q quiet: do not output PCRE version number at start\n");
800 printf(" -S <n> set stack size to <n> megabytes\n");
801 printf(" -s output store (memory) used information\n"
802 " -t time compilation and execution\n");
803 printf(" -t <n> time compilation and execution, repeating <n> times\n");
804 printf(" -tm time execution (matching) only\n");
805 printf(" -tm <n> time execution (matching) only, repeating <n> times\n");
806 }
807
808
809
810 /*************************************************
811 * Main Program *
812 *************************************************/
813
814 /* Read lines from named file or stdin and write to named file or stdout; lines
815 consist of a regular expression, in delimiters and optionally followed by
816 options, followed by a set of test data, terminated by an empty line. */
817
818 int main(int argc, char **argv)
819 {
820 FILE *infile = stdin;
821 int options = 0;
822 int study_options = 0;
823 int default_find_match_limit = FALSE;
824 int op = 1;
825 int timeit = 0;
826 int timeitm = 0;
827 int showinfo = 0;
828 int showstore = 0;
829 int quiet = 0;
830 int size_offsets = 45;
831 int size_offsets_max;
832 int *offsets = NULL;
833 #if !defined NOPOSIX
834 int posix = 0;
835 #endif
836 int debug = 0;
837 int done = 0;
838 int all_use_dfa = 0;
839 int yield = 0;
840 int stack_size;
841
842 /* These vectors store, end-to-end, a list of captured substring names. Assume
843 that 1024 is plenty long enough for the few names we'll be testing. */
844
845 uschar copynames[1024];
846 uschar getnames[1024];
847
848 uschar *copynamesptr;
849 uschar *getnamesptr;
850
851 /* Get buffers from malloc() so that Electric Fence will check their misuse
852 when I am debugging. They grow automatically when very long lines are read. */
853
854 buffer = (unsigned char *)malloc(buffer_size);
855 dbuffer = (unsigned char *)malloc(buffer_size);
856 pbuffer = (unsigned char *)malloc(buffer_size);
857
858 /* The outfile variable is static so that new_malloc can use it. */
859
860 outfile = stdout;
861
862 /* The following _setmode() stuff is some Windows magic that tells its runtime
863 library to translate CRLF into a single LF character. At least, that's what
864 I've been told: never having used Windows I take this all on trust. Originally
865 it set 0x8000, but then I was advised that _O_BINARY was better. */
866
867 #if defined(_WIN32) || defined(WIN32)
868 _setmode( _fileno( stdout ), _O_BINARY );
869 #endif
870
871 /* Scan options */
872
873 while (argc > 1 && argv[op][0] == '-')
874 {
875 unsigned char *endptr;
876
877 if (strcmp(argv[op], "-s") == 0 || strcmp(argv[op], "-m") == 0)
878 showstore = 1;
879 else if (strcmp(argv[op], "-q") == 0) quiet = 1;
880 else if (strcmp(argv[op], "-b") == 0) debug = 1;
881 else if (strcmp(argv[op], "-i") == 0) showinfo = 1;
882 else if (strcmp(argv[op], "-d") == 0) showinfo = debug = 1;
883 else if (strcmp(argv[op], "-M") == 0) default_find_match_limit = TRUE;
884 #if !defined NODFA
885 else if (strcmp(argv[op], "-dfa") == 0) all_use_dfa = 1;
886 #endif
887 else if (strcmp(argv[op], "-o") == 0 && argc > 2 &&
888 ((size_offsets = get_value((unsigned char *)argv[op+1], &endptr)),
889 *endptr == 0))
890 {
891 op++;
892 argc--;
893 }
894 else if (strcmp(argv[op], "-t") == 0 || strcmp(argv[op], "-tm") == 0)
895 {
896 int both = argv[op][2] == 0;
897 int temp;
898 if (argc > 2 && (temp = get_value((unsigned char *)argv[op+1], &endptr),
899 *endptr == 0))
900 {
901 timeitm = temp;
902 op++;
903 argc--;
904 }
905 else timeitm = LOOPREPEAT;
906 if (both) timeit = timeitm;
907 }
908 else if (strcmp(argv[op], "-S") == 0 && argc > 2 &&
909 ((stack_size = get_value((unsigned char *)argv[op+1], &endptr)),
910 *endptr == 0))
911 {
912 #if defined(_WIN32) || defined(WIN32)
913 printf("PCRE: -S not supported on this OS\n");
914 exit(1);
915 #else
916 int rc;
917 struct rlimit rlim;
918 getrlimit(RLIMIT_STACK, &rlim);
919 rlim.rlim_cur = stack_size * 1024 * 1024;
920 rc = setrlimit(RLIMIT_STACK, &rlim);
921 if (rc != 0)
922 {
923 printf("PCRE: setrlimit() failed with error %d\n", rc);
924 exit(1);
925 }
926 op++;
927 argc--;
928 #endif
929 }
930 #if !defined NOPOSIX
931 else if (strcmp(argv[op], "-p") == 0) posix = 1;
932 #endif
933 else if (strcmp(argv[op], "-C") == 0)
934 {
935 int rc;
936 unsigned long int lrc;
937 printf("PCRE version %s\n", pcre_version());
938 printf("Compiled with\n");
939 (void)pcre_config(PCRE_CONFIG_UTF8, &rc);
940 printf(" %sUTF-8 support\n", rc? "" : "No ");
941 (void)pcre_config(PCRE_CONFIG_UNICODE_PROPERTIES, &rc);
942 printf(" %sUnicode properties support\n", rc? "" : "No ");
943 (void)pcre_config(PCRE_CONFIG_NEWLINE, &rc);
944 /* Note that these values are always the ASCII values, even
945 in EBCDIC environments. CR is 13 and NL is 10. */
946 printf(" Newline sequence is %s\n", (rc == 13)? "CR" :
947 (rc == 10)? "LF" : (rc == (13<<8 | 10))? "CRLF" :
948 (rc == -2)? "ANYCRLF" :
949 (rc == -1)? "ANY" : "???");
950 (void)pcre_config(PCRE_CONFIG_BSR, &rc);
951 printf(" \\R matches %s\n", rc? "CR, LF, or CRLF only" :
952 "all Unicode newlines");
953 (void)pcre_config(PCRE_CONFIG_LINK_SIZE, &rc);
954 printf(" Internal link size = %d\n", rc);
955 (void)pcre_config(PCRE_CONFIG_POSIX_MALLOC_THRESHOLD, &rc);
956 printf(" POSIX malloc threshold = %d\n", rc);
957 (void)pcre_config(PCRE_CONFIG_MATCH_LIMIT, &lrc);
958 printf(" Default match limit = %ld\n", lrc);
959 (void)pcre_config(PCRE_CONFIG_MATCH_LIMIT_RECURSION, &lrc);
960 printf(" Default recursion depth limit = %ld\n", lrc);
961 (void)pcre_config(PCRE_CONFIG_STACKRECURSE, &rc);
962 printf(" Match recursion uses %s\n", rc? "stack" : "heap");
963 goto EXIT;
964 }
965 else if (strcmp(argv[op], "-help") == 0 ||
966 strcmp(argv[op], "--help") == 0)
967 {
968 usage();
969 goto EXIT;
970 }
971 else
972 {
973 printf("** Unknown or malformed option %s\n", argv[op]);
974 usage();
975 yield = 1;
976 goto EXIT;
977 }
978 op++;
979 argc--;
980 }
981
982 /* Get the store for the offsets vector, and remember what it was */
983
984 size_offsets_max = size_offsets;
985 offsets = (int *)malloc(size_offsets_max * sizeof(int));
986 if (offsets == NULL)
987 {
988 printf("** Failed to get %d bytes of memory for offsets vector\n",
