Commit | Line | Data |
---|---|---|
aa41d2de | 1 | /* $Cambridge: exim/src/src/pcre/pcretest.c,v 1.4 2006/11/07 16:50:36 ph10 Exp $ */ |
8ac170f3 | 2 | |
c86f6258 PH |
3 | /************************************************* |
4 | * PCRE testing program * | |
5 | *************************************************/ | |
6 | ||
7 | /* This program was hacked up as a tester for PCRE. I really should have | |
8 | written it more tidily in the first place. Will I ever learn? It has grown and | |
8ac170f3 | 9 | been extended and consequently is now rather, er, *very* untidy in places. |
c86f6258 PH |
10 | |
11 | ----------------------------------------------------------------------------- | |
12 | Redistribution and use in source and binary forms, with or without | |
13 | modification, are permitted provided that the following conditions are met: | |
14 | ||
15 | * Redistributions of source code must retain the above copyright notice, | |
16 | this list of conditions and the following disclaimer. | |
17 | ||
18 | * Redistributions in binary form must reproduce the above copyright | |
19 | notice, this list of conditions and the following disclaimer in the | |
20 | documentation and/or other materials provided with the distribution. | |
21 | ||
22 | * Neither the name of the University of Cambridge nor the names of its | |
23 | contributors may be used to endorse or promote products derived from | |
24 | this software without specific prior written permission. | |
25 | ||
26 | THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" | |
27 | AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE | |
28 | IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE | |
29 | ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE | |
30 | LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR | |
31 | CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF | |
32 | SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS | |
33 | INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN | |
34 | CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) | |
35 | ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE | |
36 | POSSIBILITY OF SUCH DAMAGE. | |
37 | ----------------------------------------------------------------------------- | |
38 | */ | |
39 | ||
40 | ||
41 | #include <ctype.h> | |
42 | #include <stdio.h> | |
43 | #include <string.h> | |
44 | #include <stdlib.h> | |
45 | #include <time.h> | |
46 | #include <locale.h> | |
47 | #include <errno.h> | |
48 | ||
aa41d2de PH |
49 | #ifndef _WIN32 |
50 | #include <sys/resource.h> | |
51 | #endif | |
52 | ||
c86f6258 | 53 | #define PCRE_SPY /* For Win32 build, import data, not export */ |
8ac170f3 | 54 | |
aa41d2de PH |
55 | /* We include pcre_internal.h because we need the internal info for displaying |
56 | the results of pcre_study() and we also need to know about the internal | |
57 | macros, structures, and other internal data values; pcretest has "inside | |
58 | information" compared to a program that strictly follows the PCRE API. */ | |
8ac170f3 PH |
59 | |
60 | #include "pcre_internal.h" | |
61 | ||
aa41d2de PH |
62 | /* We need access to the data tables that PCRE uses. So as not to have to keep |
63 | two copies, we include the source file here, changing the names of the external | |
64 | symbols to prevent clashes. */ | |
65 | ||
66 | #define _pcre_utf8_table1 utf8_table1 | |
67 | #define _pcre_utf8_table1_size utf8_table1_size | |
68 | #define _pcre_utf8_table2 utf8_table2 | |
69 | #define _pcre_utf8_table3 utf8_table3 | |
70 | #define _pcre_utf8_table4 utf8_table4 | |
71 | #define _pcre_utt utt | |
72 | #define _pcre_utt_size utt_size | |
73 | #define _pcre_OP_lengths OP_lengths | |
74 | ||
75 | #include "pcre_tables.c" | |
76 | ||
77 | /* We also need the pcre_printint() function for printing out compiled | |
78 | patterns. This function is in a separate file so that it can be included in | |
79 | pcre_compile.c when that module is compiled with debugging enabled. */ | |
80 | ||
81 | #include "pcre_printint.src" | |
82 | ||
c86f6258 PH |
83 | |
84 | /* It is possible to compile this test program without including support for | |
85 | testing the POSIX interface, though this is not available via the standard | |
86 | Makefile. */ | |
87 | ||
88 | #if !defined NOPOSIX | |
89 | #include "pcreposix.h" | |
90 | #endif | |
91 | ||
92e772ff PH |
92 | /* It is also possible, for the benefit of the version imported into Exim, to |
93 | build pcretest without support for UTF8 (define NOUTF8), without the interface | |
8ac170f3 PH |
94 | to the DFA matcher (NODFA), and without the doublecheck of the old "info" |
95 | function (define NOINFOCHECK). */ | |
96 | ||
97 | ||
aa41d2de PH |
98 | /* Other parameters */ |
99 | ||
c86f6258 PH |
100 | #ifndef CLOCKS_PER_SEC |
101 | #ifdef CLK_TCK | |
102 | #define CLOCKS_PER_SEC CLK_TCK | |
103 | #else | |
104 | #define CLOCKS_PER_SEC 100 | |
105 | #endif | |
106 | #endif | |
107 | ||
108 | #define LOOPREPEAT 500000 | |
109 | ||
aa41d2de | 110 | /* Static variables */ |
c86f6258 PH |
111 | |
112 | static FILE *outfile; | |
113 | static int log_store = 0; | |
114 | static int callout_count; | |
115 | static int callout_extra; | |
116 | static int callout_fail_count; | |
117 | static int callout_fail_id; | |
118 | static int first_callout; | |
119 | static int show_malloc; | |
120 | static int use_utf8; | |
121 | static size_t gotten_store; | |
122 | ||
aa41d2de PH |
123 | /* The buffers grow automatically if very long input lines are encountered. */ |
124 | ||
125 | static int buffer_size = 50000; | |
126 | static uschar *buffer = NULL; | |
127 | static uschar *dbuffer = NULL; | |
c86f6258 PH |
128 | static uschar *pbuffer = NULL; |
129 | ||
130 | ||
c86f6258 | 131 | |
aa41d2de PH |
132 | /************************************************* |
133 | * Read or extend an input line * | |
134 | *************************************************/ | |
135 | ||
136 | /* Input lines are read into buffer, but both patterns and data lines can be | |
137 | continued over multiple input lines. In addition, if the buffer fills up, we | |
138 | want to automatically expand it so as to be able to handle extremely large | |
139 | lines that are needed for certain stress tests. When the input buffer is | |
140 | expanded, the other two buffers must also be expanded likewise, and the | |
141 | contents of pbuffer, which are a copy of the input for callouts, must be | |
142 | preserved (for when expansion happens for a data line). This is not the most | |
143 | optimal way of handling this, but hey, this is just a test program! | |
144 | ||
145 | Arguments: | |
146 | f the file to read | |
147 | start where in buffer to start (this *must* be within buffer) | |
148 | ||
149 | Returns: pointer to the start of new data | |
150 | could be a copy of start, or could be moved | |
151 | NULL if no data read and EOF reached | |
152 | */ | |
153 | ||
154 | static uschar * | |
155 | extend_inputline(FILE *f, uschar *start) | |
156 | { | |
157 | uschar *here = start; | |
158 | ||
159 | for (;;) | |
160 | { | |
161 | int rlen = buffer_size - (here - buffer); | |
162 | if (rlen > 1000) | |
163 | { | |
164 | int dlen; | |
165 | if (fgets((char *)here, rlen, f) == NULL) | |
166 | return (here == start)? NULL : start; | |
167 | dlen = (int)strlen((char *)here); | |
168 | if (dlen > 0 && here[dlen - 1] == '\n') return start; | |
169 | here += dlen; | |
170 | } | |
171 | ||
172 | else | |
173 | { | |
174 | int new_buffer_size = 2*buffer_size; | |
175 | uschar *new_buffer = (unsigned char *)malloc(new_buffer_size); | |
176 | uschar *new_dbuffer = (unsigned char *)malloc(new_buffer_size); | |
177 | uschar *new_pbuffer = (unsigned char *)malloc(new_buffer_size); | |
178 | ||
179 | if (new_buffer == NULL || new_dbuffer == NULL || new_pbuffer == NULL) | |
180 | { | |
181 | fprintf(stderr, "pcretest: malloc(%d) failed\n", new_buffer_size); | |
182 | exit(1); | |
183 | } | |
184 | ||
185 | memcpy(new_buffer, buffer, buffer_size); | |
186 | memcpy(new_pbuffer, pbuffer, buffer_size); | |
187 | ||
188 | buffer_size = new_buffer_size; | |
189 | ||
190 | start = new_buffer + (start - buffer); | |
191 | here = new_buffer + (here - buffer); | |
192 | ||
193 | free(buffer); | |
194 | free(dbuffer); | |
195 | free(pbuffer); | |
196 | ||
197 | buffer = new_buffer; | |
198 | dbuffer = new_dbuffer; | |
199 | pbuffer = new_pbuffer; | |
200 | } | |
201 | } | |
202 | ||
203 | return NULL; /* Control never gets here */ | |
204 | } | |
205 | ||
206 | ||
207 | ||
208 | ||
209 | ||
210 | ||
211 | ||
c86f6258 PH |
212 | /************************************************* |
213 | * Read number from string * | |
214 | *************************************************/ | |
215 | ||
216 | /* We don't use strtoul() because SunOS4 doesn't have it. Rather than mess | |
217 | around with conditional compilation, just do the job by hand. It is only used | |
218 | for unpicking the -o argument, so just keep it simple. | |
219 | ||
220 | Arguments: | |
221 | str string to be converted | |
222 | endptr where to put the end pointer | |
223 | ||
224 | Returns: the unsigned long | |
225 | */ | |
226 | ||
227 | static int | |
228 | get_value(unsigned char *str, unsigned char **endptr) | |
229 | { | |
230 | int result = 0; | |
231 | while(*str != 0 && isspace(*str)) str++; | |
232 | while (isdigit(*str)) result = result * 10 + (int)(*str++ - '0'); | |
233 | *endptr = str; | |
234 | return(result); | |
235 | } | |
236 | ||
237 | ||
238 | ||
c86f6258 PH |
239 | |
240 | /************************************************* | |
241 | * Convert UTF-8 string to value * | |
