| 1 | /* $Cambridge: exim/src/src/pcre/pcretest.c,v 1.2 2005/06/15 08:57:10 ph10 Exp $ */ |
| 2 | |
| 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 |
| 9 | been extended and consequently is now rather, er, *very* untidy in places. |
| 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 | |
| 49 | #define PCRE_SPY /* For Win32 build, import data, not export */ |
| 50 | |
| 51 | /* We need the internal info for displaying the results of pcre_study() and |
| 52 | other internal data; pcretest also uses some of the fixed tables, and generally |
| 53 | has "inside information" compared to a program that strictly follows the PCRE |
| 54 | API. */ |
| 55 | |
| 56 | #include "pcre_internal.h" |
| 57 | |
| 58 | |
| 59 | /* It is possible to compile this test program without including support for |
| 60 | testing the POSIX interface, though this is not available via the standard |
| 61 | Makefile. */ |
| 62 | |
| 63 | #if !defined NOPOSIX |
| 64 | #include "pcreposix.h" |
| 65 | #endif |
| 66 | |
| 67 | /* It is also possible, for the benefit of the version imported into Exim, to |
| 68 | build pcretest without support for UTF8 (define NOUTF8), without the interface |
| 69 | to the DFA matcher (NODFA), and without the doublecheck of the old "info" |
| 70 | function (define NOINFOCHECK). */ |
| 71 | |
| 72 | |
| 73 | #ifndef CLOCKS_PER_SEC |
| 74 | #ifdef CLK_TCK |
| 75 | #define CLOCKS_PER_SEC CLK_TCK |
| 76 | #else |
| 77 | #define CLOCKS_PER_SEC 100 |
| 78 | #endif |
| 79 | #endif |
| 80 | |
| 81 | #define LOOPREPEAT 500000 |
| 82 | |
| 83 | #define BUFFER_SIZE 30000 |
| 84 | #define PBUFFER_SIZE BUFFER_SIZE |
| 85 | #define DBUFFER_SIZE BUFFER_SIZE |
| 86 | |
| 87 | |
| 88 | static FILE *outfile; |
| 89 | static int log_store = 0; |
| 90 | static int callout_count; |
| 91 | static int callout_extra; |
| 92 | static int callout_fail_count; |
| 93 | static int callout_fail_id; |
| 94 | static int first_callout; |
| 95 | static int show_malloc; |
| 96 | static int use_utf8; |
| 97 | static size_t gotten_store; |
| 98 | |
| 99 | static uschar *pbuffer = NULL; |
| 100 | |
| 101 | |
| 102 | |
| 103 | /************************************************* |
| 104 | * Read number from string * |
| 105 | *************************************************/ |
| 106 | |
| 107 | /* We don't use strtoul() because SunOS4 doesn't have it. Rather than mess |
| 108 | around with conditional compilation, just do the job by hand. It is only used |
| 109 | for unpicking the -o argument, so just keep it simple. |
| 110 | |
| 111 | Arguments: |
| 112 | str string to be converted |
| 113 | endptr where to put the end pointer |
| 114 | |
| 115 | Returns: the unsigned long |
| 116 | */ |
| 117 | |
| 118 | static int |
| 119 | get_value(unsigned char *str, unsigned char **endptr) |
| 120 | { |
| 121 | int result = 0; |
| 122 | while(*str != 0 && isspace(*str)) str++; |
| 123 | while (isdigit(*str)) result = result * 10 + (int)(*str++ - '0'); |
| 124 | *endptr = str; |
| 125 | return(result); |
| 126 | } |
| 127 | |
| 128 | |
| 129 | |
| 130 | |
| 131 | /************************************************* |
| 132 | * Convert UTF-8 string to value * |
| 133 | *************************************************/ |
| 134 | |
| 135 | /* This function takes one or more bytes that represents a UTF-8 character, |
| 136 | and returns the value of the character. |
| 137 | |
| 138 | Argument: |
| 139 | buffer a pointer to the byte vector |
| 140 | vptr a pointer to an int to receive the value |
| 141 | |
| 142 | Returns: > 0 => the number of bytes consumed |
| 143 | -6 to 0 => malformed UTF-8 character at offset = (-return) |
| 144 | */ |
| 145 | |
| 146 | #if !defined NOUTF8 |
| 147 | |
| 148 | static int |
| 149 | utf82ord(unsigned char *buffer, int *vptr) |
| 150 | { |
| 151 | int c = *buffer++; |
| 152 | int d = c; |
| 153 | int i, j, s; |
| 154 | |
| 155 | for (i = -1; i < 6; i++) /* i is number of additional bytes */ |
| 156 | { |
| 157 | if ((d & 0x80) == 0) break; |
| 158 | d <<= 1; |
| 159 | } |
| 160 | |
| 161 | if (i == -1) { *vptr = c; return 1; } /* ascii character */ |
| 162 | if (i == 0 || i == 6) return 0; /* invalid UTF-8 */ |
| 163 | |
| 164 | /* i now has a value in the range 1-5 */ |
| 165 | |
| 166 | s = 6*i; |
| 167 | d = (c & _pcre_utf8_table3[i]) << s; |
| 168 | |
| 169 | for (j = 0; j < i; j++) |
| 170 | { |
| 171 | c = *buffer++; |
| 172 | if ((c & 0xc0) != 0x80) return -(j+1); |
| 173 | s -= 6; |
| 174 | d |= (c & 0x3f) << s; |
| 175 | } |
| 176 | |
| 177 | /* Check that encoding was the correct unique one */ |
| 178 | |
| 179 | for (j = 0; j < _pcre_utf8_table1_size; j++) |
| 180 | if (d <= _pcre_utf8_table1[j]) break; |
| 181 | if (j != i) return -(i+1); |
| 182 | |
| 183 | /* Valid value */ |
| 184 | |
| 185 | *vptr = d; |
| 186 | return i+1; |
| 187 | } |
| 188 | |
| 189 | #endif |
| 190 | |
| 191 | |
| 192 | |
| 193 | /************************************************* |
| 194 | * Print character string * |
| 195 | *************************************************/ |
| 196 | |
| 197 | /* Character string printing function. Must handle UTF-8 strings in utf8 |
| 198 | mode. Yields number of characters printed. If handed a NULL file, just counts |
| 199 | chars without printing. */ |
| 200 | |
| 201 | static int pchars(unsigned char *p, int length, FILE *f) |
| 202 | { |
| 203 | int c; |
| 204 | int yield = 0; |
| 205 | |
| 206 | while (length-- > 0) |
| 207 | { |
| 208 | #if !defined NOUTF8 |
| 209 | if (use_utf8) |
| 210 | { |
| 211 | int rc = utf82ord(p, &c); |
| 212 | |
| 213 | if (rc > 0 && rc <= length + 1) /* Mustn't run over the end */ |
| 214 | { |
| 215 | length -= rc - 1; |
| 216 | p += rc; |
| 217 | if (c < 256 && isprint(c)) |
| 218 | { |
| 219 | if (f != NULL) fprintf(f, "%c", c); |
| 220 | yield++; |
| 221 | } |
| 222 | else |
| 223 | { |
| 224 | int n; |
| 225 | if (f != NULL) fprintf(f, "\\x{%02x}%n", c, &n); |
| 226 | yield += n; |
| 227 | } |
| 228 | continue; |
| 229 | } |
| 230 | } |
| 231 | #endif |
| 232 | |
| 233 | /* Not UTF-8, or malformed UTF-8 */ |
| 234 | |
| 235 | if (isprint(c = *(p++))) |
| 236 | { |
| 237 | if (f != NULL) fprintf(f, "%c", c); |
| 238 | yield++; |
| 239 | } |
| 240 | else |
| 241 | { |
| 242 | if (f != NULL) fprintf(f, "\\x%02x", c); |
| 243 | yield += 4; |
| 244 | } |
| 245 | } |
| 246 | |
| 247 | return yield; |
| 248 | } |
| 249 | |
| 250 | |
| 251 | |
| 252 | /************************************************* |
| 253 | * Callout function * |
| 254 | *************************************************/ |
| 255 | |
| 256 | /* Called from PCRE as a result of the (?C) item. We print out where we are in |
| 257 | the match. Yield zero unless more callouts than the fail count, or the callout |
| 258 | data is not zero. */ |
| 259 | |
| 260 | static int callout(pcre_callout_block *cb) |
| 261 | { |
| 262 | FILE *f = (first_callout | callout_extra)? outfile : NULL; |
| 263 | int i, pre_start, post_start, subject_length; |
| 264 | |
| 265 | if (callout_extra) |
| 266 | { |
| 267 | fprintf(f, "Callout %d: last capture = %d\n", |
| 268 | cb->callout_number, cb->capture_last); |
| 269 | |
| 270 | for (i = 0; i < cb->capture_top * 2; i += 2) |
| 271 | { |
| 272 | if (cb->offset_vector[i] < 0) |
| 273 | fprintf(f, "%2d: <unset>\n", i/2); |
| 274 | else |
| 275 | { |
| 276 | fprintf(f, "%2d: ", i/2); |
| 277 | (void)pchars((unsigned char *)cb->subject + cb->offset_vector[i], |
| 278 | cb->offset_vector[i+1] - cb->offset_vector[i], f); |
| 279 | fprintf(f, "\n"); |