989 (int)(size_offsets_max * sizeof(int)));
990 yield = 1;
991 goto EXIT;
992 }
993
994 /* Sort out the input and output files */
995
996 if (argc > 1)
997 {
998 infile = fopen(argv[op], INPUT_MODE);
999 if (infile == NULL)
1000 {
1001 printf("** Failed to open %s\n", argv[op]);
1002 yield = 1;
1003 goto EXIT;
1004 }
1005 }
1006
1007 if (argc > 2)
1008 {
1009 outfile = fopen(argv[op+1], OUTPUT_MODE);
1010 if (outfile == NULL)
1011 {
1012 printf("** Failed to open %s\n", argv[op+1]);
1013 yield = 1;
1014 goto EXIT;
1015 }
1016 }
1017
1018 /* Set alternative malloc function */
1019
1020 pcre_malloc = new_malloc;
1021 pcre_free = new_free;
1022 pcre_stack_malloc = stack_malloc;
1023 pcre_stack_free = stack_free;
1024
1025 /* Heading line unless quiet, then prompt for first regex if stdin */
1026
1027 if (!quiet) fprintf(outfile, "PCRE version %s\n\n", pcre_version());
1028
1029 /* Main loop */
1030
1031 while (!done)
1032 {
1033 pcre *re = NULL;
1034 pcre_extra *extra = NULL;
1035
1036 #if !defined NOPOSIX /* There are still compilers that require no indent */
1037 regex_t preg;
1038 int do_posix = 0;
1039 #endif
1040
1041 const char *error;
1042 unsigned char *p, *pp, *ppp;
1043 unsigned char *to_file = NULL;
1044 const unsigned char *tables = NULL;
1045 unsigned long int true_size, true_study_size = 0;
1046 size_t size, regex_gotten_store;
1047 int do_study = 0;
1048 int do_debug = debug;
1049 int do_G = 0;
1050 int do_g = 0;
1051 int do_showinfo = showinfo;
1052 int do_showrest = 0;
1053 int do_flip = 0;
1054 int erroroffset, len, delimiter, poffset;
1055
1056 use_utf8 = 0;
1057 debug_lengths = 1;
1058
1059 if (extend_inputline(infile, buffer, " re> ") == NULL) break;
1060 if (infile != stdin) fprintf(outfile, "%s", (char *)buffer);
1061 fflush(outfile);
1062
1063 p = buffer;
1064 while (isspace(*p)) p++;
1065 if (*p == 0) continue;
1066
1067 /* See if the pattern is to be loaded pre-compiled from a file. */
1068
1069 if (*p == '<' && strchr((char *)(p+1), '<') == NULL)
1070 {
1071 unsigned long int magic, get_options;
1072 uschar sbuf[8];
1073 FILE *f;
1074
1075 p++;
1076 pp = p + (int)strlen((char *)p);
1077 while (isspace(pp[-1])) pp--;
1078 *pp = 0;
1079
1080 f = fopen((char *)p, "rb");
1081 if (f == NULL)
1082 {
1083 fprintf(outfile, "Failed to open %s: %s\n", p, strerror(errno));
1084 continue;
1085 }
1086
1087 if (fread(sbuf, 1, 8, f) != 8) goto FAIL_READ;
1088
1089 true_size =
1090 (sbuf[0] << 24) | (sbuf[1] << 16) | (sbuf[2] << 8) | sbuf[3];
1091 true_study_size =
1092 (sbuf[4] << 24) | (sbuf[5] << 16) | (sbuf[6] << 8) | sbuf[7];
1093
1094 re = (real_pcre *)new_malloc(true_size);
1095 regex_gotten_store = gotten_store;
1096
1097 if (fread(re, 1, true_size, f) != true_size) goto FAIL_READ;
1098
1099 magic = ((real_pcre *)re)->magic_number;
1100 if (magic != MAGIC_NUMBER)
1101 {
1102 if (byteflip(magic, sizeof(magic)) == MAGIC_NUMBER)
1103 {
1104 do_flip = 1;
1105 }
1106 else
1107 {
1108 fprintf(outfile, "Data in %s is not a compiled PCRE regex\n", p);
1109 fclose(f);
1110 continue;
1111 }
1112 }
1113
1114 fprintf(outfile, "Compiled regex%s loaded from %s\n",
1115 do_flip? " (byte-inverted)" : "", p);
1116
1117 /* Need to know if UTF-8 for printing data strings */
1118
1119 new_info(re, NULL, PCRE_INFO_OPTIONS, &get_options);
1120 use_utf8 = (get_options & PCRE_UTF8) != 0;
1121
1122 /* Now see if there is any following study data */
1123
1124 if (true_study_size != 0)
1125 {
1126 pcre_study_data *psd;
1127
1128 extra = (pcre_extra *)new_malloc(sizeof(pcre_extra) + true_study_size);
1129 extra->flags = PCRE_EXTRA_STUDY_DATA;
1130
1131 psd = (pcre_study_data *)(((char *)extra) + sizeof(pcre_extra));
1132 extra->study_data = psd;
1133
1134 if (fread(psd, 1, true_study_size, f) != true_study_size)
1135 {
1136 FAIL_READ:
1137 fprintf(outfile, "Failed to read data from %s\n", p);
1138 if (extra != NULL) new_free(extra);
1139 if (re != NULL) new_free(re);
1140 fclose(f);
1141 continue;
1142 }
1143 fprintf(outfile, "Study data loaded from %s\n", p);
1144 do_study = 1; /* To get the data output if requested */
1145 }
1146 else fprintf(outfile, "No study data\n");
1147
1148 fclose(f);
1149 goto SHOW_INFO;
1150 }
1151
1152 /* In-line pattern (the usual case). Get the delimiter and seek the end of
1153 the pattern; if is isn't complete, read more. */
1154
1155 delimiter = *p++;
1156
1157 if (isalnum(delimiter) || delimiter == '\\')
1158 {
1159 fprintf(outfile, "** Delimiter must not be alphanumeric or \\\n");
1160 goto SKIP_DATA;
1161 }
1162
1163 pp = p;
1164 poffset = p - buffer;
1165
1166 for(;;)
1167 {
1168 while (*pp != 0)
1169 {
1170 if (*pp == '\\' && pp[1] != 0) pp++;
1171 else if (*pp == delimiter) break;
1172 pp++;
1173 }
1174 if (*pp != 0) break;
1175 if ((pp = extend_inputline(infile, pp, " > ")) == NULL)
1176 {
1177 fprintf(outfile, "** Unexpected EOF\n");
1178 done = 1;
1179 goto CONTINUE;
1180 }
1181 if (infile != stdin) fprintf(outfile, "%s", (char *)pp);
1182 }
1183
1184 /* The buffer may have moved while being extended; reset the start of data
1185 pointer to the correct relative point in the buffer. */
1186
1187 p = buffer + poffset;
1188
1189 /* If the first character after the delimiter is backslash, make
1190 the pattern end with backslash. This is purely to provide a way
1191 of testing for the error message when a pattern ends with backslash. */
1192
1193 if (pp[1] == '\\') *pp++ = '\\';
1194
1195 /* Terminate the pattern at the delimiter, and save a copy of the pattern
1196 for callouts. */
1197
1198 *pp++ = 0;
1199 strcpy((char *)pbuffer, (char *)p);
1200
1201 /* Look for options after final delimiter */
1202
1203 options = 0;
1204 study_options = 0;
1205 log_store = showstore; /* default from command line */
1206
1207 while (*pp != 0)
1208 {
1209 switch (*pp++)
1210 {
1211 case 'f': options |= PCRE_FIRSTLINE; break;
1212 case 'g': do_g = 1; break;
1213 case 'i': options |= PCRE_CASELESS; break;
1214 case 'm': options |= PCRE_MULTILINE; break;
1215 case 's': options |= PCRE_DOTALL; break;
1216 case 'x': options |= PCRE_EXTENDED; break;
1217
1218 case '+': do_showrest = 1; break;
1219 case 'A': options |= PCRE_ANCHORED; break;
1220 case 'B': do_debug = 1; break;
1221 case 'C': options |= PCRE_AUTO_CALLOUT; break;