242 | *************************************************/ | |
243 | ||
244 | /* This function takes one or more bytes that represents a UTF-8 character, | |
245 | and returns the value of the character. | |
246 | ||
247 | Argument: | |
aa41d2de PH |
248 | utf8bytes a pointer to the byte vector |
249 | vptr a pointer to an int to receive the value | |
c86f6258 | 250 | |
aa41d2de PH |
251 | Returns: > 0 => the number of bytes consumed |
252 | -6 to 0 => malformed UTF-8 character at offset = (-return) | |
c86f6258 PH |
253 | */ |
254 | ||
8ac170f3 PH |
255 | #if !defined NOUTF8 |
256 | ||
c86f6258 | 257 | static int |
aa41d2de | 258 | utf82ord(unsigned char *utf8bytes, int *vptr) |
c86f6258 | 259 | { |
aa41d2de | 260 | int c = *utf8bytes++; |
c86f6258 PH |
261 | int d = c; |
262 | int i, j, s; | |
263 | ||
264 | for (i = -1; i < 6; i++) /* i is number of additional bytes */ | |
265 | { | |
266 | if ((d & 0x80) == 0) break; | |
267 | d <<= 1; | |
268 | } | |
269 | ||
270 | if (i == -1) { *vptr = c; return 1; } /* ascii character */ | |
271 | if (i == 0 || i == 6) return 0; /* invalid UTF-8 */ | |
272 | ||
273 | /* i now has a value in the range 1-5 */ | |
274 | ||
275 | s = 6*i; | |
aa41d2de | 276 | d = (c & utf8_table3[i]) << s; |
c86f6258 PH |
277 | |
278 | for (j = 0; j < i; j++) | |
279 | { | |
aa41d2de | 280 | c = *utf8bytes++; |
c86f6258 PH |
281 | if ((c & 0xc0) != 0x80) return -(j+1); |
282 | s -= 6; | |
283 | d |= (c & 0x3f) << s; | |
284 | } | |
285 | ||
286 | /* Check that encoding was the correct unique one */ | |
287 | ||
aa41d2de PH |
288 | for (j = 0; j < utf8_table1_size; j++) |
289 | if (d <= utf8_table1[j]) break; | |
c86f6258 PH |
290 | if (j != i) return -(i+1); |
291 | ||
292 | /* Valid value */ | |
293 | ||
294 | *vptr = d; | |
295 | return i+1; | |
296 | } | |
297 | ||
8ac170f3 PH |
298 | #endif |
299 | ||
c86f6258 PH |
300 | |
301 | ||
aa41d2de PH |
302 | /************************************************* |
303 | * Convert character value to UTF-8 * | |
304 | *************************************************/ | |
305 | ||
306 | /* This function takes an integer value in the range 0 - 0x7fffffff | |
307 | and encodes it as a UTF-8 character in 0 to 6 bytes. | |
308 | ||
309 | Arguments: | |
310 | cvalue the character value | |
311 | utf8bytes pointer to buffer for result - at least 6 bytes long | |
312 | ||
313 | Returns: number of characters placed in the buffer | |
314 | */ | |
315 | ||
316 | #if !defined NOUTF8 | |
317 | ||
318 | static int | |
319 | ord2utf8(int cvalue, uschar *utf8bytes) | |
320 | { | |
321 | register int i, j; | |
322 | for (i = 0; i < utf8_table1_size; i++) | |
323 | if (cvalue <= utf8_table1[i]) break; | |
324 | utf8bytes += i; | |
325 | for (j = i; j > 0; j--) | |
326 | { | |
327 | *utf8bytes-- = 0x80 | (cvalue & 0x3f); | |
328 | cvalue >>= 6; | |
329 | } | |
330 | *utf8bytes = utf8_table2[i] | cvalue; | |
331 | return i + 1; | |
332 | } | |
333 | ||
334 | #endif | |
335 | ||
336 | ||
337 | ||
c86f6258 PH |
338 | /************************************************* |
339 | * Print character string * | |
340 | *************************************************/ | |
341 | ||
342 | /* Character string printing function. Must handle UTF-8 strings in utf8 | |
343 | mode. Yields number of characters printed. If handed a NULL file, just counts | |
344 | chars without printing. */ | |
345 | ||
346 | static int pchars(unsigned char *p, int length, FILE *f) | |
347 | { | |
aa41d2de | 348 | int c = 0; |
c86f6258 PH |
349 | int yield = 0; |
350 | ||
351 | while (length-- > 0) | |
352 | { | |
92e772ff | 353 | #if !defined NOUTF8 |
c86f6258 PH |
354 | if (use_utf8) |
355 | { | |
356 | int rc = utf82ord(p, &c); | |
357 | ||
358 | if (rc > 0 && rc <= length + 1) /* Mustn't run over the end */ | |
359 | { | |
360 | length -= rc - 1; | |
361 | p += rc; | |
362 | if (c < 256 && isprint(c)) | |
363 | { | |
364 | if (f != NULL) fprintf(f, "%c", c); | |
365 | yield++; | |
366 | } | |
367 | else | |
368 | { | |
369 | int n; | |
370 | if (f != NULL) fprintf(f, "\\x{%02x}%n", c, &n); | |
371 | yield += n; | |
372 | } | |
373 | continue; | |
374 | } | |
375 | } | |
8ac170f3 | 376 | #endif |
c86f6258 PH |
377 | |
378 | /* Not UTF-8, or malformed UTF-8 */ | |
379 | ||
380 | if (isprint(c = *(p++))) | |
381 | { | |
382 | if (f != NULL) fprintf(f, "%c", c); | |
383 | yield++; | |
384 | } | |
385 | else | |
386 | { | |
387 | if (f != NULL) fprintf(f, "\\x%02x", c); | |
388 | yield += 4; | |
389 | } | |
390 | } | |
391 | ||
392 | return yield; | |
393 | } | |
394 | ||
395 | ||
396 | ||
397 | /************************************************* | |
398 | * Callout function * | |
399 | *************************************************/ | |
400 | ||
401 | /* Called from PCRE as a result of the (?C) item. We print out where we are in | |
402 | the match. Yield zero unless more callouts than the fail count, or the callout | |
403 | data is not zero. */ | |
404 | ||
405 | static int callout(pcre_callout_block *cb) | |
406 | { | |
407 | FILE *f = (first_callout | callout_extra)? outfile : NULL; | |
408 | int i, pre_start, post_start, subject_length; | |
409 | ||
410 | if (callout_extra) | |
411 | { | |
412 | fprintf(f, "Callout %d: last capture = %d\n", | |
413 | cb->callout_number, cb->capture_last); | |
414 | ||
415 | for (i = 0; i < cb->capture_top * 2; i += 2) | |
416 | { | |
417 | if (cb->offset_vector[i] < 0) | |
418 | fprintf(f, "%2d: <unset>\n", i/2); | |
419 | else | |
420 | { | |
421 | fprintf(f, "%2d: ", i/2); | |
422 | (void)pchars((unsigned char *)cb->subject + cb->offset_vector[i], | |
423 | cb->offset_vector[i+1] - cb->offset_vector[i], f); | |
424 | fprintf(f, "\n"); | |
425 | } | |
426 | } | |
427 | } | |
428 | ||
429 | /* Re-print the subject in canonical form, the first time or if giving full | |
430 | datails. On subsequent calls in the same match, we use pchars just to find the | |
431 | printed lengths of the substrings. */ | |
432 | ||
433 | if (f != NULL) fprintf(f, "--->"); | |
434 | ||
435 | pre_start = pchars((unsigned char *)cb->subject, cb->start_match, f); | |
436 | post_start = pchars((unsigned char *)(cb->subject + cb->start_match), | |
437 | cb->current_position - cb->start_match, f); | |
438 | ||
439 | subject_length = pchars((unsigned char *)cb->subject, cb->subject_length, NULL); | |
440 | ||
441 | (void)pchars((unsigned char *)(cb->subject + cb->current_position), | |
442 | cb->subject_length - cb->current_position, f); | |
443 | ||
444 | if (f != NULL) fprintf(f, "\n"); | |
445 | ||
446 | /* Always print appropriate indicators, with callout number if not already | |
447 | shown. For automatic callouts, show the pattern offset. */ | |
448 | ||
449 | if (cb->callout_number == 255) | |
450 | { | |
451 | fprintf(outfile, "%+3d ", cb->pattern_position); | |
452 | if (cb->pattern_position > 99) fprintf(outfile, "\n "); | |
453 | } | |
454 | else | |
455 | { | |
456 | if (callout_extra) fprintf(outfile, " "); | |
457 | else fprintf(outfile, "%3d ", cb->callout_number); | |
458 | } | |
459 | ||
460 | for (i = 0; i < pre_start; i++) fprintf(outfile, " "); | |
461 | fprintf(outfile, "^"); | |
462 | ||
463 | if (post_start > 0) | |
464 | { | |
465 | for (i = 0; i < post_start - 1; i++) fprintf(outfile, " "); | |
466 | fprintf(outfile, "^"); | |
467 | } | |
468 | ||
469 | for (i = 0; i < subject_length - pre_start - post_start + 4; i++) | |
470 | fprintf(outfile, " "); | |
471 | ||
472 | fprintf(outfile, "%.*s", (cb->next_item_length == 0)? 1 : cb->next_item_length, | |
473 | pbuffer + cb->pattern_position); | |
474 | ||
475 | fprintf(outfile, "\n"); | |
476 | first_callout = 0; | |
477 | ||
478 | if (cb->callout_data != NULL) | |
479 | { | |
480 | int callout_data = *((int *)(cb->callout_data)); | |
481 | if (callout_data != 0) | |
482 | { | |
483 | fprintf(outfile, "Callout data = %d\n", callout_data); | |
484 | return callout_data; | |
485 | } | |
486 | } | |
487 | ||
488 | return (cb->callout_number != callout_fail_id)? 0 : | |
489 | (++callout_count >= callout_fail_count)? 1 : 0; | |
490 | } | |
491 | ||
492 | ||
493 | /************************************************* | |
494 | * Local malloc functions * | |
495 | *************************************************/ | |
496 | ||
497 | /* Alternative malloc function, to test functionality and show the size of the | |
498 | compiled re. */ | |
499 | ||
500 | static void *new_malloc(size_t size) | |
501 | { | |
502 | void *block = malloc(size); | |
503 | gotten_store = size; | |
504 | if (show_malloc) | |
505 | fprintf(outfile, "malloc %3d %p\n", (int)size, block); | |
506 | return block; | |
507 | } | |
508 | ||
509 | static void new_free(void *block) | |
510 | { | |
511 | if (show_malloc) | |
512 | fprintf(outfile, "free %p\n", block); | |
513 | free(block); | |
514 | } | |
515 | ||
516 | ||
517 | /* For recursion malloc/free, to test stacking calls */ | |
518 | ||
519 | static void *stack_malloc(size_t size) | |
520 | { | |
521 | void *block = malloc(size); | |
522 | if (show_malloc) | |
523 | fprintf(outfile, "stack_malloc %3d %p\n", (int)size, block); | |
524 | return block; | |
525 | } | |
526 | ||
527 | static void stack_free(void *block) | |
528 | { | |
529 | if (show_malloc) | |
530 | fprintf(outfile, "stack_free %p\n", block); | |
531 | free(block); | |
532 | } | |
533 | ||
534 | ||
535 | /************************************************* | |
536 | * Call pcre_fullinfo() * | |
537 | *************************************************/ | |
538 | ||
539 | /* Get one piece of information from the pcre_fullinfo() function */ | |
540 | ||
