Commit | Line | Data |
---|---|---|
6bf342e1 | 1 | /* $Cambridge: exim/src/src/pcre/pcretest.c,v 1.6 2007/01/23 15:08:45 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 | ||
6bf342e1 PH |
49 | |
50 | /* A number of things vary for Windows builds. Originally, pcretest opened its | |
51 | input and output without "b"; then I was told that "b" was needed in some | |
52 | environments, so it was added for release 5.0 to both the input and output. (It | |
53 | makes no difference on Unix-like systems.) Later I was told that it is wrong | |
54 | for the input on Windows. I've now abstracted the modes into two macros that | |
55 | are set here, to make it easier to fiddle with them, and removed "b" from the | |
56 | input mode under Windows. */ | |
57 | ||
58 | #if defined(_WIN32) || defined(WIN32) | |
59 | #include <io.h> /* For _setmode() */ | |
60 | #include <fcntl.h> /* For _O_BINARY */ | |
61 | #define INPUT_MODE "r" | |
62 | #define OUTPUT_MODE "wb" | |
63 | ||
64 | #else | |
65 | #include <sys/time.h> /* These two includes are needed */ | |
66 | #include <sys/resource.h> /* for setrlimit(). */ | |
67 | #define INPUT_MODE "rb" | |
68 | #define OUTPUT_MODE "wb" | |
aa41d2de PH |
69 | #endif |
70 | ||
6bf342e1 | 71 | |
c86f6258 | 72 | #define PCRE_SPY /* For Win32 build, import data, not export */ |
8ac170f3 | 73 | |
aa41d2de PH |
74 | /* We include pcre_internal.h because we need the internal info for displaying |
75 | the results of pcre_study() and we also need to know about the internal | |
76 | macros, structures, and other internal data values; pcretest has "inside | |
77 | information" compared to a program that strictly follows the PCRE API. */ | |
8ac170f3 PH |
78 | |
79 | #include "pcre_internal.h" | |
80 | ||
aa41d2de PH |
81 | /* We need access to the data tables that PCRE uses. So as not to have to keep |
82 | two copies, we include the source file here, changing the names of the external | |
83 | symbols to prevent clashes. */ | |
84 | ||
85 | #define _pcre_utf8_table1 utf8_table1 | |
86 | #define _pcre_utf8_table1_size utf8_table1_size | |
87 | #define _pcre_utf8_table2 utf8_table2 | |
88 | #define _pcre_utf8_table3 utf8_table3 | |
89 | #define _pcre_utf8_table4 utf8_table4 | |
90 | #define _pcre_utt utt | |
91 | #define _pcre_utt_size utt_size | |
92 | #define _pcre_OP_lengths OP_lengths | |
93 | ||
94 | #include "pcre_tables.c" | |
95 | ||
96 | /* We also need the pcre_printint() function for printing out compiled | |
97 | patterns. This function is in a separate file so that it can be included in | |
6bf342e1 PH |
98 | pcre_compile.c when that module is compiled with debugging enabled. |
99 | ||
100 | The definition of the macro PRINTABLE, which determines whether to print an | |
101 | output character as-is or as a hex value when showing compiled patterns, is | |
102 | contained in this file. We uses it here also, in cases when the locale has not | |
103 | been explicitly changed, so as to get consistent output from systems that | |
104 | differ in their output from isprint() even in the "C" locale. */ | |
aa41d2de PH |
105 | |
106 | #include "pcre_printint.src" | |
107 | ||
6bf342e1 PH |
108 | #define PRINTHEX(c) (locale_set? isprint(c) : PRINTABLE(c)) |
109 | ||
c86f6258 PH |
110 | |
111 | /* It is possible to compile this test program without including support for | |
112 | testing the POSIX interface, though this is not available via the standard | |
113 | Makefile. */ | |
114 | ||
115 | #if !defined NOPOSIX | |
116 | #include "pcreposix.h" | |
117 | #endif | |
118 | ||
92e772ff PH |
119 | /* It is also possible, for the benefit of the version imported into Exim, to |
120 | build pcretest without support for UTF8 (define NOUTF8), without the interface | |
8ac170f3 PH |
121 | to the DFA matcher (NODFA), and without the doublecheck of the old "info" |
122 | function (define NOINFOCHECK). */ | |
123 | ||
124 | ||
aa41d2de PH |
125 | /* Other parameters */ |
126 | ||
c86f6258 PH |
127 | #ifndef CLOCKS_PER_SEC |
128 | #ifdef CLK_TCK | |
129 | #define CLOCKS_PER_SEC CLK_TCK | |
130 | #else | |
131 | #define CLOCKS_PER_SEC 100 | |
132 | #endif | |
133 | #endif | |
134 | ||
6bf342e1 PH |
135 | /* This is the default loop count for timing. */ |
136 | ||
c86f6258 PH |
137 | #define LOOPREPEAT 500000 |
138 | ||
aa41d2de | 139 | /* Static variables */ |
c86f6258 PH |
140 | |
141 | static FILE *outfile; | |
142 | static int log_store = 0; | |
143 | static int callout_count; | |
144 | static int callout_extra; | |
145 | static int callout_fail_count; | |
146 | static int callout_fail_id; | |
147 | static int first_callout; | |
6bf342e1 | 148 | static int locale_set = 0; |
c86f6258 PH |
149 | static int show_malloc; |
150 | static int use_utf8; | |
151 | static size_t gotten_store; | |
152 | ||
aa41d2de PH |
153 | /* The buffers grow automatically if very long input lines are encountered. */ |
154 | ||
155 | static int buffer_size = 50000; | |
156 | static uschar *buffer = NULL; | |
157 | static uschar *dbuffer = NULL; | |
c86f6258 PH |
158 | static uschar *pbuffer = NULL; |
159 | ||
160 | ||
c86f6258 | 161 | |
aa41d2de PH |
162 | /************************************************* |
163 | * Read or extend an input line * | |
164 | *************************************************/ | |
165 | ||
166 | /* Input lines are read into buffer, but both patterns and data lines can be | |
167 | continued over multiple input lines. In addition, if the buffer fills up, we | |
168 | want to automatically expand it so as to be able to handle extremely large | |
169 | lines that are needed for certain stress tests. When the input buffer is | |
170 | expanded, the other two buffers must also be expanded likewise, and the | |
171 | contents of pbuffer, which are a copy of the input for callouts, must be | |
172 | preserved (for when expansion happens for a data line). This is not the most | |
173 | optimal way of handling this, but hey, this is just a test program! | |
174 | ||
175 | Arguments: | |
176 | f the file to read | |
177 | start where in buffer to start (this *must* be within buffer) | |
178 | ||
179 | Returns: pointer to the start of new data | |
180 | could be a copy of start, or could be moved | |
181 | NULL if no data read and EOF reached | |
182 | */ | |
183 | ||
184 | static uschar * | |
185 | extend_inputline(FILE *f, uschar *start) | |
186 | { | |
187 | uschar *here = start; | |
188 | ||
189 | for (;;) | |
190 | { | |
191 | int rlen = buffer_size - (here - buffer); | |
6bf342e1 | 192 | |
aa41d2de PH |
193 | if (rlen > 1000) |
194 | { | |
195 | int dlen; | |
196 | if (fgets((char *)here, rlen, f) == NULL) | |
197 | return (here == start)? NULL : start; | |
198 | dlen = (int)strlen((char *)here); | |
199 | if (dlen > 0 && here[dlen - 1] == '\n') return start; | |
200 | here += dlen; | |
201 | } | |
202 | ||
203 | else | |
204 | { | |
205 | int new_buffer_size = 2*buffer_size; | |
206 | uschar *new_buffer = (unsigned char *)malloc(new_buffer_size); | |
207 | uschar *new_dbuffer = (unsigned char *)malloc(new_buffer_size); | |
208 | uschar *new_pbuffer = (unsigned char *)malloc(new_buffer_size); | |
209 | ||
210 | if (new_buffer == NULL || new_dbuffer == NULL || new_pbuffer == NULL) | |
211 | { | |
212 | fprintf(stderr, "pcretest: malloc(%d) failed\n", new_buffer_size); | |
213 | exit(1); | |
214 | } | |
215 | ||
216 | memcpy(new_buffer, buffer, buffer_size); | |
217 | memcpy(new_pbuffer, pbuffer, buffer_size); | |
218 | ||
219 | buffer_size = new_buffer_size; | |
220 | ||
221 | start = new_buffer + (start - buffer); | |
222 | here = new_buffer + (here - buffer); | |
223 | ||
224 | free(buffer); | |
225 | free(dbuffer); | |
226 | free(pbuffer); | |
227 | ||
228 | buffer = new_buffer; | |
229 | dbuffer = new_dbuffer; | |
230 | pbuffer = new_pbuffer; | |
231 | } | |
232 | } | |
233 | ||
234 | return NULL; /* Control never gets here */ | |
235 | } | |
236 | ||
237 | ||
238 | ||
239 | ||
240 | ||
241 | ||
242 | ||
c86f6258 PH |
243 | /************************************************* |
244 | * Read number from string * | |
245 | *************************************************/ | |
246 | ||
247 | /* We don't use strtoul() because SunOS4 doesn't have it. Rather than mess | |
248 | around with conditional compilation, just do the job by hand. It is only used | |
6bf342e1 | 249 | for unpicking arguments, so just keep it simple. |
c86f6258 PH |
250 | |
251 | Arguments: | |
252 | str string to be converted | |
253 | endptr where to put the end pointer | |
254 | ||
255 | Returns: the unsigned long | |
256 | */ | |
257 | ||
258 | static int | |
259 | get_value(unsigned char *str, unsigned char **endptr) | |
260 | { | |
261 | int result = 0; | |
262 | while(*str != 0 && isspace(*str)) str++; | |
263 | while (isdigit(*str)) result = result * 10 + (int)(*str++ - '0'); | |
264 | *endptr = str; | |
265 | return(result); | |
266 | } | |
267 | ||
268 | ||
269 | ||
c86f6258 PH |
270 | |
271 | /************************************************* | |
272 | * Convert UTF-8 string to value * | |
273 | *************************************************/ | |
274 | ||
275 | /* This function takes one or more bytes that represents a UTF-8 character, | |
276 | and returns the value of the character. | |
277 | ||
278 | Argument: | |
aa41d2de PH |
279 | utf8bytes a pointer to the byte vector |
280 | vptr a pointer to an int to receive the value | |
c86f6258 | 281 | |
aa41d2de PH |
282 | Returns: > 0 => the number of bytes consumed |
283 | -6 to 0 => malformed UTF-8 character at offset = (-return) | |
c86f6258 PH |
284 | */ |
285 | ||
8ac170f3 PH |
286 | #if !defined NOUTF8 |
287 | ||
c86f6258 | 288 | static int |
aa41d2de | 289 | utf82ord(unsigned char *utf8bytes, int *vptr) |
c86f6258 | 290 | { |
aa41d2de | 291 | int c = *utf8bytes++; |
c86f6258 PH |
292 | int d = c; |
293 | int i, j, s; | |
294 | ||
295 | for (i = -1; i < 6; i++) /* i is number of additional bytes */ | |
296 | { | |
297 | if ((d & 0x80) == 0) break; | |
298 | d <<= 1; | |
299 | } | |
300 | ||