1222 case 'D': do_debug = do_showinfo = 1; break;
1223 case 'E': options |= PCRE_DOLLAR_ENDONLY; break;
1224 case 'F': do_flip = 1; break;
1225 case 'G': do_G = 1; break;
1226 case 'I': do_showinfo = 1; break;
1227 case 'J': options |= PCRE_DUPNAMES; break;
1228 case 'M': log_store = 1; break;
1229 case 'N': options |= PCRE_NO_AUTO_CAPTURE; break;
1230
1231 #if !defined NOPOSIX
1232 case 'P': do_posix = 1; break;
1233 #endif
1234
1235 case 'S': do_study = 1; break;
1236 case 'U': options |= PCRE_UNGREEDY; break;
1237 case 'X': options |= PCRE_EXTRA; break;
1238 case 'Z': debug_lengths = 0; break;
1239 case '8': options |= PCRE_UTF8; use_utf8 = 1; break;
1240 case '?': options |= PCRE_NO_UTF8_CHECK; break;
1241
1242 case 'L':
1243 ppp = pp;
1244 /* The '\r' test here is so that it works on Windows. */
1245 /* The '0' test is just in case this is an unterminated line. */
1246 while (*ppp != 0 && *ppp != '\n' && *ppp != '\r' && *ppp != ' ') ppp++;
1247 *ppp = 0;
1248 if (setlocale(LC_CTYPE, (const char *)pp) == NULL)
1249 {
1250 fprintf(outfile, "** Failed to set locale \"%s\"\n", pp);
1251 goto SKIP_DATA;
1252 }
1253 locale_set = 1;
1254 tables = pcre_maketables();
1255 pp = ppp;
1256 break;
1257
1258 case '>':
1259 to_file = pp;
1260 while (*pp != 0) pp++;
1261 while (isspace(pp[-1])) pp--;
1262 *pp = 0;
1263 break;
1264
1265 case '<':
1266 {
1267 if (strncmp((char *)pp, "JS>", 3) == 0)
1268 {
1269 options |= PCRE_JAVASCRIPT_COMPAT;
1270 pp += 3;
1271 }
1272 else
1273 {
1274 int x = check_newline(pp, outfile);
1275 if (x == 0) goto SKIP_DATA;
1276 options |= x;
1277 while (*pp++ != '>');
1278 }
1279 }
1280 break;
1281
1282 case '\r': /* So that it works in Windows */
1283 case '\n':
1284 case ' ':
1285 break;
1286
1287 default:
1288 fprintf(outfile, "** Unknown option '%c'\n", pp[-1]);
1289 goto SKIP_DATA;
1290 }
1291 }
1292
1293 /* Handle compiling via the POSIX interface, which doesn't support the
1294 timing, showing, or debugging options, nor the ability to pass over
1295 local character tables. */
1296
1297 #if !defined NOPOSIX
1298 if (posix || do_posix)
1299 {
1300 int rc;
1301 int cflags = 0;
1302
1303 if ((options & PCRE_CASELESS) != 0) cflags |= REG_ICASE;
1304 if ((options & PCRE_MULTILINE) != 0) cflags |= REG_NEWLINE;
1305 if ((options & PCRE_DOTALL) != 0) cflags |= REG_DOTALL;
1306 if ((options & PCRE_NO_AUTO_CAPTURE) != 0) cflags |= REG_NOSUB;
1307 if ((options & PCRE_UTF8) != 0) cflags |= REG_UTF8;
1308 if ((options & PCRE_UNGREEDY) != 0) cflags |= REG_UNGREEDY;
1309
1310 rc = regcomp(&preg, (char *)p, cflags);
1311
1312 /* Compilation failed; go back for another re, skipping to blank line
1313 if non-interactive. */
1314
1315 if (rc != 0)
1316 {
1317 (void)regerror(rc, &preg, (char *)buffer, buffer_size);
1318 fprintf(outfile, "Failed: POSIX code %d: %s\n", rc, buffer);
1319 goto SKIP_DATA;
1320 }
1321 }
1322
1323 /* Handle compiling via the native interface */
1324
1325 else
1326 #endif /* !defined NOPOSIX */
1327
1328 {
1329 unsigned long int get_options;
1330
1331 if (timeit > 0)
1332 {
1333 register int i;
1334 clock_t time_taken;
1335 clock_t start_time = clock();
1336 for (i = 0; i < timeit; i++)
1337 {
1338 re = pcre_compile((char *)p, options, &error, &erroroffset, tables);
1339 if (re != NULL) free(re);
1340 }
1341 time_taken = clock() - start_time;
1342 fprintf(outfile, "Compile time %.4f milliseconds\n",
1343 (((double)time_taken * 1000.0) / (double)timeit) /
1344 (double)CLOCKS_PER_SEC);
1345 }
1346
1347 re = pcre_compile((char *)p, options, &error, &erroroffset, tables);
1348
1349 /* Compilation failed; go back for another re, skipping to blank line
1350 if non-interactive. */
1351
1352 if (re == NULL)
1353 {
1354 fprintf(outfile, "Failed: %s at offset %d\n", error, erroroffset);
1355 SKIP_DATA:
1356 if (infile != stdin)
1357 {
1358 for (;;)
1359 {
1360 if (extend_inputline(infile, buffer, NULL) == NULL)
1361 {
1362 done = 1;
1363 goto CONTINUE;
1364 }
1365 len = (int)strlen((char *)buffer);
1366 while (len > 0 && isspace(buffer[len-1])) len--;
1367 if (len == 0) break;
1368 }
1369 fprintf(outfile, "\n");
1370 }
1371 goto CONTINUE;
1372 }
1373
1374 /* Compilation succeeded. It is now possible to set the UTF-8 option from
1375 within the regex; check for this so that we know how to process the data
1376 lines. */
1377
1378 new_info(re, NULL, PCRE_INFO_OPTIONS, &get_options);
1379 if ((get_options & PCRE_UTF8) != 0) use_utf8 = 1;
1380
1381 /* Print information if required. There are now two info-returning
1382 functions. The old one has a limited interface and returns only limited
1383 data. Check that it agrees with the newer one. */
1384
1385 if (log_store)
1386 fprintf(outfile, "Memory allocation (code space): %d\n",
1387 (int)(gotten_store -
1388 sizeof(real_pcre) -
1389 ((real_pcre *)re)->name_count * ((real_pcre *)re)->name_entry_size));
1390
1391 /* Extract the size for possible writing before possibly flipping it,
1392 and remember the store that was got. */
1393
1394 true_size = ((real_pcre *)re)->size;
1395 regex_gotten_store = gotten_store;
1396
1397 /* If /S was present, study the regexp to generate additional info to
1398 help with the matching. */
1399
1400 if (do_study)
1401 {
1402 if (timeit > 0)
1403 {
1404 register int i;
1405 clock_t time_taken;
1406 clock_t start_time = clock();
1407 for (i = 0; i < timeit; i++)
1408 extra = pcre_study(re, study_options, &error);
1409 time_taken = clock() - start_time;
1410 if (extra != NULL) free(extra);
1411 fprintf(outfile, " Study time %.4f milliseconds\n",
1412 (((double)time_taken * 1000.0) / (double)timeit) /
1413 (double)CLOCKS_PER_SEC);
1414 }
1415 extra = pcre_study(re, study_options, &error);
1416 if (error != NULL)
1417 fprintf(outfile, "Failed to study: %s\n", error);
1418 else if (extra != NULL)
1419 true_study_size = ((pcre_study_data *)(extra->study_data))->size;
1420 }
1421
1422 /* If the 'F' option was present, we flip the bytes of all the integer
1423 fields in the regex data block and the study block. This is to make it
1424 possible to test PCRE's handling of byte-flipped patterns, e.g. those
1425 compiled on a different architecture. */
1426
1427 if (do_flip)
1428 {
1429 real_pcre *rre = (real_pcre *)re;