541 | static void new_info(pcre *re, pcre_extra *study, int option, void *ptr) | |
542 | { | |
543 | int rc; | |
544 | if ((rc = pcre_fullinfo(re, study, option, ptr)) < 0) | |
545 | fprintf(outfile, "Error %d from pcre_fullinfo(%d)\n", rc, option); | |
546 | } | |
547 | ||
548 | ||
549 | ||
550 | /************************************************* | |
551 | * Byte flipping function * | |
552 | *************************************************/ | |
553 | ||
aa41d2de PH |
554 | static unsigned long int |
555 | byteflip(unsigned long int value, int n) | |
c86f6258 PH |
556 | { |
557 | if (n == 2) return ((value & 0x00ff) << 8) | ((value & 0xff00) >> 8); | |
558 | return ((value & 0x000000ff) << 24) | | |
559 | ((value & 0x0000ff00) << 8) | | |
560 | ((value & 0x00ff0000) >> 8) | | |
561 | ((value & 0xff000000) >> 24); | |
562 | } | |
563 | ||
564 | ||
565 | ||
566 | ||
aa41d2de PH |
567 | /************************************************* |
568 | * Check match or recursion limit * | |
569 | *************************************************/ | |
570 | ||
571 | static int | |
572 | check_match_limit(pcre *re, pcre_extra *extra, uschar *bptr, int len, | |
573 | int start_offset, int options, int *use_offsets, int use_size_offsets, | |
574 | int flag, unsigned long int *limit, int errnumber, const char *msg) | |
575 | { | |
576 | int count; | |
577 | int min = 0; | |
578 | int mid = 64; | |
579 | int max = -1; | |
580 | ||
581 | extra->flags |= flag; | |
582 | ||
583 | for (;;) | |
584 | { | |
585 | *limit = mid; | |
586 | ||
587 | count = pcre_exec(re, extra, (char *)bptr, len, start_offset, options, | |
588 | use_offsets, use_size_offsets); | |
589 | ||
590 | if (count == errnumber) | |
591 | { | |
592 | /* fprintf(outfile, "Testing %s limit = %d\n", msg, mid); */ | |
593 | min = mid; | |
594 | mid = (mid == max - 1)? max : (max > 0)? (min + max)/2 : mid*2; | |
595 | } | |
596 | ||
597 | else if (count >= 0 || count == PCRE_ERROR_NOMATCH || | |
598 | count == PCRE_ERROR_PARTIAL) | |
599 | { | |
600 | if (mid == min + 1) | |
601 | { | |
602 | fprintf(outfile, "Minimum %s limit = %d\n", msg, mid); | |
603 | break; | |
604 | } | |
605 | /* fprintf(outfile, "Testing %s limit = %d\n", msg, mid); */ | |
606 | max = mid; | |
607 | mid = (min + mid)/2; | |
608 | } | |
609 | else break; /* Some other error */ | |
610 | } | |
611 | ||
612 | extra->flags &= ~flag; | |
613 | return count; | |
614 | } | |
615 | ||
616 | ||
617 | ||
618 | /************************************************* | |
619 | * Check newline indicator * | |
620 | *************************************************/ | |
621 | ||
622 | /* This is used both at compile and run-time to check for <xxx> escapes, where | |
623 | xxx is LF, CR, or CRLF. Print a message and return 0 if there is no match. | |
624 | ||
625 | Arguments: | |
626 | p points after the leading '<' | |
627 | f file for error message | |
628 | ||
629 | Returns: appropriate PCRE_NEWLINE_xxx flags, or 0 | |
630 | */ | |
631 | ||
632 | static int | |
633 | check_newline(uschar *p, FILE *f) | |
634 | { | |
635 | if (strncmp((char *)p, "cr>", 3) == 0) return PCRE_NEWLINE_CR; | |
636 | if (strncmp((char *)p, "lf>", 3) == 0) return PCRE_NEWLINE_LF; | |
637 | if (strncmp((char *)p, "crlf>", 5) == 0) return PCRE_NEWLINE_CRLF; | |
638 | fprintf(f, "Unknown newline type at: <%s\n", p); | |
639 | return 0; | |
640 | } | |
641 | ||
642 | ||
643 | ||
c86f6258 PH |
644 | /************************************************* |
645 | * Main Program * | |
646 | *************************************************/ | |
647 | ||
648 | /* Read lines from named file or stdin and write to named file or stdout; lines | |
649 | consist of a regular expression, in delimiters and optionally followed by | |
650 | options, followed by a set of test data, terminated by an empty line. */ | |
651 | ||
652 | int main(int argc, char **argv) | |
653 | { | |
654 | FILE *infile = stdin; | |
655 | int options = 0; | |
656 | int study_options = 0; | |
657 | int op = 1; | |
658 | int timeit = 0; | |
659 | int showinfo = 0; | |
660 | int showstore = 0; | |
aa41d2de | 661 | int quiet = 0; |
c86f6258 PH |
662 | int size_offsets = 45; |
663 | int size_offsets_max; | |
8ac170f3 | 664 | int *offsets = NULL; |
c86f6258 PH |
665 | #if !defined NOPOSIX |
666 | int posix = 0; | |
667 | #endif | |
668 | int debug = 0; | |
669 | int done = 0; | |
8ac170f3 PH |
670 | int all_use_dfa = 0; |
671 | int yield = 0; | |
aa41d2de PH |
672 | int stack_size; |
673 | ||
674 | /* These vectors store, end-to-end, a list of captured substring names. Assume | |
675 | that 1024 is plenty long enough for the few names we'll be testing. */ | |
676 | ||
677 | uschar copynames[1024]; | |
678 | uschar getnames[1024]; | |
c86f6258 | 679 | |
aa41d2de PH |
680 | uschar *copynamesptr; |
681 | uschar *getnamesptr; | |
c86f6258 PH |
682 | |
683 | /* Get buffers from malloc() so that Electric Fence will check their misuse | |
aa41d2de | 684 | when I am debugging. They grow automatically when very long lines are read. */ |
c86f6258 | 685 | |
aa41d2de PH |
686 | buffer = (unsigned char *)malloc(buffer_size); |
687 | dbuffer = (unsigned char *)malloc(buffer_size); | |
688 | pbuffer = (unsigned char *)malloc(buffer_size); | |
c86f6258 PH |
689 | |
690 | /* The outfile variable is static so that new_malloc can use it. The _setmode() | |
691 | stuff is some magic that I don't understand, but which apparently does good | |
692 | things in Windows. It's related to line terminations. */ | |
693 | ||
694 | #if defined(_WIN32) || defined(WIN32) | |
695 | _setmode( _fileno( stdout ), 0x8000 ); | |
696 | #endif /* defined(_WIN32) || defined(WIN32) */ | |
697 | ||
698 | outfile = stdout; | |
699 | ||
700 | /* Scan options */ | |
701 | ||
702 | while (argc > 1 && argv[op][0] == '-') | |
703 | { | |
704 | unsigned char *endptr; | |
705 | ||
706 | if (strcmp(argv[op], "-s") == 0 || strcmp(argv[op], "-m") == 0) | |
707 | showstore = 1; | |
708 | else if (strcmp(argv[op], "-t") == 0) timeit = 1; | |
aa41d2de | 709 | else if (strcmp(argv[op], "-q") == 0) quiet = 1; |
c86f6258 PH |
710 | else if (strcmp(argv[op], "-i") == 0) showinfo = 1; |
711 | else if (strcmp(argv[op], "-d") == 0) showinfo = debug = 1; | |
92e772ff | 712 | #if !defined NODFA |
8ac170f3 PH |
713 | else if (strcmp(argv[op], "-dfa") == 0) all_use_dfa = 1; |
714 | #endif | |
c86f6258 PH |
715 | else if (strcmp(argv[op], "-o") == 0 && argc > 2 && |
716 | ((size_offsets = get_value((unsigned char *)argv[op+1], &endptr)), | |
717 | *endptr == 0)) | |
718 | { | |
719 | op++; | |
720 | argc--; | |
721 | } | |
aa41d2de PH |
722 | else if (strcmp(argv[op], "-S") == 0 && argc > 2 && |
723 | ((stack_size = get_value((unsigned char *)argv[op+1], &endptr)), | |
724 | *endptr == 0)) | |
725 | { | |
726 | #ifdef _WIN32 | |
727 | printf("PCRE: -S not supported on this OS\n"); | |
728 | exit(1); | |
729 | #else | |
730 | int rc; | |
731 | struct rlimit rlim; | |
732 | getrlimit(RLIMIT_STACK, &rlim); | |
733 | rlim.rlim_cur = stack_size * 1024 * 1024; | |
734 | rc = setrlimit(RLIMIT_STACK, &rlim); | |
735 | if (rc != 0) | |
736 | { | |
737 | printf("PCRE: setrlimit() failed with error %d\n", rc); | |
738 | exit(1); | |
739 | } | |
740 | op++; | |
741 | argc--; | |
742 | #endif | |
743 | } | |
c86f6258 PH |
744 | #if !defined NOPOSIX |
745 | else if (strcmp(argv[op], "-p") == 0) posix = 1; | |
746 | #endif | |
747 | else if (strcmp(argv[op], "-C") == 0) | |
748 | { | |
749 | int rc; | |
750 | printf("PCRE version %s\n", pcre_version()); | |
751 | printf("Compiled with\n"); | |
752 | (void)pcre_config(PCRE_CONFIG_UTF8, &rc); | |
753 | printf(" %sUTF-8 support\n", rc? "" : "No "); | |
754 | (void)pcre_config(PCRE_CONFIG_UNICODE_PROPERTIES, &rc); | |
755 | printf(" %sUnicode properties support\n", rc? "" : "No "); | |
756 | (void)pcre_config(PCRE_CONFIG_NEWLINE, &rc); | |
aa41d2de PH |
757 | printf(" Newline sequence is %s\n", (rc == '\r')? "CR" : |
758 | (rc == '\n')? "LF" : "CRLF"); | |
c86f6258 PH |
759 | (void)pcre_config(PCRE_CONFIG_LINK_SIZE, &rc); |
760 | printf(" Internal link size = %d\n", rc); | |
761 | (void)pcre_config(PCRE_CONFIG_POSIX_MALLOC_THRESHOLD, &rc); | |
762 | printf(" POSIX malloc threshold = %d\n", rc); | |
763 | (void)pcre_config(PCRE_CONFIG_MATCH_LIMIT, &rc); | |
764 | printf(" Default match limit = %d\n", rc); | |
aa41d2de PH |
765 | (void)pcre_config(PCRE_CONFIG_MATCH_LIMIT_RECURSION, &rc); |
766 | printf(" Default recursion depth limit = %d\n", rc); | |
c86f6258 PH |
767 | (void)pcre_config(PCRE_CONFIG_STACKRECURSE, &rc); |
768 | printf(" Match recursion uses %s\n", rc? "stack" : "heap"); | |
769 | exit(0); | |
770 | } | |
771 | else | |
772 | { | |
773 | printf("** Unknown or malformed option %s\n", argv[op]); | |
aa41d2de | 774 | printf("Usage: pcretest [options] [<input> [<output>]]\n"); |
c86f6258 | 775 | printf(" -C show PCRE compile-time options and exit\n"); |
8ac170f3 PH |
776 | printf(" -d debug: show compiled code; implies -i\n"); |
777 | #if !defined NODFA | |
778 | printf(" -dfa force DFA matching for all subjects\n"); | |
779 | #endif | |
780 | printf(" -i show information about compiled pattern\n" | |
c86f6258 PH |
781 | " -m output memory used information\n" |
782 | " -o <n> set size of offsets vector to <n>\n"); | |
783 | #if !defined NOPOSIX | |
784 | printf(" -p use POSIX interface\n"); | |
785 | #endif | |
aa41d2de | 786 | printf(" -S <n> set stack size to <n> megabytes\n"); |
c86f6258 PH |
787 | printf(" -s output store (memory) used information\n" |
788 | " -t time compilation and execution\n"); | |
8ac170f3 PH |
789 | yield = 1; |
790 | goto EXIT; | |
c86f6258 PH |
791 | } |
