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 |
#include <locale.h>
|
11 |
|
12 |
/* Use the internal info for displaying the results of pcre_study(). */
|
13 |
|
14 |
#include "internal.h"
|
15 |
#include "pcreposix.h"
|
16 |
|
17 |
#ifndef CLOCKS_PER_SEC
|
18 |
#ifdef CLK_TCK
|
19 |
#define CLOCKS_PER_SEC CLK_TCK
|
20 |
#else
|
21 |
#define CLOCKS_PER_SEC 100
|
22 |
#endif
|
23 |
#endif
|
24 |
|
25 |
#define LOOPREPEAT 20000
|
26 |
|
27 |
|
28 |
static FILE *outfile;
|
29 |
static int log_store = 0;
|
30 |
|
31 |
|
32 |
|
33 |
/* Debugging function to print the internal form of the regex. This is the same
|
34 |
code as contained in pcre.c under the DEBUG macro. */
|
35 |
|
36 |
static const char *OP_names[] = {
|
37 |
"End", "\\A", "\\B", "\\b", "\\D", "\\d",
|
38 |
"\\S", "\\s", "\\W", "\\w", "\\Z", "\\z",
|
39 |
"Opt", "^", "$", "Any", "chars", "not",
|
40 |
"*", "*?", "+", "+?", "?", "??", "{", "{", "{",
|
41 |
"*", "*?", "+", "+?", "?", "??", "{", "{", "{",
|
42 |
"*", "*?", "+", "+?", "?", "??", "{", "{", "{",
|
43 |
"*", "*?", "+", "+?", "?", "??", "{", "{",
|
44 |
"class", "Ref",
|
45 |
"Alt", "Ket", "KetRmax", "KetRmin", "Assert", "Assert not",
|
46 |
"AssertB", "AssertB not", "Reverse", "Once", "Cond", "Cref",
|
47 |
"Brazero", "Braminzero", "Bra"
|
48 |
};
|
49 |
|
50 |
|
51 |
static void print_internals(pcre *re, FILE *outfile)
|
52 |
{
|
53 |
unsigned char *code = ((real_pcre *)re)->code;
|
54 |
|
55 |
fprintf(outfile, "------------------------------------------------------------------\n");
|
56 |
|
57 |
for(;;)
|
58 |
{
|
59 |
int c;
|
60 |
int charlength;
|
61 |
|
62 |
fprintf(outfile, "%3d ", (int)(code - ((real_pcre *)re)->code));
|
63 |
|
64 |
if (*code >= OP_BRA)
|
65 |
{
|
66 |
fprintf(outfile, "%3d Bra %d", (code[1] << 8) + code[2], *code - OP_BRA);
|
67 |
code += 2;
|
68 |
}
|
69 |
|
70 |
else switch(*code)
|
71 |
{
|
72 |
case OP_END:
|
73 |
fprintf(outfile, " %s\n", OP_names[*code]);
|
74 |
fprintf(outfile, "------------------------------------------------------------------\n");
|
75 |
return;
|
76 |
|
77 |
case OP_OPT:
|
78 |
fprintf(outfile, " %.2x %s", code[1], OP_names[*code]);
|
79 |
code++;
|
80 |
break;
|
81 |
|
82 |
case OP_COND:
|
83 |
fprintf(outfile, "%3d Cond", (code[1] << 8) + code[2]);
|
84 |
code += 2;
|
85 |
break;
|
86 |
|
87 |
case OP_CREF:
|
88 |
fprintf(outfile, " %.2d %s", code[1], OP_names[*code]);
|
89 |
code++;
|
90 |
break;
|
91 |
|
92 |
case OP_CHARS:
|
93 |
charlength = *(++code);
|
94 |
fprintf(outfile, "%3d ", charlength);
|
95 |
while (charlength-- > 0)
|
96 |
if (isprint(c = *(++code))) fprintf(outfile, "%c", c);
|
97 |
else fprintf(outfile, "\\x%02x", c);
|
98 |
break;
|
99 |
|
100 |
case OP_KETRMAX:
|
101 |
case OP_KETRMIN:
|
102 |
case OP_ALT:
|
103 |
case OP_KET:
|
104 |
case OP_ASSERT:
|
105 |
case OP_ASSERT_NOT:
|
106 |
case OP_ASSERTBACK:
|
107 |
case OP_ASSERTBACK_NOT:
|
108 |
case OP_ONCE:
|
109 |
fprintf(outfile, "%3d %s", (code[1] << 8) + code[2], OP_names[*code]);
|
110 |
code += 2;
|
111 |
break;
|
112 |
|
113 |
case OP_REVERSE:
|
114 |
fprintf(outfile, "%3d %s", (code[1] << 8) + code[2], OP_names[*code]);
|
115 |
code += 2;
|
116 |
break;
|
117 |
|
118 |
case OP_STAR:
|
119 |