1430 rre->magic_number =
1431 byteflip(rre->magic_number, sizeof(rre->magic_number));
1432 rre->size = byteflip(rre->size, sizeof(rre->size));
1433 rre->options = byteflip(rre->options, sizeof(rre->options));
1434 rre->flags = (pcre_uint16)byteflip(rre->flags, sizeof(rre->flags));
1435 rre->top_bracket =
1436 (pcre_uint16)byteflip(rre->top_bracket, sizeof(rre->top_bracket));
1437 rre->top_backref =
1438 (pcre_uint16)byteflip(rre->top_backref, sizeof(rre->top_backref));
1439 rre->first_byte =
1440 (pcre_uint16)byteflip(rre->first_byte, sizeof(rre->first_byte));
1441 rre->req_byte =
1442 (pcre_uint16)byteflip(rre->req_byte, sizeof(rre->req_byte));
1443 rre->name_table_offset = (pcre_uint16)byteflip(rre->name_table_offset,
1444 sizeof(rre->name_table_offset));
1445 rre->name_entry_size = (pcre_uint16)byteflip(rre->name_entry_size,
1446 sizeof(rre->name_entry_size));
1447 rre->name_count = (pcre_uint16)byteflip(rre->name_count,
1448 sizeof(rre->name_count));
1449
1450 if (extra != NULL)
1451 {
1452 pcre_study_data *rsd = (pcre_study_data *)(extra->study_data);
1453 rsd->size = byteflip(rsd->size, sizeof(rsd->size));
1454 rsd->options = byteflip(rsd->options, sizeof(rsd->options));
1455 }
1456 }
1457
1458 /* Extract information from the compiled data if required */
1459
1460 SHOW_INFO:
1461
1462 if (do_debug)
1463 {
1464 fprintf(outfile, "------------------------------------------------------------------\n");
1465 pcre_printint(re, outfile, debug_lengths);
1466 }
1467
1468 /* We already have the options in get_options (see above) */
1469
1470 if (do_showinfo)
1471 {
1472 unsigned long int all_options;
1473 #if !defined NOINFOCHECK
1474 int old_first_char, old_options, old_count;
1475 #endif
1476 int count, backrefmax, first_char, need_char, okpartial, jchanged,
1477 hascrorlf;
1478 int nameentrysize, namecount;
1479 const uschar *nametable;
1480
1481 new_info(re, NULL, PCRE_INFO_SIZE, &size);
1482 new_info(re, NULL, PCRE_INFO_CAPTURECOUNT, &count);
1483 new_info(re, NULL, PCRE_INFO_BACKREFMAX, &backrefmax);
1484 new_info(re, NULL, PCRE_INFO_FIRSTBYTE, &first_char);
1485 new_info(re, NULL, PCRE_INFO_LASTLITERAL, &need_char);
1486 new_info(re, NULL, PCRE_INFO_NAMEENTRYSIZE, &nameentrysize);
1487 new_info(re, NULL, PCRE_INFO_NAMECOUNT, &namecount);
1488 new_info(re, NULL, PCRE_INFO_NAMETABLE, (void *)&nametable);
1489 new_info(re, NULL, PCRE_INFO_OKPARTIAL, &okpartial);
1490 new_info(re, NULL, PCRE_INFO_JCHANGED, &jchanged);
1491 new_info(re, NULL, PCRE_INFO_HASCRORLF, &hascrorlf);
1492
1493 #if !defined NOINFOCHECK
1494 old_count = pcre_info(re, &old_options, &old_first_char);
1495 if (count < 0) fprintf(outfile,
1496 "Error %d from pcre_info()\n", count);
1497 else
1498 {
1499 if (old_count != count) fprintf(outfile,
1500 "Count disagreement: pcre_fullinfo=%d pcre_info=%d\n", count,
1501 old_count);
1502
1503 if (old_first_char != first_char) fprintf(outfile,
1504 "First char disagreement: pcre_fullinfo=%d pcre_info=%d\n",
1505 first_char, old_first_char);
1506
1507 if (old_options != (int)get_options) fprintf(outfile,
1508 "Options disagreement: pcre_fullinfo=%ld pcre_info=%d\n",
1509 get_options, old_options);
1510 }
1511 #endif
1512
1513 if (size != regex_gotten_store) fprintf(outfile,
1514 "Size disagreement: pcre_fullinfo=%d call to malloc for %d\n",
1515 (int)size, (int)regex_gotten_store);
1516
1517 fprintf(outfile, "Capturing subpattern count = %d\n", count);
1518 if (backrefmax > 0)
1519 fprintf(outfile, "Max back reference = %d\n", backrefmax);
1520
1521 if (namecount > 0)
1522 {
1523 fprintf(outfile, "Named capturing subpatterns:\n");
1524 while (namecount-- > 0)
1525 {
1526 fprintf(outfile, " %s %*s%3d\n", nametable + 2,
1527 nameentrysize - 3 - (int)strlen((char *)nametable + 2), "",
1528 GET2(nametable, 0));
1529 nametable += nameentrysize;
1530 }
1531 }
1532
1533 if (!okpartial) fprintf(outfile, "Partial matching not supported\n");
1534 if (hascrorlf) fprintf(outfile, "Contains explicit CR or LF match\n");
1535
1536 all_options = ((real_pcre *)re)->options;
1537 if (do_flip) all_options = byteflip(all_options, sizeof(all_options));
1538
1539 if (get_options == 0) fprintf(outfile, "No options\n");
1540 else fprintf(outfile, "Options:%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s\n",
1541 ((get_options & PCRE_ANCHORED) != 0)? " anchored" : "",
1542 ((get_options & PCRE_CASELESS) != 0)? " caseless" : "",
1543 ((get_options & PCRE_EXTENDED) != 0)? " extended" : "",
1544 ((get_options & PCRE_MULTILINE) != 0)? " multiline" : "",
1545 ((get_options & PCRE_FIRSTLINE) != 0)? " firstline" : "",
1546 ((get_options & PCRE_DOTALL) != 0)? " dotall" : "",
1547 ((get_options & PCRE_BSR_ANYCRLF) != 0)? " bsr_anycrlf" : "",
1548 ((get_options & PCRE_BSR_UNICODE) != 0)? " bsr_unicode" : "",
1549 ((get_options & PCRE_DOLLAR_ENDONLY) != 0)? " dollar_endonly" : "",
1550 ((get_options & PCRE_EXTRA) != 0)? " extra" : "",
1551 ((get_options & PCRE_UNGREEDY) != 0)? " ungreedy" : "",
1552 ((get_options & PCRE_NO_AUTO_CAPTURE) != 0)? " no_auto_capture" : "",
1553 ((get_options & PCRE_UTF8) != 0)? " utf8" : "",
1554 ((get_options & PCRE_NO_UTF8_CHECK) != 0)? " no_utf8_check" : "",
1555 ((get_options & PCRE_DUPNAMES) != 0)? " dupnames" : "");
1556
1557 if (jchanged) fprintf(outfile, "Duplicate name status changes\n");
1558
1559 switch (get_options & PCRE_NEWLINE_BITS)
1560 {
1561 case PCRE_NEWLINE_CR:
1562 fprintf(outfile, "Forced newline sequence: CR\n");
1563 break;
1564
1565 case PCRE_NEWLINE_LF:
1566 fprintf(outfile, "Forced newline sequence: LF\n");
1567 break;
1568
1569 case PCRE_NEWLINE_CRLF:
1570 fprintf(outfile, "Forced newline sequence: CRLF\n");
1571 break;
1572
1573 case PCRE_NEWLINE_ANYCRLF:
1574 fprintf(outfile, "Forced newline sequence: ANYCRLF\n");
1575 break;
1576
1577 case PCRE_NEWLINE_ANY:
1578 fprintf(outfile, "Forced newline sequence: ANY\n");
1579 break;
1580
1581 default:
1582 break;
1583 }
1584
1585 if (first_char == -1)
1586 {
1587 fprintf(outfile, "First char at start or follows newline\n");
1588 }
1589 else if (first_char < 0)
1590 {
1591 fprintf(outfile, "No first char\n");
1592 }
1593 else
1594 {
1595 int ch = first_char & 255;
1596 const char *caseless = ((first_char & REQ_CASELESS) == 0)?