| 280 | } |
| 281 | } |
| 282 | } |
| 283 | |
| 284 | /* Re-print the subject in canonical form, the first time or if giving full |
| 285 | datails. On subsequent calls in the same match, we use pchars just to find the |
| 286 | printed lengths of the substrings. */ |
| 287 | |
| 288 | if (f != NULL) fprintf(f, "--->"); |
| 289 | |
| 290 | pre_start = pchars((unsigned char *)cb->subject, cb->start_match, f); |
| 291 | post_start = pchars((unsigned char *)(cb->subject + cb->start_match), |
| 292 | cb->current_position - cb->start_match, f); |
| 293 | |
| 294 | subject_length = pchars((unsigned char *)cb->subject, cb->subject_length, NULL); |
| 295 | |
| 296 | (void)pchars((unsigned char *)(cb->subject + cb->current_position), |
| 297 | cb->subject_length - cb->current_position, f); |
| 298 | |
| 299 | if (f != NULL) fprintf(f, "\n"); |
| 300 | |
| 301 | /* Always print appropriate indicators, with callout number if not already |
| 302 | shown. For automatic callouts, show the pattern offset. */ |
| 303 | |
| 304 | if (cb->callout_number == 255) |
| 305 | { |
| 306 | fprintf(outfile, "%+3d ", cb->pattern_position); |
| 307 | if (cb->pattern_position > 99) fprintf(outfile, "\n "); |
| 308 | } |
| 309 | else |
| 310 | { |
| 311 | if (callout_extra) fprintf(outfile, " "); |
| 312 | else fprintf(outfile, "%3d ", cb->callout_number); |
| 313 | } |
| 314 | |
| 315 | for (i = 0; i < pre_start; i++) fprintf(outfile, " "); |
| 316 | fprintf(outfile, "^"); |
| 317 | |
| 318 | if (post_start > 0) |
| 319 | { |
| 320 | for (i = 0; i < post_start - 1; i++) fprintf(outfile, " "); |
| 321 | fprintf(outfile, "^"); |
| 322 | } |
| 323 | |
| 324 | for (i = 0; i < subject_length - pre_start - post_start + 4; i++) |
| 325 | fprintf(outfile, " "); |
| 326 | |
| 327 | fprintf(outfile, "%.*s", (cb->next_item_length == 0)? 1 : cb->next_item_length, |
| 328 | pbuffer + cb->pattern_position); |
| 329 | |
| 330 | fprintf(outfile, "\n"); |
| 331 | first_callout = 0; |
| 332 | |
| 333 | if (cb->callout_data != NULL) |
| 334 | { |
| 335 | int callout_data = *((int *)(cb->callout_data)); |
| 336 | if (callout_data != 0) |
| 337 | { |
| 338 | fprintf(outfile, "Callout data = %d\n", callout_data); |
| 339 | return callout_data; |
| 340 | } |
| 341 | } |
| 342 | |
| 343 | return (cb->callout_number != callout_fail_id)? 0 : |
| 344 | (++callout_count >= callout_fail_count)? 1 : 0; |
| 345 | } |
| 346 | |
| 347 | |
| 348 | /************************************************* |
| 349 | * Local malloc functions * |
| 350 | *************************************************/ |
| 351 | |
| 352 | /* Alternative malloc function, to test functionality and show the size of the |
| 353 | compiled re. */ |
| 354 | |
| 355 | static void *new_malloc(size_t size) |
| 356 | { |
| 357 | void *block = malloc(size); |
| 358 | gotten_store = size; |
| 359 | if (show_malloc) |
| 360 | fprintf(outfile, "malloc %3d %p\n", (int)size, block); |
| 361 | return block; |
| 362 | } |
| 363 | |
| 364 | static void new_free(void *block) |
| 365 | { |
| 366 | if (show_malloc) |
| 367 | fprintf(outfile, "free %p\n", block); |
| 368 | free(block); |
| 369 | } |
| 370 | |
| 371 | |
| 372 | /* For recursion malloc/free, to test stacking calls */ |
| 373 | |
| 374 | static void *stack_malloc(size_t size) |
| 375 | { |
| 376 | void *block = malloc(size); |
| 377 | if (show_malloc) |
| 378 | fprintf(outfile, "stack_malloc %3d %p\n", (int)size, block); |
| 379 | return block; |
| 380 | } |
| 381 | |
| 382 | static void stack_free(void *block) |
| 383 | { |
| 384 | if (show_malloc) |
| 385 | fprintf(outfile, "stack_free %p\n", block); |
| 386 | free(block); |
| 387 | } |
| 388 | |
| 389 | |
| 390 | /************************************************* |
| 391 | * Call pcre_fullinfo() * |
| 392 | *************************************************/ |
| 393 | |
| 394 | /* Get one piece of information from the pcre_fullinfo() function */ |
| 395 | |
| 396 | static void new_info(pcre *re, pcre_extra *study, int option, void *ptr) |
| 397 | { |
| 398 | int rc; |
| 399 | if ((rc = pcre_fullinfo(re, study, option, ptr)) < 0) |
| 400 | fprintf(outfile, "Error %d from pcre_fullinfo(%d)\n", rc, option); |
| 401 | } |
| 402 | |
| 403 | |
| 404 | |
| 405 | /************************************************* |
| 406 | * Byte flipping function * |
| 407 | *************************************************/ |
| 408 | |
| 409 | static long int |
| 410 | byteflip(long int value, int n) |
| 411 | { |
| 412 | if (n == 2) return ((value & 0x00ff) << 8) | ((value & 0xff00) >> 8); |
| 413 | return ((value & 0x000000ff) << 24) | |
| 414 | ((value & 0x0000ff00) << 8) | |
| 415 | ((value & 0x00ff0000) >> 8) | |
| 416 | ((value & 0xff000000) >> 24); |
| 417 | } |
| 418 | |
| 419 | |
| 420 | |
| 421 | |
| 422 | /************************************************* |
| 423 | * Main Program * |
| 424 | *************************************************/ |
| 425 | |
| 426 | /* Read lines from named file or stdin and write to named file or stdout; lines |
| 427 | consist of a regular expression, in delimiters and optionally followed by |
| 428 | options, followed by a set of test data, terminated by an empty line. */ |
| 429 | |
| 430 | int main(int argc, char **argv) |
| 431 | { |
| 432 | FILE *infile = stdin; |
| 433 | int options = 0; |
| 434 | int study_options = 0; |
| 435 | int op = 1; |
| 436 | int timeit = 0; |
| 437 | int showinfo = 0; |
| 438 | int showstore = 0; |
| 439 | int size_offsets = 45; |
| 440 | int size_offsets_max; |
| 441 | int *offsets = NULL; |
| 442 | #if !defined NOPOSIX |
| 443 | int posix = 0; |
| 444 | #endif |
| 445 | int debug = 0; |
| 446 | int done = 0; |
| 447 | int all_use_dfa = 0; |
| 448 | int yield = 0; |
| 449 | |
| 450 | unsigned char *buffer; |
| 451 | unsigned char *dbuffer; |
| 452 | |
| 453 | /* Get buffers from malloc() so that Electric Fence will check their misuse |
| 454 | when I am debugging. */ |
| 455 | |
| 456 | buffer = (unsigned char *)malloc(BUFFER_SIZE); |
| 457 | dbuffer = (unsigned char *)malloc(DBUFFER_SIZE); |
| 458 | pbuffer = (unsigned char *)malloc(PBUFFER_SIZE); |
| 459 | |
| 460 | /* The outfile variable is static so that new_malloc can use it. The _setmode() |
| 461 | stuff is some magic that I don't understand, but which apparently does good |
| 462 | things in Windows. It's related to line terminations. */ |
| 463 | |
| 464 | #if defined(_WIN32) || defined(WIN32) |
| 465 | _setmode( _fileno( stdout ), 0x8000 ); |
| 466 | #endif /* defined(_WIN32) || defined(WIN32) */ |
| 467 | |
| 468 | outfile = stdout; |
| 469 | |
| 470 | /* Scan options */ |
| 471 | |
| 472 | while (argc > 1 && argv[op][0] == '-') |
| 473 | { |
| 474 | unsigned char *endptr; |
| 475 | |
| 476 | if (strcmp(argv[op], "-s") == 0 || strcmp(argv[op], "-m") == 0) |
| 477 | showstore = 1; |
| 478 | else if (strcmp(argv[op], "-t") == 0) timeit = 1; |
| 479 | else if (strcmp(argv[op], "-i") == 0) showinfo = 1; |
| 480 | else if (strcmp(argv[op], "-d") == 0) showinfo = debug = 1; |
| 481 | #if !defined NODFA |
| 482 | else if (strcmp(argv[op], "-dfa") == 0) all_use_dfa = 1; |
| 483 | #endif |
| 484 | else if (strcmp(argv[op], "-o") == 0 && argc > 2 && |
| 485 | ((size_offsets = get_value((unsigned char *)argv[op+1], &endptr)), |
| 486 | *endptr == 0)) |
| 487 | { |
| 488 | op++; |
| 489 | argc--; |
| 490 | } |
| 491 | #if !defined NOPOSIX |
| 492 | else if (strcmp(argv[op], "-p") == 0) posix = 1; |
| 493 | #endif |
| 494 | else if (strcmp(argv[op], "-C") == 0) |