case OP_MINSTAR:
|
120 |
case OP_PLUS:
|
121 |
case OP_MINPLUS:
|
122 |
case OP_QUERY:
|
123 |
case OP_MINQUERY:
|
124 |
case OP_TYPESTAR:
|
125 |
case OP_TYPEMINSTAR:
|
126 |
case OP_TYPEPLUS:
|
127 |
case OP_TYPEMINPLUS:
|
128 |
case OP_TYPEQUERY:
|
129 |
case OP_TYPEMINQUERY:
|
130 |
if (*code >= OP_TYPESTAR)
|
131 |
fprintf(outfile, " %s", OP_names[code[1]]);
|
132 |
else if (isprint(c = code[1])) fprintf(outfile, " %c", c);
|
133 |
else fprintf(outfile, " \\x%02x", c);
|
134 |
fprintf(outfile, "%s", OP_names[*code++]);
|
135 |
break;
|
136 |
|
137 |
case OP_EXACT:
|
138 |
case OP_UPTO:
|
139 |
case OP_MINUPTO:
|
140 |
if (isprint(c = code[3])) fprintf(outfile, " %c{", c);
|
141 |
else fprintf(outfile, " \\x%02x{", c);
|
142 |
if (*code != OP_EXACT) fprintf(outfile, ",");
|
143 |
fprintf(outfile, "%d}", (code[1] << 8) + code[2]);
|
144 |
if (*code == OP_MINUPTO) fprintf(outfile, "?");
|
145 |
code += 3;
|
146 |
break;
|
147 |
|
148 |
case OP_TYPEEXACT:
|
149 |
case OP_TYPEUPTO:
|
150 |
case OP_TYPEMINUPTO:
|
151 |
fprintf(outfile, " %s{", OP_names[code[3]]);
|
152 |
if (*code != OP_TYPEEXACT) fprintf(outfile, "0,");
|
153 |
fprintf(outfile, "%d}", (code[1] << 8) + code[2]);
|
154 |
if (*code == OP_TYPEMINUPTO) fprintf(outfile, "?");
|
155 |
code += 3;
|
156 |
break;
|
157 |
|
158 |
case OP_NOT:
|
159 |
if (isprint(c = *(++code))) fprintf(outfile, " [^%c]", c);
|
160 |
else fprintf(outfile, " [^\\x%02x]", c);
|
161 |
break;
|
162 |
|
163 |
case OP_NOTSTAR:
|
164 |
case OP_NOTMINSTAR:
|
165 |
case OP_NOTPLUS:
|
166 |
case OP_NOTMINPLUS:
|
167 |
case OP_NOTQUERY:
|
168 |
case OP_NOTMINQUERY:
|
169 |
if (isprint(c = code[1])) fprintf(outfile, " [^%c]", c);
|
170 |
else fprintf(outfile, " [^\\x%02x]", c);
|
171 |
fprintf(outfile, "%s", OP_names[*code++]);
|
172 |
break;
|
173 |
|
174 |
case OP_NOTEXACT:
|
175 |
case OP_NOTUPTO:
|
176 |
case OP_NOTMINUPTO:
|
177 |
if (isprint(c = code[3])) fprintf(outfile, " [^%c]{", c);
|
178 |
else fprintf(outfile, " [^\\x%02x]{", c);
|
179 |
if (*code != OP_NOTEXACT) fprintf(outfile, ",");
|
180 |
fprintf(outfile, "%d}", (code[1] << 8) + code[2]);
|
181 |
if (*code == OP_NOTMINUPTO) fprintf(outfile, "?");
|
182 |
code += 3;
|
183 |
break;
|
184 |
|
185 |
case OP_REF:
|
186 |
fprintf(outfile, " \\%d", *(++code));
|
187 |
code++;
|
188 |
goto CLASS_REF_REPEAT;
|
189 |
|
190 |
case OP_CLASS:
|
191 |
{
|
192 |
int i, min, max;
|
193 |
code++;
|
194 |
fprintf(outfile, " [");
|
195 |
|
196 |
for (i = 0; i < 256; i++)
|
197 |
{
|
198 |
if ((code[i/8] & (1 << (i&7))) != 0)
|
199 |
{
|
200 |
int j;
|
201 |
for (j = i+1; j < 256; j++)
|
202 |
if ((code[j/8] & (1 << (j&7))) == 0) break;
|
203 |
if (i == '-' || i == ']') fprintf(outfile, "\\");
|
204 |
if (isprint(i)) fprintf(outfile, "%c", i); else fprintf(outfile, "\\x%02x", i);
|
205 |
if (--j > i)
|
206 |
{
|
207 |
fprintf(outfile, "-");
|
208 |
if (j == '-' || j == ']') fprintf(outfile, "\\");
|
209 |
if (isprint(j)) fprintf(outfile, "%c", j); else fprintf(outfile, "\\x%02x", j);
|
210 |
}
|
211 |
i = j;
|
212 |
}
|
213 |
}
|
214 |
fprintf(outfile, "]");
|
215 |
code += 32;
|
216 |
|
217 |
CLASS_REF_REPEAT:
|
218 |
|
219 |
switch(*code)
|
220 |
{
|
221 |
case OP_CRSTAR:
|
222 |
case OP_CRMINSTAR:
