| 1 |
/*************************************************
|
| 2 |
* PCRE testing program *
|
| 3 |
*************************************************/
|
| 4 |
|
| 5 |
#include <ctype.h>
|
| 6 |
#include <stdio.h>
|
| 7 |
#include <string.h>
|
| 8 |
#include <stdlib.h>
|
| 9 |
#include <time.h>
|
| 10 |
|
| 11 |
/* Use the internal info for displaying the results of pcre_study(). */
|
| 12 |
|
| 13 |
#include "internal.h"
|
| 14 |
#include "pcreposix.h"
|
| 15 |
|
| 16 |
#ifndef CLOCKS_PER_SEC
|
| 17 |
#ifdef CLK_TCK
|
| 18 |
#define CLOCKS_PER_SEC CLK_TCK
|
| 19 |
#else
|
| 20 |
#define CLOCKS_PER_SEC 100
|
| 21 |
#endif
|
| 22 |
#endif
|
| 23 |
|
| 24 |
|
| 25 |
static FILE *outfile;
|
| 26 |
static int log_store = 0;
|
| 27 |
|
| 28 |
|
| 29 |
|
| 30 |
/* Debugging function to print the internal form of the regex. This is the same
|
| 31 |
code as contained in pcre.c under the DEBUG macro. */
|
| 32 |
|
| 33 |
static const char *OP_names[] = {
|
| 34 |
"End", "\\A", "\\B", "\\b", "\\D", "\\d",
|
| 35 |
"\\S", "\\s", "\\W", "\\w", "Cut", "\\Z", "^", "$", "Any", "chars",
|
| 36 |
"not",
|
| 37 |
"*", "*?", "+", "+?", "?", "??", "{", "{", "{",
|
| 38 |
"*", "*?", "+", "+?", "?", "??", "{", "{", "{",
|
| 39 |
"*", "*?", "+", "+?", "?", "??", "{", "{", "{",
|
| 40 |
"*", "*?", "+", "+?", "?", "??", "{", "{",
|
| 41 |
"class", "negclass", "Ref",
|
| 42 |
"Alt", "Ket", "KetRmax", "KetRmin", "Assert", "Assert not", "Once",
|
| 43 |
"Brazero", "Braminzero", "Bra"
|
| 44 |
};
|
| 45 |
|
| 46 |
|
| 47 |
static void print_internals(pcre *re)
|
| 48 |
{
|
| 49 |
unsigned char *code = ((real_pcre *)re)->code;
|
| 50 |
|
| 51 |
printf("------------------------------------------------------------------\n");
|
| 52 |
|
| 53 |
for(;;)
|
| 54 |
{
|
| 55 |
int c;
|
| 56 |
int charlength;
|
| 57 |
|
| 58 |
printf("%3d ", code - ((real_pcre *)re)->code);
|
| 59 |
|
| 60 |
if (*code >= OP_BRA)
|
| 61 |
{
|
| 62 |
printf("%3d Bra %d", (code[1] << 8) + code[2], *code - OP_BRA);
|
| 63 |
code += 2;
|
| 64 |
}
|
| 65 |
|
| 66 |
else switch(*code)
|
| 67 |
{
|
| 68 |
case OP_END:
|
| 69 |
printf(" %s\n", OP_names[*code]);
|
| 70 |
printf("------------------------------------------------------------------\n");
|
| 71 |
return;
|
| 72 |
|
| 73 |
case OP_CHARS:
|
| 74 |
charlength = *(++code);
|
| 75 |
printf("%3d ", charlength);
|
| 76 |
while (charlength-- > 0)
|
| 77 |
if (isprint(c = *(++code))) printf("%c", c); else printf("\\x%02x", c);
|
| 78 |
break;
|
| 79 |
|
| 80 |
case OP_KETRMAX:
|
| 81 |
case OP_KETRMIN:
|
| 82 |
case OP_ALT:
|
| 83 |
case OP_KET:
|
| 84 |
case OP_ASSERT:
|
| 85 |
case OP_ASSERT_NOT:
|
| 86 |
case OP_ONCE:
|
| 87 |
printf("%3d %s", (code[1] << 8) + code[2], OP_names[*code]);
|
| 88 |
code += 2;
|
| 89 |
break;
|
| 90 |
|
| 91 |
case OP_STAR:
|
| 92 |
case OP_MINSTAR:
|
| 93 |
case OP_PLUS:
|
| 94 |
case OP_MINPLUS:
|
| 95 |
case OP_QUERY:
|
| 96 |
case OP_MINQUERY:
|
| 97 |
case OP_TYPESTAR:
|
| 98 |
case OP_TYPEMINSTAR:
|
| 99 |
case OP_TYPEPLUS:
|
| 100 |
case OP_TYPEMINPLUS:
|
| 101 |
case OP_TYPEQUERY:
|
| 102 |
case OP_TYPEMINQUERY:
|
| 103 |
if (*code >= OP_TYPESTAR)
|
| 104 |
printf(" %s", OP_names[code[1]]);
|
| 105 |
else if (isprint(c = code[1])) printf(" %c", c);
|
| 106 |
else printf(" \\x%02x", c);
|
| 107 |
printf("%s", OP_names[*code++]);
|
| 108 |
break;
|
| 109 |
|
| 110 |
case OP_EXACT:
|
| 111 |
case OP_UPTO:
|
| 112 |
case OP_MINUPTO:
|
| 113 |
if (isprint(c = code[3])) printf(" %c{", c);
|
| 114 |
else printf(" \\x%02x{", c);
|
| 115 |
if (*code != OP_EXACT) printf(",");
|
| 116 |
printf("%d}", (code[1] << 8) + code[2]);
|
| 117 |
if (*code == OP_MINUPTO) printf("?");
|
| 118 |
code += 3;
|
| 119 |
break;
|
| 120 |
|
| 121 |
case OP_TYPEEXACT:
|
| 122 |
case OP_TYPEUPTO:
|
| 123 |
case OP_TYPEMINUPTO:
|
| 124 |
printf(" %s{", OP_names[code[3]]);
|
| 125 |
if (*code != OP_TYPEEXACT) printf("0,");
|
| 126 |
printf("%d}", (code[1] << 8) + code[2]);
|
| 127 |
if (*code == OP_TYPEMINUPTO) printf("?");
|
| 128 |
code += 3;
|
| 129 |
break;
|
| 130 |
|
| 131 |
case OP_NOT:
|
| 132 |
if (isprint(c = *(++code))) printf(" [^%c]", c);
|
| 133 |
else printf(" [^\\x%02x]", c);
|
| 134 |
break;
|
| 135 |
|
| 136 |
case OP_NOTSTAR:
|
| 137 |
case OP_NOTMINSTAR:
|
| 138 |
case OP_NOTPLUS:
|
| 139 |
case OP_NOTMINPLUS:
|
| 140 |
case OP_NOTQUERY:
|
| 141 |
case OP_NOTMINQUERY:
|
| 142 |
if (isprint(c = code[1])) printf(" [^%c]", c);
|
| 143 |
else printf(" [^\\x%02x]", c);
|
| 144 |
printf("%s", OP_names[*code++]);
|
| 145 |
break;
|
| 146 |
|
| 147 |
case OP_NOTEXACT:
|
| 148 |
case OP_NOTUPTO:
|
| 149 |
case OP_NOTMINUPTO:
|
| 150 |
if (isprint(c = code[3])) printf(" [^%c]{", c);
|
| 151 |
else printf(" [^\\x%02x]{", c);
|
| 152 |
if (*code != OP_NOTEXACT) printf(",");
|
| 153 |
printf("%d}", (code[1] << 8) + code[2]);
|
| 154 |
if (*code == OP_NOTMINUPTO) printf("?");
|
| 155 |
code += 3;
|
| 156 |
break;
|
| 157 |
|
| 158 |
case OP_REF:
|
| 159 |
printf(" \\%d", *(++code));
|
| 160 |
code++;
|
| 161 |
goto CLASS_REF_REPEAT;
|
| 162 |
|
| 163 |
case OP_CLASS:
|
| 164 |
case OP_NEGCLASS:
|
| 165 |
{
|
| 166 |
int i, min, max;
|
| 167 |
if (*code++ == OP_CLASS) printf(" [");
|
| 168 |
else printf(" ^[");
|
| 169 |
|
| 170 |
for (i = 0; i < 256; i++)
|
| 171 |
{
|
| 172 |
if ((code[i/8] & (1 << (i&7))) != 0)
|
| 173 |
{
|
| 174 |
int j;
|
| 175 |
for (j = i+1; j < 256; j++)
|
| 176 |
if ((code[j/8] & (1 << (j&7))) == 0) break;
|
| 177 |
if (i == '-' || i == ']') printf("\\");
|
| 178 |
if (isprint(i)) printf("%c", i); else printf("\\x%02x", i);
|
| 179 |
if (--j > i)
|
| 180 |
{
|
| 181 |
printf("-");
|
| 182 |
if (j == '-' || j == ']') printf("\\");
|
| 183 |
if (isprint(j)) printf("%c", j); else printf("\\x%02x", j);
|
| 184 |
}
|
| 185 |
i = j;
|
| 186 |
}
|
| 187 |
}
|
| 188 |
printf("]");
|
| 189 |
code += 32;
|
| 190 |
|
| 191 |
CLASS_REF_REPEAT:
|
| 192 |