1597 "" : " (caseless)";
1598 if (PRINTHEX(ch))
1599 fprintf(outfile, "First char = \'%c\'%s\n", ch, caseless);
1600 else
1601 fprintf(outfile, "First char = %d%s\n", ch, caseless);
1602 }
1603
1604 if (need_char < 0)
1605 {
1606 fprintf(outfile, "No need char\n");
1607 }
1608 else
1609 {
1610 int ch = need_char & 255;
1611 const char *caseless = ((need_char & REQ_CASELESS) == 0)?
1612 "" : " (caseless)";
1613 if (PRINTHEX(ch))
1614 fprintf(outfile, "Need char = \'%c\'%s\n", ch, caseless);
1615 else
1616 fprintf(outfile, "Need char = %d%s\n", ch, caseless);
1617 }
1618
1619 /* Don't output study size; at present it is in any case a fixed
1620 value, but it varies, depending on the computer architecture, and
1621 so messes up the test suite. (And with the /F option, it might be
1622 flipped.) */
1623
1624 if (do_study)
1625 {
1626 if (extra == NULL)
1627 fprintf(outfile, "Study returned NULL\n");
1628 else
1629 {
1630 uschar *start_bits = NULL;
1631 new_info(re, extra, PCRE_INFO_FIRSTTABLE, &start_bits);
1632
1633 if (start_bits == NULL)
1634 fprintf(outfile, "No starting byte set\n");
1635 else
1636 {
1637 int i;
1638 int c = 24;
1639 fprintf(outfile, "Starting byte set: ");
1640 for (i = 0; i < 256; i++)
1641 {
1642 if ((start_bits[i/8] & (1<<(i&7))) != 0)
1643 {
1644 if (c > 75)
1645 {
1646 fprintf(outfile, "\n ");
1647 c = 2;
1648 }
1649 if (PRINTHEX(i) && i != ' ')
1650 {
1651 fprintf(outfile, "%c ", i);
1652 c += 2;
1653 }
1654 else
1655 {
1656 fprintf(outfile, "\\x%02x ", i);
1657 c += 5;
1658 }
1659 }
1660 }
1661 fprintf(outfile, "\n");
1662 }
1663 }
1664 }
1665 }
1666
1667 /* If the '>' option was present, we write out the regex to a file, and
1668 that is all. The first 8 bytes of the file are the regex length and then
1669 the study length, in big-endian order. */
1670
1671 if (to_file != NULL)
1672 {
1673 FILE *f = fopen((char *)to_file, "wb");
1674 if (f == NULL)
1675 {
1676 fprintf(outfile, "Unable to open %s: %s\n", to_file, strerror(errno));
1677 }
1678 else
1679 {
1680 uschar sbuf[8];
1681 sbuf[0] = (uschar)((true_size >> 24) & 255);
1682 sbuf[1] = (uschar)((true_size >> 16) & 255);
1683 sbuf[2] = (uschar)((true_size >> 8) & 255);
1684 sbuf[3] = (uschar)((true_size) & 255);
1685
1686 sbuf[4] = (uschar)((true_study_size >> 24) & 255);
1687 sbuf[5] = (uschar)((true_study_size >> 16) & 255);
1688 sbuf[6] = (uschar)((true_study_size >> 8) & 255);
1689 sbuf[7] = (uschar)((true_study_size) & 255);
1690
1691 if (fwrite(sbuf, 1, 8, f) < 8 ||
1692 fwrite(re, 1, true_size, f) < true_size)
1693 {
1694 fprintf(outfile, "Write error on %s: %s\n", to_file, strerror(errno));
1695 }
1696 else
1697 {
1698 fprintf(outfile, "Compiled regex written to %s\n", to_file);
1699 if (extra != NULL)
1700 {
1701 if (fwrite(extra->study_data, 1, true_study_size, f) <
1702 true_study_size)
1703 {
1704 fprintf(outfile, "Write error on %s: %s\n", to_file,
1705 strerror(errno));
1706 }
1707 else fprintf(outfile, "Study data written to %s\n", to_file);
1708
1709 }
1710 }
1711 fclose(f);
1712 }
1713
1714 new_free(re);
1715 if (extra != NULL) new_free(extra);
1716 if (tables != NULL) new_free((void *)tables);
1717 continue; /* With next regex */
1718 }
1719 } /* End of non-POSIX compile */
1720
1721 /* Read data lines and test them */
1722
1723 for (;;)
1724 {
1725 uschar *q;
1726 uschar *bptr;
1727 int *use_offsets = offsets;
1728 int use_size_offsets = size_offsets;
1729 int callout_data = 0;
1730 int callout_data_set = 0;
1731 int count, c;
1732 int copystrings = 0;
1733 int find_match_limit = default_find_match_limit;
1734 int getstrings = 0;
1735 int getlist = 0;
1736 int gmatched = 0;
1737 int start_offset = 0;
1738 int g_notempty = 0;
1739 int use_dfa = 0;
1740
1741 options = 0;
1742
1743 *copynames = 0;
1744 *getnames = 0;
1745
1746 copynamesptr = copynames;
1747 getnamesptr = getnames;
1748
1749 pcre_callout = callout;
1750 first_callout = 1;
1751 callout_extra = 0;
1752 callout_count = 0;
1753 callout_fail_count = 999999;
1754 callout_fail_id = -1;
1755 show_malloc = 0;
1756
1757 if (extra != NULL) extra->flags &=
1758 ~(PCRE_EXTRA_MATCH_LIMIT|PCRE_EXTRA_MATCH_LIMIT_RECURSION);
1759
1760 len = 0;
1761 for (;;)
1762 {
1763 if (extend_inputline(infile, buffer + len, "data> ") == NULL)
1764 {
1765 if (len > 0) break;
1766 done = 1;
1767 goto CONTINUE;
1768 }
1769 if (infile != stdin) fprintf(outfile, "%s", (char *)buffer);
1770 len = (int)strlen((char *)buffer);
1771 if (buffer[len-1] == '\n') break;
1772 }
1773
1774 while (len > 0 && isspace(buffer[len-1])) len--;
1775 buffer[len] = 0;
1776 if (len == 0) break;
1777
1778 p = buffer;
1779 while (isspace(*p)) p++;
1780
1781 bptr = q = dbuffer;
1782 while ((c = *p++) != 0)
1783 {
1784 int i = 0;
1785 int n = 0;
1786
1787 if (c == '\\') switch ((c = *p++))
1788 {
1789 case 'a': c = 7; break;
1790 case 'b': c = '\b'; break;
1791 case 'e': c = 27; break;
1792 case 'f': c = '\f'; break;
1793 case 'n': c = '\n'; break;
1794 case 'r': c = '\r'; break;
1795 case 't': c = '\t'; break;
1796 case 'v': c = '\v'; break;
1797
1798 case '0': case '1': case '2': case '3':
1799 case '4': case '5': case '6': case '7':
1800 c -= '0';
1801 while (i++ < 2 && isdigit(*p) && *p != '8' && *p != '9')
1802 c = c * 8 + *p++ - '0';
1803
1804 #if !defined NOUTF8
1805 if (use_utf8 && c > 255)
1806 {
1807 unsigned char buff8[8];
1808 int ii, utn;
1809 utn = ord2utf8(c, buff8);
1810 for (ii = 0; ii < utn - 1; ii++) *q++ = buff8[ii];
1811 c = buff8[ii]; /* Last byte */
1812 }
1813 #endif
1814 break;
1815
1816 case 'x':
1817
1818 /* Handle \x{..} specially - new Perl thing for utf8 */
1819
1820 #if !defined NOUTF8
1821 if (*p == '{')
1822 {
1823 unsigned char *pt = p;
1824 c = 0;
1825 while (isxdigit(*(++pt)))
1826 c = c * 16 + tolower(*pt) - ((isdigit(*pt))? '0' : 'W');
1827 if (*pt == '}')
1828 {
1829 unsigned char buff8[8];
1830 int ii, utn;
1831 if (use_utf8)
1832 {
1833 utn = ord2utf8(c, buff8);
1834 for (ii = 0; ii < utn - 1; ii++) *q++ = buff8[ii];
1835 c = buff8[ii]; /* Last byte */
1836 }
1837 else
1838 {
1839 if (c > 255)
1840 fprintf(outfile, "** Character \\x{%x} is greater than 255 and "
1841 "UTF-8 mode is not enabled.\n"
1842 "** Truncation will probably give the wrong result.\n", c);
1843 }
1844 p = pt + 1;
1845 break;
1846 }
1847 /* Not correct form; fall through */
1848 }
1849 #endif
1850
1851 /* Ordinary \x */
1852
1853 c = 0;
1854 while (i++ < 2 && isxdigit(*p))
1855 {
1856 c = c * 16 + tolower(*p) - ((isdigit(*p))? '0' : 'W');
1857 p++;
1858 }
1859 break;
1860
1861 case 0: /* \ followed by EOF allows for an empty line */
1862 p--;
1863 continue;
1864
1865 case '>':
1866 while(isdigit(*p)) start_offset = start_offset * 10 + *p++ - '0';
1867 continue;
1868
1869 case 'A': /* Option setting */
1870 options |= PCRE_ANCHORED;
1871 continue;
1872
1873 case 'B':
1874 options |= PCRE_NOTBOL;
1875 continue;
1876
1877 case 'C':
1878 if (isdigit(*p)) /* Set copy string */
1879 {
1880 while(isdigit(*p)) n = n * 10 + *p++ - '0';
1881 copystrings |= 1 << n;
1882 }
1883 else if (isalnum(*p))
1884 {
1885 uschar *npp = copynamesptr;
1886 while (isalnum(*p)) *npp++ = *p++;
1887 *npp++ = 0;
1888 *npp = 0;
1889 n = pcre_get_stringnumber(re, (char *)copynamesptr);
1890 if (n < 0)
1891 fprintf(outfile, "no parentheses with name \"%s\"\n", copynamesptr);
1892 copynamesptr = npp;
1893 }
1894 else if (*p == '+')
1895 {
1896 callout_extra = 1;
1897 p++;
1898 }
1899 else if (*p == '-')
1900 {
1901 pcre_callout = NULL;
1902 p++;
1903 }
1904 else if (*p == '!')