301 | if (i == -1) { *vptr = c; return 1; } /* ascii character */ | |
302 | if (i == 0 || i == 6) return 0; /* invalid UTF-8 */ | |
303 | ||
304 | /* i now has a value in the range 1-5 */ | |
305 | ||
306 | s = 6*i; | |
aa41d2de | 307 | d = (c & utf8_table3[i]) << s; |
c86f6258 PH |
308 | |
309 | for (j = 0; j < i; j++) | |
310 | { | |
aa41d2de | 311 | c = *utf8bytes++; |
c86f6258 PH |
312 | if ((c & 0xc0) != 0x80) return -(j+1); |
313 | s -= 6; | |
314 | d |= (c & 0x3f) << s; | |
315 | } | |
316 | ||
317 | /* Check that encoding was the correct unique one */ | |
318 | ||
aa41d2de PH |
319 | for (j = 0; j < utf8_table1_size; j++) |
320 | if (d <= utf8_table1[j]) break; | |
c86f6258 PH |
321 | if (j != i) return -(i+1); |
322 | ||
323 | /* Valid value */ | |
324 | ||
325 | *vptr = d; | |
326 | return i+1; | |
327 | } | |
328 | ||
8ac170f3 PH |
329 | #endif |
330 | ||
c86f6258 PH |
331 | |
332 | ||
aa41d2de PH |
333 | /************************************************* |
334 | * Convert character value to UTF-8 * | |
335 | *************************************************/ | |
336 | ||
337 | /* This function takes an integer value in the range 0 - 0x7fffffff | |
338 | and encodes it as a UTF-8 character in 0 to 6 bytes. | |
339 | ||
340 | Arguments: | |
341 | cvalue the character value | |
342 | utf8bytes pointer to buffer for result - at least 6 bytes long | |
343 | ||
344 | Returns: number of characters placed in the buffer | |
345 | */ | |
346 | ||
347 | #if !defined NOUTF8 | |
348 | ||
349 | static int | |
350 | ord2utf8(int cvalue, uschar *utf8bytes) | |
351 | { | |
352 | register int i, j; | |
353 | for (i = 0; i < utf8_table1_size; i++) | |
354 | if (cvalue <= utf8_table1[i]) break; | |
355 | utf8bytes += i; | |
356 | for (j = i; j > 0; j--) | |
357 | { | |
358 | *utf8bytes-- = 0x80 | (cvalue & 0x3f); | |
359 | cvalue >>= 6; | |
360 | } | |
361 | *utf8bytes = utf8_table2[i] | cvalue; | |
362 | return i + 1; | |
363 | } | |
364 | ||
365 | #endif | |
366 | ||
367 | ||
368 | ||
c86f6258 PH |
369 | /************************************************* |
370 | * Print character string * | |
371 | *************************************************/ | |
372 | ||
373 | /* Character string printing function. Must handle UTF-8 strings in utf8 | |
374 | mode. Yields number of characters printed. If handed a NULL file, just counts | |
375 | chars without printing. */ | |
376 | ||
377 | static int pchars(unsigned char *p, int length, FILE *f) | |
378 | { | |
aa41d2de | 379 | int c = 0; |
c86f6258 PH |
380 | int yield = 0; |
381 | ||
382 | while (length-- > 0) | |
383 | { | |
92e772ff | 384 | #if !defined NOUTF8 |
c86f6258 PH |
385 | if (use_utf8) |
386 | { | |
387 | int rc = utf82ord(p, &c); | |
388 | ||
389 | if (rc > 0 && rc <= length + 1) /* Mustn't run over the end */ | |
390 | { | |
391 | length -= rc - 1; | |
392 | p += rc; | |
6bf342e1 | 393 | if (PRINTHEX(c)) |
c86f6258 PH |
394 | { |
395 | if (f != NULL) fprintf(f, "%c", c); | |
396 | yield++; | |
397 | } | |
398 | else | |
399 | { | |
6bf342e1 PH |
400 | int n = 4; |
401 | if (f != NULL) fprintf(f, "\\x{%02x}", c); | |
402 | yield += (n <= 0x000000ff)? 2 : | |
403 | (n <= 0x00000fff)? 3 : | |
404 | (n <= 0x0000ffff)? 4 : | |
405 | (n <= 0x000fffff)? 5 : 6; | |
c86f6258 PH |
406 | } |
407 | continue; | |
408 | } | |
409 | } | |
8ac170f3 | 410 | #endif |
c86f6258 PH |
411 | |
412 | /* Not UTF-8, or malformed UTF-8 */ | |
413 | ||
6bf342e1 PH |
414 | c = *p++; |
415 | if (PRINTHEX(c)) | |
c86f6258 PH |
416 | { |
417 | if (f != NULL) fprintf(f, "%c", c); | |
418 | yield++; | |
419 | } | |
420 | else | |
421 | { | |
422 | if (f != NULL) fprintf(f, "\\x%02x", c); | |
423 | yield += 4; | |
424 | } | |
425 | } | |
426 | ||
427 | return yield; | |
428 | } | |
429 | ||
430 | ||
431 | ||
432 | /************************************************* | |
433 | * Callout function * | |
434 | *************************************************/ | |
435 | ||
436 | /* Called from PCRE as a result of the (?C) item. We print out where we are in | |
437 | the match. Yield zero unless more callouts than the fail count, or the callout | |
438 | data is not zero. */ | |
439 | ||
440 | static int callout(pcre_callout_block *cb) | |
441 | { | |
442 | FILE *f = (first_callout | callout_extra)? outfile : NULL; | |
443 | int i, pre_start, post_start, subject_length; | |
444 | ||
445 | if (callout_extra) | |
446 | { | |
447 | fprintf(f, "Callout %d: last capture = %d\n", | |
448 | cb->callout_number, cb->capture_last); | |
449 | ||
450 | for (i = 0; i < cb->capture_top * 2; i += 2) | |
451 | { | |
452 | if (cb->offset_vector[i] < 0) | |
453 | fprintf(f, "%2d: <unset>\n", i/2); | |
454 | else | |
455 | { | |
456 | fprintf(f, "%2d: ", i/2); | |
457 | (void)pchars((unsigned char *)cb->subject + cb->offset_vector[i], | |
458 | cb->offset_vector[i+1] - cb->offset_vector[i], f); | |
459 | fprintf(f, "\n"); | |
460 | } | |
461 | } | |
462 | } | |
463 | ||
464 | /* Re-print the subject in canonical form, the first time or if giving full | |
465 | datails. On subsequent calls in the same match, we use pchars just to find the | |
466 | printed lengths of the substrings. */ | |
467 | ||
468 | if (f != NULL) fprintf(f, "--->"); | |
469 | ||
470 | pre_start = pchars((unsigned char *)cb->subject, cb->start_match, f); | |
471 | post_start = pchars((unsigned char *)(cb->subject + cb->start_match), | |
472 | cb->current_position - cb->start_match, f); | |
473 | ||
474 | subject_length = pchars((unsigned char *)cb->subject, cb->subject_length, NULL); | |
475 | ||
476 | (void)pchars((unsigned char *)(cb->subject + cb->current_position), | |
477 | cb->subject_length - cb->current_position, f); | |
478 | ||
479 | if (f != NULL) fprintf(f, "\n"); | |
480 | ||
481 | /* Always print appropriate indicators, with callout number if not already | |
482 | shown. For automatic callouts, show the pattern offset. */ | |
483 | ||
484 | if (cb->callout_number == 255) | |
485 | { | |
486 | fprintf(outfile, "%+3d ", cb->pattern_position); | |
487 | if (cb->pattern_position > 99) fprintf(outfile, "\n "); | |
488 | } | |
489 | else | |
490 | { | |
491 | if (callout_extra) fprintf(outfile, " "); | |
492 | else fprintf(outfile, "%3d ", cb->callout_number); | |
493 | } | |
494 | ||
495 | for (i = 0; i < pre_start; i++) fprintf(outfile, " "); | |
496 | fprintf(outfile, "^"); | |
497 | ||
498 | if (post_start > 0) | |
499 | { | |
500 | for (i = 0; i < post_start - 1; i++) fprintf(outfile, " "); | |
501 | fprintf(outfile, "^"); | |
502 | } | |
503 | ||
504 | for (i = 0; i < subject_length - pre_start - post_start + 4; i++) | |
505 | fprintf(outfile, " "); | |
506 | ||
507 | fprintf(outfile, "%.*s", (cb->next_item_length == 0)? 1 : cb->next_item_length, | |
508 | pbuffer + cb->pattern_position); | |
509 | ||
510 | fprintf(outfile, "\n"); | |
511 | first_callout = 0; | |
512 | ||
513 | if (cb->callout_data != NULL) | |
514 | { | |
515 | int callout_data = *((int *)(cb->callout_data)); | |
516 | if (callout_data != 0) | |
517 | { | |
518 | fprintf(outfile, "Callout data = %d\n", callout_data); | |
519 | return callout_data; | |
520 | } | |
521 | } | |
522 | ||
523 | return (cb->callout_number != callout_fail_id)? 0 : | |
524 | (++callout_count >= callout_fail_count)? 1 : 0; | |
525 | } | |
526 | ||
527 | ||
528 | /************************************************* | |
529 | * Local malloc functions * | |
530 | *************************************************/ | |
531 | ||
532 | /* Alternative malloc function, to test functionality and show the size of the | |
533 | compiled re. */ | |
534 | ||
535 | static void *new_malloc(size_t size) | |
536 | { | |
537 | void *block = malloc(size); | |
538 | gotten_store = size; | |
539 | if (show_malloc) | |
540 | fprintf(outfile, "malloc %3d %p\n", (int)size, block); | |
541 | return block; | |
542 | } | |
543 | ||
544 | static void new_free(void *block) | |
545 | { | |
546 | if (show_malloc) | |
547 | fprintf(outfile, "free %p\n", block); | |
548 | free(block); | |
549 | } | |
550 | ||
551 | ||
552 | /* For recursion malloc/free, to test stacking calls */ | |
553 | ||
554 | static void *stack_malloc(size_t size) | |
555 | { | |
556 | void *block = malloc(size); | |
557 | if (show_malloc) | |
558 | fprintf(outfile, "stack_malloc %3d %p\n", (int)size, block); | |
559 | return block; | |
560 | } | |
561 | ||
562 | static void stack_free(void *block) | |
563 | { | |
564 | if (show_malloc) | |
565 | fprintf(outfile, "stack_free %p\n", block); | |
566 | free(block); | |
567 | } | |
568 | ||
569 | ||
570 | /************************************************* | |
571 | * Call pcre_fullinfo() * | |
572 | *************************************************/ | |
573 | ||
574 | /* Get one piece of information from the pcre_fullinfo() function */ | |
575 | ||
576 | static void new_info(pcre *re, pcre_extra *study, int option, void *ptr) | |
577 | { | |
578 | int rc; | |
579 | if ((rc = pcre_fullinfo(re, study, option, ptr)) < 0) | |
580 | fprintf(outfile, "Error %d from pcre_fullinfo(%d)\n", rc, option); | |
581 | } | |
582 | ||
583 | ||
584 | ||
585 | /************************************************* | |
586 | * Byte flipping function * | |
587 | *************************************************/ | |
588 | ||
aa41d2de PH |
589 | static unsigned long int |
590 | byteflip(unsigned long int value, int n) | |
c86f6258 PH |
591 | { |
592 | if (n == 2) return ((value & 0x00ff) << 8) | ((value & 0xff00) >> 8); | |
593 | return ((value & 0x000000ff) << 24) | | |
594 | ((value & 0x0000ff00) << 8) | | |
595 | ((value & 0x00ff0000) >> 8) | | |
596 | ((value & 0xff000000) >> 24); | |
597 | } | |
598 | ||
599 | ||
600 | ||
601 | ||
aa41d2de PH |
602 | /************************************************* |
603 | * Check match or recursion limit * | |
604 | *************************************************/ | |
605 | ||
606 | static int | |
607 | check_match_limit(pcre *re, pcre_extra *extra, uschar *bptr, int len, | |
608 | int start_offset, int options, int *use_offsets, int use_size_offsets, | |