|
| 193 |
switch(*code)
|
| 194 |
{
|
| 195 |
case OP_CRSTAR:
|
| 196 |
case OP_CRMINSTAR:
|
| 197 |
case OP_CRPLUS:
|
| 198 |
case OP_CRMINPLUS:
|
| 199 |
case OP_CRQUERY:
|
| 200 |
case OP_CRMINQUERY:
|
| 201 |
printf("%s", OP_names[*code]);
|
| 202 |
break;
|
| 203 |
|
| 204 |
case OP_CRRANGE:
|
| 205 |
case OP_CRMINRANGE:
|
| 206 |
min = (code[1] << 8) + code[2];
|
| 207 |
max = (code[3] << 8) + code[4];
|
| 208 |
if (max == 0) printf("{%d,}", min);
|
| 209 |
else printf("{%d,%d}", min, max);
|
| 210 |
if (*code == OP_CRMINRANGE) printf("?");
|
| 211 |
code += 4;
|
| 212 |
break;
|
| 213 |
|
| 214 |
default:
|
| 215 |
code--;
|
| 216 |
}
|
| 217 |
}
|
| 218 |
break;
|
| 219 |
|
| 220 |
/* Anything else is just a one-node item */
|
| 221 |
|
| 222 |
default:
|
| 223 |
printf(" %s", OP_names[*code]);
|
| 224 |
break;
|
| 225 |
}
|
| 226 |
|
| 227 |
code++;
|
| 228 |
printf("\n");
|
| 229 |
}
|
| 230 |
}
|
| 231 |
|
| 232 |
|
| 233 |
|
| 234 |
/* Character string printing function. */
|
| 235 |
|
| 236 |
static void pchars(unsigned char *p, int length)
|
| 237 |
{
|
| 238 |
int c;
|
| 239 |
while (length-- > 0)
|
| 240 |
if (isprint(c = *(p++))) fprintf(outfile, "%c", c);
|
| 241 |
else fprintf(outfile, "\\x%02x", c);
|
| 242 |
}
|
| 243 |
|
| 244 |
|
| 245 |
|
| 246 |
/* Alternative malloc function, to test functionality and show the size of the
|
| 247 |
compiled re. */
|
| 248 |
|
| 249 |
static void *new_malloc(size_t size)
|
| 250 |
{
|
| 251 |
if (log_store) fprintf(outfile, "Store size request: %d\n", (int)size);
|
| 252 |
return malloc(size);
|
| 253 |
}
|
| 254 |
|
| 255 |
|
| 256 |
|
| 257 |
/* Read lines from named file or stdin and write to named file or stdout; lines
|
| 258 |
consist of a regular expression, in delimiters and optionally followed by
|
| 259 |
options, followed by a set of test data, terminated by an empty line. */
|
| 260 |
|
| 261 |
int main(int argc, char **argv)
|
| 262 |
{
|
| 263 |
FILE *infile = stdin;
|
| 264 |
int options = 0;
|
| 265 |
int study_options = 0;
|
| 266 |
int op = 1;
|
| 267 |
int timeit = 0;
|
| 268 |
int showinfo = 0;
|
| 269 |
int posix = 0;
|
| 270 |
int debug = 0;
|
| 271 |
int done = 0;
|
| 272 |
unsigned char buffer[30000];
|
| 273 |
unsigned char dbuffer[1024];
|
| 274 |
|
| 275 |
/* Static so that new_malloc can use it. */
|
| 276 |
|
| 277 |
outfile = stdout;
|
| 278 |
|
| 279 |
/* Scan options */
|
| 280 |
|
| 281 |
while (argc > 1 && argv[op][0] == '-')
|
| 282 |
{
|
| 283 |
if (strcmp(argv[op], "-s") == 0) log_store = 1;
|
| 284 |
else if (strcmp(argv[op], "-t") == 0) timeit = 1;
|
| 285 |
else if (strcmp(argv[op], "-i") == 0) showinfo = 1;
|
| 286 |
else if (strcmp(argv[op], "-d") == 0) showinfo = debug = 1;
|
| 287 |
else if (strcmp(argv[op], "-p") == 0) posix = 1;
|
| 288 |
else
|
| 289 |
{
|
| 290 |
printf("*** Unknown option %s\n", argv[op]);
|
| 291 |
return 1;
|
| 292 |
}
|
| 293 |
op++;
|
| 294 |
argc--;
|