| 495 | { |
| 496 | int rc; |
| 497 | printf("PCRE version %s\n", pcre_version()); |
| 498 | printf("Compiled with\n"); |
| 499 | (void)pcre_config(PCRE_CONFIG_UTF8, &rc); |
| 500 | printf(" %sUTF-8 support\n", rc? "" : "No "); |
| 501 | (void)pcre_config(PCRE_CONFIG_UNICODE_PROPERTIES, &rc); |
| 502 | printf(" %sUnicode properties support\n", rc? "" : "No "); |
| 503 | (void)pcre_config(PCRE_CONFIG_NEWLINE, &rc); |
| 504 | printf(" Newline character is %s\n", (rc == '\r')? "CR" : "LF"); |
| 505 | (void)pcre_config(PCRE_CONFIG_LINK_SIZE, &rc); |
| 506 | printf(" Internal link size = %d\n", rc); |
| 507 | (void)pcre_config(PCRE_CONFIG_POSIX_MALLOC_THRESHOLD, &rc); |
| 508 | printf(" POSIX malloc threshold = %d\n", rc); |
| 509 | (void)pcre_config(PCRE_CONFIG_MATCH_LIMIT, &rc); |
| 510 | printf(" Default match limit = %d\n", rc); |
| 511 | (void)pcre_config(PCRE_CONFIG_STACKRECURSE, &rc); |
| 512 | printf(" Match recursion uses %s\n", rc? "stack" : "heap"); |
| 513 | exit(0); |
| 514 | } |
| 515 | else |
| 516 | { |
| 517 | printf("** Unknown or malformed option %s\n", argv[op]); |
| 518 | printf("Usage: pcretest [-d] [-i] [-o <n>] [-p] [-s] [-t] [<input> [<output>]]\n"); |
| 519 | printf(" -C show PCRE compile-time options and exit\n"); |
| 520 | printf(" -d debug: show compiled code; implies -i\n"); |
| 521 | #if !defined NODFA |
| 522 | printf(" -dfa force DFA matching for all subjects\n"); |
| 523 | #endif |
| 524 | printf(" -i show information about compiled pattern\n" |
| 525 | " -m output memory used information\n" |
| 526 | " -o <n> set size of offsets vector to <n>\n"); |
| 527 | #if !defined NOPOSIX |
| 528 | printf(" -p use POSIX interface\n"); |
| 529 | #endif |
| 530 | printf(" -s output store (memory) used information\n" |
| 531 | " -t time compilation and execution\n"); |
| 532 | yield = 1; |
| 533 | goto EXIT; |
| 534 | } |
| 535 | op++; |
| 536 | argc--; |
| 537 | } |
| 538 | |
| 539 | /* Get the store for the offsets vector, and remember what it was */ |
| 540 | |
| 541 | size_offsets_max = size_offsets; |
| 542 | offsets = (int *)malloc(size_offsets_max * sizeof(int)); |
| 543 | if (offsets == NULL) |
| 544 | { |
| 545 | printf("** Failed to get %d bytes of memory for offsets vector\n", |
| 546 | size_offsets_max * sizeof(int)); |
| 547 | yield = 1; |
| 548 | goto EXIT; |
| 549 | } |
| 550 | |
| 551 | /* Sort out the input and output files */ |
| 552 | |
| 553 | if (argc > 1) |
| 554 | { |
| 555 | infile = fopen(argv[op], "rb"); |
| 556 | if (infile == NULL) |
| 557 | { |
| 558 | printf("** Failed to open %s\n", argv[op]); |
| 559 | yield = 1; |
| 560 | goto EXIT; |
| 561 | } |
| 562 | } |
| 563 | |
| 564 | if (argc > 2) |
| 565 | { |
| 566 | outfile = fopen(argv[op+1], "wb"); |
| 567 | if (outfile == NULL) |
| 568 | { |
| 569 | printf("** Failed to open %s\n", argv[op+1]); |
| 570 | yield = 1; |
| 571 | goto EXIT; |
| 572 | } |
| 573 | } |
| 574 | |
| 575 | /* Set alternative malloc function */ |
| 576 | |
| 577 | pcre_malloc = new_malloc; |
| 578 | pcre_free = new_free; |
| 579 | pcre_stack_malloc = stack_malloc; |
| 580 | pcre_stack_free = stack_free; |
| 581 | |
| 582 | /* Heading line, then prompt for first regex if stdin */ |
| 583 | |
| 584 | fprintf(outfile, "PCRE version %s\n\n", pcre_version()); |
| 585 | |
| 586 | /* Main loop */ |
| 587 | |
| 588 | while (!done) |
| 589 | { |
| 590 | pcre *re = NULL; |
| 591 | pcre_extra *extra = NULL; |
| 592 | |
| 593 | #if !defined NOPOSIX /* There are still compilers that require no indent */ |
| 594 | regex_t preg; |
| 595 | int do_posix = 0; |
| 596 | #endif |
| 597 | |
| 598 | const char *error; |
| 599 | unsigned char *p, *pp, *ppp; |
| 600 | unsigned char *to_file = NULL; |
| 601 | const unsigned char *tables = NULL; |
| 602 | unsigned long int true_size, true_study_size = 0; |
| 603 | size_t size, regex_gotten_store; |
| 604 | int do_study = 0; |
| 605 | int do_debug = debug; |
| 606 | int do_G = 0; |
| 607 | int do_g = 0; |
| 608 | int do_showinfo = showinfo; |
| 609 | int do_showrest = 0; |
| 610 | int do_flip = 0; |
| 611 | int erroroffset, len, delimiter; |
| 612 | |
| 613 | use_utf8 = 0; |
| 614 | |
| 615 | if (infile == stdin) printf(" re> "); |
| 616 | if (fgets((char *)buffer, BUFFER_SIZE, infile) == NULL) break; |
| 617 | if (infile != stdin) fprintf(outfile, "%s", (char *)buffer); |
| 618 | fflush(outfile); |
| 619 | |
| 620 | p = buffer; |
| 621 | while (isspace(*p)) p++; |
| 622 | if (*p == 0) continue; |
| 623 | |
| 624 | /* See if the pattern is to be loaded pre-compiled from a file. */ |
| 625 | |
| 626 | if (*p == '<' && strchr((char *)(p+1), '<') == NULL) |
| 627 | { |
| 628 | unsigned long int magic; |
| 629 | uschar sbuf[8]; |
| 630 | FILE *f; |
| 631 | |
| 632 | p++; |
| 633 | pp = p + (int)strlen((char *)p); |
| 634 | while (isspace(pp[-1])) pp--; |
| 635 | *pp = 0; |
| 636 | |
| 637 | f = fopen((char *)p, "rb"); |
| 638 | if (f == NULL) |
| 639 | { |
| 640 | fprintf(outfile, "Failed to open %s: %s\n", p, strerror(errno)); |
| 641 | continue; |
| 642 | } |
| 643 | |
| 644 | if (fread(sbuf, 1, 8, f) != 8) goto FAIL_READ; |
| 645 | |
| 646 | true_size = |
| 647 | (sbuf[0] << 24) | (sbuf[1] << 16) | (sbuf[2] << 8) | sbuf[3]; |
| 648 | true_study_size = |
| 649 | (sbuf[4] << 24) | (sbuf[5] << 16) | (sbuf[6] << 8) | sbuf[7]; |
| 650 | |
| 651 | re = (real_pcre *)new_malloc(true_size); |
| 652 | regex_gotten_store = gotten_store; |
| 653 | |
| 654 | if (fread(re, 1, true_size, f) != true_size) goto FAIL_READ; |
| 655 | |
| 656 | magic = ((real_pcre *)re)->magic_number; |
| 657 | if (magic != MAGIC_NUMBER) |
| 658 | { |
| 659 | if (byteflip(magic, sizeof(magic)) == MAGIC_NUMBER) |
| 660 | { |
| 661 | do_flip = 1; |
| 662 | } |
| 663 | else |
| 664 | { |
| 665 | fprintf(outfile, "Data in %s is not a compiled PCRE regex\n", p); |
| 666 | fclose(f); |
| 667 | continue; |
| 668 | } |
| 669 | } |
| 670 | |
| 671 | fprintf(outfile, "Compiled regex%s loaded from %s\n", |
| 672 | do_flip? " (byte-inverted)" : "", p); |
| 673 | |
| 674 | /* Need to know if UTF-8 for printing data strings */ |
| 675 | |
| 676 | new_info(re, NULL, PCRE_INFO_OPTIONS, &options); |
| 677 | use_utf8 = (options & PCRE_UTF8) != 0; |
| 678 | |
| 679 | /* Now see if there is any following study data */ |
| 680 | |
| 681 | if (true_study_size != 0) |
| 682 | { |
| 683 | pcre_study_data *psd; |
| 684 | |
| 685 | extra = (pcre_extra *)new_malloc(sizeof(pcre_extra) + true_study_size); |
| 686 | extra->flags = PCRE_EXTRA_STUDY_DATA; |
| 687 | |
| 688 | psd = (pcre_study_data *)(((char *)extra) + sizeof(pcre_extra)); |
| 689 | extra->study_data = psd; |
| 690 | |
| 691 | if (fread(psd, 1, true_study_size, f) != true_study_size) |
| 692 | { |
| 693 | FAIL_READ: |
| 694 | fprintf(outfile, "Failed to read data from %s\n", p); |
| 695 | if (extra != NULL) new_free(extra); |
| 696 | if (re != NULL) new_free(re); |
| 697 | fclose(f); |
| 698 | continue; |
| 699 | } |
| 700 | fprintf(outfile, "Study data loaded from %s\n", p); |
| 701 | do_study = 1; /* To get the data output if requested */ |
| 702 | } |
| 703 | else fprintf(outfile, "No study data\n"); |
| 704 | |
| 705 | fclose(f); |
| 706 | goto SHOW_INFO; |
| 707 | } |
| 708 | |
| 709 | /* In-line pattern (the usual case). Get the delimiter and seek the end of |
| 710 | the pattern; if is isn't complete, read more. */ |
| 711 | |
| 712 | delimiter = *p++; |
| 713 | |
| 714 | if (isalnum(delimiter) || delimiter == '\\') |
| 715 | { |
| 716 | fprintf(outfile, "** Delimiter must not be alphameric or \\\n"); |