|
223 |
case OP_CRPLUS:
|
224 |
case OP_CRMINPLUS:
|
225 |
case OP_CRQUERY:
|
226 |
case OP_CRMINQUERY:
|
227 |
fprintf(outfile, "%s", OP_names[*code]);
|
228 |
break;
|
229 |
|
230 |
case OP_CRRANGE:
|
231 |
case OP_CRMINRANGE:
|
232 |
min = (code[1] << 8) + code[2];
|
233 |
max = (code[3] << 8) + code[4];
|
234 |
if (max == 0) fprintf(outfile, "{%d,}", min);
|
235 |
else fprintf(outfile, "{%d,%d}", min, max);
|
236 |
if (*code == OP_CRMINRANGE) fprintf(outfile, "?");
|
237 |
code += 4;
|
238 |
break;
|
239 |
|
240 |
default:
|
241 |
code--;
|
242 |
}
|
243 |
}
|
244 |
break;
|
245 |
|
246 |
/* Anything else is just a one-node item */
|
247 |
|
248 |
default:
|
249 |
fprintf(outfile, " %s", OP_names[*code]);
|
250 |
break;
|
251 |
}
|
252 |
|
253 |
code++;
|
254 |
fprintf(outfile, "\n");
|
255 |
}
|
256 |
}
|
257 |
|
258 |
|
259 |
|
260 |
/* Character string printing function. */
|
261 |
|
262 |
static void pchars(unsigned char *p, int length)
|
263 |
{
|
264 |
int c;
|
265 |
while (length-- > 0)
|
266 |
if (isprint(c = *(p++))) fprintf(outfile, "%c", c);
|
267 |
else fprintf(outfile, "\\x%02x", c);
|
268 |
}
|
269 |
|
270 |
|
271 |
|
272 |
/* Alternative malloc function, to test functionality and show the size of the
|
273 |
compiled re. */
|
274 |
|
275 |
static void *new_malloc(size_t size)
|
276 |
{
|
277 |
if (log_store) fprintf(outfile, "Store size request: %d\n", (int)size);
|
278 |
return malloc(size);
|
279 |
}
|
280 |
|
281 |
|
282 |
|
283 |
/* Read lines from named file or stdin and write to named file or stdout; lines
|
284 |
consist of a regular expression, in delimiters and optionally followed by
|
285 |
options, followed by a set of test data, terminated by an empty line. */
|
286 |
|
287 |
int main(int argc, char **argv)
|
288 |
{
|
289 |
FILE *infile = stdin;
|
290 |
int options = 0;
|
291 |
int study_options = 0;
|
292 |
int op = 1;
|
293 |
int timeit = 0;
|
294 |
int showinfo = 0;
|
295 |
int posix = 0;
|
296 |
int debug = 0;
|
297 |
int done = 0;
|
298 |
unsigned char buffer[30000];
|
299 |
unsigned char dbuffer[1024];
|
300 |
|
301 |
/* Static so that new_malloc can use it. */
|
302 |
|
303 |
outfile = stdout;
|
304 |
|
305 |
/* Scan options */
|
306 |
|
307 |
while (argc > 1 && argv[op][0] == '-')
|
308 |
{
|
309 |
if (strcmp(argv[op], "-s") == 0) log_store = 1;
|
310 |
else if (strcmp(argv[op], "-t") == 0) timeit = 1;
|
311 |
else if (strcmp(argv[op], "-i") == 0) showinfo = 1;
|
312 |
else if (strcmp(argv[op], "-d") == 0) showinfo = debug = 1;
|
313 |
else if (strcmp(argv[op], "-p") == 0) posix = 1;
|
314 |
else
|
315 |
{
|
316 |
printf("*** Unknown option %s\n", argv[op]);
|
317 |
printf("Usage: pcretest [-d] [-i] [-p] [-s] [-t] [<input> [<output>]]\n");
|
318 |
printf(" -d debug: show compiled code; implies -i\n"
|
319 |
" -i show information about compiled pattern\n"
|
320 |
" -p use POSIX interface\n"
|
321 |
" -s output store information\n"
|
322 |
" -t time compilation and execution\n");
|
323 |
return 1;
|
324 |
}
|
325 |
op++;
|
326 |
argc--;
|
327 |
}
|
328 |
|
329 |
/* Sort out the input and output files */
|
330 |
|
331 |
if (argc > 1)
|
332 |
{
|
333 |
infile = fopen(argv[op], "r");
|
334 |
if (infile == NULL)
|
335 |
{