| 295 |
}
|
| 296 |
|
| 297 |
/* Sort out the input and output files */
|
| 298 |
|
| 299 |
if (argc > 1)
|
| 300 |
{
|
| 301 |
infile = fopen(argv[op], "r");
|
| 302 |
if (infile == NULL)
|
| 303 |
{
|
| 304 |
printf("** Failed to open %s\n", argv[op]);
|
| 305 |
return 1;
|
| 306 |
}
|
| 307 |
}
|
| 308 |
|
| 309 |
if (argc > 2)
|
| 310 |
{
|
| 311 |
outfile = fopen(argv[op+1], "w");
|
| 312 |
if (outfile == NULL)
|
| 313 |
{
|
| 314 |
printf("** Failed to open %s\n", argv[op+1]);
|
| 315 |
return 1;
|
| 316 |
}
|
| 317 |
}
|
| 318 |
|
| 319 |
/* Set alternative malloc function */
|
| 320 |
|
| 321 |
pcre_malloc = new_malloc;
|
| 322 |
|
| 323 |
/* Heading line, then prompt for first re if stdin */
|
| 324 |
|
| 325 |
fprintf(outfile, "Testing Perl-Compatible Regular Expressions\n");
|
| 326 |
fprintf(outfile, "PCRE version %s\n\n", pcre_version());
|
| 327 |
|
| 328 |
/* Main loop */
|
| 329 |
|
| 330 |
while (!done)
|
| 331 |
{
|
| 332 |
pcre *re = NULL;
|
| 333 |
pcre_extra *extra = NULL;
|
| 334 |
regex_t preg;
|
| 335 |
const char *error;
|
| 336 |
unsigned char *p, *pp;
|
| 337 |
int do_study = 0;
|
| 338 |
int do_debug = 0;
|
| 339 |
int do_posix = 0;
|
| 340 |
int erroroffset, len, delimiter;
|
| 341 |
|
| 342 |
if (infile == stdin) printf(" re> ");
|
| 343 |
if (fgets((char *)buffer, sizeof(buffer), infile) == NULL) break;
|
| 344 |
if (infile != stdin) fprintf(outfile, (char *)buffer);
|
| 345 |
|
| 346 |
p = buffer;
|
| 347 |
while (isspace(*p)) p++;
|
| 348 |
if (*p == 0) continue;
|
| 349 |
|
| 350 |
/* Get the delimiter and seek the end of the pattern; if is isn't
|
| 351 |
complete, read more. */
|
| 352 |
|
| 353 |
delimiter = *p++;
|
| 354 |
|
| 355 |
if (isalnum(delimiter))
|
| 356 |
{
|
| 357 |
fprintf(outfile, "** Delimiter must not be alphameric\n");
|
| 358 |
goto SKIP_DATA;
|
| 359 |
}
|
| 360 |
|
| 361 |
pp = p;
|
| 362 |
|
| 363 |
for(;;)
|
| 364 |
{
|
| 365 |
while (*pp != 0 && *pp != delimiter) pp++;
|
| 366 |
if (*pp != 0) break;
|
| 367 |
|
| 368 |
len = sizeof(buffer) - (pp - buffer);
|
| 369 |
if (len < 256)
|
| 370 |
{
|
| 371 |
fprintf(outfile, "** Expression too long - missing delimiter?\n");
|
| 372 |
goto SKIP_DATA;
|
| 373 |
}
|
| 374 |
|
| 375 |
if (infile == stdin) printf(" > ");
|
| 376 |
if (fgets((char *)pp, len, infile) == NULL)
|
| 377 |
{
|
| 378 |
fprintf(outfile, "** Unexpected EOF\n");
|
| 379 |
done = 1;
|
| 380 |
goto CONTINUE;
|
| 381 |
}
|
| 382 |
if (infile != stdin) fprintf(outfile, (char *)pp);
|
| 383 |
}
|
| 384 |
|
| 385 |
/* Terminate the pattern at the delimiter */
|
| 386 |
|
| 387 |
*pp++ = 0;
|
| 388 |
|
| 389 |
/* Look for options after final delimiter */
|
| 390 |
|
| 391 |
options = 0;
|
| 392 |
study_options = 0;
|
| 393 |
while (*pp != 0)
|
| 394 |
{
|
| 395 |
switch (*pp++)
|
| 396 |
{
|
| 397 |
case 'i': options |= PCRE_CASELESS; break;
|
| 398 |
case 'm': options |= PCRE_MULTILINE; break;
|
| 399 |