| 717 | goto SKIP_DATA; |
| 718 | } |
| 719 | |
| 720 | pp = p; |
| 721 | |
| 722 | for(;;) |
| 723 | { |
| 724 | while (*pp != 0) |
| 725 | { |
| 726 | if (*pp == '\\' && pp[1] != 0) pp++; |
| 727 | else if (*pp == delimiter) break; |
| 728 | pp++; |
| 729 | } |
| 730 | if (*pp != 0) break; |
| 731 | |
| 732 | len = BUFFER_SIZE - (pp - buffer); |
| 733 | if (len < 256) |
| 734 | { |
| 735 | fprintf(outfile, "** Expression too long - missing delimiter?\n"); |
| 736 | goto SKIP_DATA; |
| 737 | } |
| 738 | |
| 739 | if (infile == stdin) printf(" > "); |
| 740 | if (fgets((char *)pp, len, infile) == NULL) |
| 741 | { |
| 742 | fprintf(outfile, "** Unexpected EOF\n"); |
| 743 | done = 1; |
| 744 | goto CONTINUE; |
| 745 | } |
| 746 | if (infile != stdin) fprintf(outfile, "%s", (char *)pp); |
| 747 | } |
| 748 | |
| 749 | /* If the first character after the delimiter is backslash, make |
| 750 | the pattern end with backslash. This is purely to provide a way |
| 751 | of testing for the error message when a pattern ends with backslash. */ |
| 752 | |
| 753 | if (pp[1] == '\\') *pp++ = '\\'; |
| 754 | |
| 755 | /* Terminate the pattern at the delimiter, and save a copy of the pattern |
| 756 | for callouts. */ |
| 757 | |
| 758 | *pp++ = 0; |
| 759 | strcpy((char *)pbuffer, (char *)p); |
| 760 | |
| 761 | /* Look for options after final delimiter */ |
| 762 | |
| 763 | options = 0; |
| 764 | study_options = 0; |
| 765 | log_store = showstore; /* default from command line */ |
| 766 | |
| 767 | while (*pp != 0) |
| 768 | { |
| 769 | switch (*pp++) |
| 770 | { |
| 771 | case 'f': options |= PCRE_FIRSTLINE; break; |
| 772 | case 'g': do_g = 1; break; |
| 773 | case 'i': options |= PCRE_CASELESS; break; |
| 774 | case 'm': options |= PCRE_MULTILINE; break; |
| 775 | case 's': options |= PCRE_DOTALL; break; |
| 776 | case 'x': options |= PCRE_EXTENDED; break; |
| 777 | |
| 778 | case '+': do_showrest = 1; break; |
| 779 | case 'A': options |= PCRE_ANCHORED; break; |
| 780 | case 'C': options |= PCRE_AUTO_CALLOUT; break; |
| 781 | case 'D': do_debug = do_showinfo = 1; break; |
| 782 | case 'E': options |= PCRE_DOLLAR_ENDONLY; break; |
| 783 | case 'F': do_flip = 1; break; |
| 784 | case 'G': do_G = 1; break; |
| 785 | case 'I': do_showinfo = 1; break; |
| 786 | case 'M': log_store = 1; break; |
| 787 | case 'N': options |= PCRE_NO_AUTO_CAPTURE; break; |
| 788 | |
| 789 | #if !defined NOPOSIX |
| 790 | case 'P': do_posix = 1; break; |
| 791 | #endif |
| 792 | |
| 793 | case 'S': do_study = 1; break; |
| 794 | case 'U': options |= PCRE_UNGREEDY; break; |
| 795 | case 'X': options |= PCRE_EXTRA; break; |
| 796 | case '8': options |= PCRE_UTF8; use_utf8 = 1; break; |
| 797 | case '?': options |= PCRE_NO_UTF8_CHECK; break; |
| 798 | |
| 799 | case 'L': |
| 800 | ppp = pp; |
| 801 | /* The '\r' test here is so that it works on Windows */ |
| 802 | while (*ppp != '\n' && *ppp != '\r' && *ppp != ' ') ppp++; |
| 803 | *ppp = 0; |
| 804 | if (setlocale(LC_CTYPE, (const char *)pp) == NULL) |
| 805 | { |
| 806 | fprintf(outfile, "** Failed to set locale \"%s\"\n", pp); |
| 807 | goto SKIP_DATA; |
| 808 | } |
| 809 | tables = pcre_maketables(); |
| 810 | pp = ppp; |
| 811 | break; |
| 812 | |
| 813 | case '>': |
| 814 | to_file = pp; |
| 815 | while (*pp != 0) pp++; |
| 816 | while (isspace(pp[-1])) pp--; |
| 817 | *pp = 0; |
| 818 | break; |
| 819 | |
| 820 | case '\r': /* So that it works in Windows */ |
| 821 | case '\n': |
| 822 | case ' ': |
| 823 | break; |
| 824 | |
| 825 | default: |
| 826 | fprintf(outfile, "** Unknown option '%c'\n", pp[-1]); |
| 827 | goto SKIP_DATA; |
| 828 | } |
| 829 | } |
| 830 | |
| 831 | /* Handle compiling via the POSIX interface, which doesn't support the |
| 832 | timing, showing, or debugging options, nor the ability to pass over |
| 833 | local character tables. */ |
| 834 | |
| 835 | #if !defined NOPOSIX |
| 836 | if (posix || do_posix) |
| 837 | { |
| 838 | int rc; |
| 839 | int cflags = 0; |
| 840 | |
| 841 | if ((options & PCRE_CASELESS) != 0) cflags |= REG_ICASE; |
| 842 | if ((options & PCRE_MULTILINE) != 0) cflags |= REG_NEWLINE; |
| 843 | if ((options & PCRE_DOTALL) != 0) cflags |= REG_DOTALL; |
| 844 | rc = regcomp(&preg, (char *)p, cflags); |
| 845 | |
| 846 | /* Compilation failed; go back for another re, skipping to blank line |
| 847 | if non-interactive. */ |
| 848 | |
| 849 | if (rc != 0) |
| 850 | { |
| 851 | (void)regerror(rc, &preg, (char *)buffer, BUFFER_SIZE); |
| 852 | fprintf(outfile, "Failed: POSIX code %d: %s\n", rc, buffer); |
| 853 | goto SKIP_DATA; |
| 854 | } |
| 855 | } |
| 856 | |
| 857 | /* Handle compiling via the native interface */ |
| 858 | |
| 859 | else |
| 860 | #endif /* !defined NOPOSIX */ |
| 861 | |
| 862 | { |
| 863 | if (timeit) |
| 864 | { |
| 865 | register int i; |
| 866 | clock_t time_taken; |
| 867 | clock_t start_time = clock(); |
| 868 | for (i = 0; i < LOOPREPEAT; i++) |
| 869 | { |
| 870 | re = pcre_compile((char *)p, options, &error, &erroroffset, tables); |
| 871 | if (re != NULL) free(re); |
| 872 | } |
| 873 | time_taken = clock() - start_time; |
| 874 | fprintf(outfile, "Compile time %.3f milliseconds\n", |
| 875 | (((double)time_taken * 1000.0) / (double)LOOPREPEAT) / |
| 876 | (double)CLOCKS_PER_SEC); |
| 877 | } |
| 878 | |
| 879 | re = pcre_compile((char *)p, options, &error, &erroroffset, tables); |
| 880 | |
| 881 | /* Compilation failed; go back for another re, skipping to blank line |
| 882 | if non-interactive. */ |
| 883 | |
| 884 | if (re == NULL) |
| 885 | { |
| 886 | fprintf(outfile, "Failed: %s at offset %d\n", error, erroroffset); |
| 887 | SKIP_DATA: |
| 888 | if (infile != stdin) |
| 889 | { |
| 890 | for (;;) |
| 891 | { |
| 892 | if (fgets((char *)buffer, BUFFER_SIZE, infile) == NULL) |
| 893 | { |
| 894 | done = 1; |
| 895 | goto CONTINUE; |
| 896 | } |
| 897 | len = (int)strlen((char *)buffer); |
| 898 | while (len > 0 && isspace(buffer[len-1])) len--; |
| 899 | if (len == 0) break; |
| 900 | } |
| 901 | fprintf(outfile, "\n"); |
| 902 | } |
| 903 | goto CONTINUE; |
| 904 | } |
| 905 | |
| 906 | /* Compilation succeeded; print data if required. There are now two |
| 907 | info-returning functions. The old one has a limited interface and |
| 908 | returns only limited data. Check that it agrees with the newer one. */ |
| 909 | |
| 910 | if (log_store) |
| 911 | fprintf(outfile, "Memory allocation (code space): %d\n", |
| 912 | (int)(gotten_store - |
| 913 | sizeof(real_pcre) - |
| 914 | ((real_pcre *)re)->name_count * ((real_pcre *)re)->name_entry_size)); |
| 915 | |
| 916 | /* Extract the size for possible writing before possibly flipping it, |
| 917 | and remember the store that was got. */ |
| 918 | |
| 919 | true_size = ((real_pcre *)re)->size; |
| 920 | regex_gotten_store = gotten_store; |
| 921 | |
| 922 | /* If /S was present, study the regexp to generate additional info to |
| 923 | help with the matching. */ |
| 924 | |
| 925 | if (do_study) |
| 926 | { |
| 927 | if (timeit) |
| 928 | { |
| 929 | register int i; |
| 930 | clock_t time_taken; |
| 931 | clock_t start_time = clock(); |
| 932 | for (i = 0; i < LOOPREPEAT; i++) |
| 933 | extra = pcre_study(re, study_options, &error); |
| 934 | time_taken = clock() - start_time; |
| 935 | if (extra != NULL) free(extra); |
| 936 | fprintf(outfile, " Study time %.3f milliseconds\n", |
| 937 | (((double)time_taken * 1000.0) / (double)LOOPREPEAT) / |
| 938 | (double)CLOCKS_PER_SEC); |
| 939 | } |