792 | op++; | |
793 | argc--; | |
794 | } | |
795 | ||
796 | /* Get the store for the offsets vector, and remember what it was */ | |
797 | ||
798 | size_offsets_max = size_offsets; | |
799 | offsets = (int *)malloc(size_offsets_max * sizeof(int)); | |
800 | if (offsets == NULL) | |
801 | { | |
802 | printf("** Failed to get %d bytes of memory for offsets vector\n", | |
803 | size_offsets_max * sizeof(int)); | |
8ac170f3 PH |
804 | yield = 1; |
805 | goto EXIT; | |
c86f6258 PH |
806 | } |
807 | ||
808 | /* Sort out the input and output files */ | |
809 | ||
810 | if (argc > 1) | |
811 | { | |
812 | infile = fopen(argv[op], "rb"); | |
813 | if (infile == NULL) | |
814 | { | |
815 | printf("** Failed to open %s\n", argv[op]); | |
8ac170f3 PH |
816 | yield = 1; |
817 | goto EXIT; | |
c86f6258 PH |
818 | } |
819 | } | |
820 | ||
821 | if (argc > 2) | |
822 | { | |
823 | outfile = fopen(argv[op+1], "wb"); | |
824 | if (outfile == NULL) | |
825 | { | |
826 | printf("** Failed to open %s\n", argv[op+1]); | |
8ac170f3 PH |
827 | yield = 1; |
828 | goto EXIT; | |
c86f6258 PH |
829 | } |
830 | } | |
831 | ||
832 | /* Set alternative malloc function */ | |
833 | ||
834 | pcre_malloc = new_malloc; | |
835 | pcre_free = new_free; | |
836 | pcre_stack_malloc = stack_malloc; | |
837 | pcre_stack_free = stack_free; | |
838 | ||
aa41d2de | 839 | /* Heading line unless quiet, then prompt for first regex if stdin */ |
c86f6258 | 840 | |
aa41d2de | 841 | if (!quiet) fprintf(outfile, "PCRE version %s\n\n", pcre_version()); |
c86f6258 PH |
842 | |
843 | /* Main loop */ | |
844 | ||
845 | while (!done) | |
846 | { | |
847 | pcre *re = NULL; | |
848 | pcre_extra *extra = NULL; | |
849 | ||
850 | #if !defined NOPOSIX /* There are still compilers that require no indent */ | |
851 | regex_t preg; | |
852 | int do_posix = 0; | |
853 | #endif | |
854 | ||
855 | const char *error; | |
856 | unsigned char *p, *pp, *ppp; | |
857 | unsigned char *to_file = NULL; | |
858 | const unsigned char *tables = NULL; | |
859 | unsigned long int true_size, true_study_size = 0; | |
860 | size_t size, regex_gotten_store; | |
861 | int do_study = 0; | |
862 | int do_debug = debug; | |
863 | int do_G = 0; | |
864 | int do_g = 0; | |
865 | int do_showinfo = showinfo; | |
866 | int do_showrest = 0; | |
867 | int do_flip = 0; | |
868 | int erroroffset, len, delimiter; | |
869 | ||
870 | use_utf8 = 0; | |
871 | ||
872 | if (infile == stdin) printf(" re> "); | |
aa41d2de | 873 | if (extend_inputline(infile, buffer) == NULL) break; |
c86f6258 PH |
874 | if (infile != stdin) fprintf(outfile, "%s", (char *)buffer); |
875 | fflush(outfile); | |
876 | ||
877 | p = buffer; | |
878 | while (isspace(*p)) p++; | |
879 | if (*p == 0) continue; | |
880 | ||
881 | /* See if the pattern is to be loaded pre-compiled from a file. */ | |
882 | ||
883 | if (*p == '<' && strchr((char *)(p+1), '<') == NULL) | |
884 | { | |
aa41d2de | 885 | unsigned long int magic, get_options; |
c86f6258 PH |
886 | uschar sbuf[8]; |
887 | FILE *f; | |
888 | ||
889 | p++; | |
890 | pp = p + (int)strlen((char *)p); | |
891 | while (isspace(pp[-1])) pp--; | |
892 | *pp = 0; | |
893 | ||
894 | f = fopen((char *)p, "rb"); | |
895 | if (f == NULL) | |
896 | { | |
897 | fprintf(outfile, "Failed to open %s: %s\n", p, strerror(errno)); | |
898 | continue; | |
899 | } | |
900 | ||
901 | if (fread(sbuf, 1, 8, f) != 8) goto FAIL_READ; | |
902 | ||
903 | true_size = | |
904 | (sbuf[0] << 24) | (sbuf[1] << 16) | (sbuf[2] << 8) | sbuf[3]; | |
905 | true_study_size = | |
906 | (sbuf[4] << 24) | (sbuf[5] << 16) | (sbuf[6] << 8) | sbuf[7]; | |
907 | ||
908 | re = (real_pcre *)new_malloc(true_size); | |
909 | regex_gotten_store = gotten_store; | |
910 | ||
911 | if (fread(re, 1, true_size, f) != true_size) goto FAIL_READ; | |
912 | ||
913 | magic = ((real_pcre *)re)->magic_number; | |
914 | if (magic != MAGIC_NUMBER) | |
915 | { | |
916 | if (byteflip(magic, sizeof(magic)) == MAGIC_NUMBER) | |
917 | { | |
918 | do_flip = 1; | |
919 | } | |
920 | else | |
921 | { | |
922 | fprintf(outfile, "Data in %s is not a compiled PCRE regex\n", p); | |
923 | fclose(f); | |
924 | continue; | |
925 | } | |
926 | } | |
927 | ||
928 | fprintf(outfile, "Compiled regex%s loaded from %s\n", | |
929 | do_flip? " (byte-inverted)" : "", p); | |
930 | ||
931 | /* Need to know if UTF-8 for printing data strings */ | |
932 | ||
aa41d2de PH |
933 | new_info(re, NULL, PCRE_INFO_OPTIONS, &get_options); |
934 | use_utf8 = (get_options & PCRE_UTF8) != 0; | |
c86f6258 PH |
935 | |
936 | /* Now see if there is any following study data */ | |
937 | ||
938 | if (true_study_size != 0) | |
939 | { | |
940 | pcre_study_data *psd; | |
941 | ||
942 | extra = (pcre_extra *)new_malloc(sizeof(pcre_extra) + true_study_size); | |
943 | extra->flags = PCRE_EXTRA_STUDY_DATA; | |
944 | ||
945 | psd = (pcre_study_data *)(((char *)extra) + sizeof(pcre_extra)); | |
946 | extra->study_data = psd; | |
947 | ||
948 | if (fread(psd, 1, true_study_size, f) != true_study_size) | |
949 | { | |
950 | FAIL_READ: | |
951 | fprintf(outfile, "Failed to read data from %s\n", p); | |
952 | if (extra != NULL) new_free(extra); | |
953 | if (re != NULL) new_free(re); | |
954 | fclose(f); | |
955 | continue; | |
956 | } | |
957 | fprintf(outfile, "Study data loaded from %s\n", p); | |
958 | do_study = 1; /* To get the data output if requested */ | |
959 | } | |
960 | else fprintf(outfile, "No study data\n"); | |
961 | ||
962 | fclose(f); | |
963 | goto SHOW_INFO; | |
964 | } | |
965 | ||
966 | /* In-line pattern (the usual case). Get the delimiter and seek the end of | |
967 | the pattern; if is isn't complete, read more. */ | |
968 | ||
969 | delimiter = *p++; | |
970 | ||
971 | if (isalnum(delimiter) || delimiter == '\\') | |
972 | { | |
973 | fprintf(outfile, "** Delimiter must not be alphameric or \\\n"); | |
974 | goto SKIP_DATA; | |
975 | } | |
976 | ||
977 | pp = p; | |
978 | ||
979 | for(;;) | |
980 | { | |
981 | while (*pp != 0) | |
982 | { | |
983 | if (*pp == '\\' && pp[1] != 0) pp++; | |
984 | else if (*pp == delimiter) break; | |
985 | pp++; | |
986 | } | |
987 | if (*pp != 0) break; | |
c86f6258 | 988 | if (infile == stdin) printf(" > "); |
aa41d2de | 989 | if ((pp = extend_inputline(infile, pp)) == NULL) |
c86f6258 PH |
990 | { |
991 | fprintf(outfile, "** Unexpected EOF\n"); | |
992 | done = 1; | |
993 | goto CONTINUE; | |
994 | } | |
995 | if (infile != stdin) fprintf(outfile, "%s", (char *)pp); | |
996 | } | |
997 | ||
998 | /* If the first character after the delimiter is backslash, make | |
999 | the pattern end with backslash. This is purely to provide a way | |
1000 | of testing for the error message when a pattern ends with backslash. */ | |
1001 | ||
1002 | if (pp[1] == '\\') *pp++ = '\\'; | |
1003 | ||
1004 | /* Terminate the pattern at the delimiter, and save a copy of the pattern | |
1005 | for callouts. */ | |
1006 | ||
1007 | *pp++ = 0; | |
1008 | strcpy((char *)pbuffer, (char *)p); | |
1009 | ||
1010 | /* Look for options after final delimiter */ | |
1011 | ||
1012 | options = 0; | |
1013 | study_options = 0; | |
1014 | log_store = showstore; /* default from command line */ | |
1015 | ||
1016 | while (*pp != 0) | |
1017 | { | |
1018 | switch (*pp++) | |
1019 | { | |
8ac170f3 | 1020 | case 'f': options |= PCRE_FIRSTLINE; break; |
c86f6258 PH |
1021 | case 'g': do_g = 1; break; |
1022 | case 'i': options |= PCRE_CASELESS; break; | |
1023 | case 'm': options |= PCRE_MULTILINE; break; | |
1024 | case 's': options |= PCRE_DOTALL; break; | |
1025 | case 'x': options |= PCRE_EXTENDED; break; | |
1026 | ||
1027 | case '+': do_showrest = 1; break; | |
1028 | case 'A': options |= PCRE_ANCHORED; break; | |
1029 | case 'C': options |= PCRE_AUTO_CALLOUT; break; | |
1030 | case 'D': do_debug = do_showinfo = 1; break; | |
1031 | case 'E': options |= PCRE_DOLLAR_ENDONLY; break; | |
1032 | case 'F': do_flip = 1; break; | |
1033 | case 'G': do_G = 1; break; | |
1034 | case 'I': do_showinfo = 1; break; | |
aa41d2de | 1035 | case 'J': options |= PCRE_DUPNAMES; break; |
c86f6258 PH |
1036 | case 'M': log_store = 1; break; |
1037 | case 'N': options |= PCRE_NO_AUTO_CAPTURE; break; | |
1038 | ||
1039 | #if !defined NOPOSIX | |
1040 | case 'P': do_posix = 1; break; | |
1041 | #endif | |
1042 | ||
1043 | case 'S': do_study = 1; break; | |
1044 | case 'U': options |= PCRE_UNGREEDY; break; | |
1045 | case 'X': options |= PCRE_EXTRA; break; | |
1046 | case '8': options |= PCRE_UTF8; use_utf8 = 1; break; | |
1047 | case '?': options |= PCRE_NO_UTF8_CHECK; break; | |
1048 | ||
1049 | case 'L': | |
1050 | ppp = pp; | |
8ac170f3 PH |
1051 | /* The '\r' test here is so that it works on Windows */ |
1052 | while (*ppp != '\n' && *ppp != '\r' && *ppp != ' ') ppp++; | |
c86f6258 PH |
1053 | *ppp = 0; |
1054 | if (setlocale(LC_CTYPE, (const char *)pp) == NULL) | |
1055 | { | |
1056 | fprintf(outfile, "** Failed to set locale \"%s\"\n", pp); | |
1057 | goto SKIP_DATA; | |
1058 | } | |
1059 | tables = pcre_maketables(); | |
1060 | pp = ppp; | |
1061 | break; | |
1062 | ||
1063 | case '>': | |
1064 | to_file = pp; | |
1065 | while (*pp != 0) pp++; | |
1066 | while (isspace(pp[-1])) pp--; | |
1067 | *pp = 0; | |
1068 | break; | |
1069 | ||
aa41d2de PH |
1070 | case '<': |
1071 | { | |
1072 | int x = check_newline(pp, outfile); | |
1073 | if (x == 0) goto SKIP_DATA; | |
1074 | options |= x; | |
1075 | while (*pp++ != '>'); | |
1076 | } | |
1077 | break; | |
1078 | ||
8ac170f3 PH |
1079 | case '\r': /* So that it works in Windows */ |
1080 | case '\n': | |
1081 | case ' ': | |
1082 | break; | |
c86f6258 PH |
1083 | |
1084 | default: | |