case 's': options |= PCRE_DOTALL; break;
|
| 400 |
case 'x': options |= PCRE_EXTENDED; break;
|
| 401 |
case 'A': options |= PCRE_ANCHORED; break;
|
| 402 |
case 'D': do_debug = 1; break;
|
| 403 |
case 'E': options |= PCRE_DOLLAR_ENDONLY; break;
|
| 404 |
case 'P': do_posix = 1; break;
|
| 405 |
case 'S': do_study = 1; break;
|
| 406 |
case 'I': study_options |= PCRE_CASELESS; break;
|
| 407 |
case 'U': options |= PCRE_UNGREEDY; break;
|
| 408 |
case 'X': options |= PCRE_EXTRA; break;
|
| 409 |
case '\n': case ' ': break;
|
| 410 |
default:
|
| 411 |
fprintf(outfile, "** Unknown option '%c'\n", pp[-1]);
|
| 412 |
goto SKIP_DATA;
|
| 413 |
}
|
| 414 |
}
|
| 415 |
|
| 416 |
/* Handle compiling via the POSIX interface, which doesn't support the
|
| 417 |
timing, showing, or debugging options. */
|
| 418 |
|
| 419 |
if (posix || do_posix)
|
| 420 |
{
|
| 421 |
int rc;
|
| 422 |
int cflags = 0;
|
| 423 |
if ((options & PCRE_CASELESS) != 0) cflags |= REG_ICASE;
|
| 424 |
if ((options & PCRE_MULTILINE) != 0) cflags |= REG_NEWLINE;
|
| 425 |
rc = regcomp(&preg, (char *)p, cflags);
|
| 426 |
|
| 427 |
/* Compilation failed; go back for another re, skipping to blank line
|
| 428 |
if non-interactive. */
|
| 429 |
|
| 430 |
if (rc != 0)
|
| 431 |
{
|
| 432 |
(void)regerror(rc, &preg, (char *)buffer, sizeof(buffer));
|
| 433 |
fprintf(outfile, "Failed: POSIX code %d: %s\n", rc, buffer);
|
| 434 |
goto SKIP_DATA;
|
| 435 |
}
|
| 436 |
}
|
| 437 |
|
| 438 |
/* Handle compiling via the native interface */
|
| 439 |
|
| 440 |
else
|
| 441 |
{
|
| 442 |
if (timeit)
|
| 443 |
{
|
| 444 |
register int i;
|
| 445 |
clock_t time_taken;
|
| 446 |
clock_t start_time = clock();
|
| 447 |
for (i = 0; i < 4000; i++)
|
| 448 |
{
|
| 449 |
re = pcre_compile((char *)p, options, &error, &erroroffset);
|
| 450 |
if (re != NULL) free(re);
|
| 451 |
}
|
| 452 |
time_taken = clock() - start_time;
|
| 453 |
fprintf(outfile, "Compile time %.2f milliseconds\n",
|
| 454 |
((double)time_taken)/(4 * CLOCKS_PER_SEC));
|
| 455 |
}
|
| 456 |
|
| 457 |
re = pcre_compile((char *)p, options, &error, &erroroffset);
|
| 458 |
|
| 459 |
/* Compilation failed; go back for another re, skipping to blank line
|
| 460 |
if non-interactive. */
|
| 461 |
|
| 462 |
if (re == NULL)
|
| 463 |
{
|
| 464 |
fprintf(outfile, "Failed: %s at offset %d\n", error, erroroffset);
|
| 465 |
SKIP_DATA:
|
| 466 |
if (infile != stdin)
|
| 467 |
{
|
| 468 |
for (;;)
|
| 469 |
{
|
| 470 |
if (fgets((char *)buffer, sizeof(buffer), infile) == NULL)
|
| 471 |
{
|
| 472 |
done = 1;
|
| 473 |
goto CONTINUE;
|
| 474 |
}
|
| 475 |
len = (int)strlen((char *)buffer);
|
| 476 |
while (len > 0 && isspace(buffer[len-1])) len--;
|
| 477 |
if (len == 0) break;
|
| 478 |
}
|
| 479 |
fprintf(outfile, "\n");
|
| 480 |
}
|
| 481 |
continue;
|
| 482 |
}
|
| 483 |
|
| 484 |
/* Compilation succeeded; print data if required */
|
| 485 |
|
| 486 |
if (showinfo || do_debug)
|
| 487 |
{
|
| 488 |
