12 |
/* Use the internal info for displaying the results of pcre_study(). */ |
/* Use the internal info for displaying the results of pcre_study(). */ |
13 |
|
|
14 |
#include "internal.h" |
#include "internal.h" |
15 |
|
|
16 |
|
/* It is possible to compile this test program without including support for |
17 |
|
testing the POSIX interface, though this is not available via the standard |
18 |
|
Makefile. */ |
19 |
|
|
20 |
|
#if !defined NOPOSIX |
21 |
#include "pcreposix.h" |
#include "pcreposix.h" |
22 |
|
#endif |
23 |
|
|
24 |
#ifndef CLOCKS_PER_SEC |
#ifndef CLOCKS_PER_SEC |
25 |
#ifdef CLK_TCK |
#ifdef CLK_TCK |
55 |
}; |
}; |
56 |
|
|
57 |
|
|
58 |
static void print_internals(pcre *re, FILE *outfile) |
static void print_internals(pcre *re) |
59 |
{ |
{ |
60 |
unsigned char *code = ((real_pcre *)re)->code; |
unsigned char *code = ((real_pcre *)re)->code; |
61 |
|
|
282 |
static void *new_malloc(size_t size) |
static void *new_malloc(size_t size) |
283 |
{ |
{ |
284 |
if (log_store) |
if (log_store) |
285 |
fprintf(outfile, "Memory allocation request: %d (code space %d)\n", |
fprintf(outfile, "Memory allocation (code space): %d\n", |
286 |
(int)size, (int)size - offsetof(real_pcre, code[0])); |
(int)((int)size - offsetof(real_pcre, code[0]))); |
287 |
return malloc(size); |
return malloc(size); |
288 |
} |
} |
289 |
|
|
373 |
{ |
{ |
374 |
pcre *re = NULL; |
pcre *re = NULL; |
375 |
pcre_extra *extra = NULL; |
pcre_extra *extra = NULL; |
376 |
|
|
377 |
|
#if !defined NOPOSIX /* There are still compilers that require no indent */ |
378 |
regex_t preg; |
regex_t preg; |
379 |
|
#endif |
380 |
|
|
381 |
const char *error; |
const char *error; |
382 |
unsigned char *p, *pp, *ppp; |
unsigned char *p, *pp, *ppp; |
383 |
unsigned const char *tables = NULL; |
unsigned const char *tables = NULL; |
384 |
int do_study = 0; |
int do_study = 0; |
385 |
int do_debug = debug; |
int do_debug = debug; |
386 |
|
int do_G = 0; |
387 |
|
int do_g = 0; |
388 |
int do_showinfo = showinfo; |
int do_showinfo = showinfo; |
389 |
|
int do_showrest = 0; |
390 |
int do_posix = 0; |
int do_posix = 0; |
391 |
int erroroffset, len, delimiter; |
int erroroffset, len, delimiter; |
392 |
|
|
458 |
{ |
{ |
459 |
switch (*pp++) |
switch (*pp++) |
460 |
{ |
{ |
461 |
|
case 'g': do_g = 1; break; |
462 |
case 'i': options |= PCRE_CASELESS; break; |
case 'i': options |= PCRE_CASELESS; break; |
463 |
case 'm': options |= PCRE_MULTILINE; break; |
case 'm': options |= PCRE_MULTILINE; break; |
464 |
case 's': options |= PCRE_DOTALL; break; |
case 's': options |= PCRE_DOTALL; break; |
465 |
case 'x': options |= PCRE_EXTENDED; break; |
case 'x': options |= PCRE_EXTENDED; break; |
466 |
|
|
467 |
|
case '+': do_showrest = 1; break; |
468 |
case 'A': options |= PCRE_ANCHORED; break; |
case 'A': options |= PCRE_ANCHORED; break; |
469 |
case 'D': do_debug = do_showinfo = 1; break; |