609 | int flag, unsigned long int *limit, int errnumber, const char *msg) | |
610 | { | |
611 | int count; | |
612 | int min = 0; | |
613 | int mid = 64; | |
614 | int max = -1; | |
615 | ||
616 | extra->flags |= flag; | |
617 | ||
618 | for (;;) | |
619 | { | |
620 | *limit = mid; | |
621 | ||
622 | count = pcre_exec(re, extra, (char *)bptr, len, start_offset, options, | |
623 | use_offsets, use_size_offsets); | |
624 | ||
625 | if (count == errnumber) | |
626 | { | |
627 | /* fprintf(outfile, "Testing %s limit = %d\n", msg, mid); */ | |
628 | min = mid; | |
629 | mid = (mid == max - 1)? max : (max > 0)? (min + max)/2 : mid*2; | |
630 | } | |
631 | ||
632 | else if (count >= 0 || count == PCRE_ERROR_NOMATCH || | |
633 | count == PCRE_ERROR_PARTIAL) | |
634 | { | |
635 | if (mid == min + 1) | |
636 | { | |
637 | fprintf(outfile, "Minimum %s limit = %d\n", msg, mid); | |
638 | break; | |
639 | } | |
640 | /* fprintf(outfile, "Testing %s limit = %d\n", msg, mid); */ | |
641 | max = mid; | |
642 | mid = (min + mid)/2; | |
643 | } | |
644 | else break; /* Some other error */ | |
645 | } | |
646 | ||
647 | extra->flags &= ~flag; | |
648 | return count; | |
649 | } | |
650 | ||
651 | ||
652 | ||
653 | /************************************************* | |
654 | * Check newline indicator * | |
655 | *************************************************/ | |
656 | ||
657 | /* This is used both at compile and run-time to check for <xxx> escapes, where | |
6bf342e1 | 658 | xxx is LF, CR, CRLF, or ANY. Print a message and return 0 if there is no match. |
aa41d2de PH |
659 | |
660 | Arguments: | |
661 | p points after the leading '<' | |
662 | f file for error message | |
663 | ||
664 | Returns: appropriate PCRE_NEWLINE_xxx flags, or 0 | |
665 | */ | |
666 | ||
667 | static int | |
668 | check_newline(uschar *p, FILE *f) | |
669 | { | |
670 | if (strncmp((char *)p, "cr>", 3) == 0) return PCRE_NEWLINE_CR; | |
671 | if (strncmp((char *)p, "lf>", 3) == 0) return PCRE_NEWLINE_LF; | |
672 | if (strncmp((char *)p, "crlf>", 5) == 0) return PCRE_NEWLINE_CRLF; | |
6bf342e1 | 673 | if (strncmp((char *)p, "any>", 4) == 0) return PCRE_NEWLINE_ANY; |
aa41d2de PH |
674 | fprintf(f, "Unknown newline type at: <%s\n", p); |
675 | return 0; | |
676 | } | |
677 | ||
678 | ||
679 | ||
6bf342e1 PH |
680 | /************************************************* |
681 | * Usage function * | |
682 | *************************************************/ | |
683 | ||
684 | static void | |
685 | usage(void) | |
686 | { | |
687 | printf("Usage: pcretest [options] [<input> [<output>]]\n"); | |
688 | printf(" -b show compiled code (bytecode)\n"); | |
689 | printf(" -C show PCRE compile-time options and exit\n"); | |
690 | printf(" -d debug: show compiled code and information (-b and -i)\n"); | |
691 | #if !defined NODFA | |
692 | printf(" -dfa force DFA matching for all subjects\n"); | |
693 | #endif | |
694 | printf(" -help show usage information\n"); | |
695 | printf(" -i show information about compiled patterns\n" | |
696 | " -m output memory used information\n" | |
697 | " -o <n> set size of offsets vector to <n>\n"); | |
698 | #if !defined NOPOSIX | |
699 | printf(" -p use POSIX interface\n"); | |
700 | #endif | |
701 | printf(" -q quiet: do not output PCRE version number at start\n"); | |
702 | printf(" -S <n> set stack size to <n> megabytes\n"); | |
703 | printf(" -s output store (memory) used information\n" | |
704 | " -t time compilation and execution\n"); | |
705 | printf(" -t <n> time compilation and execution, repeating <n> times\n"); | |
706 | printf(" -tm time execution (matching) only\n"); | |
707 | printf(" -tm <n> time execution (matching) only, repeating <n> times\n"); | |
708 | } | |
709 | ||
710 | ||
711 | ||
c86f6258 PH |
712 | /************************************************* |
713 | * Main Program * | |
714 | *************************************************/ | |
715 | ||
716 | /* Read lines from named file or stdin and write to named file or stdout; lines | |
717 | consist of a regular expression, in delimiters and optionally followed by | |
718 | options, followed by a set of test data, terminated by an empty line. */ | |
719 | ||
720 | int main(int argc, char **argv) | |
721 | { | |
722 | FILE *infile = stdin; | |
723 | int options = 0; | |
724 | int study_options = 0; | |
725 | int op = 1; | |
726 | int timeit = 0; | |
6bf342e1 | 727 | int timeitm = 0; |
c86f6258 PH |
728 | int showinfo = 0; |
729 | int showstore = 0; | |
aa41d2de | 730 | int quiet = 0; |
c86f6258 PH |
731 | int size_offsets = 45; |
732 | int size_offsets_max; | |
8ac170f3 | 733 | int *offsets = NULL; |
c86f6258 PH |
734 | #if !defined NOPOSIX |
735 | int posix = 0; | |
736 | #endif | |
737 | int debug = 0; | |
738 | int done = 0; | |
8ac170f3 PH |
739 | int all_use_dfa = 0; |
740 | int yield = 0; | |
aa41d2de PH |
741 | int stack_size; |
742 | ||
743 | /* These vectors store, end-to-end, a list of captured substring names. Assume | |
744 | that 1024 is plenty long enough for the few names we'll be testing. */ | |
745 | ||
746 | uschar copynames[1024]; | |
747 | uschar getnames[1024]; | |
c86f6258 | 748 | |
aa41d2de PH |
749 | uschar *copynamesptr; |
750 | uschar *getnamesptr; | |
c86f6258 PH |
751 | |
752 | /* Get buffers from malloc() so that Electric Fence will check their misuse | |
aa41d2de | 753 | when I am debugging. They grow automatically when very long lines are read. */ |
c86f6258 | 754 | |
aa41d2de PH |
755 | buffer = (unsigned char *)malloc(buffer_size); |
756 | dbuffer = (unsigned char *)malloc(buffer_size); | |
757 | pbuffer = (unsigned char *)malloc(buffer_size); | |
c86f6258 | 758 | |
6bf342e1 | 759 | /* The outfile variable is static so that new_malloc can use it. */ |
c86f6258 PH |
760 | |
761 | outfile = stdout; | |
762 | ||
6bf342e1 PH |
763 | /* The following _setmode() stuff is some Windows magic that tells its runtime |
764 | library to translate CRLF into a single LF character. At least, that's what | |
765 | I've been told: never having used Windows I take this all on trust. Originally | |
766 | it set 0x8000, but then I was advised that _O_BINARY was better. */ | |
767 | ||
768 | #if defined(_WIN32) || defined(WIN32) | |
769 | _setmode( _fileno( stdout ), _O_BINARY ); | |
770 | #endif | |
771 | ||
c86f6258 PH |
772 | /* Scan options */ |
773 | ||
774 | while (argc > 1 && argv[op][0] == '-') | |
775 | { | |
776 | unsigned char *endptr; | |
777 | ||
778 | if (strcmp(argv[op], "-s") == 0 || strcmp(argv[op], "-m") == 0) | |
779 | showstore = 1; | |
aa41d2de | 780 | else if (strcmp(argv[op], "-q") == 0) quiet = 1; |
6bf342e1 | 781 | else if (strcmp(argv[op], "-b") == 0) debug = 1; |
c86f6258 PH |
782 | else if (strcmp(argv[op], "-i") == 0) showinfo = 1; |
783 | else if (strcmp(argv[op], "-d") == 0) showinfo = debug = 1; | |
92e772ff | 784 | #if !defined NODFA |
8ac170f3 PH |
785 | else if (strcmp(argv[op], "-dfa") == 0) all_use_dfa = 1; |
786 | #endif | |
c86f6258 PH |
787 | else if (strcmp(argv[op], "-o") == 0 && argc > 2 && |
788 | ((size_offsets = get_value((unsigned char *)argv[op+1], &endptr)), | |
789 | *endptr == 0)) | |
790 | { | |
791 | op++; | |
792 | argc--; | |
793 | } | |
6bf342e1 PH |
794 | else if (strcmp(argv[op], "-t") == 0 || strcmp(argv[op], "-tm") == 0) |
795 | { | |
796 | int both = argv[op][2] == 0; | |
797 | int temp; | |
798 | if (argc > 2 && (temp = get_value((unsigned char *)argv[op+1], &endptr), | |
799 | *endptr == 0)) | |
800 | { | |
801 | timeitm = temp; | |
802 | op++; | |
803 | argc--; | |
804 | } | |
805 | else timeitm = LOOPREPEAT; | |
806 | if (both) timeit = timeitm; | |
807 | } | |
aa41d2de PH |
808 | else if (strcmp(argv[op], "-S") == 0 && argc > 2 && |
809 | ((stack_size = get_value((unsigned char *)argv[op+1], &endptr)), | |
810 | *endptr == 0)) | |
811 | { | |
6bf342e1 | 812 | #if defined(_WIN32) || defined(WIN32) |
aa41d2de PH |
813 | printf("PCRE: -S not supported on this OS\n"); |
814 | exit(1); | |
815 | #else | |
816 | int rc; | |
817 | struct rlimit rlim; | |
818 | getrlimit(RLIMIT_STACK, &rlim); | |
819 | rlim.rlim_cur = stack_size * 1024 * 1024; | |
820 | rc = setrlimit(RLIMIT_STACK, &rlim); | |
821 | if (rc != 0) | |
822 | { | |
823 | printf("PCRE: setrlimit() failed with error %d\n", rc); | |
824 | exit(1); | |
825 | } | |
826 | op++; | |
827 | argc--; | |
828 | #endif | |
829 | } | |
c86f6258 PH |
830 | #if !defined NOPOSIX |
831 | else if (strcmp(argv[op], "-p") == 0) posix = 1; | |
832 | #endif | |
833 | else if (strcmp(argv[op], "-C") == 0) | |
834 | { | |
835 | int rc; | |
836 | printf("PCRE version %s\n", pcre_version()); | |
837 | printf("Compiled with\n"); | |
838 | (void)pcre_config(PCRE_CONFIG_UTF8, &rc); | |
839 | printf(" %sUTF-8 support\n", rc? "" : "No "); | |
840 | (void)pcre_config(PCRE_CONFIG_UNICODE_PROPERTIES, &rc); | |
841 | printf(" %sUnicode properties support\n", rc? "" : "No "); | |
842 | (void)pcre_config(PCRE_CONFIG_NEWLINE, &rc); | |
aa41d2de | 843 | printf(" Newline sequence is %s\n", (rc == '\r')? "CR" : |
6bf342e1 PH |
844 | (rc == '\n')? "LF" : (rc == ('\r'<<8 | '\n'))? "CRLF" : |
845 | (rc == -1)? "ANY" : "???"); | |
c86f6258 PH |
846 | (void)pcre_config(PCRE_CONFIG_LINK_SIZE, &rc); |
847 | printf(" Internal link size = %d\n", rc); | |
848 | (void)pcre_config(PCRE_CONFIG_POSIX_MALLOC_THRESHOLD, &rc); | |
849 | printf(" POSIX malloc threshold = %d\n", rc); | |
850 | (void)pcre_config(PCRE_CONFIG_MATCH_LIMIT, &rc); | |
851 | printf(" Default match limit = %d\n", rc); | |
aa41d2de PH |
852 | (void)pcre_config(PCRE_CONFIG_MATCH_LIMIT_RECURSION, &rc); |
853 | printf(" Default recursion depth limit = %d\n", rc); | |
c86f6258 PH |
854 | (void)pcre_config(PCRE_CONFIG_STACKRECURSE, &rc); |
855 | printf(" Match recursion uses %s\n", rc? "stack" : "heap"); | |
856 | exit(0); | |
857 | } | |
6bf342e1 PH |
858 | else if (strcmp(argv[op], "-help") == 0 || |
859 | strcmp(argv[op], "--help") == 0) | |
860 | { | |
861 | usage(); | |
862 | goto EXIT; | |
863 | } | |
c86f6258 PH |
864 | else |
865 | { | |