| 940 | extra = pcre_study(re, study_options, &error); |
| 941 | if (error != NULL) |
| 942 | fprintf(outfile, "Failed to study: %s\n", error); |
| 943 | else if (extra != NULL) |
| 944 | true_study_size = ((pcre_study_data *)(extra->study_data))->size; |
| 945 | } |
| 946 | |
| 947 | /* If the 'F' option was present, we flip the bytes of all the integer |
| 948 | fields in the regex data block and the study block. This is to make it |
| 949 | possible to test PCRE's handling of byte-flipped patterns, e.g. those |
| 950 | compiled on a different architecture. */ |
| 951 | |
| 952 | if (do_flip) |
| 953 | { |
| 954 | real_pcre *rre = (real_pcre *)re; |
| 955 | rre->magic_number = byteflip(rre->magic_number, sizeof(rre->magic_number)); |
| 956 | rre->size = byteflip(rre->size, sizeof(rre->size)); |
| 957 | rre->options = byteflip(rre->options, sizeof(rre->options)); |
| 958 | rre->top_bracket = byteflip(rre->top_bracket, sizeof(rre->top_bracket)); |
| 959 | rre->top_backref = byteflip(rre->top_backref, sizeof(rre->top_backref)); |
| 960 | rre->first_byte = byteflip(rre->first_byte, sizeof(rre->first_byte)); |
| 961 | rre->req_byte = byteflip(rre->req_byte, sizeof(rre->req_byte)); |
| 962 | rre->name_table_offset = byteflip(rre->name_table_offset, |
| 963 | sizeof(rre->name_table_offset)); |
| 964 | rre->name_entry_size = byteflip(rre->name_entry_size, |
| 965 | sizeof(rre->name_entry_size)); |
| 966 | rre->name_count = byteflip(rre->name_count, sizeof(rre->name_count)); |
| 967 | |
| 968 | if (extra != NULL) |
| 969 | { |
| 970 | pcre_study_data *rsd = (pcre_study_data *)(extra->study_data); |
| 971 | rsd->size = byteflip(rsd->size, sizeof(rsd->size)); |
| 972 | rsd->options = byteflip(rsd->options, sizeof(rsd->options)); |
| 973 | } |
| 974 | } |
| 975 | |
| 976 | /* Extract information from the compiled data if required */ |
| 977 | |
| 978 | SHOW_INFO: |
| 979 | |
| 980 | if (do_showinfo) |
| 981 | { |
| 982 | unsigned long int get_options, all_options; |
| 983 | #if !defined NOINFOCHECK |
| 984 | int old_first_char, old_options, old_count; |
| 985 | #endif |
| 986 | int count, backrefmax, first_char, need_char; |
| 987 | int nameentrysize, namecount; |
| 988 | const uschar *nametable; |
| 989 | |
| 990 | if (do_debug) |
| 991 | { |
| 992 | fprintf(outfile, "------------------------------------------------------------------\n"); |
| 993 | _pcre_printint(re, outfile); |
| 994 | } |
| 995 | |
| 996 | new_info(re, NULL, PCRE_INFO_OPTIONS, &get_options); |
| 997 | new_info(re, NULL, PCRE_INFO_SIZE, &size); |
| 998 | new_info(re, NULL, PCRE_INFO_CAPTURECOUNT, &count); |
| 999 | new_info(re, NULL, PCRE_INFO_BACKREFMAX, &backrefmax); |
| 1000 | new_info(re, NULL, PCRE_INFO_FIRSTBYTE, &first_char); |
| 1001 | new_info(re, NULL, PCRE_INFO_LASTLITERAL, &need_char); |
| 1002 | new_info(re, NULL, PCRE_INFO_NAMEENTRYSIZE, &nameentrysize); |
| 1003 | new_info(re, NULL, PCRE_INFO_NAMECOUNT, &namecount); |
| 1004 | new_info(re, NULL, PCRE_INFO_NAMETABLE, (void *)&nametable); |
| 1005 | |
| 1006 | #if !defined NOINFOCHECK |
| 1007 | old_count = pcre_info(re, &old_options, &old_first_char); |
| 1008 | if (count < 0) fprintf(outfile, |
| 1009 | "Error %d from pcre_info()\n", count); |
| 1010 | else |
| 1011 | { |
| 1012 | if (old_count != count) fprintf(outfile, |
| 1013 | "Count disagreement: pcre_fullinfo=%d pcre_info=%d\n", count, |
| 1014 | old_count); |
| 1015 | |
| 1016 | if (old_first_char != first_char) fprintf(outfile, |
| 1017 | "First char disagreement: pcre_fullinfo=%d pcre_info=%d\n", |
| 1018 | first_char, old_first_char); |
| 1019 | |
| 1020 | if (old_options != (int)get_options) fprintf(outfile, |
| 1021 | "Options disagreement: pcre_fullinfo=%ld pcre_info=%d\n", |
| 1022 | get_options, old_options); |
| 1023 | } |
| 1024 | #endif |
| 1025 | |
| 1026 | if (size != regex_gotten_store) fprintf(outfile, |
| 1027 | "Size disagreement: pcre_fullinfo=%d call to malloc for %d\n", |
| 1028 | (int)size, (int)regex_gotten_store); |
| 1029 | |
| 1030 | fprintf(outfile, "Capturing subpattern count = %d\n", count); |
| 1031 | if (backrefmax > 0) |
| 1032 | fprintf(outfile, "Max back reference = %d\n", backrefmax); |
| 1033 | |
| 1034 | if (namecount > 0) |
| 1035 | { |
| 1036 | fprintf(outfile, "Named capturing subpatterns:\n"); |
| 1037 | while (namecount-- > 0) |
| 1038 | { |
| 1039 | fprintf(outfile, " %s %*s%3d\n", nametable + 2, |
| 1040 | nameentrysize - 3 - (int)strlen((char *)nametable + 2), "", |
| 1041 | GET2(nametable, 0)); |
| 1042 | nametable += nameentrysize; |
| 1043 | } |
| 1044 | } |
| 1045 | |
| 1046 | /* The NOPARTIAL bit is a private bit in the options, so we have |
| 1047 | to fish it out via out back door */ |
| 1048 | |
| 1049 | all_options = ((real_pcre *)re)->options; |
| 1050 | if (do_flip) |
| 1051 | { |
| 1052 | all_options = byteflip(all_options, sizeof(all_options)); |
| 1053 | } |
| 1054 | |
| 1055 | if ((all_options & PCRE_NOPARTIAL) != 0) |
| 1056 | fprintf(outfile, "Partial matching not supported\n"); |
| 1057 | |
| 1058 | if (get_options == 0) fprintf(outfile, "No options\n"); |
| 1059 | else fprintf(outfile, "Options:%s%s%s%s%s%s%s%s%s%s%s\n", |
| 1060 | ((get_options & PCRE_ANCHORED) != 0)? " anchored" : "", |
| 1061 | ((get_options & PCRE_CASELESS) != 0)? " caseless" : "", |
| 1062 | ((get_options & PCRE_EXTENDED) != 0)? " extended" : "", |
| 1063 | ((get_options & PCRE_MULTILINE) != 0)? " multiline" : "", |
| 1064 | ((get_options & PCRE_FIRSTLINE) != 0)? " firstline" : "", |
| 1065 | ((get_options & PCRE_DOTALL) != 0)? " dotall" : "", |
| 1066 | ((get_options & PCRE_DOLLAR_ENDONLY) != 0)? " dollar_endonly" : "", |
| 1067 | ((get_options & PCRE_EXTRA) != 0)? " extra" : "", |
| 1068 | ((get_options & PCRE_UNGREEDY) != 0)? " ungreedy" : "", |
| 1069 | ((get_options & PCRE_UTF8) != 0)? " utf8" : "", |
| 1070 | ((get_options & PCRE_NO_UTF8_CHECK) != 0)? " no_utf8_check" : ""); |
| 1071 | |
| 1072 | if (((((real_pcre *)re)->options) & PCRE_ICHANGED) != 0) |
| 1073 | fprintf(outfile, "Case state changes\n"); |
| 1074 | |
| 1075 | if (first_char == -1) |
| 1076 | { |
| 1077 | fprintf(outfile, "First char at start or follows \\n\n"); |
| 1078 | } |
| 1079 | else if (first_char < 0) |
| 1080 | { |
| 1081 | fprintf(outfile, "No first char\n"); |
| 1082 | } |
| 1083 | else |
| 1084 | { |
| 1085 | int ch = first_char & 255; |
| 1086 | const char *caseless = ((first_char & REQ_CASELESS) == 0)? |
| 1087 | "" : " (caseless)"; |
| 1088 | if (isprint(ch)) |
| 1089 | fprintf(outfile, "First char = \'%c\'%s\n", ch, caseless); |
| 1090 | else |
| 1091 | fprintf(outfile, "First char = %d%s\n", ch, caseless); |
| 1092 | } |
| 1093 | |
| 1094 | if (need_char < 0) |
| 1095 | { |
| 1096 | fprintf(outfile, "No need char\n"); |
| 1097 | } |
| 1098 | else |
| 1099 | { |
| 1100 | int ch = need_char & 255; |
| 1101 | const char *caseless = ((need_char & REQ_CASELESS) == 0)? |
| 1102 | "" : " (caseless)"; |
| 1103 | if (isprint(ch)) |
| 1104 | fprintf(outfile, "Need char = \'%c\'%s\n", ch, caseless); |
| 1105 | else |
| 1106 | fprintf(outfile, "Need char = %d%s\n", ch, caseless); |
| 1107 | } |
| 1108 | |
| 1109 | /* Don't output study size; at present it is in any case a fixed |
| 1110 | value, but it varies, depending on the computer architecture, and |
| 1111 | so messes up the test suite. (And with the /F option, it might be |
| 1112 | flipped.) */ |
| 1113 | |
| 1114 | if (do_study) |
| 1115 | { |
| 1116 | if (extra == NULL) |
| 1117 | fprintf(outfile, "Study returned NULL\n"); |