1085 | fprintf(outfile, "** Unknown option '%c'\n", pp[-1]); | |
1086 | goto SKIP_DATA; | |
1087 | } | |
1088 | } | |
1089 | ||
1090 | /* Handle compiling via the POSIX interface, which doesn't support the | |
1091 | timing, showing, or debugging options, nor the ability to pass over | |
1092 | local character tables. */ | |
1093 | ||
1094 | #if !defined NOPOSIX | |
1095 | if (posix || do_posix) | |
1096 | { | |
1097 | int rc; | |
1098 | int cflags = 0; | |
1099 | ||
1100 | if ((options & PCRE_CASELESS) != 0) cflags |= REG_ICASE; | |
1101 | if ((options & PCRE_MULTILINE) != 0) cflags |= REG_NEWLINE; | |
8ac170f3 | 1102 | if ((options & PCRE_DOTALL) != 0) cflags |= REG_DOTALL; |
aa41d2de PH |
1103 | if ((options & PCRE_NO_AUTO_CAPTURE) != 0) cflags |= REG_NOSUB; |
1104 | if ((options & PCRE_UTF8) != 0) cflags |= REG_UTF8; | |
1105 | ||
c86f6258 PH |
1106 | rc = regcomp(&preg, (char *)p, cflags); |
1107 | ||
1108 | /* Compilation failed; go back for another re, skipping to blank line | |
1109 | if non-interactive. */ | |
1110 | ||
1111 | if (rc != 0) | |
1112 | { | |
aa41d2de | 1113 | (void)regerror(rc, &preg, (char *)buffer, buffer_size); |
c86f6258 PH |
1114 | fprintf(outfile, "Failed: POSIX code %d: %s\n", rc, buffer); |
1115 | goto SKIP_DATA; | |
1116 | } | |
1117 | } | |
1118 | ||
1119 | /* Handle compiling via the native interface */ | |
1120 | ||
1121 | else | |
1122 | #endif /* !defined NOPOSIX */ | |
1123 | ||
1124 | { | |
1125 | if (timeit) | |
1126 | { | |
1127 | register int i; | |
1128 | clock_t time_taken; | |
1129 | clock_t start_time = clock(); | |
1130 | for (i = 0; i < LOOPREPEAT; i++) | |
1131 | { | |
1132 | re = pcre_compile((char *)p, options, &error, &erroroffset, tables); | |
1133 | if (re != NULL) free(re); | |
1134 | } | |
1135 | time_taken = clock() - start_time; | |
1136 | fprintf(outfile, "Compile time %.3f milliseconds\n", | |
1137 | (((double)time_taken * 1000.0) / (double)LOOPREPEAT) / | |
1138 | (double)CLOCKS_PER_SEC); | |
1139 | } | |
1140 | ||
1141 | re = pcre_compile((char *)p, options, &error, &erroroffset, tables); | |
1142 | ||
1143 | /* Compilation failed; go back for another re, skipping to blank line | |
1144 | if non-interactive. */ | |
1145 | ||
1146 | if (re == NULL) | |
1147 | { | |
1148 | fprintf(outfile, "Failed: %s at offset %d\n", error, erroroffset); | |
1149 | SKIP_DATA: | |
1150 | if (infile != stdin) | |
1151 | { | |
1152 | for (;;) | |
1153 | { | |
aa41d2de | 1154 | if (extend_inputline(infile, buffer) == NULL) |
c86f6258 PH |
1155 | { |
1156 | done = 1; | |
1157 | goto CONTINUE; | |
1158 | } | |
1159 | len = (int)strlen((char *)buffer); | |
1160 | while (len > 0 && isspace(buffer[len-1])) len--; | |
1161 | if (len == 0) break; | |
1162 | } | |
1163 | fprintf(outfile, "\n"); | |
1164 | } | |
1165 | goto CONTINUE; | |
1166 | } | |
1167 | ||
1168 | /* Compilation succeeded; print data if required. There are now two | |
1169 | info-returning functions. The old one has a limited interface and | |
1170 | returns only limited data. Check that it agrees with the newer one. */ | |
1171 | ||
1172 | if (log_store) | |
1173 | fprintf(outfile, "Memory allocation (code space): %d\n", | |
1174 | (int)(gotten_store - | |
1175 | sizeof(real_pcre) - | |
1176 | ((real_pcre *)re)->name_count * ((real_pcre *)re)->name_entry_size)); | |
1177 | ||
1178 | /* Extract the size for possible writing before possibly flipping it, | |
1179 | and remember the store that was got. */ | |
1180 | ||
1181 | true_size = ((real_pcre *)re)->size; | |
1182 | regex_gotten_store = gotten_store; | |
1183 | ||
1184 | /* If /S was present, study the regexp to generate additional info to | |
1185 | help with the matching. */ | |
1186 | ||
1187 | if (do_study) | |
1188 | { | |
1189 | if (timeit) | |
1190 | { | |
1191 | register int i; | |
1192 | clock_t time_taken; | |
1193 | clock_t start_time = clock(); | |
1194 | for (i = 0; i < LOOPREPEAT; i++) | |
1195 | extra = pcre_study(re, study_options, &error); | |
1196 | time_taken = clock() - start_time; | |
1197 | if (extra != NULL) free(extra); | |
1198 | fprintf(outfile, " Study time %.3f milliseconds\n", | |
1199 | (((double)time_taken * 1000.0) / (double)LOOPREPEAT) / | |
1200 | (double)CLOCKS_PER_SEC); | |
1201 | } | |
1202 | extra = pcre_study(re, study_options, &error); | |
1203 | if (error != NULL) | |
1204 | fprintf(outfile, "Failed to study: %s\n", error); | |
1205 | else if (extra != NULL) | |
1206 | true_study_size = ((pcre_study_data *)(extra->study_data))->size; | |
1207 | } | |
1208 | ||
1209 | /* If the 'F' option was present, we flip the bytes of all the integer | |
1210 | fields in the regex data block and the study block. This is to make it | |
1211 | possible to test PCRE's handling of byte-flipped patterns, e.g. those | |
1212 | compiled on a different architecture. */ | |
1213 | ||
1214 | if (do_flip) | |
1215 | { | |
1216 | real_pcre *rre = (real_pcre *)re; | |
1217 | rre->magic_number = byteflip(rre->magic_number, sizeof(rre->magic_number)); | |
1218 | rre->size = byteflip(rre->size, sizeof(rre->size)); | |
1219 | rre->options = byteflip(rre->options, sizeof(rre->options)); | |
1220 | rre->top_bracket = byteflip(rre->top_bracket, sizeof(rre->top_bracket)); | |
1221 | rre->top_backref = byteflip(rre->top_backref, sizeof(rre->top_backref)); | |
1222 | rre->first_byte = byteflip(rre->first_byte, sizeof(rre->first_byte)); | |
1223 | rre->req_byte = byteflip(rre->req_byte, sizeof(rre->req_byte)); | |
1224 | rre->name_table_offset = byteflip(rre->name_table_offset, | |
1225 | sizeof(rre->name_table_offset)); | |
1226 | rre->name_entry_size = byteflip(rre->name_entry_size, | |
1227 | sizeof(rre->name_entry_size)); | |
1228 | rre->name_count = byteflip(rre->name_count, sizeof(rre->name_count)); | |
1229 | ||
1230 | if (extra != NULL) | |
1231 | { | |
1232 | pcre_study_data *rsd = (pcre_study_data *)(extra->study_data); | |
1233 | rsd->size = byteflip(rsd->size, sizeof(rsd->size)); | |
1234 | rsd->options = byteflip(rsd->options, sizeof(rsd->options)); | |
1235 | } | |
1236 | } | |
1237 | ||
1238 | /* Extract information from the compiled data if required */ | |
1239 | ||
1240 | SHOW_INFO: | |
1241 | ||
1242 | if (do_showinfo) | |
1243 | { | |
1244 | unsigned long int get_options, all_options; | |
8ac170f3 | 1245 | #if !defined NOINFOCHECK |
c86f6258 | 1246 | int old_first_char, old_options, old_count; |
8ac170f3 | 1247 | #endif |
c86f6258 PH |
1248 | int count, backrefmax, first_char, need_char; |
1249 | int nameentrysize, namecount; | |
1250 | const uschar *nametable; | |
1251 | ||
1252 | if (do_debug) | |
1253 | { | |
1254 | fprintf(outfile, "------------------------------------------------------------------\n"); | |
aa41d2de | 1255 | pcre_printint(re, outfile); |
c86f6258 PH |
1256 | } |
1257 | ||
1258 | new_info(re, NULL, PCRE_INFO_OPTIONS, &get_options); | |
1259 | new_info(re, NULL, PCRE_INFO_SIZE, &size); | |
1260 | new_info(re, NULL, PCRE_INFO_CAPTURECOUNT, &count); | |
1261 | new_info(re, NULL, PCRE_INFO_BACKREFMAX, &backrefmax); | |
1262 | new_info(re, NULL, PCRE_INFO_FIRSTBYTE, &first_char); | |
1263 | new_info(re, NULL, PCRE_INFO_LASTLITERAL, &need_char); | |
1264 | new_info(re, NULL, PCRE_INFO_NAMEENTRYSIZE, &nameentrysize); | |
1265 | new_info(re, NULL, PCRE_INFO_NAMECOUNT, &namecount); | |
1266 | new_info(re, NULL, PCRE_INFO_NAMETABLE, (void *)&nametable); | |
1267 | ||
8ac170f3 | 1268 | #if !defined NOINFOCHECK |
c86f6258 PH |
1269 | old_count = pcre_info(re, &old_options, &old_first_char); |
1270 | if (count < 0) fprintf(outfile, | |
1271 | "Error %d from pcre_info()\n", count); | |
1272 | else | |
1273 | { | |
1274 | if (old_count != count) fprintf(outfile, | |
1275 | "Count disagreement: pcre_fullinfo=%d pcre_info=%d\n", count, | |
1276 | old_count); | |
1277 | ||
1278 | if (old_first_char != first_char) fprintf(outfile, | |
1279 | "First char disagreement: pcre_fullinfo=%d pcre_info=%d\n", | |
1280 | first_char, old_first_char); | |
1281 | ||
1282 | if (old_options != (int)get_options) fprintf(outfile, | |
1283 | "Options disagreement: pcre_fullinfo=%ld pcre_info=%d\n", | |
1284 | get_options, old_options); | |
1285 | } | |
8ac170f3 | 1286 | #endif |
c86f6258 PH |
1287 | |
1288 | if (size != regex_gotten_store) fprintf(outfile, | |
1289 | "Size disagreement: pcre_fullinfo=%d call to malloc for %d\n", | |
1290 | (int)size, (int)regex_gotten_store); | |
1291 | ||
1292 | fprintf(outfile, "Capturing subpattern count = %d\n", count); | |
1293 | if (backrefmax > 0) | |
1294 | fprintf(outfile, "Max back reference = %d\n", backrefmax); | |
1295 | ||
1296 | if (namecount > 0) | |
1297 | { | |
1298 | fprintf(outfile, "Named capturing subpatterns:\n"); | |
1299 | while (namecount-- > 0) | |
1300 | { | |
1301 | fprintf(outfile, " %s %*s%3d\n", nametable + 2, | |
1302 | nameentrysize - 3 - (int)strlen((char *)nametable + 2), "", | |
1303 | GET2(nametable, 0)); | |
1304 | nametable += nameentrysize; | |
1305 | } | |
1306 | } | |
1307 | ||
1308 | /* The NOPARTIAL bit is a private bit in the options, so we have | |
1309 | to fish it out via out back door */ | |
1310 | ||
1311 | all_options = ((real_pcre *)re)->options; | |
1312 | if (do_flip) | |
1313 | { | |
1314 | all_options = byteflip(all_options, sizeof(all_options)); | |
aa41d2de | 1315 | } |
c86f6258 PH |
1316 | |
1317 | if ((all_options & PCRE_NOPARTIAL) != 0) | |
1318 | fprintf(outfile, "Partial matching not supported\n"); | |
1319 | ||
1320 | if (get_options == 0) fprintf(outfile, "No options\n"); | |
aa41d2de | 1321 | else fprintf(outfile, "Options:%s%s%s%s%s%s%s%s%s%s%s%s%s\n", |
c86f6258 PH |