866 | printf("** Unknown or malformed option %s\n", argv[op]); | |
6bf342e1 | 867 | usage(); |
8ac170f3 PH |
868 | yield = 1; |
869 | goto EXIT; | |
c86f6258 PH |
870 | } |
871 | op++; | |
872 | argc--; | |
873 | } | |
874 | ||
875 | /* Get the store for the offsets vector, and remember what it was */ | |
876 | ||
877 | size_offsets_max = size_offsets; | |
878 | offsets = (int *)malloc(size_offsets_max * sizeof(int)); | |
879 | if (offsets == NULL) | |
880 | { | |
881 | printf("** Failed to get %d bytes of memory for offsets vector\n", | |
882 | size_offsets_max * sizeof(int)); | |
8ac170f3 PH |
883 | yield = 1; |
884 | goto EXIT; | |
c86f6258 PH |
885 | } |
886 | ||
887 | /* Sort out the input and output files */ | |
888 | ||
889 | if (argc > 1) | |
890 | { | |
6bf342e1 | 891 | infile = fopen(argv[op], INPUT_MODE); |
c86f6258 PH |
892 | if (infile == NULL) |
893 | { | |
894 | printf("** Failed to open %s\n", argv[op]); | |
8ac170f3 PH |
895 | yield = 1; |
896 | goto EXIT; | |
c86f6258 PH |
897 | } |
898 | } | |
899 | ||
900 | if (argc > 2) | |
901 | { | |
6bf342e1 | 902 | outfile = fopen(argv[op+1], OUTPUT_MODE); |
c86f6258 PH |
903 | if (outfile == NULL) |
904 | { | |
905 | printf("** Failed to open %s\n", argv[op+1]); | |
8ac170f3 PH |
906 | yield = 1; |
907 | goto EXIT; | |
c86f6258 PH |
908 | } |
909 | } | |
910 | ||
911 | /* Set alternative malloc function */ | |
912 | ||
913 | pcre_malloc = new_malloc; | |
914 | pcre_free = new_free; | |
915 | pcre_stack_malloc = stack_malloc; | |
916 | pcre_stack_free = stack_free; | |
917 | ||
aa41d2de | 918 | /* Heading line unless quiet, then prompt for first regex if stdin */ |
c86f6258 | 919 | |
aa41d2de | 920 | if (!quiet) fprintf(outfile, "PCRE version %s\n\n", pcre_version()); |
c86f6258 PH |
921 | |
922 | /* Main loop */ | |
923 | ||
924 | while (!done) | |
925 | { | |
926 | pcre *re = NULL; | |
927 | pcre_extra *extra = NULL; | |
928 | ||
929 | #if !defined NOPOSIX /* There are still compilers that require no indent */ | |
930 | regex_t preg; | |
931 | int do_posix = 0; | |
932 | #endif | |
933 | ||
934 | const char *error; | |
935 | unsigned char *p, *pp, *ppp; | |
936 | unsigned char *to_file = NULL; | |
937 | const unsigned char *tables = NULL; | |
938 | unsigned long int true_size, true_study_size = 0; | |
939 | size_t size, regex_gotten_store; | |
940 | int do_study = 0; | |
941 | int do_debug = debug; | |
942 | int do_G = 0; | |
943 | int do_g = 0; | |
944 | int do_showinfo = showinfo; | |
945 | int do_showrest = 0; | |
946 | int do_flip = 0; | |
6bf342e1 | 947 | int erroroffset, len, delimiter, poffset; |
c86f6258 PH |
948 | |
949 | use_utf8 = 0; | |
950 | ||
951 | if (infile == stdin) printf(" re> "); | |
aa41d2de | 952 | if (extend_inputline(infile, buffer) == NULL) break; |
c86f6258 PH |
953 | if (infile != stdin) fprintf(outfile, "%s", (char *)buffer); |
954 | fflush(outfile); | |
955 | ||
956 | p = buffer; | |
957 | while (isspace(*p)) p++; | |
958 | if (*p == 0) continue; | |
959 | ||
960 | /* See if the pattern is to be loaded pre-compiled from a file. */ | |
961 | ||
962 | if (*p == '<' && strchr((char *)(p+1), '<') == NULL) | |
963 | { | |
aa41d2de | 964 | unsigned long int magic, get_options; |
c86f6258 PH |
965 | uschar sbuf[8]; |
966 | FILE *f; | |
967 | ||
968 | p++; | |
969 | pp = p + (int)strlen((char *)p); | |
970 | while (isspace(pp[-1])) pp--; | |
971 | *pp = 0; | |
972 | ||
973 | f = fopen((char *)p, "rb"); | |
974 | if (f == NULL) | |
975 | { | |
976 | fprintf(outfile, "Failed to open %s: %s\n", p, strerror(errno)); | |
977 | continue; | |
978 | } | |
979 | ||
980 | if (fread(sbuf, 1, 8, f) != 8) goto FAIL_READ; | |
981 | ||
982 | true_size = | |
983 | (sbuf[0] << 24) | (sbuf[1] << 16) | (sbuf[2] << 8) | sbuf[3]; | |
984 | true_study_size = | |
985 | (sbuf[4] << 24) | (sbuf[5] << 16) | (sbuf[6] << 8) | sbuf[7]; | |
986 | ||
987 | re = (real_pcre *)new_malloc(true_size); | |
988 | regex_gotten_store = gotten_store; | |
989 | ||
990 | if (fread(re, 1, true_size, f) != true_size) goto FAIL_READ; | |
991 | ||
992 | magic = ((real_pcre *)re)->magic_number; | |
993 | if (magic != MAGIC_NUMBER) | |
994 | { | |
995 | if (byteflip(magic, sizeof(magic)) == MAGIC_NUMBER) | |
996 | { | |
997 | do_flip = 1; | |
998 | } | |
999 | else | |
1000 | { | |
1001 | fprintf(outfile, "Data in %s is not a compiled PCRE regex\n", p); | |
1002 | fclose(f); | |
1003 | continue; | |
1004 | } | |
1005 | } | |
1006 | ||
1007 | fprintf(outfile, "Compiled regex%s loaded from %s\n", | |
1008 | do_flip? " (byte-inverted)" : "", p); | |
1009 | ||
1010 | /* Need to know if UTF-8 for printing data strings */ | |
1011 | ||
aa41d2de PH |
1012 | new_info(re, NULL, PCRE_INFO_OPTIONS, &get_options); |
1013 | use_utf8 = (get_options & PCRE_UTF8) != 0; | |
c86f6258 PH |
1014 | |
1015 | /* Now see if there is any following study data */ | |
1016 | ||
1017 | if (true_study_size != 0) | |
1018 | { | |
1019 | pcre_study_data *psd; | |
1020 | ||
1021 | extra = (pcre_extra *)new_malloc(sizeof(pcre_extra) + true_study_size); | |
1022 | extra->flags = PCRE_EXTRA_STUDY_DATA; | |
1023 | ||
1024 | psd = (pcre_study_data *)(((char *)extra) + sizeof(pcre_extra)); | |
1025 | extra->study_data = psd; | |
1026 | ||
1027 | if (fread(psd, 1, true_study_size, f) != true_study_size) | |
1028 | { | |
1029 | FAIL_READ: | |
1030 | fprintf(outfile, "Failed to read data from %s\n", p); | |
1031 | if (extra != NULL) new_free(extra); | |
1032 | if (re != NULL) new_free(re); | |
1033 | fclose(f); | |
1034 | continue; | |
1035 | } | |
1036 | fprintf(outfile, "Study data loaded from %s\n", p); | |
1037 | do_study = 1; /* To get the data output if requested */ | |
1038 | } | |
1039 | else fprintf(outfile, "No study data\n"); | |
1040 | ||
1041 | fclose(f); | |
1042 | goto SHOW_INFO; | |
1043 | } | |
1044 | ||
1045 | /* In-line pattern (the usual case). Get the delimiter and seek the end of | |
1046 | the pattern; if is isn't complete, read more. */ | |
1047 | ||
1048 | delimiter = *p++; | |
1049 | ||
1050 | if (isalnum(delimiter) || delimiter == '\\') | |
1051 | { | |
1052 | fprintf(outfile, "** Delimiter must not be alphameric or \\\n"); | |
1053 | goto SKIP_DATA; | |
1054 | } | |
1055 | ||
1056 | pp = p; | |
6bf342e1 | 1057 | poffset = p - buffer; |
c86f6258 PH |
1058 | |
1059 | for(;;) | |
1060 | { | |
1061 | while (*pp != 0) | |
1062 | { | |
1063 | if (*pp == '\\' && pp[1] != 0) pp++; | |
1064 | else if (*pp == delimiter) break; | |
1065 | pp++; | |
1066 | } | |
1067 | if (*pp != 0) break; | |
c86f6258 | 1068 | if (infile == stdin) printf(" > "); |
aa41d2de | 1069 | if ((pp = extend_inputline(infile, pp)) == NULL) |
c86f6258 PH |
1070 | { |
1071 | fprintf(outfile, "** Unexpected EOF\n"); | |
1072 | done = 1; | |
1073 | goto CONTINUE; | |
1074 | } | |
1075 | if (infile != stdin) fprintf(outfile, "%s", (char *)pp); | |
1076 | } | |
1077 | ||
6bf342e1 PH |
1078 | /* The buffer may have moved while being extended; reset the start of data |
1079 | pointer to the correct relative point in the buffer. */ | |
1080 | ||
1081 | p = buffer + poffset; | |
1082 | ||
c86f6258 PH |
1083 | /* If the first character after the delimiter is backslash, make |
1084 | the pattern end with backslash. This is purely to provide a way | |
1085 | of testing for the error message when a pattern ends with backslash. */ | |
1086 | ||
1087 | if (pp[1] == '\\') *pp++ = '\\'; | |
1088 | ||
1089 | /* Terminate the pattern at the delimiter, and save a copy of the pattern | |
1090 | for callouts. */ | |
1091 | ||
1092 | *pp++ = 0; | |
1093 | strcpy((char *)pbuffer, (char *)p); | |
1094 | ||
1095 | /* Look for options after final delimiter */ | |
1096 | ||
1097 | options = 0; | |
1098 | study_options = 0; | |
1099 | log_store = showstore; /* default from command line */ | |
1100 | ||
1101 | while (*pp != 0) | |
1102 | { | |
1103 | switch (*pp++) | |
1104 | { | |
8ac170f3 | 1105 | case 'f': options |= PCRE_FIRSTLINE; break; |
c86f6258 PH |
1106 | case 'g': do_g = 1; break; |
1107 | case 'i': options |= PCRE_CASELESS; break; | |
1108 | case 'm': options |= PCRE_MULTILINE; break; | |
1109 | case 's': options |= PCRE_DOTALL; break; | |
1110 | case 'x': options |= PCRE_EXTENDED; break; | |
1111 | ||
1112 | case '+': do_showrest = 1; break; | |
1113 | case 'A': options |= PCRE_ANCHORED; break; | |
6bf342e1 | 1114 | case 'B': do_debug = 1; break; |
c86f6258 PH |
1115 | case 'C': options |= PCRE_AUTO_CALLOUT; break; |
1116 | case 'D': do_debug = do_showinfo = 1; break; | |
1117 | case 'E': options |= PCRE_DOLLAR_ENDONLY; break; | |
1118 | case 'F': do_flip = 1; break; | |
1119 | case 'G': do_G = 1; break; | |
1120 | case 'I': do_showinfo = 1; break; | |
aa41d2de | 1121 | case 'J': options |= PCRE_DUPNAMES; break; |
c86f6258 PH |
1122 | case 'M': log_store = 1; break; |
1123 | case 'N': options |= PCRE_NO_AUTO_CAPTURE; break; | |
1124 | ||
1125 | #if !defined NOPOSIX | |
1126 | case 'P': do_posix = 1; break; | |
1127 | #endif | |
1128 | ||
1129 | case 'S': do_study = 1; break; | |
1130 | case 'U': options |= PCRE_UNGREEDY; break; | |
1131 | case 'X': options |= PCRE_EXTRA; break; | |
1132 | case '8': options |= PCRE_UTF8; use_utf8 = 1; break; | |
1133 | case '?': options |= PCRE_NO_UTF8_CHECK; break; | |
1134 | ||
1135 | case 'L': | |
1136 | ppp = pp; | |
6bf342e1 PH |
1137 | /* The '\r' test here is so that it works on Windows. */ |
1138 | /* The '0' test is just in case this is an unterminated line. */ | |
1139 | while (*ppp != 0 && *ppp != '\n' && *ppp != '\r' && *ppp != ' ') ppp++; | |
c86f6258 PH |
1140 | *ppp = 0; |
1141 | if (setlocale(LC_CTYPE, (const char *)pp) == NULL) | |
1142 | { | |
1143 | fprintf(outfile, "** Failed to set locale \"%s\"\n", pp); | |
1144 | goto SKIP_DATA; | |
1145 | } | |
6bf342e1 | 1146 | locale_set = 1; |
c86f6258 PH |
1147 | tables = pcre_maketables(); |
1148 | pp = ppp; | |
1149 | break; | |
1150 | ||
1151 | case '>': | |
1152 | to_file = pp; | |
1153 | while (*pp != 0) pp++; | |
1154 | while (isspace(pp[-1])) pp--; | |
1155 | *pp = 0; | |
1156 | break; | |
1157 | ||
aa41d2de PH |