| 1118 | else |
| 1119 | { |
| 1120 | uschar *start_bits = NULL; |
| 1121 | new_info(re, extra, PCRE_INFO_FIRSTTABLE, &start_bits); |
| 1122 | |
| 1123 | if (start_bits == NULL) |
| 1124 | fprintf(outfile, "No starting byte set\n"); |
| 1125 | else |
| 1126 | { |
| 1127 | int i; |
| 1128 | int c = 24; |
| 1129 | fprintf(outfile, "Starting byte set: "); |
| 1130 | for (i = 0; i < 256; i++) |
| 1131 | { |
| 1132 | if ((start_bits[i/8] & (1<<(i&7))) != 0) |
| 1133 | { |
| 1134 | if (c > 75) |
| 1135 | { |
| 1136 | fprintf(outfile, "\n "); |
| 1137 | c = 2; |
| 1138 | } |
| 1139 | if (isprint(i) && i != ' ') |
| 1140 | { |
| 1141 | fprintf(outfile, "%c ", i); |
| 1142 | c += 2; |
| 1143 | } |
| 1144 | else |
| 1145 | { |
| 1146 | fprintf(outfile, "\\x%02x ", i); |
| 1147 | c += 5; |
| 1148 | } |
| 1149 | } |
| 1150 | } |
| 1151 | fprintf(outfile, "\n"); |
| 1152 | } |
| 1153 | } |
| 1154 | } |
| 1155 | } |
| 1156 | |
| 1157 | /* If the '>' option was present, we write out the regex to a file, and |
| 1158 | that is all. The first 8 bytes of the file are the regex length and then |
| 1159 | the study length, in big-endian order. */ |
| 1160 | |
| 1161 | if (to_file != NULL) |
| 1162 | { |
| 1163 | FILE *f = fopen((char *)to_file, "wb"); |
| 1164 | if (f == NULL) |
| 1165 | { |
| 1166 | fprintf(outfile, "Unable to open %s: %s\n", to_file, strerror(errno)); |
| 1167 | } |
| 1168 | else |
| 1169 | { |
| 1170 | uschar sbuf[8]; |
| 1171 | sbuf[0] = (true_size >> 24) & 255; |
| 1172 | sbuf[1] = (true_size >> 16) & 255; |
| 1173 | sbuf[2] = (true_size >> 8) & 255; |
| 1174 | sbuf[3] = (true_size) & 255; |
| 1175 | |
| 1176 | sbuf[4] = (true_study_size >> 24) & 255; |
| 1177 | sbuf[5] = (true_study_size >> 16) & 255; |
| 1178 | sbuf[6] = (true_study_size >> 8) & 255; |
| 1179 | sbuf[7] = (true_study_size) & 255; |
| 1180 | |
| 1181 | if (fwrite(sbuf, 1, 8, f) < 8 || |
| 1182 | fwrite(re, 1, true_size, f) < true_size) |
| 1183 | { |
| 1184 | fprintf(outfile, "Write error on %s: %s\n", to_file, strerror(errno)); |
| 1185 | } |
| 1186 | else |
| 1187 | { |
| 1188 | fprintf(outfile, "Compiled regex written to %s\n", to_file); |
| 1189 | if (extra != NULL) |
| 1190 | { |
| 1191 | if (fwrite(extra->study_data, 1, true_study_size, f) < |
| 1192 | true_study_size) |
| 1193 | { |
| 1194 | fprintf(outfile, "Write error on %s: %s\n", to_file, |
| 1195 | strerror(errno)); |
| 1196 | } |
| 1197 | else fprintf(outfile, "Study data written to %s\n", to_file); |
| 1198 | } |
| 1199 | } |
| 1200 | fclose(f); |
| 1201 | } |
| 1202 | |
| 1203 | new_free(re); |
| 1204 | if (extra != NULL) new_free(extra); |
| 1205 | if (tables != NULL) new_free((void *)tables); |
| 1206 | continue; /* With next regex */ |
| 1207 | } |
| 1208 | } /* End of non-POSIX compile */ |
| 1209 | |
| 1210 | /* Read data lines and test them */ |
| 1211 | |
| 1212 | for (;;) |
| 1213 | { |
| 1214 | unsigned char *q; |
| 1215 | unsigned char *bptr = dbuffer; |
| 1216 | int *use_offsets = offsets; |
| 1217 | int use_size_offsets = size_offsets; |
| 1218 | int callout_data = 0; |
| 1219 | int callout_data_set = 0; |
| 1220 | int count, c; |
| 1221 | int copystrings = 0; |
| 1222 | int find_match_limit = 0; |
| 1223 | int getstrings = 0; |
| 1224 | int getlist = 0; |
| 1225 | int gmatched = 0; |
| 1226 | int start_offset = 0; |
| 1227 | int g_notempty = 0; |
| 1228 | int use_dfa = 0; |
| 1229 | |
| 1230 | options = 0; |
| 1231 | |
| 1232 | pcre_callout = callout; |
| 1233 | first_callout = 1; |
| 1234 | callout_extra = 0; |
| 1235 | callout_count = 0; |
| 1236 | callout_fail_count = 999999; |
| 1237 | callout_fail_id = -1; |
| 1238 | show_malloc = 0; |
| 1239 | |
| 1240 | if (infile == stdin) printf("data> "); |
| 1241 | if (fgets((char *)buffer, BUFFER_SIZE, infile) == NULL) |
| 1242 | { |
| 1243 | done = 1; |
| 1244 | goto CONTINUE; |
| 1245 | } |
| 1246 | if (infile != stdin) fprintf(outfile, "%s", (char *)buffer); |
| 1247 | |
| 1248 | len = (int)strlen((char *)buffer); |
| 1249 | while (len > 0 && isspace(buffer[len-1])) len--; |
| 1250 | buffer[len] = 0; |
| 1251 | if (len == 0) break; |
| 1252 | |
| 1253 | p = buffer; |
| 1254 | while (isspace(*p)) p++; |
| 1255 | |
| 1256 | q = dbuffer; |
| 1257 | while ((c = *p++) != 0) |
| 1258 | { |
| 1259 | int i = 0; |
| 1260 | int n = 0; |
| 1261 | |
| 1262 | if (c == '\\') switch ((c = *p++)) |
| 1263 | { |
| 1264 | case 'a': c = 7; break; |
| 1265 | case 'b': c = '\b'; break; |
| 1266 | case 'e': c = 27; break; |
| 1267 | case 'f': c = '\f'; break; |
| 1268 | case 'n': c = '\n'; break; |
| 1269 | case 'r': c = '\r'; break; |
| 1270 | case 't': c = '\t'; break; |
| 1271 | case 'v': c = '\v'; break; |
| 1272 | |
| 1273 | case '0': case '1': case '2': case '3': |
| 1274 | case '4': case '5': case '6': case '7': |
| 1275 | c -= '0'; |
| 1276 | while (i++ < 2 && isdigit(*p) && *p != '8' && *p != '9') |
| 1277 | c = c * 8 + *p++ - '0'; |
| 1278 | break; |
| 1279 | |
| 1280 | case 'x': |
| 1281 | |
| 1282 | /* Handle \x{..} specially - new Perl thing for utf8 */ |
| 1283 | |
| 1284 | #if !defined NOUTF8 |
| 1285 | if (*p == '{') |
| 1286 | { |
| 1287 | unsigned char *pt = p; |
| 1288 | c = 0; |
| 1289 | while (isxdigit(*(++pt))) |
| 1290 | c = c * 16 + tolower(*pt) - ((isdigit(*pt))? '0' : 'W'); |
| 1291 | if (*pt == '}') |
| 1292 | { |
| 1293 | unsigned char buff8[8]; |
| 1294 | int ii, utn; |
| 1295 | utn = _pcre_ord2utf8(c, buff8); |
| 1296 | for (ii = 0; ii < utn - 1; ii++) *q++ = buff8[ii]; |
| 1297 | c = buff8[ii]; /* Last byte */ |
| 1298 | p = pt + 1; |
| 1299 | break; |
| 1300 | } |
| 1301 | /* Not correct form; fall through */ |
| 1302 | } |
| 1303 | #endif |
| 1304 | |
| 1305 | /* Ordinary \x */ |
| 1306 | |
| 1307 | c = 0; |
| 1308 | while (i++ < 2 && isxdigit(*p)) |
| 1309 | { |
| 1310 | c = c * 16 + tolower(*p) - ((isdigit(*p))? '0' : 'W'); |
| 1311 | p++; |
| 1312 | } |
| 1313 | break; |
| 1314 | |
| 1315 | case 0: /* \ followed by EOF allows for an empty line */ |
| 1316 | p--; |
| 1317 | continue; |
| 1318 | |
| 1319 | case '>': |
| 1320 | while(isdigit(*p)) start_offset = start_offset * 10 + *p++ - '0'; |
| 1321 | continue; |
| 1322 | |
| 1323 | case 'A': /* Option setting */ |
| 1324 | options |= PCRE_ANCHORED; |
| 1325 | continue; |
| 1326 | |
| 1327 | case 'B': |
| 1328 | options |= PCRE_NOTBOL; |
| 1329 | continue; |
| 1330 | |
| 1331 | case 'C': |
| 1332 | if (isdigit(*p)) /* Set copy string */ |
| 1333 | { |
| 1334 | while(isdigit(*p)) n = n * 10 + *p++ - '0'; |
| 1335 | copystrings |= 1 << n; |
| 1336 | } |
| 1337 | else if (isalnum(*p)) |
| 1338 | { |
| 1339 | uschar name[256]; |
| 1340 | uschar *npp = name; |
| 1341 | while (isalnum(*p)) *npp++ = *p++; |
| 1342 | *npp = 0; |
| 1343 | n = pcre_get_stringnumber(re, (char *)name); |
| 1344 | if (n < 0) |
| 1345 | fprintf(outfile, "no parentheses with name \"%s\"\n", name); |
| 1346 | else copystrings |= 1 << n; |
| 1347 | } |
| 1348 | else if (*p == '+') |
| 1349 | { |
| 1350 | callout_extra = 1; |
| 1351 | p++; |
| 1352 | } |
| 1353 | else if (*p == '-') |
| 1354 | { |
| 1355 | pcre_callout = NULL; |
| 1356 | p++; |
| 1357 | } |
| 1358 | else if (*p == '!') |
| 1359 | { |
| 1360 | callout_fail_id = 0; |
| 1361 | p++; |
| 1362 | while(isdigit(*p)) |
| 1363 | callout_fail_id = callout_fail_id * 10 + *p++ - '0'; |
| 1364 | callout_fail_count = 0; |
| 1365 | if (*p == '!') |
| 1366 | { |
| 1367 | p++; |
| 1368 | while(isdigit(*p)) |
| 1369 | callout_fail_count = callout_fail_count * 10 + *p++ - '0'; |
| 1370 | } |
| 1371 | } |