|
336 |
printf("** Failed to open %s\n", argv[op]);
|
337 |
return 1;
|
338 |
}
|
339 |
}
|
340 |
|
341 |
if (argc > 2)
|
342 |
{
|
343 |
outfile = fopen(argv[op+1], "w");
|
344 |
if (outfile == NULL)
|
345 |
{
|
346 |
printf("** Failed to open %s\n", argv[op+1]);
|
347 |
return 1;
|
348 |
}
|
349 |
}
|
350 |
|
351 |
/* Set alternative malloc function */
|
352 |
|
353 |
pcre_malloc = new_malloc;
|
354 |
|
355 |
/* Heading line, then prompt for first regex if stdin */
|
356 |
|
357 |
fprintf(outfile, "PCRE version %s\n\n", pcre_version());
|
358 |
|
359 |
/* Main loop */
|
360 |
|
361 |
while (!done)
|
362 |
{
|
363 |
pcre *re = NULL;
|
364 |
pcre_extra *extra = NULL;
|
365 |
regex_t preg;
|
366 |
const char *error;
|
367 |
unsigned char *p, *pp, *ppp;
|
368 |
unsigned const char *tables = NULL;
|
369 |
int do_study = 0;
|
370 |
int do_debug = debug;
|
371 |
int do_showinfo = showinfo;
|
372 |
int do_posix = 0;
|
373 |
int erroroffset, len, delimiter;
|
374 |
|
375 |
if (infile == stdin) printf(" re> ");
|
376 |
if (fgets((char *)buffer, sizeof(buffer), infile) == NULL) break;
|
377 |
if (infile != stdin) fprintf(outfile, "%s", (char *)buffer);
|
378 |
|
379 |
p = buffer;
|
380 |
while (isspace(*p)) p++;
|
381 |
if (*p == 0) continue;
|
382 |
|
383 |
/* Get the delimiter and seek the end of the pattern; if is isn't
|
384 |
complete, read more. */
|
385 |
|
386 |
delimiter = *p++;
|
387 |
|
388 |
if (isalnum(delimiter) || delimiter == '\\')
|
389 |
{
|
390 |
fprintf(outfile, "** Delimiter must not be alphameric or \\\n");
|
391 |
goto SKIP_DATA;
|
392 |
}
|
393 |
|
394 |
pp = p;
|
395 |
|
396 |
for(;;)
|
397 |
{
|
398 |
while (*pp != 0)
|
399 |
{
|
400 |
if (*pp == '\\' && pp[1] != 0) pp++;
|
401 |
else if (*pp == delimiter) break;
|
402 |
pp++;
|
403 |
}
|
404 |
if (*pp != 0) break;
|
405 |
|
406 |
len = sizeof(buffer) - (pp - buffer);
|
407 |
if (len < 256)
|
408 |
{
|
409 |
fprintf(outfile, "** Expression too long - missing delimiter?\n");
|
410 |
goto SKIP_DATA;
|
411 |
}
|
412 |
|
413 |
if (infile == stdin) printf(" > ");
|
414 |
if (fgets((char *)pp, len, infile) == NULL)
|
415 |
{
|
416 |
fprintf(outfile, "** Unexpected EOF\n");
|
417 |
done = 1;
|
418 |
goto CONTINUE;
|
419 |
}
|
420 |
if (infile != stdin) fprintf(outfile, "%s", (char *)pp);
|
421 |
}
|
422 |
|
423 |
/* If the first character after the delimiter is backslash, make
|
424 |
the pattern end with backslash. This is purely to provide a way
|
425 |
of testing for the error message when a pattern ends with backslash. */
|
426 |
|
427 |
if (pp[1] == '\\') *pp++ = '\\';
|
428 |
|
429 |
/* Terminate the pattern at the delimiter */
|
430 |
|
431 |
*pp++ = 0;
|
432 |
|
433 |
/* Look for options after final delimiter */
|
434 |
|
435 |
options = 0;
|
436 |
study_options = 0;
|
437 |
while (*pp != 0)
|
438 |
{
|
439 |
switch (*pp++)
|
440 |
{
|
441 |
case 'i': options |= PCRE_CASELESS; break;
|
442 |
case 'm': options |= PCRE_MULTILINE; break;
|
443 |
case 's': options |= PCRE_DOTALL; break;
|
444 |
case 'x': options |= PCRE_EXTENDED; break;
|
445 |
|
446 |
case 'A': options |= PCRE_ANCHORED; break;
|
447 |
case 'D': do_debug = do_showinfo = 1; break;
|
448 |
case 'E': options |= PCRE_DOLLAR_ENDONLY; break;