int first_char, count;
|
| 489 |
|
| 490 |
if (debug || do_debug) print_internals(re);
|
| 491 |
|
| 492 |
count = pcre_info(re, &options, &first_char);
|
| 493 |
if (count < 0) fprintf(outfile,
|
| 494 |
"Error %d while reading info\n", count);
|
| 495 |
else
|
| 496 |
{
|
| 497 |
fprintf(outfile, "Identifying subpattern count = %d\n", count);
|
| 498 |
if (options == 0) fprintf(outfile, "No options\n");
|
| 499 |
else fprintf(outfile, "Options:%s%s%s%s%s%s%s%s\n",
|
| 500 |
((options & PCRE_ANCHORED) != 0)? " anchored" : "",
|
| 501 |
((options & PCRE_CASELESS) != 0)? " caseless" : "",
|
| 502 |
((options & PCRE_EXTENDED) != 0)? " extended" : "",
|
| 503 |
((options & PCRE_MULTILINE) != 0)? " multiline" : "",
|
| 504 |
((options & PCRE_DOTALL) != 0)? " dotall" : "",
|
| 505 |
((options & PCRE_DOLLAR_ENDONLY) != 0)? " dollar_endonly" : "",
|
| 506 |
((options & PCRE_EXTRA) != 0)? " extra" : "",
|
| 507 |
((options & PCRE_UNGREEDY) != 0)? " ungreedy" : "");
|
| 508 |
if (first_char == -1)
|
| 509 |
{
|
| 510 |
fprintf(outfile, "First char at start or follows \\n\n");
|
| 511 |
}
|
| 512 |
else if (first_char < 0)
|
| 513 |
{
|
| 514 |
fprintf(outfile, "No first char\n");
|
| 515 |
}
|
| 516 |
else
|
| 517 |
{
|
| 518 |
if (isprint(first_char))
|
| 519 |
fprintf(outfile, "First char = \'%c\'\n", first_char);
|
| 520 |
else
|
| 521 |
fprintf(outfile, "First char = %d\n", first_char);
|
| 522 |
}
|
| 523 |
}
|
| 524 |
}
|
| 525 |
|
| 526 |
/* If /S was present, study the regexp to generate additional info to
|
| 527 |
help with the matching. */
|
| 528 |
|
| 529 |
if (do_study)
|
| 530 |
{
|
| 531 |
if (timeit)
|
| 532 |
{
|
| 533 |
register int i;
|
| 534 |
clock_t time_taken;
|
| 535 |
clock_t start_time = clock();
|
| 536 |
for (i = 0; i < 4000; i++)
|
| 537 |
extra = pcre_study(re, study_options, &error);
|
| 538 |
time_taken = clock() - start_time;
|
| 539 |
if (extra != NULL) free(extra);
|
| 540 |
fprintf(outfile, " Study time %.2f milliseconds\n",
|
| 541 |
((double)time_taken)/(4 * CLOCKS_PER_SEC));
|
| 542 |
}
|
| 543 |
|
| 544 |
extra = pcre_study(re, study_options, &error);
|
| 545 |
if (error != NULL)
|
| 546 |
fprintf(outfile, "Failed to study: %s\n", error);
|
| 547 |
else if (extra == NULL)
|
| 548 |
fprintf(outfile, "Study returned NULL\n");
|
| 549 |
|
| 550 |
/* This looks at internal information. A bit kludgy to do it this
|
| 551 |
way, but it is useful for testing. */
|
| 552 |
|
| 553 |
else if (showinfo || do_debug)
|
| 554 |
{
|
| 555 |
real_pcre_extra *xx = (real_pcre_extra *)extra;
|
| 556 |
if ((xx->options & PCRE_STUDY_MAPPED) == 0)
|
| 557 |
fprintf(outfile, "No starting character set\n");
|
| 558 |
else
|
| 559 |
{
|
| 560 |
int i;
|
| 561 |
int c = 24;
|
| 562 |
fprintf(outfile, "Starting character set: ");
|
| 563 |
for (i = 0; i < 256; i++)
|
| 564 |
{
|
| 565 |
if ((xx->start_bits[i/8] & (1<<(i%8))) != 0)
|
| 566 |
{
|
| 567 |
if (c > 75)
|
| 568 |
{
|