1322 | ((get_options & PCRE_ANCHORED) != 0)? " anchored" : "", |
1323 | ((get_options & PCRE_CASELESS) != 0)? " caseless" : "", | |
1324 | ((get_options & PCRE_EXTENDED) != 0)? " extended" : "", | |
1325 | ((get_options & PCRE_MULTILINE) != 0)? " multiline" : "", | |
8ac170f3 | 1326 | ((get_options & PCRE_FIRSTLINE) != 0)? " firstline" : "", |
c86f6258 PH |
1327 | ((get_options & PCRE_DOTALL) != 0)? " dotall" : "", |
1328 | ((get_options & PCRE_DOLLAR_ENDONLY) != 0)? " dollar_endonly" : "", | |
1329 | ((get_options & PCRE_EXTRA) != 0)? " extra" : "", | |
1330 | ((get_options & PCRE_UNGREEDY) != 0)? " ungreedy" : "", | |
aa41d2de | 1331 | ((get_options & PCRE_NO_AUTO_CAPTURE) != 0)? " no_auto_capture" : "", |
c86f6258 | 1332 | ((get_options & PCRE_UTF8) != 0)? " utf8" : "", |
aa41d2de PH |
1333 | ((get_options & PCRE_NO_UTF8_CHECK) != 0)? " no_utf8_check" : "", |
1334 | ((get_options & PCRE_DUPNAMES) != 0)? " dupnames" : ""); | |
1335 | ||
1336 | switch (get_options & PCRE_NEWLINE_CRLF) | |
1337 | { | |
1338 | case PCRE_NEWLINE_CR: | |
1339 | fprintf(outfile, "Forced newline sequence: CR\n"); | |
1340 | break; | |
1341 | ||
1342 | case PCRE_NEWLINE_LF: | |
1343 | fprintf(outfile, "Forced newline sequence: LF\n"); | |
1344 | break; | |
c86f6258 | 1345 | |
aa41d2de PH |
1346 | case PCRE_NEWLINE_CRLF: |
1347 | fprintf(outfile, "Forced newline sequence: CRLF\n"); | |
1348 | break; | |
1349 | ||
1350 | default: | |
1351 | break; | |
1352 | } | |
c86f6258 PH |
1353 | |
1354 | if (first_char == -1) | |
1355 | { | |
aa41d2de | 1356 | fprintf(outfile, "First char at start or follows newline\n"); |
c86f6258 PH |
1357 | } |
1358 | else if (first_char < 0) | |
1359 | { | |
1360 | fprintf(outfile, "No first char\n"); | |
1361 | } | |
1362 | else | |
1363 | { | |
1364 | int ch = first_char & 255; | |
1365 | const char *caseless = ((first_char & REQ_CASELESS) == 0)? | |
1366 | "" : " (caseless)"; | |
1367 | if (isprint(ch)) | |
1368 | fprintf(outfile, "First char = \'%c\'%s\n", ch, caseless); | |
1369 | else | |
1370 | fprintf(outfile, "First char = %d%s\n", ch, caseless); | |
1371 | } | |
1372 | ||
1373 | if (need_char < 0) | |
1374 | { | |
1375 | fprintf(outfile, "No need char\n"); | |
1376 | } | |
1377 | else | |
1378 | { | |
1379 | int ch = need_char & 255; | |
1380 | const char *caseless = ((need_char & REQ_CASELESS) == 0)? | |
1381 | "" : " (caseless)"; | |
1382 | if (isprint(ch)) | |
1383 | fprintf(outfile, "Need char = \'%c\'%s\n", ch, caseless); | |
1384 | else | |
1385 | fprintf(outfile, "Need char = %d%s\n", ch, caseless); | |
1386 | } | |
1387 | ||
1388 | /* Don't output study size; at present it is in any case a fixed | |
1389 | value, but it varies, depending on the computer architecture, and | |
1390 | so messes up the test suite. (And with the /F option, it might be | |
1391 | flipped.) */ | |
1392 | ||
1393 | if (do_study) | |
1394 | { | |
1395 | if (extra == NULL) | |
1396 | fprintf(outfile, "Study returned NULL\n"); | |
1397 | else | |
1398 | { | |
1399 | uschar *start_bits = NULL; | |
1400 | new_info(re, extra, PCRE_INFO_FIRSTTABLE, &start_bits); | |
1401 | ||
1402 | if (start_bits == NULL) | |
1403 | fprintf(outfile, "No starting byte set\n"); | |
1404 | else | |
1405 | { | |
1406 | int i; | |
1407 | int c = 24; | |
1408 | fprintf(outfile, "Starting byte set: "); | |
1409 | for (i = 0; i < 256; i++) | |
1410 | { | |
1411 | if ((start_bits[i/8] & (1<<(i&7))) != 0) | |
1412 | { | |
1413 | if (c > 75) | |
1414 | { | |
1415 | fprintf(outfile, "\n "); | |
1416 | c = 2; | |
1417 | } | |
1418 | if (isprint(i) && i != ' ') | |
1419 | { | |
1420 | fprintf(outfile, "%c ", i); | |
1421 | c += 2; | |
1422 | } | |
1423 | else | |
1424 | { | |
1425 | fprintf(outfile, "\\x%02x ", i); | |
1426 | c += 5; | |
1427 | } | |
1428 | } | |
1429 | } | |
1430 | fprintf(outfile, "\n"); | |
1431 | } | |
1432 | } | |
1433 | } | |
1434 | } | |
1435 | ||
1436 | /* If the '>' option was present, we write out the regex to a file, and | |
1437 | that is all. The first 8 bytes of the file are the regex length and then | |
1438 | the study length, in big-endian order. */ | |
1439 | ||
1440 | if (to_file != NULL) | |
1441 | { | |
1442 | FILE *f = fopen((char *)to_file, "wb"); | |
1443 | if (f == NULL) | |
1444 | { | |
1445 | fprintf(outfile, "Unable to open %s: %s\n", to_file, strerror(errno)); | |
1446 | } | |
1447 | else | |
1448 | { | |
1449 | uschar sbuf[8]; | |
1450 | sbuf[0] = (true_size >> 24) & 255; | |
1451 | sbuf[1] = (true_size >> 16) & 255; | |
1452 | sbuf[2] = (true_size >> 8) & 255; | |
1453 | sbuf[3] = (true_size) & 255; | |
1454 | ||
1455 | sbuf[4] = (true_study_size >> 24) & 255; | |
1456 | sbuf[5] = (true_study_size >> 16) & 255; | |
1457 | sbuf[6] = (true_study_size >> 8) & 255; | |
1458 | sbuf[7] = (true_study_size) & 255; | |
1459 | ||
1460 | if (fwrite(sbuf, 1, 8, f) < 8 || | |
1461 | fwrite(re, 1, true_size, f) < true_size) | |
1462 | { | |
1463 | fprintf(outfile, "Write error on %s: %s\n", to_file, strerror(errno)); | |
1464 | } | |
1465 | else | |
1466 | { | |
1467 | fprintf(outfile, "Compiled regex written to %s\n", to_file); | |
1468 | if (extra != NULL) | |
1469 | { | |
1470 | if (fwrite(extra->study_data, 1, true_study_size, f) < | |
1471 | true_study_size) | |
1472 | { | |
1473 | fprintf(outfile, "Write error on %s: %s\n", to_file, | |
1474 | strerror(errno)); | |
1475 | } | |
1476 | else fprintf(outfile, "Study data written to %s\n", to_file); | |
1477 | } | |
1478 | } | |
1479 | fclose(f); | |
1480 | } | |
8ac170f3 PH |
1481 | |
1482 | new_free(re); | |
1483 | if (extra != NULL) new_free(extra); | |
1484 | if (tables != NULL) new_free((void *)tables); | |
c86f6258 PH |
1485 | continue; /* With next regex */ |
1486 | } | |
1487 | } /* End of non-POSIX compile */ | |
1488 | ||
1489 | /* Read data lines and test them */ | |
1490 | ||
1491 | for (;;) | |
1492 | { | |
aa41d2de PH |
1493 | uschar *q; |
1494 | uschar *bptr = dbuffer; | |
c86f6258 PH |
1495 | int *use_offsets = offsets; |
1496 | int use_size_offsets = size_offsets; | |
1497 | int callout_data = 0; | |
1498 | int callout_data_set = 0; | |
1499 | int count, c; | |
1500 | int copystrings = 0; | |
1501 | int find_match_limit = 0; | |
1502 | int getstrings = 0; | |
1503 | int getlist = 0; | |
1504 | int gmatched = 0; | |
1505 | int start_offset = 0; | |
1506 | int g_notempty = 0; | |
8ac170f3 | 1507 | int use_dfa = 0; |
c86f6258 PH |
1508 | |
1509 | options = 0; | |
1510 | ||
aa41d2de PH |
1511 | *copynames = 0; |
1512 | *getnames = 0; | |
1513 | ||
1514 | copynamesptr = copynames; | |
1515 | getnamesptr = getnames; | |
1516 | ||
c86f6258 PH |
1517 | pcre_callout = callout; |
1518 | first_callout = 1; | |
1519 | callout_extra = 0; | |
1520 | callout_count = 0; | |
1521 | callout_fail_count = 999999; | |
1522 | callout_fail_id = -1; | |
1523 | show_malloc = 0; | |
1524 | ||
aa41d2de PH |
1525 | if (extra != NULL) extra->flags &= |
1526 | ~(PCRE_EXTRA_MATCH_LIMIT|PCRE_EXTRA_MATCH_LIMIT_RECURSION); | |
1527 | ||
1528 | len = 0; | |
1529 | for (;;) | |
c86f6258 | 1530 | { |
aa41d2de PH |
1531 | if (infile == stdin) printf("data> "); |
1532 | if (extend_inputline(infile, buffer + len) == NULL) | |
1533 | { | |
1534 | if (len > 0) break; | |
1535 | done = 1; | |
1536 | goto CONTINUE; | |
1537 | } | |
1538 | if (infile != stdin) fprintf(outfile, "%s", (char *)buffer); | |
1539 | len = (int)strlen((char *)buffer); | |
1540 | if (buffer[len-1] == '\n') break; | |
c86f6258 | 1541 | } |
c86f6258 | 1542 | |
c86f6258 PH |
1543 | while (len > 0 && isspace(buffer[len-1])) len--; |
1544 | buffer[len] = 0; | |
1545 | if (len == 0) break; | |
1546 | ||
1547 | p = buffer; | |
1548 | while (isspace(*p)) p++; | |
1549 | ||
1550 | q = dbuffer; | |
1551 | while ((c = *p++) != 0) | |
1552 | { | |
1553 | int i = 0; | |
1554 | int n = 0; | |
1555 | ||
1556 | if (c == '\\') switch ((c = *p++)) | |
1557 | { | |
1558 | case 'a': c = 7; break; | |
1559 | case 'b': c = '\b'; break; | |
1560 | case 'e': c = 27; break; | |
1561 | case 'f': c = '\f'; break; | |
1562 | case 'n': c = '\n'; break; | |
1563 | case 'r': c = '\r'; break; | |
1564 | case 't': c = '\t'; break; | |
1565 | case 'v': c = '\v'; break; | |
1566 | ||
1567 | case '0': case '1': case '2': case '3': | |
1568 | case '4': case '5': case '6': case '7': | |
1569 | c -= '0'; | |
1570 | while (i++ < 2 && isdigit(*p) && *p != '8' && *p != '9') | |
1571 | c = c * 8 + *p++ - '0'; | |
aa41d2de PH |
1572 | |
1573 | #if !defined NOUTF8 | |
1574 | if (use_utf8 && c > 255) | |
1575 | { | |
1576 | unsigned char buff8[8]; | |
1577 | int ii, utn; | |
1578 | utn = ord2utf8(c, buff8); | |
1579 | for (ii = 0; ii < utn - 1; ii++) *q++ = buff8[ii]; | |
1580 | c = buff8[ii]; /* Last byte */ | |
1581 | } | |
1582 | #endif | |
c86f6258 PH |
1583 | break; |
1584 | ||
1585 | case 'x': | |
1586 | ||
1587 | /* Handle \x{..} specially - new Perl thing for utf8 */ | |
1588 | ||
8ac170f3 | 1589 | #if !defined NOUTF8 |
c86f6258 PH |
1590 | if (*p == '{') |
1591 | { | |
1592 | unsigned char *pt = p; | |
1593 | c = 0; | |
1594 | while (isxdigit(*(++pt))) | |
1595 | c = c * 16 + tolower(*pt) - ((isdigit(*pt))? '0' : 'W'); | |
1596 | if (*pt == '}') | |
1597 | { | |
1598 | unsigned char buff8[8]; | |
1599 | int ii, utn; | |
aa41d2de | 1600 | utn = ord2utf8(c, buff8); |
c86f6258 PH |
1601 | for (ii = 0; ii < utn - 1; ii++) *q++ = buff8[ii]; |
1602 | c = buff8[ii]; /* Last byte */ | |
1603 | p = pt + 1; | |
1604 | break; | |
1605 | } | |
1606 | /* Not correct form; fall through */ | |
1607 | } | |
8ac170f3 | 1608 | #endif |
c86f6258 PH |
1609 | |
1610 | /* Ordinary \x */ | |
1611 | ||
1612 | c = 0; | |
1613 | while (i++ < 2 && isxdigit(*p)) | |