|
449 |
case 'I': do_showinfo = 1; break;
|
450 |
case 'P': do_posix = 1; break;
|
451 |
case 'S': do_study = 1; break;
|
452 |
case 'U': options |= PCRE_UNGREEDY; break;
|
453 |
case 'X': options |= PCRE_EXTRA; break;
|
454 |
|
455 |
case 'L':
|
456 |
ppp = pp;
|
457 |
while (*ppp != '\n' && *ppp != ' ') ppp++;
|
458 |
*ppp = 0;
|
459 |
if (setlocale(LC_CTYPE, (const char *)pp) == NULL)
|
460 |
{
|
461 |
fprintf(outfile, "** Failed to set locale \"%s\"\n", pp);
|
462 |
goto SKIP_DATA;
|
463 |
}
|
464 |
tables = pcre_maketables();
|
465 |
pp = ppp;
|
466 |
break;
|
467 |
|
468 |
case '\n': case ' ': break;
|
469 |
default:
|
470 |
fprintf(outfile, "** Unknown option '%c'\n", pp[-1]);
|
471 |
goto SKIP_DATA;
|
472 |
}
|
473 |
}
|
474 |
|
475 |
/* Handle compiling via the POSIX interface, which doesn't support the
|
476 |
timing, showing, or debugging options, nor the ability to pass over
|
477 |
local character tables. */
|
478 |
|
479 |
if (posix || do_posix)
|
480 |
{
|
481 |
int rc;
|
482 |
int cflags = 0;
|
483 |
if ((options & PCRE_CASELESS) != 0) cflags |= REG_ICASE;
|
484 |
if ((options & PCRE_MULTILINE) != 0) cflags |= REG_NEWLINE;
|
485 |
rc = regcomp(&preg, (char *)p, cflags);
|
486 |
|
487 |
/* Compilation failed; go back for another re, skipping to blank line
|
488 |
if non-interactive. */
|
489 |
|
490 |
if (rc != 0)
|
491 |
{
|
492 |
(void)regerror(rc, &preg, (char *)buffer, sizeof(buffer));
|
493 |
fprintf(outfile, "Failed: POSIX code %d: %s\n", rc, buffer);
|
494 |
goto SKIP_DATA;
|
495 |
}
|
496 |
}
|
497 |
|
498 |
/* Handle compiling via the native interface */
|
499 |
|
500 |
else
|
501 |
{
|
502 |
if (timeit)
|
503 |
{
|
504 |
register int i;
|
505 |
clock_t time_taken;
|
506 |
clock_t start_time = clock();
|
507 |
for (i = 0; i < LOOPREPEAT; i++)
|
508 |
{
|
509 |
re = pcre_compile((char *)p, options, &error, &erroroffset, tables);
|
510 |
if (re != NULL) free(re);
|
511 |
}
|
512 |
time_taken = clock() - start_time;
|
513 |
fprintf(outfile, "Compile time %.3f milliseconds\n",
|
514 |
((double)time_taken * 1000.0) /
|
515 |
((double)LOOPREPEAT * (double)CLOCKS_PER_SEC));
|
516 |
}
|
517 |
|
518 |
re = pcre_compile((char *)p, options, &error, &erroroffset, tables);
|
519 |
|
520 |
/* Compilation failed; go back for another re, skipping to blank line
|
521 |
if non-interactive. */
|
522 |
|
523 |
if (re == NULL)
|
524 |
{
|
525 |
fprintf(outfile, "Failed: %s at offset %d\n", error, erroroffset);
|
526 |
SKIP_DATA:
|
527 |
if (infile != stdin)
|
528 |
{
|
529 |
for (;;)
|
530 |
{
|
531 |
if (fgets((char *)buffer, sizeof(buffer), infile) == NULL)
|
532 |
{
|
533 |
done = 1;
|
534 |
goto CONTINUE;
|
535 |
}
|
536 |
len = (int)strlen((char *)buffer);
|
537 |
while (len > 0 && isspace(buffer[len-1])) len--;
|
538 |
if (len == 0) break;
|
539 |
}
|
540 |
fprintf(outfile, "\n");
|
541 |
}
|
542 |
goto CONTINUE;
|
543 |
}
|
544 |
|
545 |
/* Compilation succeeded; print data if required */
|
546 |
|
547 |
if (do_showinfo)
|
548 |
{
|
549 |
int first_char, count;
|
550 |
|
551 |
if (do_debug) print_internals(re, outfile);
|
552 |
|
553 |
count = pcre_info(re, &options, &first_char);
|
554 |
if (count < 0) fprintf(outfile,
|
555 |