1905 {
1906 callout_fail_id = 0;
1907 p++;
1908 while(isdigit(*p))
1909 callout_fail_id = callout_fail_id * 10 + *p++ - '0';
1910 callout_fail_count = 0;
1911 if (*p == '!')
1912 {
1913 p++;
1914 while(isdigit(*p))
1915 callout_fail_count = callout_fail_count * 10 + *p++ - '0';
1916 }
1917 }
1918 else if (*p == '*')
1919 {
1920 int sign = 1;
1921 callout_data = 0;
1922 if (*(++p) == '-') { sign = -1; p++; }
1923 while(isdigit(*p))
1924 callout_data = callout_data * 10 + *p++ - '0';
1925 callout_data *= sign;
1926 callout_data_set = 1;
1927 }
1928 continue;
1929
1930 #if !defined NODFA
1931 case 'D':
1932 #if !defined NOPOSIX
1933 if (posix || do_posix)
1934 printf("** Can't use dfa matching in POSIX mode: \\D ignored\n");
1935 else
1936 #endif
1937 use_dfa = 1;
1938 continue;
1939
1940 case 'F':
1941 options |= PCRE_DFA_SHORTEST;
1942 continue;
1943 #endif
1944
1945 case 'G':
1946 if (isdigit(*p))
1947 {
1948 while(isdigit(*p)) n = n * 10 + *p++ - '0';
1949 getstrings |= 1 << n;
1950 }
1951 else if (isalnum(*p))
1952 {
1953 uschar *npp = getnamesptr;
1954 while (isalnum(*p)) *npp++ = *p++;
1955 *npp++ = 0;
1956 *npp = 0;
1957 n = pcre_get_stringnumber(re, (char *)getnamesptr);
1958 if (n < 0)
1959 fprintf(outfile, "no parentheses with name \"%s\"\n", getnamesptr);
1960 getnamesptr = npp;
1961 }
1962 continue;
1963
1964 case 'L':
1965 getlist = 1;
1966 continue;
1967
1968 case 'M':
1969 find_match_limit = 1;
1970 continue;
1971
1972 case 'N':
1973 if ((options & PCRE_NOTEMPTY) != 0)
1974 options = (options & ~PCRE_NOTEMPTY) | PCRE_NOTEMPTY_ATSTART;
1975 else
1976 options |= PCRE_NOTEMPTY;
1977 continue;
1978
1979 case 'O':
1980 while(isdigit(*p)) n = n * 10 + *p++ - '0';
1981 if (n > size_offsets_max)
1982 {
1983 size_offsets_max = n;
1984 free(offsets);
1985 use_offsets = offsets = (int *)malloc(size_offsets_max * sizeof(int));
1986 if (offsets == NULL)
1987 {
1988 printf("** Failed to get %d bytes of memory for offsets vector\n",
1989 (int)(size_offsets_max * sizeof(int)));
1990 yield = 1;
1991 goto EXIT;
1992 }
1993 }
1994 use_size_offsets = n;
1995 if (n == 0) use_offsets = NULL; /* Ensures it can't write to it */
1996 continue;
1997
1998 case 'P':
1999 options |= ((options & PCRE_PARTIAL_SOFT) == 0)?
2000 PCRE_PARTIAL_SOFT : PCRE_PARTIAL_HARD;
2001 continue;
2002
2003 case 'Q':
2004 while(isdigit(*p)) n = n * 10 + *p++ - '0';
2005 if (extra == NULL)
2006 {
2007 extra = (pcre_extra *)malloc(sizeof(pcre_extra));
2008 extra->flags = 0;
2009 }
2010 extra->flags |= PCRE_EXTRA_MATCH_LIMIT_RECURSION;
2011 extra->match_limit_recursion = n;
2012 continue;
2013
2014 case 'q':
2015 while(isdigit(*p)) n = n * 10 + *p++ - '0';
2016 if (extra == NULL)
2017 {
2018 extra = (pcre_extra *)malloc(sizeof(pcre_extra));
2019 extra->flags = 0;
2020 }
2021 extra->flags |= PCRE_EXTRA_MATCH_LIMIT;
2022 extra->match_limit = n;
2023 continue;
2024
2025 #if !defined NODFA
2026 case 'R':
2027 options |= PCRE_DFA_RESTART;
2028 continue;
2029 #endif
2030
2031 case 'S':
2032 show_malloc = 1;
2033 continue;
2034
2035 case 'Y':
2036 options |= PCRE_NO_START_OPTIMIZE;
2037 continue;
2038
2039 case 'Z':
2040 options |= PCRE_NOTEOL;
2041 continue;
2042
2043 case '?':
2044 options |= PCRE_NO_UTF8_CHECK;
2045 continue;
2046
2047 case '<':
2048 {
2049 int x = check_newline(p, outfile);
2050 if (x == 0) goto NEXT_DATA;
2051 options |= x;
2052 while (*p++ != '>');
2053 }
2054 continue;
2055 }
2056 *q++ = c;
2057 }
2058 *q = 0;
2059 len = q - dbuffer;
2060
2061 /* Move the data to the end of the buffer so that a read over the end of
2062 the buffer will be seen by valgrind, even if it doesn't cause a crash. If
2063 we are using the POSIX interface, we must include the terminating zero. */
2064
2065 #if !defined NOPOSIX
2066 if (posix || do_posix)
2067 {
2068 memmove(bptr + buffer_size - len - 1, bptr, len + 1);
2069 bptr += buffer_size - len - 1;
2070 }
2071 else
2072 #endif
2073 {
2074 memmove(bptr + buffer_size - len, bptr, len);
2075 bptr += buffer_size - len;
2076 }
2077
2078 if ((all_use_dfa || use_dfa) && find_match_limit)
2079 {
2080 printf("**Match limit not relevant for DFA matching: ignored\n");
2081 find_match_limit = 0;
2082 }
2083
2084 /* Handle matching via the POSIX interface, which does not
2085 support timing or playing with the match limit or callout data. */
2086
2087 #if !defined NOPOSIX
2088 if (posix || do_posix)
2089 {
2090 int rc;
2091 int eflags = 0;
2092 regmatch_t *pmatch = NULL;
2093 if (use_size_offsets > 0)
2094 pmatch = (regmatch_t *)malloc(sizeof(regmatch_t) * use_size_offsets);
2095 if ((options & PCRE_NOTBOL) != 0) eflags |= REG_NOTBOL;
2096 if ((options & PCRE_NOTEOL) != 0) eflags |= REG_NOTEOL;
2097 if ((options & PCRE_NOTEMPTY) != 0) eflags |= REG_NOTEMPTY;
2098
2099 rc = regexec(&preg, (const char *)bptr, use_size_offsets, pmatch, eflags);
2100
2101 if (rc != 0)
2102 {
2103 (void)regerror(rc, &preg, (char *)buffer, buffer_size);
2104 fprintf(outfile, "No match: POSIX code %d: %s\n", rc, buffer);
2105 }
2106 else if ((((const pcre *)preg.re_pcre)->options & PCRE_NO_AUTO_CAPTURE)
2107 != 0)
2108 {
2109 fprintf(outfile, "Matched with REG_NOSUB\n");
2110 }
2111 else
2112 {
2113 size_t i;
2114 for (i = 0; i < (size_t)use_size_offsets; i++)
2115 {
2116 if (pmatch[i].rm_so >= 0)
2117 {
2118 fprintf(outfile, "%2d: ", (int)i);
2119 (void)pchars(dbuffer + pmatch[i].rm_so,
2120 pmatch[i].rm_eo - pmatch[i].rm_so, outfile);
2121 fprintf(outfile, "\n");
2122 if (i == 0 && do_showrest)
2123 {
2124 fprintf(outfile, " 0+ ");