| 1372 | else if (*p == '*') |
| 1373 | { |
| 1374 | int sign = 1; |
| 1375 | callout_data = 0; |
| 1376 | if (*(++p) == '-') { sign = -1; p++; } |
| 1377 | while(isdigit(*p)) |
| 1378 | callout_data = callout_data * 10 + *p++ - '0'; |
| 1379 | callout_data *= sign; |
| 1380 | callout_data_set = 1; |
| 1381 | } |
| 1382 | continue; |
| 1383 | |
| 1384 | #if !defined NODFA |
| 1385 | case 'D': |
| 1386 | #if !defined NOPOSIX |
| 1387 | if (posix || do_posix) |
| 1388 | printf("** Can't use dfa matching in POSIX mode: \\D ignored\n"); |
| 1389 | else |
| 1390 | #endif |
| 1391 | use_dfa = 1; |
| 1392 | continue; |
| 1393 | |
| 1394 | case 'F': |
| 1395 | options |= PCRE_DFA_SHORTEST; |
| 1396 | continue; |
| 1397 | #endif |
| 1398 | |
| 1399 | case 'G': |
| 1400 | if (isdigit(*p)) |
| 1401 | { |
| 1402 | while(isdigit(*p)) n = n * 10 + *p++ - '0'; |
| 1403 | getstrings |= 1 << n; |
| 1404 | } |
| 1405 | else if (isalnum(*p)) |
| 1406 | { |
| 1407 | uschar name[256]; |
| 1408 | uschar *npp = name; |
| 1409 | while (isalnum(*p)) *npp++ = *p++; |
| 1410 | *npp = 0; |
| 1411 | n = pcre_get_stringnumber(re, (char *)name); |
| 1412 | if (n < 0) |
| 1413 | fprintf(outfile, "no parentheses with name \"%s\"\n", name); |
| 1414 | else getstrings |= 1 << n; |
| 1415 | } |
| 1416 | continue; |
| 1417 | |
| 1418 | case 'L': |
| 1419 | getlist = 1; |
| 1420 | continue; |
| 1421 | |
| 1422 | case 'M': |
| 1423 | find_match_limit = 1; |
| 1424 | continue; |
| 1425 | |
| 1426 | case 'N': |
| 1427 | options |= PCRE_NOTEMPTY; |
| 1428 | continue; |
| 1429 | |
| 1430 | case 'O': |
| 1431 | while(isdigit(*p)) n = n * 10 + *p++ - '0'; |
| 1432 | if (n > size_offsets_max) |
| 1433 | { |
| 1434 | size_offsets_max = n; |
| 1435 | free(offsets); |
| 1436 | use_offsets = offsets = (int *)malloc(size_offsets_max * sizeof(int)); |
| 1437 | if (offsets == NULL) |
| 1438 | { |
| 1439 | printf("** Failed to get %d bytes of memory for offsets vector\n", |
| 1440 | size_offsets_max * sizeof(int)); |
| 1441 | yield = 1; |
| 1442 | goto EXIT; |
| 1443 | } |
| 1444 | } |
| 1445 | use_size_offsets = n; |
| 1446 | if (n == 0) use_offsets = NULL; /* Ensures it can't write to it */ |
| 1447 | continue; |
| 1448 | |
| 1449 | case 'P': |
| 1450 | options |= PCRE_PARTIAL; |
| 1451 | continue; |
| 1452 | |
| 1453 | #if !defined NODFA |
| 1454 | case 'R': |
| 1455 | options |= PCRE_DFA_RESTART; |
| 1456 | continue; |
| 1457 | #endif |
| 1458 | |
| 1459 | case 'S': |
| 1460 | show_malloc = 1; |
| 1461 | continue; |
| 1462 | |
| 1463 | case 'Z': |
| 1464 | options |= PCRE_NOTEOL; |
| 1465 | continue; |
| 1466 | |
| 1467 | case '?': |
| 1468 | options |= PCRE_NO_UTF8_CHECK; |
| 1469 | continue; |
| 1470 | } |
| 1471 | *q++ = c; |
| 1472 | } |
| 1473 | *q = 0; |
| 1474 | len = q - dbuffer; |
| 1475 | |
| 1476 | if ((all_use_dfa || use_dfa) && find_match_limit) |
| 1477 | { |
| 1478 | printf("**Match limit not relevant for DFA matching: ignored\n"); |
| 1479 | find_match_limit = 0; |
| 1480 | } |
| 1481 | |
| 1482 | /* Handle matching via the POSIX interface, which does not |
| 1483 | support timing or playing with the match limit or callout data. */ |
| 1484 | |
| 1485 | #if !defined NOPOSIX |
| 1486 | if (posix || do_posix) |
| 1487 | { |
| 1488 | int rc; |
| 1489 | int eflags = 0; |
| 1490 | regmatch_t *pmatch = NULL; |
| 1491 | if (use_size_offsets > 0) |
| 1492 | pmatch = (regmatch_t *)malloc(sizeof(regmatch_t) * use_size_offsets); |
| 1493 | if ((options & PCRE_NOTBOL) != 0) eflags |= REG_NOTBOL; |
| 1494 | if ((options & PCRE_NOTEOL) != 0) eflags |= REG_NOTEOL; |
| 1495 | |
| 1496 | rc = regexec(&preg, (const char *)bptr, use_size_offsets, pmatch, eflags); |
| 1497 | |
| 1498 | if (rc != 0) |
| 1499 | { |
| 1500 | (void)regerror(rc, &preg, (char *)buffer, BUFFER_SIZE); |
| 1501 | fprintf(outfile, "No match: POSIX code %d: %s\n", rc, buffer); |
| 1502 | } |
| 1503 | else |
| 1504 | { |
| 1505 | size_t i; |
| 1506 | for (i = 0; i < (size_t)use_size_offsets; i++) |
| 1507 | { |
| 1508 | if (pmatch[i].rm_so >= 0) |
| 1509 | { |
| 1510 | fprintf(outfile, "%2d: ", (int)i); |
| 1511 | (void)pchars(dbuffer + pmatch[i].rm_so, |
| 1512 | pmatch[i].rm_eo - pmatch[i].rm_so, outfile); |
| 1513 | fprintf(outfile, "\n"); |
| 1514 | if (i == 0 && do_showrest) |
| 1515 | { |
| 1516 | fprintf(outfile, " 0+ "); |
| 1517 | (void)pchars(dbuffer + pmatch[i].rm_eo, len - pmatch[i].rm_eo, |
| 1518 | outfile); |
| 1519 | fprintf(outfile, "\n"); |
| 1520 | } |
| 1521 | } |
| 1522 | } |
| 1523 | } |
| 1524 | free(pmatch); |
| 1525 | } |
| 1526 | |
| 1527 | /* Handle matching via the native interface - repeats for /g and /G */ |
| 1528 | |
| 1529 | else |
| 1530 | #endif /* !defined NOPOSIX */ |
| 1531 | |
| 1532 | for (;; gmatched++) /* Loop for /g or /G */ |
| 1533 | { |
| 1534 | if (timeit) |
| 1535 | { |
| 1536 | register int i; |
| 1537 | clock_t time_taken; |
| 1538 | clock_t start_time = clock(); |
| 1539 | |
| 1540 | #if !defined NODFA |
| 1541 | if (all_use_dfa || use_dfa) |
| 1542 | { |
| 1543 | int workspace[1000]; |
| 1544 | for (i = 0; i < LOOPREPEAT; i++) |
| 1545 | count = pcre_dfa_exec(re, NULL, (char *)bptr, len, start_offset, |
| 1546 | options | g_notempty, use_offsets, use_size_offsets, workspace, |
| 1547 | sizeof(workspace)/sizeof(int)); |
| 1548 | } |
| 1549 | else |
| 1550 | #endif |
| 1551 | |
| 1552 | for (i = 0; i < LOOPREPEAT; i++) |
| 1553 | count = pcre_exec(re, extra, (char *)bptr, len, |
| 1554 | start_offset, options | g_notempty, use_offsets, use_size_offsets); |
| 1555 | |
| 1556 | time_taken = clock() - start_time; |
| 1557 | fprintf(outfile, "Execute time %.3f milliseconds\n", |
| 1558 | (((double)time_taken * 1000.0) / (double)LOOPREPEAT) / |
| 1559 | (double)CLOCKS_PER_SEC); |
| 1560 | } |
| 1561 | |
| 1562 | /* If find_match_limit is set, we want to do repeated matches with |
| 1563 | varying limits in order to find the minimum value. */ |
| 1564 | |
| 1565 | if (find_match_limit) |
| 1566 | { |
| 1567 | int min = 0; |
| 1568 | int mid = 64; |
| 1569 | int max = -1; |
| 1570 | |
| 1571 | if (extra == NULL) |
| 1572 | { |
| 1573 | extra = (pcre_extra *)malloc(sizeof(pcre_extra)); |
| 1574 | extra->flags = 0; |
| 1575 | } |
| 1576 | extra->flags |= PCRE_EXTRA_MATCH_LIMIT; |
| 1577 | |
| 1578 | for (;;) |
| 1579 | { |
| 1580 | extra->match_limit = mid; |
| 1581 | count = pcre_exec(re, extra, (char *)bptr, len, start_offset, |
| 1582 | options | g_notempty, use_offsets, use_size_offsets); |
| 1583 | if (count == PCRE_ERROR_MATCHLIMIT) |
| 1584 | { |
| 1585 | /* fprintf(outfile, "Testing match limit = %d\n", mid); */ |
| 1586 | min = mid; |
| 1587 | mid = (mid == max - 1)? max : (max > 0)? (min + max)/2 : mid*2; |
| 1588 | } |
| 1589 | else if (count >= 0 || count == PCRE_ERROR_NOMATCH || |
| 1590 | count == PCRE_ERROR_PARTIAL) |
| 1591 | { |
| 1592 | if (mid == min + 1) |
| 1593 | { |
| 1594 | fprintf(outfile, "Minimum match limit = %d\n", mid); |
| 1595 | break; |
| 1596 | } |
| 1597 | /* fprintf(outfile, "Testing match limit = %d\n", mid); */ |
| 1598 | max = mid; |
| 1599 | mid = (min + mid)/2; |
| 1600 | } |
| 1601 | else break; /* Some other error */ |
| 1602 | } |
| 1603 | |
| 1604 | extra->flags &= ~PCRE_EXTRA_MATCH_LIMIT; |
| 1605 | } |
| 1606 | |
| 1607 | /* If callout_data is set, use the interface with additional data */ |
| 1608 | |
| 1609 | else if (callout_data_set) |
| 1610 | { |
| 1611 | if (extra == NULL) |
| 1612 | { |
| 1613 | extra = (pcre_extra *)malloc(sizeof(pcre_extra)); |