"Error %d while reading info\n", count);
|
556 |
else
|
557 |
{
|
558 |
fprintf(outfile, "Identifying subpattern count = %d\n", count);
|
559 |
if (options == 0) fprintf(outfile, "No options\n");
|
560 |
else fprintf(outfile, "Options:%s%s%s%s%s%s%s%s\n",
|
561 |
((options & PCRE_ANCHORED) != 0)? " anchored" : "",
|
562 |
((options & PCRE_CASELESS) != 0)? " caseless" : "",
|
563 |
((options & PCRE_EXTENDED) != 0)? " extended" : "",
|
564 |
((options & PCRE_MULTILINE) != 0)? " multiline" : "",
|
565 |
((options & PCRE_DOTALL) != 0)? " dotall" : "",
|
566 |
((options & PCRE_DOLLAR_ENDONLY) != 0)? " dollar_endonly" : "",
|
567 |
((options & PCRE_EXTRA) != 0)? " extra" : "",
|
568 |
((options & PCRE_UNGREEDY) != 0)? " ungreedy" : "");
|
569 |
if (first_char == -1)
|
570 |
{
|
571 |
fprintf(outfile, "First char at start or follows \\n\n");
|
572 |
}
|
573 |
else if (first_char < 0)
|
574 |
{
|
575 |
fprintf(outfile, "No first char\n");
|
576 |
}
|
577 |
else
|
578 |
{
|
579 |
if (isprint(first_char))
|
580 |
fprintf(outfile, "First char = \'%c\'\n", first_char);
|
581 |
else
|
582 |
fprintf(outfile, "First char = %d\n", first_char);
|
583 |
}
|
584 |
}
|
585 |
}
|
586 |
|
587 |
/* If /S was present, study the regexp to generate additional info to
|
588 |
help with the matching. */
|
589 |
|
590 |
if (do_study)
|
591 |
{
|
592 |
if (timeit)
|
593 |
{
|
594 |
register int i;
|
595 |
clock_t time_taken;
|
596 |
clock_t start_time = clock();
|
597 |
for (i = 0; i < LOOPREPEAT; i++)
|
598 |
extra = pcre_study(re, study_options, &error);
|
599 |
time_taken = clock() - start_time;
|
600 |
if (extra != NULL) free(extra);
|
601 |
fprintf(outfile, " Study time %.3f milliseconds\n",
|
602 |
((double)time_taken * 1000.0)/
|
603 |
((double)LOOPREPEAT * (double)CLOCKS_PER_SEC));
|
604 |
}
|
605 |
|
606 |
extra = pcre_study(re, study_options, &error);
|
607 |
if (error != NULL)
|
608 |
fprintf(outfile, "Failed to study: %s\n", error);
|
609 |
else if (extra == NULL)
|
610 |
fprintf(outfile, "Study returned NULL\n");
|
611 |
|
612 |
/* This looks at internal information. A bit kludgy to do it this
|
613 |
way, but it is useful for testing. */
|
614 |
|
615 |
else if (do_showinfo)
|
616 |
{
|
617 |
real_pcre_extra *xx = (real_pcre_extra *)extra;
|
618 |
if ((xx->options & PCRE_STUDY_MAPPED) == 0)
|
619 |
fprintf(outfile, "No starting character set\n");
|
620 |
else
|
621 |
{
|
622 |
int i;
|
623 |
int c = 24;
|
624 |
fprintf(outfile, "Starting character set: ");
|
625 |
for (i = 0; i < 256; i++)
|
626 |
{
|
627 |
if ((xx->start_bits[i/8] & (1<<(i%8))) != 0)
|
628 |
{
|
629 |
if (c > 75)
|
630 |
{
|
631 |
fprintf(outfile, "\n ");
|
632 |
c = 2;
|
633 |
}
|
634 |
if (isprint(i) && i != ' ')
|
635 |
{
|
636 |
fprintf(outfile, "%c ", i);
|
637 |
c += 2;
|
638 |
}
|
639 |
else
|
640 |
{
|
641 |
fprintf(outfile, "\\x%02x ", i);
|
642 |
c += 5;
|
643 |
}
|
644 |
}
|
645 |
}
|
646 |
fprintf(outfile, "\n");
|
647 |
}
|
648 |
}
|
649 |
}
|
650 |
}
|
651 |
|
652 |
/* Read data lines and test them */
|
653 |
|
654 |
for (;;)
|
655 |
{
|
656 |
unsigned char *q;
|
657 |
int count, c;
|
658 |
int copystrings = 0;
|
659 |
int getstrings = 0;
|
660 |
int getlist = 0;
|
661 |
int offsets[45];
|
662 |