case 'D': do_debug = do_showinfo = 1; break; |
470 |
case 'E': options |= PCRE_DOLLAR_ENDONLY; break; |
case 'E': options |= PCRE_DOLLAR_ENDONLY; break; |
471 |
|
case 'G': do_G = 1; break; |
472 |
case 'I': do_showinfo = 1; break; |
case 'I': do_showinfo = 1; break; |
473 |
case 'M': log_store = 1; break; |
case 'M': log_store = 1; break; |
474 |
|
|
475 |
|
#if !defined NOPOSIX |
476 |
case 'P': do_posix = 1; break; |
case 'P': do_posix = 1; break; |
477 |
|
#endif |
478 |
|
|
479 |
case 'S': do_study = 1; break; |
case 'S': do_study = 1; break; |
480 |
case 'U': options |= PCRE_UNGREEDY; break; |
case 'U': options |= PCRE_UNGREEDY; break; |
481 |
case 'X': options |= PCRE_EXTRA; break; |
case 'X': options |= PCRE_EXTRA; break; |
504 |
timing, showing, or debugging options, nor the ability to pass over |
timing, showing, or debugging options, nor the ability to pass over |
505 |
local character tables. */ |
local character tables. */ |
506 |
|
|
507 |
|
#if !defined NOPOSIX |
508 |
if (posix || do_posix) |
if (posix || do_posix) |
509 |
{ |
{ |
510 |
int rc; |
int rc; |
527 |
/* Handle compiling via the native interface */ |
/* Handle compiling via the native interface */ |
528 |
|
|
529 |
else |
else |
530 |
|
#endif /* !defined NOPOSIX */ |
531 |
|
|
532 |
{ |
{ |
533 |
if (timeit) |
if (timeit) |
534 |
{ |
{ |
579 |
{ |
{ |
580 |
int first_char, count; |
int first_char, count; |
581 |
|
|
582 |
if (do_debug) print_internals(re, outfile); |
if (do_debug) print_internals(re); |
583 |
|
|
584 |
count = pcre_info(re, &options, &first_char); |
count = pcre_info(re, &options, &first_char); |
585 |
if (count < 0) fprintf(outfile, |
if (count < 0) fprintf(outfile, |
597 |
((options & PCRE_DOLLAR_ENDONLY) != 0)? " dollar_endonly" : "", |
((options & PCRE_DOLLAR_ENDONLY) != 0)? " dollar_endonly" : "", |
598 |
((options & PCRE_EXTRA) != 0)? " extra" : "", |
((options & PCRE_EXTRA) != 0)? " extra" : "", |
599 |
((options & PCRE_UNGREEDY) != 0)? " ungreedy" : ""); |
((options & PCRE_UNGREEDY) != 0)? " ungreedy" : ""); |
600 |
|
|
601 |
|
if (((((real_pcre *)re)->options) & PCRE_ICHANGED) != 0) |
602 |
|
fprintf(outfile, "Case state changes\n"); |
603 |
|
|
604 |
if (first_char == -1) |
if (first_char == -1) |
605 |
{ |
{ |
606 |
fprintf(outfile, "First char at start or follows \\n\n"); |
fprintf(outfile, "First char at start or follows \\n\n"); |
616 |
else |
else |
617 |
fprintf(outfile, "First char = %d\n", first_char); |
fprintf(outfile, "First char = %d\n", first_char); |
618 |
} |
} |
619 |
|
|
620 |
|
if (((((real_pcre *)re)->options) & PCRE_REQCHSET) != 0) |
621 |
|
{ |
622 |
|
int req_char = ((real_pcre *)re)->req_char; |
623 |
|
if (isprint(req_char)) |
624 |
|
fprintf(outfile, "Req char = \'%c\'\n", req_char); |
625 |
|
else |
626 |
|
fprintf(outfile, "Req char = %d\n", req_char); |
627 |
|
} |
628 |
|
else fprintf(outfile, "No req char\n"); |