| 1614 | extra->flags = 0; |
| 1615 | } |
| 1616 | extra->flags |= PCRE_EXTRA_CALLOUT_DATA; |
| 1617 | extra->callout_data = &callout_data; |
| 1618 | count = pcre_exec(re, extra, (char *)bptr, len, start_offset, |
| 1619 | options | g_notempty, use_offsets, use_size_offsets); |
| 1620 | extra->flags &= ~PCRE_EXTRA_CALLOUT_DATA; |
| 1621 | } |
| 1622 | |
| 1623 | /* The normal case is just to do the match once, with the default |
| 1624 | value of match_limit. */ |
| 1625 | |
| 1626 | #if !defined NODFA |
| 1627 | else if (all_use_dfa || use_dfa) |
| 1628 | { |
| 1629 | int workspace[1000]; |
| 1630 | count = pcre_dfa_exec(re, NULL, (char *)bptr, len, start_offset, |
| 1631 | options | g_notempty, use_offsets, use_size_offsets, workspace, |
| 1632 | sizeof(workspace)/sizeof(int)); |
| 1633 | if (count == 0) |
| 1634 | { |
| 1635 | fprintf(outfile, "Matched, but too many subsidiary matches\n"); |
| 1636 | count = use_size_offsets/2; |
| 1637 | } |
| 1638 | } |
| 1639 | #endif |
| 1640 | |
| 1641 | else |
| 1642 | { |
| 1643 | count = pcre_exec(re, extra, (char *)bptr, len, |
| 1644 | start_offset, options | g_notempty, use_offsets, use_size_offsets); |
| 1645 | if (count == 0) |
| 1646 | { |
| 1647 | fprintf(outfile, "Matched, but too many substrings\n"); |
| 1648 | count = use_size_offsets/3; |
| 1649 | } |
| 1650 | } |
| 1651 | |
| 1652 | /* Matched */ |
| 1653 | |
| 1654 | if (count >= 0) |
| 1655 | { |
| 1656 | int i; |
| 1657 | for (i = 0; i < count * 2; i += 2) |
| 1658 | { |
| 1659 | if (use_offsets[i] < 0) |
| 1660 | fprintf(outfile, "%2d: <unset>\n", i/2); |
| 1661 | else |
| 1662 | { |
| 1663 | fprintf(outfile, "%2d: ", i/2); |
| 1664 | (void)pchars(bptr + use_offsets[i], |
| 1665 | use_offsets[i+1] - use_offsets[i], outfile); |
| 1666 | fprintf(outfile, "\n"); |
| 1667 | if (i == 0) |
| 1668 | { |
| 1669 | if (do_showrest) |
| 1670 | { |
| 1671 | fprintf(outfile, " 0+ "); |
| 1672 | (void)pchars(bptr + use_offsets[i+1], len - use_offsets[i+1], |
| 1673 | outfile); |
| 1674 | fprintf(outfile, "\n"); |
| 1675 | } |
| 1676 | } |
| 1677 | } |
| 1678 | } |
| 1679 | |
| 1680 | for (i = 0; i < 32; i++) |
| 1681 | { |
| 1682 | if ((copystrings & (1 << i)) != 0) |
| 1683 | { |
| 1684 | char copybuffer[16]; |
| 1685 | int rc = pcre_copy_substring((char *)bptr, use_offsets, count, |
| 1686 | i, copybuffer, sizeof(copybuffer)); |
| 1687 | if (rc < 0) |
| 1688 | fprintf(outfile, "copy substring %d failed %d\n", i, rc); |
| 1689 | else |
| 1690 | fprintf(outfile, "%2dC %s (%d)\n", i, copybuffer, rc); |
| 1691 | } |
| 1692 | } |
| 1693 | |
| 1694 | for (i = 0; i < 32; i++) |
| 1695 | { |
| 1696 | if ((getstrings & (1 << i)) != 0) |
| 1697 | { |
| 1698 | const char *substring; |
| 1699 | int rc = pcre_get_substring((char *)bptr, use_offsets, count, |
| 1700 | i, &substring); |
| 1701 | if (rc < 0) |
| 1702 | fprintf(outfile, "get substring %d failed %d\n", i, rc); |
| 1703 | else |
| 1704 | { |
| 1705 | fprintf(outfile, "%2dG %s (%d)\n", i, substring, rc); |
| 1706 | /* free((void *)substring); */ |
| 1707 | pcre_free_substring(substring); |
| 1708 | } |
| 1709 | } |
| 1710 | } |
| 1711 | |
| 1712 | if (getlist) |
| 1713 | { |
| 1714 | const char **stringlist; |
| 1715 | int rc = pcre_get_substring_list((char *)bptr, use_offsets, count, |
| 1716 | &stringlist); |
| 1717 | if (rc < 0) |
| 1718 | fprintf(outfile, "get substring list failed %d\n", rc); |
| 1719 | else |
| 1720 | { |
| 1721 | for (i = 0; i < count; i++) |
| 1722 | fprintf(outfile, "%2dL %s\n", i, stringlist[i]); |
| 1723 | if (stringlist[i] != NULL) |
| 1724 | fprintf(outfile, "string list not terminated by NULL\n"); |
| 1725 | /* free((void *)stringlist); */ |
| 1726 | pcre_free_substring_list(stringlist); |
| 1727 | } |
| 1728 | } |
| 1729 | } |
| 1730 | |
| 1731 | /* There was a partial match */ |
| 1732 | |
| 1733 | else if (count == PCRE_ERROR_PARTIAL) |
| 1734 | { |
| 1735 | fprintf(outfile, "Partial match"); |
| 1736 | #if !defined NODFA |
| 1737 | if ((all_use_dfa || use_dfa) && use_size_offsets > 2) |
| 1738 | fprintf(outfile, ": %.*s", use_offsets[1] - use_offsets[0], |
| 1739 | bptr + use_offsets[0]); |
| 1740 | #endif |
| 1741 | fprintf(outfile, "\n"); |
| 1742 | break; /* Out of the /g loop */ |
| 1743 | } |
| 1744 | |
| 1745 | /* Failed to match. If this is a /g or /G loop and we previously set |
| 1746 | g_notempty after a null match, this is not necessarily the end. |
| 1747 | We want to advance the start offset, and continue. In the case of UTF-8 |
| 1748 | matching, the advance must be one character, not one byte. Fudge the |
| 1749 | offset values to achieve this. We won't be at the end of the string - |
| 1750 | that was checked before setting g_notempty. */ |
| 1751 | |
| 1752 | else |
| 1753 | { |
| 1754 | if (g_notempty != 0) |
| 1755 | { |
| 1756 | int onechar = 1; |
| 1757 | use_offsets[0] = start_offset; |
| 1758 | if (use_utf8) |
| 1759 | { |
| 1760 | while (start_offset + onechar < len) |
| 1761 | { |
| 1762 | int tb = bptr[start_offset+onechar]; |
| 1763 | if (tb <= 127) break; |
| 1764 | tb &= 0xc0; |
| 1765 | if (tb != 0 && tb != 0xc0) onechar++; |
| 1766 | } |
| 1767 | } |
| 1768 | use_offsets[1] = start_offset + onechar; |
| 1769 | } |
| 1770 | else |
| 1771 | { |
| 1772 | if (count == PCRE_ERROR_NOMATCH) |
| 1773 | { |
| 1774 | if (gmatched == 0) fprintf(outfile, "No match\n"); |
| 1775 | } |
| 1776 | else fprintf(outfile, "Error %d\n", count); |
| 1777 | break; /* Out of the /g loop */ |
| 1778 | } |
| 1779 | } |
| 1780 | |
| 1781 | /* If not /g or /G we are done */ |
| 1782 | |
| 1783 | if (!do_g && !do_G) break; |
| 1784 | |
| 1785 | /* If we have matched an empty string, first check to see if we are at |
| 1786 | the end of the subject. If so, the /g loop is over. Otherwise, mimic |
| 1787 | what Perl's /g options does. This turns out to be rather cunning. First |
| 1788 | we set PCRE_NOTEMPTY and PCRE_ANCHORED and try the match again at the |
| 1789 | same point. If this fails (picked up above) we advance to the next |
| 1790 | character. */ |
| 1791 | |
| 1792 | g_notempty = 0; |
| 1793 | if (use_offsets[0] == use_offsets[1]) |
| 1794 | { |
| 1795 | if (use_offsets[0] == len) break; |
| 1796 | g_notempty = PCRE_NOTEMPTY | PCRE_ANCHORED; |
| 1797 | } |
| 1798 | |
| 1799 | /* For /g, update the start offset, leaving the rest alone */ |
| 1800 | |
| 1801 | if (do_g) start_offset = use_offsets[1]; |
| 1802 | |
| 1803 | /* For /G, update the pointer and length */ |
| 1804 | |
| 1805 | else |
| 1806 | { |
| 1807 | bptr += use_offsets[1]; |
| 1808 | len -= use_offsets[1]; |
| 1809 | } |
| 1810 | } /* End of loop for /g and /G */ |
| 1811 | } /* End of loop for data lines */ |
| 1812 | |
| 1813 | CONTINUE: |
| 1814 | |
| 1815 | #if !defined NOPOSIX |
| 1816 | if (posix || do_posix) regfree(&preg); |
| 1817 | #endif |
| 1818 | |
| 1819 | if (re != NULL) new_free(re); |
| 1820 | if (extra != NULL) new_free(extra); |
| 1821 | if (tables != NULL) |
| 1822 | { |
| 1823 | new_free((void *)tables); |
| 1824 | setlocale(LC_CTYPE, "C"); |
| 1825 | } |
| 1826 | } |
| 1827 | |
| 1828 | if (infile == stdin) fprintf(outfile, "\n"); |
| 1829 | |
| 1830 | EXIT: |
| 1831 | |
| 1832 | if (infile != NULL && infile != stdin) fclose(infile); |
| 1833 | if (outfile != NULL && outfile != stdout) fclose(outfile); |
| 1834 | |
| 1835 | free(buffer); |
| 1836 | free(dbuffer); |
| 1837 | free(pbuffer); |
| 1838 | free(offsets); |
| 1839 | |
| 1840 | return yield; |
| 1841 | } |
| 1842 | |
| 1843 | /* End of pcretest.c */ |