int size_offsets = sizeof(offsets)/sizeof(int);
|
663 |
|
664 |
options = 0;
|
665 |
|
666 |
if (infile == stdin) printf(" data> ");
|
667 |
if (fgets((char *)buffer, sizeof(buffer), infile) == NULL)
|
668 |
{
|
669 |
done = 1;
|
670 |
goto CONTINUE;
|
671 |
}
|
672 |
if (infile != stdin) fprintf(outfile, "%s", (char *)buffer);
|
673 |
|
674 |
len = (int)strlen((char *)buffer);
|
675 |
while (len > 0 && isspace(buffer[len-1])) len--;
|
676 |
buffer[len] = 0;
|
677 |
if (len == 0) break;
|
678 |
|
679 |
p = buffer;
|
680 |
while (isspace(*p)) p++;
|
681 |
|
682 |
q = dbuffer;
|
683 |
while ((c = *p++) != 0)
|
684 |
{
|
685 |
int i = 0;
|
686 |
int n = 0;
|
687 |
if (c == '\\') switch ((c = *p++))
|
688 |
{
|
689 |
case 'a': c = 7; break;
|
690 |
case 'b': c = '\b'; break;
|
691 |
case 'e': c = 27; break;
|
692 |
case 'f': c = '\f'; break;
|
693 |
case 'n': c = '\n'; break;
|
694 |
case 'r': c = '\r'; break;
|
695 |
case 't': c = '\t'; break;
|
696 |
case 'v': c = '\v'; break;
|
697 |
|
698 |
case '0': case '1': case '2': case '3':
|
699 |
case '4': case '5': case '6': case '7':
|
700 |
c -= '0';
|
701 |
while (i++ < 2 && isdigit(*p) && *p != '8' && *p != '9')
|
702 |
c = c * 8 + *p++ - '0';
|
703 |
break;
|
704 |
|
705 |
case 'x':
|
706 |
c = 0;
|
707 |
while (i++ < 2 && isxdigit(*p))
|
708 |
{
|
709 |
c = c * 16 + tolower(*p) - ((isdigit(*p))? '0' : 'W');
|
710 |
p++;
|
711 |
}
|
712 |
break;
|
713 |
|
714 |
case 0: /* Allows for an empty line */
|
715 |
p--;
|
716 |
continue;
|
717 |
|
718 |
case 'A': /* Option setting */
|
719 |
options |= PCRE_ANCHORED;
|
720 |
continue;
|
721 |
|
722 |
case 'B':
|
723 |
options |= PCRE_NOTBOL;
|
724 |
continue;
|
725 |
|
726 |
case 'C':
|
727 |
while(isdigit(*p)) n = n * 10 + *p++ - '0';
|
728 |
copystrings |= 1 << n;
|
729 |
continue;
|
730 |
|
731 |
case 'G':
|
732 |
while(isdigit(*p)) n = n * 10 + *p++ - '0';
|
733 |
getstrings |= 1 << n;
|
734 |
continue;
|
735 |
|
736 |
case 'L':
|
737 |
getlist = 1;
|
738 |
continue;
|
739 |
|
740 |
case 'O':
|
741 |
while(isdigit(*p)) n = n * 10 + *p++ - '0';
|
742 |
if (n <= (int)(sizeof(offsets)/sizeof(int))) size_offsets = n;
|
743 |
continue;
|
744 |
|
745 |
case 'Z':
|
746 |
options |= PCRE_NOTEOL;
|
747 |
continue;
|
748 |
}
|
749 |
*q++ = c;
|
750 |
}
|
751 |
*q = 0;
|
752 |
len = q - dbuffer;
|
753 |
|
754 |
/* Handle matching via the POSIX interface, which does not
|
755 |
support timing. */
|
756 |
|
757 |
if (posix || do_posix)
|
758 |
{
|
759 |
int rc;
|
760 |
int eflags = 0;
|
761 |
regmatch_t pmatch[30];
|
762 |
if ((options & PCRE_NOTBOL) != 0) eflags |= REG_NOTBOL;
|
763 |
if ((options & PCRE_NOTEOL) != 0) eflags |= REG_NOTEOL;
|
764 |
|
765 |
rc = regexec(&preg, (char *)dbuffer, sizeof(pmatch)/sizeof(regmatch_t),
|
766 |
pmatch, eflags);
|
767 |
|
768 |
if (rc != 0)
|
769 |
{
|
770 |
(void)regerror(rc, &preg, (char *)buffer, sizeof(buffer));
|
771 |
fprintf(outfile, "No match: POSIX code %d: %s\n", rc, buffer);
|
772 |
}
|
773 |
else
|
774 |
{
|
775 |
size_t i;
|
776 |
for (i = 0; i < sizeof(pmatch)/sizeof(regmatch_t); i++)
|
777 |
{
|
778 |
if (pmatch[i].rm_so >= 0)
|
779 |
{
|
780 |
fprintf(outfile, "%2d: ", (int)i);
|
781 |
pchars(dbuffer + pmatch[i].rm_so,
|
782 |