2125 (void)pchars(dbuffer + pmatch[i].rm_eo, len - pmatch[i].rm_eo,
2126 outfile);
2127 fprintf(outfile, "\n");
2128 }
2129 }
2130 }
2131 }
2132 free(pmatch);
2133 }
2134
2135 /* Handle matching via the native interface - repeats for /g and /G */
2136
2137 else
2138 #endif /* !defined NOPOSIX */
2139
2140 for (;; gmatched++) /* Loop for /g or /G */
2141 {
2142 if (timeitm > 0)
2143 {
2144 register int i;
2145 clock_t time_taken;
2146 clock_t start_time = clock();
2147
2148 #if !defined NODFA
2149 if (all_use_dfa || use_dfa)
2150 {
2151 int workspace[1000];
2152 for (i = 0; i < timeitm; i++)
2153 count = pcre_dfa_exec(re, NULL, (char *)bptr, len, start_offset,
2154 options | g_notempty, use_offsets, use_size_offsets, workspace,
2155 sizeof(workspace)/sizeof(int));
2156 }
2157 else
2158 #endif
2159
2160 for (i = 0; i < timeitm; i++)
2161 count = pcre_exec(re, extra, (char *)bptr, len,
2162 start_offset, options | g_notempty, use_offsets, use_size_offsets);
2163
2164 time_taken = clock() - start_time;
2165 fprintf(outfile, "Execute time %.4f milliseconds\n",
2166 (((double)time_taken * 1000.0) / (double)timeitm) /
2167 (double)CLOCKS_PER_SEC);
2168 }
2169
2170 /* If find_match_limit is set, we want to do repeated matches with
2171 varying limits in order to find the minimum value for the match limit and
2172 for the recursion limit. */
2173
2174 if (find_match_limit)
2175 {
2176 if (extra == NULL)
2177 {
2178 extra = (pcre_extra *)malloc(sizeof(pcre_extra));
2179 extra->flags = 0;
2180 }
2181
2182 (void)check_match_limit(re, extra, bptr, len, start_offset,
2183 options|g_notempty, use_offsets, use_size_offsets,
2184 PCRE_EXTRA_MATCH_LIMIT, &(extra->match_limit),
2185 PCRE_ERROR_MATCHLIMIT, "match()");
2186
2187 count = check_match_limit(re, extra, bptr, len, start_offset,
2188 options|g_notempty, use_offsets, use_size_offsets,
2189 PCRE_EXTRA_MATCH_LIMIT_RECURSION, &(extra->match_limit_recursion),
2190 PCRE_ERROR_RECURSIONLIMIT, "match() recursion");
2191 }
2192
2193 /* If callout_data is set, use the interface with additional data */
2194
2195 else if (callout_data_set)
2196 {
2197 if (extra == NULL)
2198 {
2199 extra = (pcre_extra *)malloc(sizeof(pcre_extra));
2200 extra->flags = 0;
2201 }
2202 extra->flags |= PCRE_EXTRA_CALLOUT_DATA;
2203 extra->callout_data = &callout_data;
2204 count = pcre_exec(re, extra, (char *)bptr, len, start_offset,
2205 options | g_notempty, use_offsets, use_size_offsets);
2206 extra->flags &= ~PCRE_EXTRA_CALLOUT_DATA;
2207 }
2208
2209 /* The normal case is just to do the match once, with the default
2210 value of match_limit. */
2211
2212 #if !defined NODFA
2213 else if (all_use_dfa || use_dfa)
2214 {
2215 int workspace[1000];
2216 count = pcre_dfa_exec(re, NULL, (char *)bptr, len, start_offset,
2217 options | g_notempty, use_offsets, use_size_offsets, workspace,
2218 sizeof(workspace)/sizeof(int));
2219 if (count == 0)
2220 {
2221 fprintf(outfile, "Matched, but too many subsidiary matches\n");
2222 count = use_size_offsets/2;
2223 }
2224 }
2225 #endif
2226
2227 else
2228 {
2229 count = pcre_exec(re, extra, (char *)bptr, len,
2230 start_offset, options | g_notempty, use_offsets, use_size_offsets);
2231 if (count == 0)
2232 {
2233 fprintf(outfile, "Matched, but too many substrings\n");
2234 count = use_size_offsets/3;
2235 }
2236 }
2237
2238 /* Matched */
2239
2240 if (count >= 0)
2241 {
2242 int i, maxcount;
2243
2244 #if !defined NODFA
2245 if (all_use_dfa || use_dfa) maxcount = use_size_offsets/2; else
2246 #endif
2247 maxcount = use_size_offsets/3;
2248
2249 /* This is a check against a lunatic return value. */
2250
2251 if (count > maxcount)
2252 {
2253 fprintf(outfile,
2254 "** PCRE error: returned count %d is too big for offset size %d\n",
2255 count, use_size_offsets);
2256 count = use_size_offsets/3;
2257 if (do_g || do_G)
2258 {
2259 fprintf(outfile, "** /%c loop abandoned\n", do_g? 'g' : 'G');
2260 do_g = do_G = FALSE; /* Break g/G loop */
2261 }
2262 }
2263
2264 for (i = 0; i < count * 2; i += 2)
2265 {
2266 if (use_offsets[i] < 0)
2267 fprintf(outfile, "%2d: <unset>\n", i/2);
2268 else
2269 {
2270 fprintf(outfile, "%2d: ", i/2);
2271 (void)pchars(bptr + use_offsets[i],
2272 use_offsets[i+1] - use_offsets[i], outfile);
2273 fprintf(outfile, "\n");
2274 if (i == 0)
2275 {
2276 if (do_showrest)
2277 {
2278 fprintf(outfile, " 0+ ");
2279 (void)pchars(bptr + use_offsets[i+1], len - use_offsets[i+1],
2280 outfile);
2281 fprintf(outfile, "\n");
2282 }
2283 }
2284 }
2285 }
2286
2287 for (i = 0; i < 32; i++)
2288 {
2289 if ((copystrings & (1 << i)) != 0)
2290 {
2291 char copybuffer[256];
2292 int rc = pcre_copy_substring((char *)bptr, use_offsets, count,
2293 i, copybuffer, sizeof(copybuffer));
2294 if (rc < 0)
2295 fprintf(outfile, "copy substring %d failed %d\n", i, rc);
2296 else
2297 fprintf(outfile, "%2dC %s (%d)\n", i, copybuffer, rc);
2298 }
2299 }
2300
2301 for (copynamesptr = copynames;
2302 *copynamesptr != 0;
2303 copynamesptr += (int)strlen((char*)copynamesptr) + 1)
2304 {
2305 char copybuffer[256];
2306 int rc = pcre_copy_named_substring(re, (char *)bptr, use_offsets,
2307 count, (char *)copynamesptr, copybuffer, sizeof(copybuffer));
2308 if (rc < 0)
2309 fprintf(outfile, "copy substring %s failed %d\n", copynamesptr, rc);
2310 else
2311 fprintf(outfile, " C %s (%d) %s\n", copybuffer, rc, copynamesptr);
2312 }
2313
2314 for (i = 0; i < 32; i++)
2315 {
2316 if ((getstrings & (1 << i)) != 0)
2317 {
2318 const char *substring;
2319 int rc = pcre_get_substring((char *)bptr, use_offsets, count,
2320 i, &substring);