1158 | case '<': |
1159 | { | |
1160 | int x = check_newline(pp, outfile); | |
1161 | if (x == 0) goto SKIP_DATA; | |
1162 | options |= x; | |
1163 | while (*pp++ != '>'); | |
1164 | } | |
1165 | break; | |
1166 | ||
8ac170f3 PH |
1167 | case '\r': /* So that it works in Windows */ |
1168 | case '\n': | |
1169 | case ' ': | |
1170 | break; | |
c86f6258 PH |
1171 | |
1172 | default: | |
1173 | fprintf(outfile, "** Unknown option '%c'\n", pp[-1]); | |
1174 | goto SKIP_DATA; | |
1175 | } | |
1176 | } | |
1177 | ||
1178 | /* Handle compiling via the POSIX interface, which doesn't support the | |
1179 | timing, showing, or debugging options, nor the ability to pass over | |
1180 | local character tables. */ | |
1181 | ||
1182 | #if !defined NOPOSIX | |
1183 | if (posix || do_posix) | |
1184 | { | |
1185 | int rc; | |
1186 | int cflags = 0; | |
1187 | ||
1188 | if ((options & PCRE_CASELESS) != 0) cflags |= REG_ICASE; | |
1189 | if ((options & PCRE_MULTILINE) != 0) cflags |= REG_NEWLINE; | |
8ac170f3 | 1190 | if ((options & PCRE_DOTALL) != 0) cflags |= REG_DOTALL; |
aa41d2de PH |
1191 | if ((options & PCRE_NO_AUTO_CAPTURE) != 0) cflags |= REG_NOSUB; |
1192 | if ((options & PCRE_UTF8) != 0) cflags |= REG_UTF8; | |
1193 | ||
c86f6258 PH |
1194 | rc = regcomp(&preg, (char *)p, cflags); |
1195 | ||
1196 | /* Compilation failed; go back for another re, skipping to blank line | |
1197 | if non-interactive. */ | |
1198 | ||
1199 | if (rc != 0) | |
1200 | { | |
aa41d2de | 1201 | (void)regerror(rc, &preg, (char *)buffer, buffer_size); |
c86f6258 PH |
1202 | fprintf(outfile, "Failed: POSIX code %d: %s\n", rc, buffer); |
1203 | goto SKIP_DATA; | |
1204 | } | |
1205 | } | |
1206 | ||
1207 | /* Handle compiling via the native interface */ | |
1208 | ||
1209 | else | |
1210 | #endif /* !defined NOPOSIX */ | |
1211 | ||
1212 | { | |
6bf342e1 | 1213 | if (timeit > 0) |
c86f6258 PH |
1214 | { |
1215 | register int i; | |
1216 | clock_t time_taken; | |
1217 | clock_t start_time = clock(); | |
6bf342e1 | 1218 | for (i = 0; i < timeit; i++) |
c86f6258 PH |
1219 | { |
1220 | re = pcre_compile((char *)p, options, &error, &erroroffset, tables); | |
1221 | if (re != NULL) free(re); | |
1222 | } | |
1223 | time_taken = clock() - start_time; | |
6bf342e1 PH |
1224 | fprintf(outfile, "Compile time %.4f milliseconds\n", |
1225 | (((double)time_taken * 1000.0) / (double)timeit) / | |
c86f6258 PH |
1226 | (double)CLOCKS_PER_SEC); |
1227 | } | |
1228 | ||
1229 | re = pcre_compile((char *)p, options, &error, &erroroffset, tables); | |
1230 | ||
1231 | /* Compilation failed; go back for another re, skipping to blank line | |
1232 | if non-interactive. */ | |
1233 | ||
1234 | if (re == NULL) | |
1235 | { | |
1236 | fprintf(outfile, "Failed: %s at offset %d\n", error, erroroffset); | |
1237 | SKIP_DATA: | |
1238 | if (infile != stdin) | |
1239 | { | |
1240 | for (;;) | |
1241 | { | |
aa41d2de | 1242 | if (extend_inputline(infile, buffer) == NULL) |
c86f6258 PH |
1243 | { |
1244 | done = 1; | |
1245 | goto CONTINUE; | |
1246 | } | |
1247 | len = (int)strlen((char *)buffer); | |
1248 | while (len > 0 && isspace(buffer[len-1])) len--; | |
1249 | if (len == 0) break; | |
1250 | } | |
1251 | fprintf(outfile, "\n"); | |
1252 | } | |
1253 | goto CONTINUE; | |
1254 | } | |
1255 | ||
1256 | /* Compilation succeeded; print data if required. There are now two | |
1257 | info-returning functions. The old one has a limited interface and | |
1258 | returns only limited data. Check that it agrees with the newer one. */ | |
1259 | ||
1260 | if (log_store) | |
1261 | fprintf(outfile, "Memory allocation (code space): %d\n", | |
1262 | (int)(gotten_store - | |
1263 | sizeof(real_pcre) - | |
1264 | ((real_pcre *)re)->name_count * ((real_pcre *)re)->name_entry_size)); | |
1265 | ||
1266 | /* Extract the size for possible writing before possibly flipping it, | |
1267 | and remember the store that was got. */ | |
1268 | ||
1269 | true_size = ((real_pcre *)re)->size; | |
1270 | regex_gotten_store = gotten_store; | |
1271 | ||
1272 | /* If /S was present, study the regexp to generate additional info to | |
1273 | help with the matching. */ | |
1274 | ||
1275 | if (do_study) | |
1276 | { | |
6bf342e1 | 1277 | if (timeit > 0) |
c86f6258 PH |
1278 | { |
1279 | register int i; | |
1280 | clock_t time_taken; | |
1281 | clock_t start_time = clock(); | |
6bf342e1 | 1282 | for (i = 0; i < timeit; i++) |
c86f6258 PH |
1283 | extra = pcre_study(re, study_options, &error); |
1284 | time_taken = clock() - start_time; | |
1285 | if (extra != NULL) free(extra); | |
6bf342e1 PH |
1286 | fprintf(outfile, " Study time %.4f milliseconds\n", |
1287 | (((double)time_taken * 1000.0) / (double)timeit) / | |
c86f6258 PH |
1288 | (double)CLOCKS_PER_SEC); |
1289 | } | |
1290 | extra = pcre_study(re, study_options, &error); | |
1291 | if (error != NULL) | |
1292 | fprintf(outfile, "Failed to study: %s\n", error); | |
1293 | else if (extra != NULL) | |
1294 | true_study_size = ((pcre_study_data *)(extra->study_data))->size; | |
1295 | } | |
1296 | ||
1297 | /* If the 'F' option was present, we flip the bytes of all the integer | |
1298 | fields in the regex data block and the study block. This is to make it | |
1299 | possible to test PCRE's handling of byte-flipped patterns, e.g. those | |
1300 | compiled on a different architecture. */ | |
1301 | ||
1302 | if (do_flip) | |
1303 | { | |
1304 | real_pcre *rre = (real_pcre *)re; | |
1305 | rre->magic_number = byteflip(rre->magic_number, sizeof(rre->magic_number)); | |
1306 | rre->size = byteflip(rre->size, sizeof(rre->size)); | |
1307 | rre->options = byteflip(rre->options, sizeof(rre->options)); | |
1308 | rre->top_bracket = byteflip(rre->top_bracket, sizeof(rre->top_bracket)); | |
1309 | rre->top_backref = byteflip(rre->top_backref, sizeof(rre->top_backref)); | |
1310 | rre->first_byte = byteflip(rre->first_byte, sizeof(rre->first_byte)); | |
1311 | rre->req_byte = byteflip(rre->req_byte, sizeof(rre->req_byte)); | |
1312 | rre->name_table_offset = byteflip(rre->name_table_offset, | |
1313 | sizeof(rre->name_table_offset)); | |
1314 | rre->name_entry_size = byteflip(rre->name_entry_size, | |
1315 | sizeof(rre->name_entry_size)); | |
1316 | rre->name_count = byteflip(rre->name_count, sizeof(rre->name_count)); | |
1317 | ||
1318 | if (extra != NULL) | |
1319 | { | |
1320 | pcre_study_data *rsd = (pcre_study_data *)(extra->study_data); | |
1321 | rsd->size = byteflip(rsd->size, sizeof(rsd->size)); | |
1322 | rsd->options = byteflip(rsd->options, sizeof(rsd->options)); | |
1323 | } | |
1324 | } | |
1325 | ||
1326 | /* Extract information from the compiled data if required */ | |
1327 | ||
1328 | SHOW_INFO: | |
1329 | ||
6bf342e1 PH |
1330 | if (do_debug) |
1331 | { | |
1332 | fprintf(outfile, "------------------------------------------------------------------\n"); | |
1333 | pcre_printint(re, outfile); | |
1334 | } | |
1335 | ||
c86f6258 PH |
1336 | if (do_showinfo) |
1337 | { | |
1338 | unsigned long int get_options, all_options; | |
8ac170f3 | 1339 | #if !defined NOINFOCHECK |
c86f6258 | 1340 | int old_first_char, old_options, old_count; |
8ac170f3 | 1341 | #endif |
c86f6258 PH |
1342 | int count, backrefmax, first_char, need_char; |
1343 | int nameentrysize, namecount; | |
1344 | const uschar *nametable; | |
1345 | ||
c86f6258 PH |
1346 | new_info(re, NULL, PCRE_INFO_OPTIONS, &get_options); |
1347 | new_info(re, NULL, PCRE_INFO_SIZE, &size); | |
1348 | new_info(re, NULL, PCRE_INFO_CAPTURECOUNT, &count); | |
1349 | new_info(re, NULL, PCRE_INFO_BACKREFMAX, &backrefmax); | |
1350 | new_info(re, NULL, PCRE_INFO_FIRSTBYTE, &first_char); | |
1351 | new_info(re, NULL, PCRE_INFO_LASTLITERAL, &need_char); | |
1352 | new_info(re, NULL, PCRE_INFO_NAMEENTRYSIZE, &nameentrysize); | |
1353 | new_info(re, NULL, PCRE_INFO_NAMECOUNT, &namecount); | |
1354 | new_info(re, NULL, PCRE_INFO_NAMETABLE, (void *)&nametable); | |
1355 | ||
8ac170f3 | 1356 | #if !defined NOINFOCHECK |
c86f6258 PH |
1357 | old_count = pcre_info(re, &old_options, &old_first_char); |
1358 | if (count < 0) fprintf(outfile, | |
1359 | "Error %d from pcre_info()\n", count); | |
1360 | else | |
1361 | { | |
1362 | if (old_count != count) fprintf(outfile, | |
1363 | "Count disagreement: pcre_fullinfo=%d pcre_info=%d\n", count, | |
1364 | old_count); | |
1365 | ||
1366 | if (old_first_char != first_char) fprintf(outfile, | |
1367 | "First char disagreement: pcre_fullinfo=%d pcre_info=%d\n", | |
1368 | first_char, old_first_char); | |
1369 | ||
1370 | if (old_options != (int)get_options) fprintf(outfile, | |
1371 | "Options disagreement: pcre_fullinfo=%ld pcre_info=%d\n", | |
1372 | get_options, old_options); | |
1373 | } | |
8ac170f3 | 1374 | #endif |
c86f6258 PH |
1375 | |
1376 | if (size != regex_gotten_store) fprintf(outfile, | |
1377 | "Size disagreement: pcre_fullinfo=%d call to malloc for %d\n", | |
1378 | (int)size, (int)regex_gotten_store); | |
1379 | ||
1380 | fprintf(outfile, "Capturing subpattern count = %d\n", count); | |
1381 | if (backrefmax > 0) | |
1382 | fprintf(outfile, "Max back reference = %d\n", backrefmax); | |
1383 | ||
1384 | if (namecount > 0) | |
1385 | { | |
1386 | fprintf(outfile, "Named capturing subpatterns:\n"); | |
1387 | while (namecount-- > 0) | |
1388 | { | |
1389 | fprintf(outfile, " %s %*s%3d\n", nametable + 2, | |
1390 | nameentrysize - 3 - (int)strlen((char *)nametable + 2), "", | |
1391 | GET2(nametable, 0)); | |
1392 | nametable += nameentrysize; | |
1393 | } | |
1394 | } | |
1395 | ||
1396 | /* The NOPARTIAL bit is a private bit in the options, so we have | |
1397 | to fish it out via out back door */ | |
1398 | ||
1399 | all_options = ((real_pcre *)re)->options; | |
1400 | if (do_flip) | |
1401 | { | |
1402 | all_options = byteflip(all_options, sizeof(all_options)); | |
aa41d2de | 1403 | } |
c86f6258 PH |
1404 | |
1405 | if ((all_options & PCRE_NOPARTIAL) != 0) | |
1406 | fprintf(outfile, "Partial matching not supported\n"); | |
1407 | ||
1408 | if (get_options == 0) fprintf(outfile, "No options\n"); | |