pmatch[i].rm_eo - pmatch[i].rm_so);
|
783 |
fprintf(outfile, "\n");
|
784 |
}
|
785 |
}
|
786 |
}
|
787 |
}
|
788 |
|
789 |
/* Handle matching via the native interface */
|
790 |
|
791 |
else
|
792 |
{
|
793 |
if (timeit)
|
794 |
{
|
795 |
register int i;
|
796 |
clock_t time_taken;
|
797 |
clock_t start_time = clock();
|
798 |
for (i = 0; i < LOOPREPEAT; i++)
|
799 |
count = pcre_exec(re, extra, (char *)dbuffer, len, options, offsets,
|
800 |
size_offsets);
|
801 |
time_taken = clock() - start_time;
|
802 |
fprintf(outfile, "Execute time %.3f milliseconds\n",
|
803 |
((double)time_taken * 1000.0)/
|
804 |
((double)LOOPREPEAT * (double)CLOCKS_PER_SEC));
|
805 |
}
|
806 |
|
807 |
count = pcre_exec(re, extra, (char *)dbuffer, len, options, offsets,
|
808 |
size_offsets);
|
809 |
|
810 |
if (count == 0)
|
811 |
{
|
812 |
fprintf(outfile, "Matched, but too many substrings\n");
|
813 |
count = size_offsets/3;
|
814 |
}
|
815 |
|
816 |
if (count >= 0)
|
817 |
{
|
818 |
int i;
|
819 |
for (i = 0; i < count * 2; i += 2)
|
820 |
{
|
821 |
if (offsets[i] < 0)
|
822 |
fprintf(outfile, "%2d: <unset>\n", i/2);
|
823 |
else
|
824 |
{
|
825 |
fprintf(outfile, "%2d: ", i/2);
|
826 |
pchars(dbuffer + offsets[i], offsets[i+1] - offsets[i]);
|
827 |
fprintf(outfile, "\n");
|
828 |
}
|
829 |
}
|
830 |
|
831 |
for (i = 0; i < 32; i++)
|
832 |
{
|
833 |
if ((copystrings & (1 << i)) != 0)
|
834 |
{
|
835 |
char buffer[16];
|
836 |
int rc = pcre_copy_substring((char *)dbuffer, offsets, count,
|
837 |
i, buffer, sizeof(buffer));
|
838 |
if (rc < 0)
|
839 |
fprintf(outfile, "copy substring %d failed %d\n", i, rc);
|
840 |
else
|
841 |
fprintf(outfile, "%2dC %s (%d)\n", i, buffer, rc);
|
842 |
}
|
843 |
}
|
844 |
|
845 |
for (i = 0; i < 32; i++)
|
846 |
{
|
847 |
if ((getstrings & (1 << i)) != 0)
|
848 |
{
|
849 |
const char *substring;
|
850 |
int rc = pcre_get_substring((char *)dbuffer, offsets, count,
|
851 |
i, &substring);
|
852 |
if (rc < 0)
|
853 |
fprintf(outfile, "get substring %d failed %d\n", i, rc);
|
854 |
else
|
855 |
{
|
856 |
fprintf(outfile, "%2dG %s (%d)\n", i, substring, rc);
|
857 |
free((void *)substring);
|
858 |
}
|
859 |
}
|
860 |
}
|
861 |
|
862 |
if (getlist)
|
863 |
{
|
864 |
const char **stringlist;
|
865 |
int rc = pcre_get_substring_list((char *)dbuffer, offsets, count,
|
866 |
&stringlist);
|
867 |
if (rc < 0)
|
868 |
fprintf(outfile, "get substring list failed %d\n", rc);
|
869 |
else
|
870 |
{
|
871 |
for (i = 0; i < count; i++)
|
872 |
fprintf(outfile, "%2dL %s\n", i, stringlist[i]);
|
873 |
if (stringlist[i] != NULL)
|
874 |
fprintf(outfile, "string list not terminated by NULL\n");
|
875 |
free((void *)stringlist);
|
876 |
}
|
877 |
}
|
878 |
|
879 |
}
|
880 |
else
|
881 |
{
|
882 |
if (count == -1) fprintf(outfile, "No match\n");
|
883 |
else fprintf(outfile, "Error %d\n", count);
|
884 |
}
|
885 |
}
|
886 |
}
|
887 |
|
888 |
CONTINUE:
|
889 |
if (posix || do_posix) regfree(&preg);
|
890 |
if (re != NULL) free(re);
|
891 |
if (extra != NULL) free(extra);
|
892 |
if (tables != NULL)
|
893 |
{
|
894 |
free((void *)tables);
|
895 |
setlocale(LC_CTYPE, "C");
|
896 |
}
|
897 |
}
|
898 |
|
899 |
fprintf(outfile, "\n");
|
900 |
return 0;
|
901 |
}
|
902 |
|
903 |
/* End */
|