1614 | { | |
1615 | c = c * 16 + tolower(*p) - ((isdigit(*p))? '0' : 'W'); | |
1616 | p++; | |
1617 | } | |
1618 | break; | |
1619 | ||
1620 | case 0: /* \ followed by EOF allows for an empty line */ | |
1621 | p--; | |
1622 | continue; | |
1623 | ||
1624 | case '>': | |
1625 | while(isdigit(*p)) start_offset = start_offset * 10 + *p++ - '0'; | |
1626 | continue; | |
1627 | ||
1628 | case 'A': /* Option setting */ | |
1629 | options |= PCRE_ANCHORED; | |
1630 | continue; | |
1631 | ||
1632 | case 'B': | |
1633 | options |= PCRE_NOTBOL; | |
1634 | continue; | |
1635 | ||
1636 | case 'C': | |
1637 | if (isdigit(*p)) /* Set copy string */ | |
1638 | { | |
1639 | while(isdigit(*p)) n = n * 10 + *p++ - '0'; | |
1640 | copystrings |= 1 << n; | |
1641 | } | |
1642 | else if (isalnum(*p)) | |
1643 | { | |
aa41d2de | 1644 | uschar *npp = copynamesptr; |
c86f6258 | 1645 | while (isalnum(*p)) *npp++ = *p++; |
aa41d2de | 1646 | *npp++ = 0; |
c86f6258 | 1647 | *npp = 0; |
aa41d2de | 1648 | n = pcre_get_stringnumber(re, (char *)copynamesptr); |
c86f6258 | 1649 | if (n < 0) |
aa41d2de PH |
1650 | fprintf(outfile, "no parentheses with name \"%s\"\n", copynamesptr); |
1651 | copynamesptr = npp; | |
c86f6258 PH |
1652 | } |
1653 | else if (*p == '+') | |
1654 | { | |
1655 | callout_extra = 1; | |
1656 | p++; | |
1657 | } | |
1658 | else if (*p == '-') | |
1659 | { | |
1660 | pcre_callout = NULL; | |
1661 | p++; | |
1662 | } | |
1663 | else if (*p == '!') | |
1664 | { | |
1665 | callout_fail_id = 0; | |
1666 | p++; | |
1667 | while(isdigit(*p)) | |
1668 | callout_fail_id = callout_fail_id * 10 + *p++ - '0'; | |
1669 | callout_fail_count = 0; | |
1670 | if (*p == '!') | |
1671 | { | |
1672 | p++; | |
1673 | while(isdigit(*p)) | |
1674 | callout_fail_count = callout_fail_count * 10 + *p++ - '0'; | |
1675 | } | |
1676 | } | |
1677 | else if (*p == '*') | |
1678 | { | |
1679 | int sign = 1; | |
1680 | callout_data = 0; | |
1681 | if (*(++p) == '-') { sign = -1; p++; } | |
1682 | while(isdigit(*p)) | |
1683 | callout_data = callout_data * 10 + *p++ - '0'; | |
1684 | callout_data *= sign; | |
1685 | callout_data_set = 1; | |
1686 | } | |
1687 | continue; | |
1688 | ||
8ac170f3 PH |
1689 | #if !defined NODFA |
1690 | case 'D': | |
1691 | #if !defined NOPOSIX | |
1692 | if (posix || do_posix) | |
1693 | printf("** Can't use dfa matching in POSIX mode: \\D ignored\n"); | |
1694 | else | |
1695 | #endif | |
1696 | use_dfa = 1; | |
1697 | continue; | |
1698 | ||
1699 | case 'F': | |
1700 | options |= PCRE_DFA_SHORTEST; | |
1701 | continue; | |
1702 | #endif | |
1703 | ||
c86f6258 PH |
1704 | case 'G': |
1705 | if (isdigit(*p)) | |
1706 | { | |
1707 | while(isdigit(*p)) n = n * 10 + *p++ - '0'; | |
1708 | getstrings |= 1 << n; | |
1709 | } | |
1710 | else if (isalnum(*p)) | |
1711 | { | |
aa41d2de | 1712 | uschar *npp = getnamesptr; |
c86f6258 | 1713 | while (isalnum(*p)) *npp++ = *p++; |
aa41d2de | 1714 | *npp++ = 0; |
c86f6258 | 1715 | *npp = 0; |
aa41d2de | 1716 | n = pcre_get_stringnumber(re, (char *)getnamesptr); |
c86f6258 | 1717 | if (n < 0) |
aa41d2de PH |
1718 | fprintf(outfile, "no parentheses with name \"%s\"\n", getnamesptr); |
1719 | getnamesptr = npp; | |
c86f6258 PH |
1720 | } |
1721 | continue; | |
1722 | ||
1723 | case 'L': | |
1724 | getlist = 1; | |
1725 | continue; | |
1726 | ||
1727 | case 'M': | |
1728 | find_match_limit = 1; | |
1729 | continue; | |
1730 | ||
1731 | case 'N': | |
1732 | options |= PCRE_NOTEMPTY; | |
1733 | continue; | |
1734 | ||
1735 | case 'O': | |
1736 | while(isdigit(*p)) n = n * 10 + *p++ - '0'; | |
1737 | if (n > size_offsets_max) | |
1738 | { | |
1739 | size_offsets_max = n; | |
1740 | free(offsets); | |
1741 | use_offsets = offsets = (int *)malloc(size_offsets_max * sizeof(int)); | |
1742 | if (offsets == NULL) | |
1743 | { | |
1744 | printf("** Failed to get %d bytes of memory for offsets vector\n", | |
1745 | size_offsets_max * sizeof(int)); | |
8ac170f3 PH |
1746 | yield = 1; |
1747 | goto EXIT; | |
c86f6258 PH |
1748 | } |
1749 | } | |
1750 | use_size_offsets = n; | |
1751 | if (n == 0) use_offsets = NULL; /* Ensures it can't write to it */ | |
1752 | continue; | |
1753 | ||
1754 | case 'P': | |
1755 | options |= PCRE_PARTIAL; | |
1756 | continue; | |
1757 | ||
aa41d2de PH |
1758 | case 'Q': |
1759 | while(isdigit(*p)) n = n * 10 + *p++ - '0'; | |
1760 | if (extra == NULL) | |
1761 | { | |
1762 | extra = (pcre_extra *)malloc(sizeof(pcre_extra)); | |
1763 | extra->flags = 0; | |
1764 | } | |
1765 | extra->flags |= PCRE_EXTRA_MATCH_LIMIT_RECURSION; | |
1766 | extra->match_limit_recursion = n; | |
1767 | continue; | |
1768 | ||
1769 | case 'q': | |
1770 | while(isdigit(*p)) n = n * 10 + *p++ - '0'; | |
1771 | if (extra == NULL) | |
1772 | { | |
1773 | extra = (pcre_extra *)malloc(sizeof(pcre_extra)); | |
1774 | extra->flags = 0; | |
1775 | } | |
1776 | extra->flags |= PCRE_EXTRA_MATCH_LIMIT; | |
1777 | extra->match_limit = n; | |
1778 | continue; | |
1779 | ||
8ac170f3 PH |
1780 | #if !defined NODFA |
1781 | case 'R': | |
1782 | options |= PCRE_DFA_RESTART; | |
1783 | continue; | |
1784 | #endif | |
1785 | ||
c86f6258 PH |
1786 | case 'S': |
1787 | show_malloc = 1; | |
1788 | continue; | |
1789 | ||
1790 | case 'Z': | |
1791 | options |= PCRE_NOTEOL; | |
1792 | continue; | |
1793 | ||
1794 | case '?': | |
1795 | options |= PCRE_NO_UTF8_CHECK; | |
1796 | continue; | |
aa41d2de PH |
1797 | |
1798 | case '<': | |
1799 | { | |
1800 | int x = check_newline(p, outfile); | |
1801 | if (x == 0) goto NEXT_DATA; | |
1802 | options |= x; | |
1803 | while (*p++ != '>'); | |
1804 | } | |
1805 | continue; | |
c86f6258 PH |
1806 | } |
1807 | *q++ = c; | |
1808 | } | |
1809 | *q = 0; | |
1810 | len = q - dbuffer; | |
1811 | ||
8ac170f3 PH |
1812 | if ((all_use_dfa || use_dfa) && find_match_limit) |
1813 | { | |
1814 | printf("**Match limit not relevant for DFA matching: ignored\n"); | |
1815 | find_match_limit = 0; | |
1816 | } | |
1817 | ||
c86f6258 PH |
1818 | /* Handle matching via the POSIX interface, which does not |
1819 | support timing or playing with the match limit or callout data. */ | |
1820 | ||
1821 | #if !defined NOPOSIX | |
1822 | if (posix || do_posix) | |
1823 | { | |
1824 | int rc; | |
1825 | int eflags = 0; | |
1826 | regmatch_t *pmatch = NULL; | |
1827 | if (use_size_offsets > 0) | |
1828 | pmatch = (regmatch_t *)malloc(sizeof(regmatch_t) * use_size_offsets); | |
1829 | if ((options & PCRE_NOTBOL) != 0) eflags |= REG_NOTBOL; | |
1830 | if ((options & PCRE_NOTEOL) != 0) eflags |= REG_NOTEOL; | |
1831 | ||
1832 | rc = regexec(&preg, (const char *)bptr, use_size_offsets, pmatch, eflags); | |
1833 | ||
1834 | if (rc != 0) | |
1835 | { | |
aa41d2de | 1836 | (void)regerror(rc, &preg, (char *)buffer, buffer_size); |
c86f6258 PH |
1837 | fprintf(outfile, "No match: POSIX code %d: %s\n", rc, buffer); |
1838 | } | |
aa41d2de PH |
1839 | else if ((((const pcre *)preg.re_pcre)->options & PCRE_NO_AUTO_CAPTURE) |
1840 | != 0) | |
1841 | { | |
1842 | fprintf(outfile, "Matched with REG_NOSUB\n"); | |
1843 | } | |
c86f6258 PH |
1844 | else |
1845 | { | |
1846 | size_t i; | |
1847 | for (i = 0; i < (size_t)use_size_offsets; i++) | |
1848 | { | |
1849 | if (pmatch[i].rm_so >= 0) | |
1850 | { | |
1851 | fprintf(outfile, "%2d: ", (int)i); | |
1852 | (void)pchars(dbuffer + pmatch[i].rm_so, | |
1853 | pmatch[i].rm_eo - pmatch[i].rm_so, outfile); | |
1854 | fprintf(outfile, "\n"); | |
1855 | if (i == 0 && do_showrest) | |
1856 | { | |
1857 | fprintf(outfile, " 0+ "); | |
1858 | (void)pchars(dbuffer + pmatch[i].rm_eo, len - pmatch[i].rm_eo, | |
1859 | outfile); | |
1860 | fprintf(outfile, "\n"); | |
1861 | } | |
1862 | } | |
1863 | } | |
1864 | } | |
1865 | free(pmatch); | |
1866 | } | |
1867 | ||
1868 | /* Handle matching via the native interface - repeats for /g and /G */ | |
1869 | ||
1870 | else | |
1871 | #endif /* !defined NOPOSIX */ | |
1872 | ||
1873 | for (;; gmatched++) /* Loop for /g or /G */ | |
1874 | { | |
1875 | if (timeit) | |
1876 | { | |
1877 | register int i; | |
1878 | clock_t time_taken; | |
1879 | clock_t start_time = clock(); | |
8ac170f3 PH |
1880 | |
1881 | #if !defined NODFA | |
1882 | if (all_use_dfa || use_dfa) | |
1883 | { | |
1884 | int workspace[1000]; | |
1885 | for (i = 0; i < LOOPREPEAT; i++) | |
1886 | count = pcre_dfa_exec(re, NULL, (char *)bptr, len, start_offset, | |
1887 | options | g_notempty, use_offsets, use_size_offsets, workspace, | |
1888 | sizeof(workspace)/sizeof(int)); | |
1889 | } | |
1890 | else | |
1891 | #endif | |
1892 | ||
c86f6258 PH |
1893 | for (i = 0; i < LOOPREPEAT; i++) |
1894 | count = pcre_exec(re, extra, (char *)bptr, len, | |
1895 | start_offset, options | g_notempty, use_offsets, use_size_offsets); | |
8ac170f3 | 1896 | |
c86f6258 PH |
1897 | time_taken = clock() - start_time; |
1898 | fprintf(outfile, "Execute time %.3f milliseconds\n", | |
1899 | (((double)time_taken * 1000.0) / (double)LOOPREPEAT) / | |
1900 | (double)CLOCKS_PER_SEC); | |
1901 | } | |
1902 | ||
1903 | /* If find_match_limit is set, we want to do repeated matches with | |
aa41d2de PH |
1904 | varying limits in order to find the minimum value for the match limit and |
1905 | for the recursion limit. */ | |
c86f6258 PH |
1906 | |
1907 | if (find_match_limit) | |
1908 | { | |
c86f6258 PH |
1909 | if (extra == NULL) |
1910 | { | |
1911 | extra = (pcre_extra *)malloc(sizeof(pcre_extra)); | |
1912 | extra->flags = 0; | |
1913 | } | |
c86f6258 | 1914 | |
aa41d2de PH |
1915 | (void)check_match_limit(re, extra, bptr, len, start_offset, |
1916 | options|g_notempty, use_offsets, use_size_offsets, | |