2321 if (rc < 0)
2322 fprintf(outfile, "get substring %d failed %d\n", i, rc);
2323 else
2324 {
2325 fprintf(outfile, "%2dG %s (%d)\n", i, substring, rc);
2326 pcre_free_substring(substring);
2327 }
2328 }
2329 }
2330
2331 for (getnamesptr = getnames;
2332 *getnamesptr != 0;
2333 getnamesptr += (int)strlen((char*)getnamesptr) + 1)
2334 {
2335 const char *substring;
2336 int rc = pcre_get_named_substring(re, (char *)bptr, use_offsets,
2337 count, (char *)getnamesptr, &substring);
2338 if (rc < 0)
2339 fprintf(outfile, "copy substring %s failed %d\n", getnamesptr, rc);
2340 else
2341 {
2342 fprintf(outfile, " G %s (%d) %s\n", substring, rc, getnamesptr);
2343 pcre_free_substring(substring);
2344 }
2345 }
2346
2347 if (getlist)
2348 {
2349 const char **stringlist;
2350 int rc = pcre_get_substring_list((char *)bptr, use_offsets, count,
2351 &stringlist);
2352 if (rc < 0)
2353 fprintf(outfile, "get substring list failed %d\n", rc);
2354 else
2355 {
2356 for (i = 0; i < count; i++)
2357 fprintf(outfile, "%2dL %s\n", i, stringlist[i]);
2358 if (stringlist[i] != NULL)
2359 fprintf(outfile, "string list not terminated by NULL\n");
2360 /* free((void *)stringlist); */
2361 pcre_free_substring_list(stringlist);
2362 }
2363 }
2364 }
2365
2366 /* There was a partial match */
2367
2368 else if (count == PCRE_ERROR_PARTIAL)
2369 {
2370 fprintf(outfile, "Partial match");
2371 if (use_size_offsets > 1)
2372 {
2373 fprintf(outfile, ": ");
2374 pchars(bptr + use_offsets[0], use_offsets[1] - use_offsets[0],
2375 outfile);
2376 }
2377 fprintf(outfile, "\n");
2378 break; /* Out of the /g loop */
2379 }
2380
2381 /* Failed to match. If this is a /g or /G loop and we previously set
2382 g_notempty after a null match, this is not necessarily the end. We want
2383 to advance the start offset, and continue. We won't be at the end of the
2384 string - that was checked before setting g_notempty.
2385
2386 Complication arises in the case when the newline option is "any" or
2387 "anycrlf". If the previous match was at the end of a line terminated by
2388 CRLF, an advance of one character just passes the \r, whereas we should
2389 prefer the longer newline sequence, as does the code in pcre_exec().
2390 Fudge the offset value to achieve this.
2391
2392 Otherwise, in the case of UTF-8 matching, the advance must be one
2393 character, not one byte. */
2394
2395 else
2396 {
2397 if (g_notempty != 0)
2398 {
2399 int onechar = 1;
2400 unsigned int obits = ((real_pcre *)re)->options;
2401 use_offsets[0] = start_offset;
2402 if ((obits & PCRE_NEWLINE_BITS) == 0)
2403 {
2404 int d;
2405 (void)pcre_config(PCRE_CONFIG_NEWLINE, &d);
2406 /* Note that these values are always the ASCII ones, even in
2407 EBCDIC environments. CR = 13, NL = 10. */
2408 obits = (d == 13)? PCRE_NEWLINE_CR :
2409 (d == 10)? PCRE_NEWLINE_LF :
2410 (d == (13<<8 | 10))? PCRE_NEWLINE_CRLF :
2411 (d == -2)? PCRE_NEWLINE_ANYCRLF :
2412 (d == -1)? PCRE_NEWLINE_ANY : 0;
2413 }
2414 if (((obits & PCRE_NEWLINE_BITS) == PCRE_NEWLINE_ANY ||
2415 (obits & PCRE_NEWLINE_BITS) == PCRE_NEWLINE_ANYCRLF)
2416 &&
2417 start_offset < len - 1 &&
2418 bptr[start_offset] == '\r' &&
2419 bptr[start_offset+1] == '\n')
2420 onechar++;
2421 else if (use_utf8)
2422 {
2423 while (start_offset + onechar < len)
2424 {
2425 int tb = bptr[start_offset+onechar];
2426 if (tb <= 127) break;
2427 tb &= 0xc0;
2428 if (tb != 0 && tb != 0xc0) onechar++;
2429 }
2430 }
2431 use_offsets[1] = start_offset + onechar;
2432 }
2433 else
2434 {
2435 if (count == PCRE_ERROR_NOMATCH)
2436 {
2437 if (gmatched == 0) fprintf(outfile, "No match\n");
2438 }
2439 else fprintf(outfile, "Error %d\n", count);
2440 break; /* Out of the /g loop */
2441 }
2442 }
2443
2444 /* If not /g or /G we are done */
2445
2446 if (!do_g && !do_G) break;
2447
2448 /* If we have matched an empty string, first check to see if we are at
2449 the end of the subject. If so, the /g loop is over. Otherwise, mimic what
2450 Perl's /g options does. This turns out to be rather cunning. First we set
2451 PCRE_NOTEMPTY_ATSTART and PCRE_ANCHORED and try the match again at the
2452 same point. If this fails (picked up above) we advance to the next
2453 character. */
2454
2455 g_notempty = 0;
2456
2457 if (use_offsets[0] == use_offsets[1])
2458 {
2459 if (use_offsets[0] == len) break;
2460 g_notempty = PCRE_NOTEMPTY_ATSTART | PCRE_ANCHORED;
2461 }
2462
2463 /* For /g, update the start offset, leaving the rest alone */
2464
2465 if (do_g) start_offset = use_offsets[1];
2466
2467 /* For /G, update the pointer and length */
2468
2469 else
2470 {
2471 bptr += use_offsets[1];
2472 len -= use_offsets[1];
2473 }
2474 } /* End of loop for /g and /G */
2475
2476 NEXT_DATA: continue;
2477 } /* End of loop for data lines */
2478
2479 CONTINUE:
2480
2481 #if !defined NOPOSIX
2482 if (posix || do_posix) regfree(&preg);
2483 #endif
2484
2485 if (re != NULL) new_free(re);
2486 if (extra != NULL) new_free(extra);
2487 if (tables != NULL)
2488 {
2489 new_free((void *)tables);
2490 setlocale(LC_CTYPE, "C");
2491 locale_set = 0;
2492 }
2493 }
2494
2495 if (infile == stdin) fprintf(outfile, "\n");
2496
2497 EXIT:
2498
2499 if (infile != NULL && infile != stdin) fclose(infile);
2500 if (outfile != NULL && outfile != stdout) fclose(outfile);
2501
2502 free(buffer);
2503 free(dbuffer);
2504 free(pbuffer);
2505 free(offsets);
2506
2507 return yield;
2508 }
2509
2510 /* End of pcretest.c */

Properties

Name Value
svn:eol-style native
svn:keywords "Author Date Id Revision Url"

  ViewVC Help
Powered by ViewVC 1.1.5