| 569 |
fprintf(outfile, "\n ");
|
| 570 |
c = 2;
|
| 571 |
}
|
| 572 |
if (isprint(i) && i != ' ')
|
| 573 |
{
|
| 574 |
fprintf(outfile, "%c ", i);
|
| 575 |
c += 2;
|
| 576 |
}
|
| 577 |
else
|
| 578 |
{
|
| 579 |
fprintf(outfile, "\\x%02x ", i);
|
| 580 |
c += 5;
|
| 581 |
}
|
| 582 |
}
|
| 583 |
}
|
| 584 |
fprintf(outfile, "\n");
|
| 585 |
}
|
| 586 |
}
|
| 587 |
}
|
| 588 |
}
|
| 589 |
|
| 590 |
/* Read data lines and test them */
|
| 591 |
|
| 592 |
for (;;)
|
| 593 |
{
|
| 594 |
unsigned char *q;
|
| 595 |
int count, c;
|
| 596 |
int offsets[30];
|
| 597 |
int size_offsets = sizeof(offsets)/sizeof(int);
|
| 598 |
|
| 599 |
options = 0;
|
| 600 |
|
| 601 |
if (infile == stdin) printf(" data> ");
|
| 602 |
if (fgets((char *)buffer, sizeof(buffer), infile) == NULL)
|
| 603 |
{
|
| 604 |
done = 1;
|
| 605 |
goto CONTINUE;
|
| 606 |
}
|
| 607 |
if (infile != stdin) fprintf(outfile, (char *)buffer);
|
| 608 |
|
| 609 |
len = (int)strlen((char *)buffer);
|
| 610 |
while (len > 0 && isspace(buffer[len-1])) len--;
|
| 611 |
buffer[len] = 0;
|
| 612 |
if (len == 0) break;
|
| 613 |
|
| 614 |
p = buffer;
|
| 615 |
while (isspace(*p)) p++;
|
| 616 |
|
| 617 |
q = dbuffer;
|
| 618 |
while ((c = *p++) != 0)
|
| 619 |
{
|
| 620 |
int i = 0;
|
| 621 |
int n = 0;
|
| 622 |
if (c == '\\') switch ((c = *p++))
|
| 623 |
{
|
| 624 |
case 'a': c = 7; break;
|
| 625 |
case 'b': c = '\b'; break;
|
| 626 |
case 'e': c = 27; break;
|
| 627 |
case 'f': c = '\f'; break;
|
| 628 |
case 'n': c = '\n'; break;
|
| 629 |
case 'r': c = '\r'; break;
|
| 630 |
case 't': c = '\t'; break;
|
| 631 |
case 'v': c = '\v'; break;
|
| 632 |
|
| 633 |
case '0': case '1': case '2': case '3':
|
| 634 |
case '4': case '5': case '6': case '7':
|
| 635 |
c -= '0';
|
| 636 |
while (i++ < 2 && isdigit(*p) && *p != '8' && *p != '9')
|
| 637 |
c = c * 8 + *p++ - '0';
|
| 638 |
break;
|
| 639 |
|
| 640 |
case 'x':
|
| 641 |
c = 0;
|
| 642 |
while (i++ < 2 && isxdigit(*p))
|
| 643 |
{
|
| 644 |
c = c * 16 + tolower(*p) - ((isdigit(*p))? '0' : 'W');
|
| 645 |
p++;
|
| 646 |
}
|
| 647 |
break;
|
| 648 |
|
| 649 |
case 0: /* Allows for an empty line */
|
| 650 |
p--;
|
| 651 |
continue;
|
| 652 |
|
| 653 |
case 'A': /* Option setting */
|
| 654 |
options |= PCRE_ANCHORED;
|
| 655 |
continue;
|
| 656 |
|
| 657 |
case 'B':
|
| 658 |
options |= PCRE_NOTBOL;
|
| 659 |
continue;
|
| 660 |
|
| 661 |
case 'E':
|
| 662 |
options |= PCRE_DOLLAR_ENDONLY;
|
| 663 |
continue;
|
| 664 |
|
| 665 |
case 'I':
|
| 666 |
options |= PCRE_CASELESS;
|
| 667 |
continue;
|
| 668 |
|
| 669 |
case 'M':
|
| 670 |
options |= PCRE_MULTILINE;
|
| 671 |
continue;
|
| 672 |
|
| 673 |
case 'S':
|
| 674 |
options |= PCRE_DOTALL;
|
| 675 |
continue;
|
| 676 |
|
| 677 |
case 'O':
|
| 678 |
while(isdigit(*p)) n = n * 10 + *p++ - '0';
|
| 679 |
if (n <= (int)(sizeof(offsets)/sizeof(int))) size_offsets = n;
|
| 680 |
continue;
|
| 681 |
|
| 682 |
case 'Z':
|
| 683 |