aa41d2de | 1409 | else fprintf(outfile, "Options:%s%s%s%s%s%s%s%s%s%s%s%s%s\n", |
c86f6258 PH |
1410 | ((get_options & PCRE_ANCHORED) != 0)? " anchored" : "", |
1411 | ((get_options & PCRE_CASELESS) != 0)? " caseless" : "", | |
1412 | ((get_options & PCRE_EXTENDED) != 0)? " extended" : "", | |
1413 | ((get_options & PCRE_MULTILINE) != 0)? " multiline" : "", | |
8ac170f3 | 1414 | ((get_options & PCRE_FIRSTLINE) != 0)? " firstline" : "", |
c86f6258 PH |
1415 | ((get_options & PCRE_DOTALL) != 0)? " dotall" : "", |
1416 | ((get_options & PCRE_DOLLAR_ENDONLY) != 0)? " dollar_endonly" : "", | |
1417 | ((get_options & PCRE_EXTRA) != 0)? " extra" : "", | |
1418 | ((get_options & PCRE_UNGREEDY) != 0)? " ungreedy" : "", | |
aa41d2de | 1419 | ((get_options & PCRE_NO_AUTO_CAPTURE) != 0)? " no_auto_capture" : "", |
c86f6258 | 1420 | ((get_options & PCRE_UTF8) != 0)? " utf8" : "", |
aa41d2de PH |
1421 | ((get_options & PCRE_NO_UTF8_CHECK) != 0)? " no_utf8_check" : "", |
1422 | ((get_options & PCRE_DUPNAMES) != 0)? " dupnames" : ""); | |
1423 | ||
6bf342e1 | 1424 | switch (get_options & PCRE_NEWLINE_BITS) |
aa41d2de PH |
1425 | { |
1426 | case PCRE_NEWLINE_CR: | |
1427 | fprintf(outfile, "Forced newline sequence: CR\n"); | |
1428 | break; | |
1429 | ||
1430 | case PCRE_NEWLINE_LF: | |
1431 | fprintf(outfile, "Forced newline sequence: LF\n"); | |
1432 | break; | |
c86f6258 | 1433 | |
aa41d2de PH |
1434 | case PCRE_NEWLINE_CRLF: |
1435 | fprintf(outfile, "Forced newline sequence: CRLF\n"); | |
1436 | break; | |
1437 | ||
6bf342e1 PH |
1438 | case PCRE_NEWLINE_ANY: |
1439 | fprintf(outfile, "Forced newline sequence: ANY\n"); | |
1440 | break; | |
1441 | ||
aa41d2de PH |
1442 | default: |
1443 | break; | |
1444 | } | |
c86f6258 PH |
1445 | |
1446 | if (first_char == -1) | |
1447 | { | |
aa41d2de | 1448 | fprintf(outfile, "First char at start or follows newline\n"); |
c86f6258 PH |
1449 | } |
1450 | else if (first_char < 0) | |
1451 | { | |
1452 | fprintf(outfile, "No first char\n"); | |
1453 | } | |
1454 | else | |
1455 | { | |
1456 | int ch = first_char & 255; | |
1457 | const char *caseless = ((first_char & REQ_CASELESS) == 0)? | |
1458 | "" : " (caseless)"; | |
6bf342e1 | 1459 | if (PRINTHEX(ch)) |
c86f6258 PH |
1460 | fprintf(outfile, "First char = \'%c\'%s\n", ch, caseless); |
1461 | else | |
1462 | fprintf(outfile, "First char = %d%s\n", ch, caseless); | |
1463 | } | |
1464 | ||
1465 | if (need_char < 0) | |
1466 | { | |
1467 | fprintf(outfile, "No need char\n"); | |
1468 | } | |
1469 | else | |
1470 | { | |
1471 | int ch = need_char & 255; | |
1472 | const char *caseless = ((need_char & REQ_CASELESS) == 0)? | |
1473 | "" : " (caseless)"; | |
6bf342e1 | 1474 | if (PRINTHEX(ch)) |
c86f6258 PH |
1475 | fprintf(outfile, "Need char = \'%c\'%s\n", ch, caseless); |
1476 | else | |
1477 | fprintf(outfile, "Need char = %d%s\n", ch, caseless); | |
1478 | } | |
1479 | ||
1480 | /* Don't output study size; at present it is in any case a fixed | |
1481 | value, but it varies, depending on the computer architecture, and | |
1482 | so messes up the test suite. (And with the /F option, it might be | |
1483 | flipped.) */ | |
1484 | ||
1485 | if (do_study) | |
1486 | { | |
1487 | if (extra == NULL) | |
1488 | fprintf(outfile, "Study returned NULL\n"); | |
1489 | else | |
1490 | { | |
1491 | uschar *start_bits = NULL; | |
1492 | new_info(re, extra, PCRE_INFO_FIRSTTABLE, &start_bits); | |
1493 | ||
1494 | if (start_bits == NULL) | |
1495 | fprintf(outfile, "No starting byte set\n"); | |
1496 | else | |
1497 | { | |
1498 | int i; | |
1499 | int c = 24; | |
1500 | fprintf(outfile, "Starting byte set: "); | |
1501 | for (i = 0; i < 256; i++) | |
1502 | { | |
1503 | if ((start_bits[i/8] & (1<<(i&7))) != 0) | |
1504 | { | |
1505 | if (c > 75) | |
1506 | { | |
1507 | fprintf(outfile, "\n "); | |
1508 | c = 2; | |
1509 | } | |
6bf342e1 | 1510 | if (PRINTHEX(i) && i != ' ') |
c86f6258 PH |
1511 | { |
1512 | fprintf(outfile, "%c ", i); | |
1513 | c += 2; | |
1514 | } | |
1515 | else | |
1516 | { | |
1517 | fprintf(outfile, "\\x%02x ", i); | |
1518 | c += 5; | |
1519 | } | |
1520 | } | |
1521 | } | |
1522 | fprintf(outfile, "\n"); | |
1523 | } | |
1524 | } | |
1525 | } | |
1526 | } | |
1527 | ||
1528 | /* If the '>' option was present, we write out the regex to a file, and | |
1529 | that is all. The first 8 bytes of the file are the regex length and then | |
1530 | the study length, in big-endian order. */ | |
1531 | ||
1532 | if (to_file != NULL) | |
1533 | { | |
1534 | FILE *f = fopen((char *)to_file, "wb"); | |
1535 | if (f == NULL) | |
1536 | { | |
1537 | fprintf(outfile, "Unable to open %s: %s\n", to_file, strerror(errno)); | |
1538 | } | |
1539 | else | |
1540 | { | |
1541 | uschar sbuf[8]; | |
1542 | sbuf[0] = (true_size >> 24) & 255; | |
1543 | sbuf[1] = (true_size >> 16) & 255; | |
1544 | sbuf[2] = (true_size >> 8) & 255; | |
1545 | sbuf[3] = (true_size) & 255; | |
1546 | ||
1547 | sbuf[4] = (true_study_size >> 24) & 255; | |
1548 | sbuf[5] = (true_study_size >> 16) & 255; | |
1549 | sbuf[6] = (true_study_size >> 8) & 255; | |
1550 | sbuf[7] = (true_study_size) & 255; | |
1551 | ||
1552 | if (fwrite(sbuf, 1, 8, f) < 8 || | |
1553 | fwrite(re, 1, true_size, f) < true_size) | |
1554 | { | |
1555 | fprintf(outfile, "Write error on %s: %s\n", to_file, strerror(errno)); | |
1556 | } | |
1557 | else | |
1558 | { | |
1559 | fprintf(outfile, "Compiled regex written to %s\n", to_file); | |
1560 | if (extra != NULL) | |
1561 | { | |
1562 | if (fwrite(extra->study_data, 1, true_study_size, f) < | |
1563 | true_study_size) | |
1564 | { | |
1565 | fprintf(outfile, "Write error on %s: %s\n", to_file, | |
1566 | strerror(errno)); | |
1567 | } | |
1568 | else fprintf(outfile, "Study data written to %s\n", to_file); | |
6bf342e1 | 1569 | |
c86f6258 PH |
1570 | } |
1571 | } | |
1572 | fclose(f); | |
1573 | } | |
8ac170f3 PH |
1574 | |
1575 | new_free(re); | |
1576 | if (extra != NULL) new_free(extra); | |
1577 | if (tables != NULL) new_free((void *)tables); | |
c86f6258 PH |
1578 | continue; /* With next regex */ |
1579 | } | |
1580 | } /* End of non-POSIX compile */ | |
1581 | ||
1582 | /* Read data lines and test them */ | |
1583 | ||
1584 | for (;;) | |
1585 | { | |
aa41d2de PH |
1586 | uschar *q; |
1587 | uschar *bptr = dbuffer; | |
c86f6258 PH |
1588 | int *use_offsets = offsets; |
1589 | int use_size_offsets = size_offsets; | |
1590 | int callout_data = 0; | |
1591 | int callout_data_set = 0; | |
1592 | int count, c; | |
1593 | int copystrings = 0; | |
1594 | int find_match_limit = 0; | |
1595 | int getstrings = 0; | |
1596 | int getlist = 0; | |
1597 | int gmatched = 0; | |
1598 | int start_offset = 0; | |
1599 | int g_notempty = 0; | |
8ac170f3 | 1600 | int use_dfa = 0; |
c86f6258 PH |
1601 | |
1602 | options = 0; | |
1603 | ||
aa41d2de PH |
1604 | *copynames = 0; |
1605 | *getnames = 0; | |
1606 | ||
1607 | copynamesptr = copynames; | |
1608 | getnamesptr = getnames; | |
1609 | ||
c86f6258 PH |
1610 | pcre_callout = callout; |
1611 | first_callout = 1; | |
1612 | callout_extra = 0; | |
1613 | callout_count = 0; | |
1614 | callout_fail_count = 999999; | |
1615 | callout_fail_id = -1; | |
1616 | show_malloc = 0; | |
1617 | ||
aa41d2de PH |
1618 | if (extra != NULL) extra->flags &= |
1619 | ~(PCRE_EXTRA_MATCH_LIMIT|PCRE_EXTRA_MATCH_LIMIT_RECURSION); | |
1620 | ||
1621 | len = 0; | |
1622 | for (;;) | |
c86f6258 | 1623 | { |
aa41d2de PH |
1624 | if (infile == stdin) printf("data> "); |
1625 | if (extend_inputline(infile, buffer + len) == NULL) | |
1626 | { | |
1627 | if (len > 0) break; | |
1628 | done = 1; | |
1629 | goto CONTINUE; | |
1630 | } | |
1631 | if (infile != stdin) fprintf(outfile, "%s", (char *)buffer); | |
1632 | len = (int)strlen((char *)buffer); | |
1633 | if (buffer[len-1] == '\n') break; | |
c86f6258 | 1634 | } |
c86f6258 | 1635 | |
c86f6258 PH |
1636 | while (len > 0 && isspace(buffer[len-1])) len--; |
1637 | buffer[len] = 0; | |
1638 | if (len == 0) break; | |
1639 | ||
1640 | p = buffer; | |
1641 | while (isspace(*p)) p++; | |
1642 | ||
1643 | q = dbuffer; | |
1644 | while ((c = *p++) != 0) | |
1645 | { | |
1646 | int i = 0; | |
1647 | int n = 0; | |
1648 | ||
1649 | if (c == '\\') switch ((c = *p++)) | |
1650 | { | |
1651 | case 'a': c = 7; break; | |
1652 | case 'b': c = '\b'; break; | |
1653 | case 'e': c = 27; break; | |
1654 | case 'f': c = '\f'; break; | |
1655 | case 'n': c = '\n'; break; | |
1656 | case 'r': c = '\r'; break; | |
1657 | case 't': c = '\t'; break; | |
1658 | case 'v': c = '\v'; break; | |
1659 | ||
1660 | case '0': case '1': case '2': case '3': | |
1661 | case '4': case '5': case '6': case '7': | |
1662 | c -= '0'; | |
1663 | while (i++ < 2 && isdigit(*p) && *p != '8' && *p != '9') | |
1664 | c = c * 8 + *p++ - '0'; | |
aa41d2de PH |
1665 | |
1666 | #if !defined NOUTF8 | |
1667 | if (use_utf8 && c > 255) | |
1668 | { | |
1669 | unsigned char buff8[8]; | |
1670 | int ii, utn; | |
1671 | utn = ord2utf8(c, buff8); | |
1672 | for (ii = 0; ii < utn - 1; ii++) *q++ = buff8[ii]; | |
1673 | c = buff8[ii]; /* Last byte */ | |
1674 | } | |
1675 | #endif | |
c86f6258 PH |
1676 | break; |
1677 | ||
1678 | case 'x': | |
1679 | ||
1680 | /* Handle \x{..} specially - new Perl thing for utf8 */ | |
1681 | ||
8ac170f3 | 1682 | #if !defined NOUTF8 |
c86f6258 PH |
1683 | if (*p == '{') |
1684 | { | |
1685 | unsigned char *pt = p; | |
1686 | c = 0; | |
1687 | while (isxdigit(*(++pt))) | |
1688 | c = c * 16 + tolower(*pt) - ((isdigit(*pt))? '0' : 'W'); | |
1689 | if (*pt == '}') | |
1690 | { | |
1691 | unsigned char buff8[8]; | |
1692 | int ii, utn; | |
aa41d2de | 1693 | utn = ord2utf8(c, buff8); |
c86f6258 PH |
1694 | for (ii = 0; ii < utn - 1; ii++) *q++ = buff8[ii]; |
1695 | c = buff8[ii]; /* Last byte */ | |
1696 | p = pt + 1; | |
1697 | break; | |
1698 | } | |
1699 | /* Not correct form; fall through */ | |
1700 | } | |
8ac170f3 | 1701 | #endif |
c86f6258 PH |
1702 | |
1703 | /* Ordinary \x */ | |
1704 | ||
1705 | c = 0; | |
1706 | while (i++ < 2 && isxdigit(*p)) | |