629 |
} |
} |
630 |
} |
} |
631 |
|
|
699 |
for (;;) |
for (;;) |
700 |
{ |
{ |
701 |
unsigned char *q; |
unsigned char *q; |
702 |
|
unsigned char *bptr = dbuffer; |
703 |
int count, c; |
int count, c; |
704 |
int copystrings = 0; |
int copystrings = 0; |
705 |
int getstrings = 0; |
int getstrings = 0; |
706 |
int getlist = 0; |
int getlist = 0; |
707 |
|
int start_offset = 0; |
708 |
int offsets[45]; |
int offsets[45]; |
709 |
int size_offsets = sizeof(offsets)/sizeof(int); |
int size_offsets = sizeof(offsets)/sizeof(int); |
710 |
|
|
711 |
options = 0; |
options = 0; |
712 |
|
|
713 |
if (infile == stdin) printf(" data> "); |
if (infile == stdin) printf("data> "); |
714 |
if (fgets((char *)buffer, sizeof(buffer), infile) == NULL) |
if (fgets((char *)buffer, sizeof(buffer), infile) == NULL) |
715 |
{ |
{ |
716 |
done = 1; |
done = 1; |
784 |
getlist = 1; |
getlist = 1; |
785 |
continue; |
continue; |
786 |
|
|
787 |
|
case 'N': |
788 |
|
options |= PCRE_NOTEMPTY; |
789 |
|
continue; |
790 |
|
|
791 |
case 'O': |
case 'O': |
792 |
while(isdigit(*p)) n = n * 10 + *p++ - '0'; |
while(isdigit(*p)) n = n * 10 + *p++ - '0'; |
793 |
if (n <= (int)(sizeof(offsets)/sizeof(int))) size_offsets = n; |
if (n <= (int)(sizeof(offsets)/sizeof(int))) size_offsets = n; |
805 |
/* Handle matching via the POSIX interface, which does not |
/* Handle matching via the POSIX interface, which does not |
806 |
support timing. */ |
support timing. */ |
807 |
|
|
808 |
|
#if !defined NOPOSIX |
809 |
if (posix || do_posix) |
if (posix || do_posix) |
810 |
{ |
{ |
811 |
int rc; |
int rc; |
814 |
if ((options & PCRE_NOTBOL) != 0) eflags |= REG_NOTBOL; |
if ((options & PCRE_NOTBOL) != 0) eflags |= REG_NOTBOL; |
815 |
if ((options & PCRE_NOTEOL) != 0) eflags |= REG_NOTEOL; |
if ((options & PCRE_NOTEOL) != 0) eflags |= REG_NOTEOL; |
816 |
|
|
817 |
rc = regexec(&preg, (char *)dbuffer, sizeof(pmatch)/sizeof(regmatch_t), |
rc = regexec(&preg, (const char *)bptr, |
818 |
pmatch, eflags); |
sizeof(pmatch)/sizeof(regmatch_t), pmatch, eflags); |
819 |
|
|
820 |
if (rc != 0) |
if (rc != 0) |
821 |
{ |
{ |
833 |
pchars(dbuffer + pmatch[i].rm_so, |
pchars(dbuffer + pmatch[i].rm_so, |
834 |
pmatch[i].rm_eo - pmatch[i].rm_so); |
pmatch[i].rm_eo - pmatch[i].rm_so); |
835 |
fprintf(outfile, "\n"); |
fprintf(outfile, "\n"); |
836 |
|
if (i == 0 && do_showrest) |
837 |
|
{ |
838 |
|
fprintf(outfile, " 0+ "); |
839 |
|
pchars(dbuffer + pmatch[i].rm_eo, len - pmatch[i].rm_eo); |
840 |
|
fprintf(outfile, "\n"); |
841 |
|
} |
842 |
} |
} |
843 |
} |
} |
844 |
} |
} |
845 |
} |
} |
846 |
|
|
847 |
/* Handle matching via the native interface */ |
/* Handle matching via the native interface - repeats for /g and /G */ |
848 |
|
|
849 |
else |
else |
850 |
|
#endif /* !defined NOPOSIX */ |
851 |
|
|
852 |
|
for (;;) |
853 |
{ |
{ |
854 |
if (timeit) |
if (timeit) |
855 |
{ |
{ |
857 |