options |= PCRE_NOTEOL;
|
| 684 |
continue;
|
| 685 |
}
|
| 686 |
*q++ = c;
|
| 687 |
}
|
| 688 |
*q = 0;
|
| 689 |
len = q - dbuffer;
|
| 690 |
|
| 691 |
/* Handle matching via the POSIX interface, which does not
|
| 692 |
support timing. */
|
| 693 |
|
| 694 |
if (posix || do_posix)
|
| 695 |
{
|
| 696 |
int rc;
|
| 697 |
int eflags = 0;
|
| 698 |
regmatch_t pmatch[30];
|
| 699 |
if ((options & PCRE_NOTBOL) != 0) eflags |= REG_NOTBOL;
|
| 700 |
if ((options & PCRE_NOTEOL) != 0) eflags |= REG_NOTEOL;
|
| 701 |
|
| 702 |
rc = regexec(&preg, (char *)dbuffer, sizeof(pmatch)/sizeof(regmatch_t),
|
| 703 |
pmatch, eflags);
|
| 704 |
|
| 705 |
if (rc != 0)
|
| 706 |
{
|
| 707 |
(void)regerror(rc, &preg, (char *)buffer, sizeof(buffer));
|
| 708 |
fprintf(outfile, "No match: POSIX code %d: %s\n", rc, buffer);
|
| 709 |
}
|
| 710 |
else
|
| 711 |
{
|
| 712 |
size_t i;
|
| 713 |
for (i = 0; i < sizeof(pmatch)/sizeof(regmatch_t); i++)
|
| 714 |
{
|
| 715 |
if (pmatch[i].rm_so >= 0)
|
| 716 |
{
|
| 717 |
fprintf(outfile, "%2d: ", i);
|
| 718 |
pchars(dbuffer + pmatch[i].rm_so,
|
| 719 |
pmatch[i].rm_eo - pmatch[i].rm_so);
|
| 720 |
fprintf(outfile, "\n");
|
| 721 |
}
|
| 722 |
}
|
| 723 |
}
|
| 724 |
}
|
| 725 |
|
| 726 |
/* Handle matching via the native interface */
|
| 727 |
|
| 728 |
else
|
| 729 |
{
|
| 730 |
if (timeit)
|
| 731 |
{
|
| 732 |
register int i;
|
| 733 |
clock_t time_taken;
|
| 734 |
clock_t start_time = clock();
|
| 735 |
for (i = 0; i < 4000; i++)
|
| 736 |
count = pcre_exec(re, extra, (char *)dbuffer, len, options, offsets,
|
| 737 |
size_offsets);
|
| 738 |
time_taken = clock() - start_time;
|
| 739 |
fprintf(outfile, "Execute time %.2f milliseconds\n",
|
| 740 |
((double)time_taken)/(4 * CLOCKS_PER_SEC));
|
| 741 |
}
|
| 742 |
|
| 743 |
count = pcre_exec(re, extra, (char *)dbuffer, len, options, offsets,
|
| 744 |
size_offsets);
|
| 745 |
|
| 746 |
if (count == 0)
|
| 747 |
{
|
| 748 |
fprintf(outfile, "Matched, but too many substrings\n");
|
| 749 |
count = size_offsets/2;
|
| 750 |
}
|
| 751 |
|
| 752 |
if (count >= 0)
|
| 753 |
{
|
| 754 |
int i;
|
| 755 |
count *= 2;
|
| 756 |
for (i = 0; i < count; i += 2)
|
| 757 |
{
|
| 758 |
if (offsets[i] < 0)
|
| 759 |
fprintf(outfile, "%2d: <unset>\n", i/2);
|
| 760 |
else
|
| 761 |
{
|
| 762 |
fprintf(outfile, "%2d: ", i/2);
|
| 763 |
pchars(dbuffer + offsets[i], offsets[i+1] - offsets[i]);
|
| 764 |
fprintf(outfile, "\n");
|
| 765 |
}
|
| 766 |
}
|
| 767 |
}
|
| 768 |
else
|
| 769 |
{
|
| 770 |
if (count == -1) fprintf(outfile, "No match\n");
|
| 771 |
else fprintf(outfile, "Error %d\n", count);
|
| 772 |
}
|
| 773 |
}
|
| 774 |
}
|
| 775 |
|
| 776 |
CONTINUE:
|
| 777 |
if (posix || do_posix) regfree(&preg);
|
| 778 |
if (re != NULL) free(re);
|
| 779 |
if (extra != NULL) free(extra);
|
| 780 |
}
|
| 781 |
|
| 782 |
fprintf(outfile, "\n");
|
| 783 |
return 0;
|
| 784 |
}
|
| 785 |
|
| 786 |
/* End */
|
Parent Directory
| 
Revision Log