1707 | { | |
1708 | c = c * 16 + tolower(*p) - ((isdigit(*p))? '0' : 'W'); | |
1709 | p++; | |
1710 | } | |
1711 | break; | |
1712 | ||
1713 | case 0: /* \ followed by EOF allows for an empty line */ | |
1714 | p--; | |
1715 | continue; | |
1716 | ||
1717 | case '>': | |
1718 | while(isdigit(*p)) start_offset = start_offset * 10 + *p++ - '0'; | |
1719 | continue; | |
1720 | ||
1721 | case 'A': /* Option setting */ | |
1722 | options |= PCRE_ANCHORED; | |
1723 | continue; | |
1724 | ||
1725 | case 'B': | |
1726 | options |= PCRE_NOTBOL; | |
1727 | continue; | |
1728 | ||
1729 | case 'C': | |
1730 | if (isdigit(*p)) /* Set copy string */ | |
1731 | { | |
1732 | while(isdigit(*p)) n = n * 10 + *p++ - '0'; | |
1733 | copystrings |= 1 << n; | |
1734 | } | |
1735 | else if (isalnum(*p)) | |
1736 | { | |
aa41d2de | 1737 | uschar *npp = copynamesptr; |
c86f6258 | 1738 | while (isalnum(*p)) *npp++ = *p++; |
aa41d2de | 1739 | *npp++ = 0; |
c86f6258 | 1740 | *npp = 0; |
aa41d2de | 1741 | n = pcre_get_stringnumber(re, (char *)copynamesptr); |
c86f6258 | 1742 | if (n < 0) |
aa41d2de PH |
1743 | fprintf(outfile, "no parentheses with name \"%s\"\n", copynamesptr); |
1744 | copynamesptr = npp; | |
c86f6258 PH |
1745 | } |
1746 | else if (*p == '+') | |
1747 | { | |
1748 | callout_extra = 1; | |
1749 | p++; | |
1750 | } | |
1751 | else if (*p == '-') | |
1752 | { | |
1753 | pcre_callout = NULL; | |
1754 | p++; | |
1755 | } | |
1756 | else if (*p == '!') | |
1757 | { | |
1758 | callout_fail_id = 0; | |
1759 | p++; | |
1760 | while(isdigit(*p)) | |
1761 | callout_fail_id = callout_fail_id * 10 + *p++ - '0'; | |
1762 | callout_fail_count = 0; | |
1763 | if (*p == '!') | |
1764 | { | |
1765 | p++; | |
1766 | while(isdigit(*p)) | |
1767 | callout_fail_count = callout_fail_count * 10 + *p++ - '0'; | |
1768 | } | |
1769 | } | |
1770 | else if (*p == '*') | |
1771 | { | |
1772 | int sign = 1; | |
1773 | callout_data = 0; | |
1774 | if (*(++p) == '-') { sign = -1; p++; } | |
1775 | while(isdigit(*p)) | |
1776 | callout_data = callout_data * 10 + *p++ - '0'; | |
1777 | callout_data *= sign; | |
1778 | callout_data_set = 1; | |
1779 | } | |
1780 | continue; | |
1781 | ||
8ac170f3 PH |
1782 | #if !defined NODFA |
1783 | case 'D': | |
1784 | #if !defined NOPOSIX | |
1785 | if (posix || do_posix) | |
1786 | printf("** Can't use dfa matching in POSIX mode: \\D ignored\n"); | |
1787 | else | |
1788 | #endif | |
1789 | use_dfa = 1; | |
1790 | continue; | |
1791 | ||
1792 | case 'F': | |
1793 | options |= PCRE_DFA_SHORTEST; | |
1794 | continue; | |
1795 | #endif | |
1796 | ||
c86f6258 PH |
1797 | case 'G': |
1798 | if (isdigit(*p)) | |
1799 | { | |
1800 | while(isdigit(*p)) n = n * 10 + *p++ - '0'; | |
1801 | getstrings |= 1 << n; | |
1802 | } | |
1803 | else if (isalnum(*p)) | |
1804 | { | |
aa41d2de | 1805 | uschar *npp = getnamesptr; |
c86f6258 | 1806 | while (isalnum(*p)) *npp++ = *p++; |
aa41d2de | 1807 | *npp++ = 0; |
c86f6258 | 1808 | *npp = 0; |
aa41d2de | 1809 | n = pcre_get_stringnumber(re, (char *)getnamesptr); |
c86f6258 | 1810 | if (n < 0) |
aa41d2de PH |
1811 | fprintf(outfile, "no parentheses with name \"%s\"\n", getnamesptr); |
1812 | getnamesptr = npp; | |
c86f6258 PH |
1813 | } |
1814 | continue; | |
1815 | ||
1816 | case 'L': | |
1817 | getlist = 1; | |
1818 | continue; | |
1819 | ||
1820 | case 'M': | |
1821 | find_match_limit = 1; | |
1822 | continue; | |
1823 | ||
1824 | case 'N': | |
1825 | options |= PCRE_NOTEMPTY; | |
1826 | continue; | |
1827 | ||
1828 | case 'O': | |
1829 | while(isdigit(*p)) n = n * 10 + *p++ - '0'; | |
1830 | if (n > size_offsets_max) | |
1831 | { | |
1832 | size_offsets_max = n; | |
1833 | free(offsets); | |
1834 | use_offsets = offsets = (int *)malloc(size_offsets_max * sizeof(int)); | |
1835 | if (offsets == NULL) | |
1836 | { | |
1837 | printf("** Failed to get %d bytes of memory for offsets vector\n", | |
1838 | size_offsets_max * sizeof(int)); | |
8ac170f3 PH |
1839 | yield = 1; |
1840 | goto EXIT; | |
c86f6258 PH |
1841 | } |
1842 | } | |
1843 | use_size_offsets = n; | |
1844 | if (n == 0) use_offsets = NULL; /* Ensures it can't write to it */ | |
1845 | continue; | |
1846 | ||
1847 | case 'P': | |
1848 | options |= PCRE_PARTIAL; | |
1849 | continue; | |
1850 | ||
aa41d2de PH |
1851 | case 'Q': |
1852 | while(isdigit(*p)) n = n * 10 + *p++ - '0'; | |
1853 | if (extra == NULL) | |
1854 | { | |
1855 | extra = (pcre_extra *)malloc(sizeof(pcre_extra)); | |
1856 | extra->flags = 0; | |
1857 | } | |
1858 | extra->flags |= PCRE_EXTRA_MATCH_LIMIT_RECURSION; | |
1859 | extra->match_limit_recursion = n; | |
1860 | continue; | |
1861 | ||
1862 | case 'q': | |
1863 | while(isdigit(*p)) n = n * 10 + *p++ - '0'; | |
1864 | if (extra == NULL) | |
1865 | { | |
1866 | extra = (pcre_extra *)malloc(sizeof(pcre_extra)); | |
1867 | extra->flags = 0; | |
1868 | } | |
1869 | extra->flags |= PCRE_EXTRA_MATCH_LIMIT; | |
1870 | extra->match_limit = n; | |
1871 | continue; | |
1872 | ||
8ac170f3 PH |
1873 | #if !defined NODFA |
1874 | case 'R': | |
1875 | options |= PCRE_DFA_RESTART; | |
1876 | continue; | |
1877 | #endif | |
1878 | ||
c86f6258 PH |
1879 | case 'S': |
1880 | show_malloc = 1; | |
1881 | continue; | |
1882 | ||
1883 | case 'Z': | |
1884 | options |= PCRE_NOTEOL; | |
1885 | continue; | |
1886 | ||
1887 | case '?': | |
1888 | options |= PCRE_NO_UTF8_CHECK; | |
1889 | continue; | |
aa41d2de PH |
1890 | |
1891 | case '<': | |
1892 | { | |
1893 | int x = check_newline(p, outfile); | |
1894 | if (x == 0) goto NEXT_DATA; | |
1895 | options |= x; | |
1896 | while (*p++ != '>'); | |
1897 | } | |
1898 | continue; | |
c86f6258 PH |
1899 | } |
1900 | *q++ = c; | |
1901 | } | |
1902 | *q = 0; | |
1903 | len = q - dbuffer; | |
1904 | ||
8ac170f3 PH |
1905 | if ((all_use_dfa || use_dfa) && find_match_limit) |
1906 | { | |
1907 | printf("**Match limit not relevant for DFA matching: ignored\n"); | |
1908 | find_match_limit = 0; | |
1909 | } | |
1910 | ||
c86f6258 PH |
1911 | /* Handle matching via the POSIX interface, which does not |
1912 | support timing or playing with the match limit or callout data. */ | |
1913 | ||
1914 | #if !defined NOPOSIX | |
1915 | if (posix || do_posix) | |
1916 | { | |
1917 | int rc; | |
1918 | int eflags = 0; | |
1919 | regmatch_t *pmatch = NULL; | |
1920 | if (use_size_offsets > 0) | |
1921 | pmatch = (regmatch_t *)malloc(sizeof(regmatch_t) * use_size_offsets); | |
1922 | if ((options & PCRE_NOTBOL) != 0) eflags |= REG_NOTBOL; | |
1923 | if ((options & PCRE_NOTEOL) != 0) eflags |= REG_NOTEOL; | |
1924 | ||
1925 | rc = regexec(&preg, (const char *)bptr, use_size_offsets, pmatch, eflags); | |
1926 | ||
1927 | if (rc != 0) | |
1928 | { | |
aa41d2de | 1929 | (void)regerror(rc, &preg, (char *)buffer, buffer_size); |
c86f6258 PH |
1930 | fprintf(outfile, "No match: POSIX code %d: %s\n", rc, buffer); |
1931 | } | |
aa41d2de PH |
1932 | else if ((((const pcre *)preg.re_pcre)->options & PCRE_NO_AUTO_CAPTURE) |
1933 | != 0) | |
1934 | { | |
1935 | fprintf(outfile, "Matched with REG_NOSUB\n"); | |
1936 | } | |
c86f6258 PH |
1937 | else |
1938 | { | |
1939 | size_t i; | |
1940 | for (i = 0; i < (size_t)use_size_offsets; i++) | |
1941 | { | |
1942 | if (pmatch[i].rm_so >= 0) | |
1943 | { | |
1944 | fprintf(outfile, "%2d: ", (int)i); | |
1945 | (void)pchars(dbuffer + pmatch[i].rm_so, | |
1946 | pmatch[i].rm_eo - pmatch[i].rm_so, outfile); | |
1947 | fprintf(outfile, "\n"); | |
1948 | if (i == 0 && do_showrest) | |
1949 | { | |
1950 | fprintf(outfile, " 0+ "); | |
1951 | (void)pchars(dbuffer + pmatch[i].rm_eo, len - pmatch[i].rm_eo, | |
1952 | outfile); | |
1953 | fprintf(outfile, "\n"); | |
1954 | } | |
1955 | } | |
1956 | } | |
1957 | } | |
1958 | free(pmatch); | |
1959 | } | |
1960 | ||
1961 | /* Handle matching via the native interface - repeats for /g and /G */ | |
1962 | ||
1963 | else | |
1964 | #endif /* !defined NOPOSIX */ | |
1965 | ||
1966 | for (;; gmatched++) /* Loop for /g or /G */ | |
1967 | { | |
6bf342e1 | 1968 | if (timeitm > 0) |
c86f6258 PH |
1969 | { |
1970 | register int i; | |
1971 | clock_t time_taken; | |
1972 | clock_t start_time = clock(); | |
8ac170f3 PH |
1973 | |
1974 | #if !defined NODFA | |
1975 | if (all_use_dfa || use_dfa) | |
1976 | { | |
1977 | int workspace[1000]; | |
6bf342e1 | 1978 | for (i = 0; i < timeitm; i++) |
8ac170f3 PH |
1979 | count = pcre_dfa_exec(re, NULL, (char *)bptr, len, start_offset, |
1980 | options | g_notempty, use_offsets, use_size_offsets, workspace, | |
1981 | sizeof(workspace)/sizeof(int)); | |
1982 | } | |
1983 | else | |
1984 | #endif | |
1985 | ||
6bf342e1 | 1986 | for (i = 0; i < timeitm; i++) |
c86f6258 PH |
1987 | count = pcre_exec(re, extra, (char *)bptr, len, |
1988 | start_offset, options | g_notempty, use_offsets, use_size_offsets); | |
8ac170f3 | 1989 | |
c86f6258 | 1990 | time_taken = clock() - start_time; |
6bf342e1 PH |
1991 | fprintf(outfile, "Execute time %.4f milliseconds\n", |
1992 | (((double)time_taken * 1000.0) / (double)timeitm) / | |
c86f6258 PH |
1993 | (double)CLOCKS_PER_SEC); |
1994 | } | |
1995 | ||
1996 | /* If find_match_limit is set, we want to do repeated matches with | |
aa41d2de PH |
1997 | varying limits in order to find the minimum value for the match limit and |
1998 | for the recursion limit. */ | |
c86f6258 PH |
1999 | |
2000 | if (find_match_limit) | |
2001 | { | |
c86f6258 PH |
2002 | if (extra == NULL) |
2003 | { | |
2004 | extra = (pcre_extra *)malloc(sizeof(pcre_extra)); | |
2005 | extra->flags = 0; | |
2006 | } | |
c86f6258 | 2007 | |
aa41d2de PH |
2008 | (void)check_match_limit(re, extra, bptr, len, start_offset, |
2009 | options|g_notempty, use_offsets, use_size_offsets, | |
2010 | PCRE_EXTRA_MATCH_LIMIT, &(extra->match_limit), | |
2011 | PCRE_ERROR_MATCHLIMIT, "match()"); | |
c86f6258 | 2012 | |
aa41d2de PH |
2013 | count = check_match_limit(re, extra, bptr, len, start_offset, |
2014 | options|g_notempty, use_offsets, use_size_offsets, | |
2015 | PCRE_EXTRA_MATCH_LIMIT_RECURSION, &(extra->match_limit_recursion), | |