1917 | PCRE_EXTRA_MATCH_LIMIT, &(extra->match_limit), | |
1918 | PCRE_ERROR_MATCHLIMIT, "match()"); | |
c86f6258 | 1919 | |
aa41d2de PH |
1920 | count = check_match_limit(re, extra, bptr, len, start_offset, |
1921 | options|g_notempty, use_offsets, use_size_offsets, | |
1922 | PCRE_EXTRA_MATCH_LIMIT_RECURSION, &(extra->match_limit_recursion), | |
1923 | PCRE_ERROR_RECURSIONLIMIT, "match() recursion"); | |
c86f6258 PH |
1924 | } |
1925 | ||
1926 | /* If callout_data is set, use the interface with additional data */ | |
1927 | ||
1928 | else if (callout_data_set) | |
1929 | { | |
1930 | if (extra == NULL) | |
1931 | { | |
1932 | extra = (pcre_extra *)malloc(sizeof(pcre_extra)); | |
1933 | extra->flags = 0; | |
1934 | } | |
1935 | extra->flags |= PCRE_EXTRA_CALLOUT_DATA; | |
1936 | extra->callout_data = &callout_data; | |
1937 | count = pcre_exec(re, extra, (char *)bptr, len, start_offset, | |
1938 | options | g_notempty, use_offsets, use_size_offsets); | |
1939 | extra->flags &= ~PCRE_EXTRA_CALLOUT_DATA; | |
1940 | } | |
1941 | ||
1942 | /* The normal case is just to do the match once, with the default | |
1943 | value of match_limit. */ | |
1944 | ||
8ac170f3 PH |
1945 | #if !defined NODFA |
1946 | else if (all_use_dfa || use_dfa) | |
c86f6258 | 1947 | { |
8ac170f3 PH |
1948 | int workspace[1000]; |
1949 | count = pcre_dfa_exec(re, NULL, (char *)bptr, len, start_offset, | |
1950 | options | g_notempty, use_offsets, use_size_offsets, workspace, | |
1951 | sizeof(workspace)/sizeof(int)); | |
1952 | if (count == 0) | |
1953 | { | |
1954 | fprintf(outfile, "Matched, but too many subsidiary matches\n"); | |
1955 | count = use_size_offsets/2; | |
1956 | } | |
c86f6258 | 1957 | } |
8ac170f3 | 1958 | #endif |
c86f6258 | 1959 | |
8ac170f3 | 1960 | else |
c86f6258 | 1961 | { |
8ac170f3 PH |
1962 | count = pcre_exec(re, extra, (char *)bptr, len, |
1963 | start_offset, options | g_notempty, use_offsets, use_size_offsets); | |
1964 | if (count == 0) | |
1965 | { | |
1966 | fprintf(outfile, "Matched, but too many substrings\n"); | |
1967 | count = use_size_offsets/3; | |
1968 | } | |
c86f6258 PH |
1969 | } |
1970 | ||
1971 | /* Matched */ | |
1972 | ||
1973 | if (count >= 0) | |
1974 | { | |
1975 | int i; | |
1976 | for (i = 0; i < count * 2; i += 2) | |
1977 | { | |
1978 | if (use_offsets[i] < 0) | |
1979 | fprintf(outfile, "%2d: <unset>\n", i/2); | |
1980 | else | |
1981 | { | |
1982 | fprintf(outfile, "%2d: ", i/2); | |
1983 | (void)pchars(bptr + use_offsets[i], | |
1984 | use_offsets[i+1] - use_offsets[i], outfile); | |
1985 | fprintf(outfile, "\n"); | |
1986 | if (i == 0) | |
1987 | { | |
1988 | if (do_showrest) | |
1989 | { | |
1990 | fprintf(outfile, " 0+ "); | |
1991 | (void)pchars(bptr + use_offsets[i+1], len - use_offsets[i+1], | |
1992 | outfile); | |
1993 | fprintf(outfile, "\n"); | |
1994 | } | |
1995 | } | |
1996 | } | |
1997 | } | |
1998 | ||
1999 | for (i = 0; i < 32; i++) | |
2000 | { | |
2001 | if ((copystrings & (1 << i)) != 0) | |
2002 | { | |
aa41d2de | 2003 | char copybuffer[256]; |
c86f6258 PH |
2004 | int rc = pcre_copy_substring((char *)bptr, use_offsets, count, |
2005 | i, copybuffer, sizeof(copybuffer)); | |
2006 | if (rc < 0) | |
2007 | fprintf(outfile, "copy substring %d failed %d\n", i, rc); | |
2008 | else | |
2009 | fprintf(outfile, "%2dC %s (%d)\n", i, copybuffer, rc); | |
2010 | } | |
2011 | } | |
2012 | ||
aa41d2de PH |
2013 | for (copynamesptr = copynames; |
2014 | *copynamesptr != 0; | |
2015 | copynamesptr += (int)strlen((char*)copynamesptr) + 1) | |
2016 | { | |
2017 | char copybuffer[256]; | |
2018 | int rc = pcre_copy_named_substring(re, (char *)bptr, use_offsets, | |
2019 | count, (char *)copynamesptr, copybuffer, sizeof(copybuffer)); | |
2020 | if (rc < 0) | |
2021 | fprintf(outfile, "copy substring %s failed %d\n", copynamesptr, rc); | |
2022 | else | |
2023 | fprintf(outfile, " C %s (%d) %s\n", copybuffer, rc, copynamesptr); | |
2024 | } | |
2025 | ||
c86f6258 PH |
2026 | for (i = 0; i < 32; i++) |
2027 | { | |
2028 | if ((getstrings & (1 << i)) != 0) | |
2029 | { | |
2030 | const char *substring; | |
2031 | int rc = pcre_get_substring((char *)bptr, use_offsets, count, | |
2032 | i, &substring); | |
2033 | if (rc < 0) | |
2034 | fprintf(outfile, "get substring %d failed %d\n", i, rc); | |
2035 | else | |
2036 | { | |
2037 | fprintf(outfile, "%2dG %s (%d)\n", i, substring, rc); | |
c86f6258 PH |
2038 | pcre_free_substring(substring); |
2039 | } | |
2040 | } | |
2041 | } | |
2042 | ||
aa41d2de PH |
2043 | for (getnamesptr = getnames; |
2044 | *getnamesptr != 0; | |
2045 | getnamesptr += (int)strlen((char*)getnamesptr) + 1) | |
2046 | { | |
2047 | const char *substring; | |
2048 | int rc = pcre_get_named_substring(re, (char *)bptr, use_offsets, | |
2049 | count, (char *)getnamesptr, &substring); | |
2050 | if (rc < 0) | |
2051 | fprintf(outfile, "copy substring %s failed %d\n", getnamesptr, rc); | |
2052 | else | |
2053 | { | |
2054 | fprintf(outfile, " G %s (%d) %s\n", substring, rc, getnamesptr); | |
2055 | pcre_free_substring(substring); | |
2056 | } | |
2057 | } | |
2058 | ||
c86f6258 PH |
2059 | if (getlist) |
2060 | { | |
2061 | const char **stringlist; | |
2062 | int rc = pcre_get_substring_list((char *)bptr, use_offsets, count, | |
2063 | &stringlist); | |
2064 | if (rc < 0) | |
2065 | fprintf(outfile, "get substring list failed %d\n", rc); | |
2066 | else | |
2067 | { | |
2068 | for (i = 0; i < count; i++) | |
2069 | fprintf(outfile, "%2dL %s\n", i, stringlist[i]); | |
2070 | if (stringlist[i] != NULL) | |
2071 | fprintf(outfile, "string list not terminated by NULL\n"); | |
2072 | /* free((void *)stringlist); */ | |
2073 | pcre_free_substring_list(stringlist); | |
2074 | } | |
2075 | } | |
2076 | } | |
2077 | ||
2078 | /* There was a partial match */ | |
2079 | ||
2080 | else if (count == PCRE_ERROR_PARTIAL) | |
2081 | { | |
8ac170f3 PH |
2082 | fprintf(outfile, "Partial match"); |
2083 | #if !defined NODFA | |
2084 | if ((all_use_dfa || use_dfa) && use_size_offsets > 2) | |
2085 | fprintf(outfile, ": %.*s", use_offsets[1] - use_offsets[0], | |
2086 | bptr + use_offsets[0]); | |
2087 | #endif | |
2088 | fprintf(outfile, "\n"); | |
c86f6258 PH |
2089 | break; /* Out of the /g loop */ |
2090 | } | |
2091 | ||
2092 | /* Failed to match. If this is a /g or /G loop and we previously set | |
2093 | g_notempty after a null match, this is not necessarily the end. | |
2094 | We want to advance the start offset, and continue. In the case of UTF-8 | |
2095 | matching, the advance must be one character, not one byte. Fudge the | |
2096 | offset values to achieve this. We won't be at the end of the string - | |
2097 | that was checked before setting g_notempty. */ | |
2098 | ||
2099 | else | |
2100 | { | |
2101 | if (g_notempty != 0) | |
2102 | { | |
2103 | int onechar = 1; | |
2104 | use_offsets[0] = start_offset; | |
2105 | if (use_utf8) | |
2106 | { | |
2107 | while (start_offset + onechar < len) | |
2108 | { | |
2109 | int tb = bptr[start_offset+onechar]; | |
2110 | if (tb <= 127) break; | |
2111 | tb &= 0xc0; | |
2112 | if (tb != 0 && tb != 0xc0) onechar++; | |
2113 | } | |
2114 | } | |
2115 | use_offsets[1] = start_offset + onechar; | |
2116 | } | |
2117 | else | |
2118 | { | |
2119 | if (count == PCRE_ERROR_NOMATCH) | |
2120 | { | |
2121 | if (gmatched == 0) fprintf(outfile, "No match\n"); | |
2122 | } | |
2123 | else fprintf(outfile, "Error %d\n", count); | |
2124 | break; /* Out of the /g loop */ | |
2125 | } | |
2126 | } | |
2127 | ||
2128 | /* If not /g or /G we are done */ | |
2129 | ||
2130 | if (!do_g && !do_G) break; | |
2131 | ||
2132 | /* If we have matched an empty string, first check to see if we are at | |
2133 | the end of the subject. If so, the /g loop is over. Otherwise, mimic | |
2134 | what Perl's /g options does. This turns out to be rather cunning. First | |
2135 | we set PCRE_NOTEMPTY and PCRE_ANCHORED and try the match again at the | |
2136 | same point. If this fails (picked up above) we advance to the next | |
2137 | character. */ | |
2138 | ||
2139 | g_notempty = 0; | |
2140 | if (use_offsets[0] == use_offsets[1]) | |
2141 | { | |
2142 | if (use_offsets[0] == len) break; | |
2143 | g_notempty = PCRE_NOTEMPTY | PCRE_ANCHORED; | |
2144 | } | |
2145 | ||
2146 | /* For /g, update the start offset, leaving the rest alone */ | |
2147 | ||
2148 | if (do_g) start_offset = use_offsets[1]; | |
2149 | ||
2150 | /* For /G, update the pointer and length */ | |
2151 | ||
2152 | else | |
2153 | { | |
2154 | bptr += use_offsets[1]; | |
2155 | len -= use_offsets[1]; | |
2156 | } | |
2157 | } /* End of loop for /g and /G */ | |
aa41d2de PH |
2158 | |
2159 | NEXT_DATA: continue; | |
c86f6258 PH |
2160 | } /* End of loop for data lines */ |
2161 | ||
2162 | CONTINUE: | |
2163 | ||
2164 | #if !defined NOPOSIX | |
2165 | if (posix || do_posix) regfree(&preg); | |
2166 | #endif | |
2167 | ||
8ac170f3 PH |
2168 | if (re != NULL) new_free(re); |
2169 | if (extra != NULL) new_free(extra); | |
c86f6258 PH |
2170 | if (tables != NULL) |
2171 | { | |
8ac170f3 | 2172 | new_free((void *)tables); |
c86f6258 PH |
2173 | setlocale(LC_CTYPE, "C"); |
2174 | } | |
2175 | } | |
2176 | ||
2177 | if (infile == stdin) fprintf(outfile, "\n"); | |
8ac170f3 PH |
2178 | |
2179 | EXIT: | |
2180 | ||
2181 | if (infile != NULL && infile != stdin) fclose(infile); | |
2182 | if (outfile != NULL && outfile != stdout) fclose(outfile); | |
2183 | ||
2184 | free(buffer); | |
2185 | free(dbuffer); | |
2186 | free(pbuffer); | |
2187 | free(offsets); | |
2188 | ||
2189 | return yield; | |
c86f6258 PH |
2190 | } |
2191 | ||
8ac170f3 | 2192 | /* End of pcretest.c */ |