2016 | PCRE_ERROR_RECURSIONLIMIT, "match() recursion"); | |
c86f6258 PH |
2017 | } |
2018 | ||
2019 | /* If callout_data is set, use the interface with additional data */ | |
2020 | ||
2021 | else if (callout_data_set) | |
2022 | { | |
2023 | if (extra == NULL) | |
2024 | { | |
2025 | extra = (pcre_extra *)malloc(sizeof(pcre_extra)); | |
2026 | extra->flags = 0; | |
2027 | } | |
2028 | extra->flags |= PCRE_EXTRA_CALLOUT_DATA; | |
2029 | extra->callout_data = &callout_data; | |
2030 | count = pcre_exec(re, extra, (char *)bptr, len, start_offset, | |
2031 | options | g_notempty, use_offsets, use_size_offsets); | |
2032 | extra->flags &= ~PCRE_EXTRA_CALLOUT_DATA; | |
2033 | } | |
2034 | ||
2035 | /* The normal case is just to do the match once, with the default | |
2036 | value of match_limit. */ | |
2037 | ||
8ac170f3 PH |
2038 | #if !defined NODFA |
2039 | else if (all_use_dfa || use_dfa) | |
c86f6258 | 2040 | { |
8ac170f3 PH |
2041 | int workspace[1000]; |
2042 | count = pcre_dfa_exec(re, NULL, (char *)bptr, len, start_offset, | |
2043 | options | g_notempty, use_offsets, use_size_offsets, workspace, | |
2044 | sizeof(workspace)/sizeof(int)); | |
2045 | if (count == 0) | |
2046 | { | |
2047 | fprintf(outfile, "Matched, but too many subsidiary matches\n"); | |
2048 | count = use_size_offsets/2; | |
2049 | } | |
c86f6258 | 2050 | } |
8ac170f3 | 2051 | #endif |
c86f6258 | 2052 | |
8ac170f3 | 2053 | else |
c86f6258 | 2054 | { |
8ac170f3 PH |
2055 | count = pcre_exec(re, extra, (char *)bptr, len, |
2056 | start_offset, options | g_notempty, use_offsets, use_size_offsets); | |
2057 | if (count == 0) | |
2058 | { | |
2059 | fprintf(outfile, "Matched, but too many substrings\n"); | |
2060 | count = use_size_offsets/3; | |
2061 | } | |
c86f6258 PH |
2062 | } |
2063 | ||
2064 | /* Matched */ | |
2065 | ||
2066 | if (count >= 0) | |
2067 | { | |
6bf342e1 PH |
2068 | int i, maxcount; |
2069 | ||
2070 | #if !defined NODFA | |
2071 | if (all_use_dfa || use_dfa) maxcount = use_size_offsets/2; else | |
2072 | #endif | |
2073 | maxcount = use_size_offsets/3; | |
2074 | ||
2075 | /* This is a check against a lunatic return value. */ | |
2076 | ||
2077 | if (count > maxcount) | |
2078 | { | |
2079 | fprintf(outfile, | |
2080 | "** PCRE error: returned count %d is too big for offset size %d\n", | |
2081 | count, use_size_offsets); | |
2082 | count = use_size_offsets/3; | |
2083 | if (do_g || do_G) | |
2084 | { | |
2085 | fprintf(outfile, "** /%c loop abandoned\n", do_g? 'g' : 'G'); | |
2086 | do_g = do_G = FALSE; /* Break g/G loop */ | |
2087 | } | |
2088 | } | |
2089 | ||
c86f6258 PH |
2090 | for (i = 0; i < count * 2; i += 2) |
2091 | { | |
2092 | if (use_offsets[i] < 0) | |
2093 | fprintf(outfile, "%2d: <unset>\n", i/2); | |
2094 | else | |
2095 | { | |
2096 | fprintf(outfile, "%2d: ", i/2); | |
2097 | (void)pchars(bptr + use_offsets[i], | |
2098 | use_offsets[i+1] - use_offsets[i], outfile); | |
2099 | fprintf(outfile, "\n"); | |
2100 | if (i == 0) | |
2101 | { | |
2102 | if (do_showrest) | |
2103 | { | |
2104 | fprintf(outfile, " 0+ "); | |
2105 | (void)pchars(bptr + use_offsets[i+1], len - use_offsets[i+1], | |
2106 | outfile); | |
2107 | fprintf(outfile, "\n"); | |
2108 | } | |
2109 | } | |
2110 | } | |
2111 | } | |
2112 | ||
2113 | for (i = 0; i < 32; i++) | |
2114 | { | |
2115 | if ((copystrings & (1 << i)) != 0) | |
2116 | { | |
aa41d2de | 2117 | char copybuffer[256]; |
c86f6258 PH |
2118 | int rc = pcre_copy_substring((char *)bptr, use_offsets, count, |
2119 | i, copybuffer, sizeof(copybuffer)); | |
2120 | if (rc < 0) | |
2121 | fprintf(outfile, "copy substring %d failed %d\n", i, rc); | |
2122 | else | |
2123 | fprintf(outfile, "%2dC %s (%d)\n", i, copybuffer, rc); | |
2124 | } | |
2125 | } | |
2126 | ||
aa41d2de PH |
2127 | for (copynamesptr = copynames; |
2128 | *copynamesptr != 0; | |
2129 | copynamesptr += (int)strlen((char*)copynamesptr) + 1) | |
2130 | { | |
2131 | char copybuffer[256]; | |
2132 | int rc = pcre_copy_named_substring(re, (char *)bptr, use_offsets, | |
2133 | count, (char *)copynamesptr, copybuffer, sizeof(copybuffer)); | |
2134 | if (rc < 0) | |
2135 | fprintf(outfile, "copy substring %s failed %d\n", copynamesptr, rc); | |
2136 | else | |
2137 | fprintf(outfile, " C %s (%d) %s\n", copybuffer, rc, copynamesptr); | |
2138 | } | |
2139 | ||
c86f6258 PH |
2140 | for (i = 0; i < 32; i++) |
2141 | { | |
2142 | if ((getstrings & (1 << i)) != 0) | |
2143 | { | |
2144 | const char *substring; | |
2145 | int rc = pcre_get_substring((char *)bptr, use_offsets, count, | |
2146 | i, &substring); | |
2147 | if (rc < 0) | |
2148 | fprintf(outfile, "get substring %d failed %d\n", i, rc); | |
2149 | else | |
2150 | { | |
2151 | fprintf(outfile, "%2dG %s (%d)\n", i, substring, rc); | |
c86f6258 PH |
2152 | pcre_free_substring(substring); |
2153 | } | |
2154 | } | |
2155 | } | |
2156 | ||
aa41d2de PH |
2157 | for (getnamesptr = getnames; |
2158 | *getnamesptr != 0; | |
2159 | getnamesptr += (int)strlen((char*)getnamesptr) + 1) | |
2160 | { | |
2161 | const char *substring; | |
2162 | int rc = pcre_get_named_substring(re, (char *)bptr, use_offsets, | |
2163 | count, (char *)getnamesptr, &substring); | |
2164 | if (rc < 0) | |
2165 | fprintf(outfile, "copy substring %s failed %d\n", getnamesptr, rc); | |
2166 | else | |
2167 | { | |
2168 | fprintf(outfile, " G %s (%d) %s\n", substring, rc, getnamesptr); | |
2169 | pcre_free_substring(substring); | |
2170 | } | |
2171 | } | |
2172 | ||
c86f6258 PH |
2173 | if (getlist) |
2174 | { | |
2175 | const char **stringlist; | |
2176 | int rc = pcre_get_substring_list((char *)bptr, use_offsets, count, | |
2177 | &stringlist); | |
2178 | if (rc < 0) | |
2179 | fprintf(outfile, "get substring list failed %d\n", rc); | |
2180 | else | |
2181 | { | |
2182 | for (i = 0; i < count; i++) | |
2183 | fprintf(outfile, "%2dL %s\n", i, stringlist[i]); | |
2184 | if (stringlist[i] != NULL) | |
2185 | fprintf(outfile, "string list not terminated by NULL\n"); | |
2186 | /* free((void *)stringlist); */ | |
2187 | pcre_free_substring_list(stringlist); | |
2188 | } | |
2189 | } | |
2190 | } | |
2191 | ||
2192 | /* There was a partial match */ | |
2193 | ||
2194 | else if (count == PCRE_ERROR_PARTIAL) | |
2195 | { | |
8ac170f3 PH |
2196 | fprintf(outfile, "Partial match"); |
2197 | #if !defined NODFA | |
2198 | if ((all_use_dfa || use_dfa) && use_size_offsets > 2) | |
2199 | fprintf(outfile, ": %.*s", use_offsets[1] - use_offsets[0], | |
2200 | bptr + use_offsets[0]); | |
2201 | #endif | |
2202 | fprintf(outfile, "\n"); | |
c86f6258 PH |
2203 | break; /* Out of the /g loop */ |
2204 | } | |
2205 | ||
2206 | /* Failed to match. If this is a /g or /G loop and we previously set | |
2207 | g_notempty after a null match, this is not necessarily the end. | |
2208 | We want to advance the start offset, and continue. In the case of UTF-8 | |
2209 | matching, the advance must be one character, not one byte. Fudge the | |
2210 | offset values to achieve this. We won't be at the end of the string - | |
2211 | that was checked before setting g_notempty. */ | |
2212 | ||
2213 | else | |
2214 | { | |
2215 | if (g_notempty != 0) | |
2216 | { | |
2217 | int onechar = 1; | |
2218 | use_offsets[0] = start_offset; | |
2219 | if (use_utf8) | |
2220 | { | |
2221 | while (start_offset + onechar < len) | |
2222 | { | |
2223 | int tb = bptr[start_offset+onechar]; | |
2224 | if (tb <= 127) break; | |
2225 | tb &= 0xc0; | |
2226 | if (tb != 0 && tb != 0xc0) onechar++; | |
2227 | } | |
2228 | } | |
2229 | use_offsets[1] = start_offset + onechar; | |
2230 | } | |
2231 | else | |
2232 | { | |
2233 | if (count == PCRE_ERROR_NOMATCH) | |
2234 | { | |
2235 | if (gmatched == 0) fprintf(outfile, "No match\n"); | |
2236 | } | |
2237 | else fprintf(outfile, "Error %d\n", count); | |
2238 | break; /* Out of the /g loop */ | |
2239 | } | |
2240 | } | |
2241 | ||
2242 | /* If not /g or /G we are done */ | |
2243 | ||
2244 | if (!do_g && !do_G) break; | |
2245 | ||
2246 | /* If we have matched an empty string, first check to see if we are at | |
2247 | the end of the subject. If so, the /g loop is over. Otherwise, mimic | |
2248 | what Perl's /g options does. This turns out to be rather cunning. First | |
2249 | we set PCRE_NOTEMPTY and PCRE_ANCHORED and try the match again at the | |
2250 | same point. If this fails (picked up above) we advance to the next | |
2251 | character. */ | |
2252 | ||
2253 | g_notempty = 0; | |
2254 | if (use_offsets[0] == use_offsets[1]) | |
2255 | { | |
2256 | if (use_offsets[0] == len) break; | |
2257 | g_notempty = PCRE_NOTEMPTY | PCRE_ANCHORED; | |
2258 | } | |
2259 | ||
2260 | /* For /g, update the start offset, leaving the rest alone */ | |
2261 | ||
2262 | if (do_g) start_offset = use_offsets[1]; | |
2263 | ||
2264 | /* For /G, update the pointer and length */ | |
2265 | ||
2266 | else | |
2267 | { | |
2268 | bptr += use_offsets[1]; | |
2269 | len -= use_offsets[1]; | |
2270 | } | |
2271 | } /* End of loop for /g and /G */ | |
aa41d2de PH |
2272 | |
2273 | NEXT_DATA: continue; | |
c86f6258 PH |
2274 | } /* End of loop for data lines */ |
2275 | ||
2276 | CONTINUE: | |
2277 | ||
2278 | #if !defined NOPOSIX | |
2279 | if (posix || do_posix) regfree(&preg); | |
2280 | #endif | |
2281 | ||
8ac170f3 PH |
2282 | if (re != NULL) new_free(re); |
2283 | if (extra != NULL) new_free(extra); | |
c86f6258 PH |
2284 | if (tables != NULL) |
2285 | { | |
8ac170f3 | 2286 | new_free((void *)tables); |
c86f6258 | 2287 | setlocale(LC_CTYPE, "C"); |
6bf342e1 | 2288 | locale_set = 0; |
c86f6258 PH |
2289 | } |
2290 | } | |
2291 | ||
2292 | if (infile == stdin) fprintf(outfile, "\n"); | |
8ac170f3 PH |
2293 | |
2294 | EXIT: | |
2295 | ||
2296 | if (infile != NULL && infile != stdin) fclose(infile); | |
2297 | if (outfile != NULL && outfile != stdout) fclose(outfile); | |
2298 | ||
2299 | free(buffer); | |
2300 | free(dbuffer); | |
2301 | free(pbuffer); | |
2302 | free(offsets); | |
2303 | ||
2304 | return yield; | |
c86f6258 PH |
2305 | } |
2306 | ||
8ac170f3 | 2307 | /* End of pcretest.c */ |