clock_t time_taken; |
clock_t time_taken; |
858 |
clock_t start_time = clock(); |
clock_t start_time = clock(); |
859 |
for (i = 0; i < LOOPREPEAT; i++) |
for (i = 0; i < LOOPREPEAT; i++) |
860 |
count = pcre_exec(re, extra, (char *)dbuffer, len, options, offsets, |
count = pcre_exec(re, extra, (char *)bptr, len, |
861 |
size_offsets); |
(do_g? start_offset : 0), options, offsets, size_offsets); |
862 |
time_taken = clock() - start_time; |
time_taken = clock() - start_time; |
863 |
fprintf(outfile, "Execute time %.3f milliseconds\n", |
fprintf(outfile, "Execute time %.3f milliseconds\n", |
864 |
((double)time_taken * 1000.0)/ |
((double)time_taken * 1000.0)/ |
865 |
((double)LOOPREPEAT * (double)CLOCKS_PER_SEC)); |
((double)LOOPREPEAT * (double)CLOCKS_PER_SEC)); |
866 |
} |
} |
867 |
|
|
868 |
count = pcre_exec(re, extra, (char *)dbuffer, len, options, offsets, |
count = pcre_exec(re, extra, (char *)bptr, len, |
869 |
size_offsets); |
(do_g? start_offset : 0), options, offsets, size_offsets); |
870 |
|
|
871 |
if (count == 0) |
if (count == 0) |
872 |
{ |
{ |
884 |
else |
else |
885 |
{ |
{ |
886 |
fprintf(outfile, "%2d: ", i/2); |
fprintf(outfile, "%2d: ", i/2); |
887 |
pchars(dbuffer + offsets[i], offsets[i+1] - offsets[i]); |
pchars(bptr + offsets[i], offsets[i+1] - offsets[i]); |
888 |
fprintf(outfile, "\n"); |
fprintf(outfile, "\n"); |
889 |
|
if (i == 0) |
890 |
|
{ |
891 |
|
start_offset = offsets[1]; |
892 |
|
if (do_showrest) |
893 |
|
{ |
894 |
|
fprintf(outfile, " 0+ "); |
895 |
|
pchars(bptr + offsets[i+1], len - offsets[i+1]); |
896 |
|
fprintf(outfile, "\n"); |
897 |
|
} |
898 |
|
} |
899 |
} |
} |
900 |
} |
} |
901 |
|
|
903 |
{ |
{ |
904 |
if ((copystrings & (1 << i)) != 0) |
if ((copystrings & (1 << i)) != 0) |
905 |
{ |
{ |
906 |
char buffer[16]; |
char copybuffer[16]; |
907 |
int rc = pcre_copy_substring((char *)dbuffer, offsets, count, |
int rc = pcre_copy_substring((char *)bptr, offsets, count, |
908 |
i, buffer, sizeof(buffer)); |
i, copybuffer, sizeof(copybuffer)); |
909 |
if (rc < 0) |
if (rc < 0) |
910 |
fprintf(outfile, "copy substring %d failed %d\n", i, rc); |
fprintf(outfile, "copy substring %d failed %d\n", i, rc); |
911 |
else |
else |
912 |
fprintf(outfile, "%2dC %s (%d)\n", i, buffer, rc); |
fprintf(outfile, "%2dC %s (%d)\n", i, copybuffer, rc); |
913 |
} |
} |
914 |
} |
} |
915 |
|
|
918 |
if ((getstrings & (1 << i)) != 0) |
if ((getstrings & (1 << i)) != 0) |
919 |
{ |
{ |
920 |
const char *substring; |
const char *substring; |
921 |
int rc = pcre_get_substring((char *)dbuffer, offsets, count, |
int rc = pcre_get_substring((char *)bptr, offsets, count, |
922 |
i, &substring); |
i, &substring); |
923 |
if (rc < 0) |
if (rc < 0) |
924 |
fprintf(outfile, "get substring %d failed %d\n", i, rc); |
fprintf(outfile, "get substring %d failed %d\n", i, rc); |
933 |
if (getlist) |
if (getlist) |
934 |
{ |
{ |
935 |
const char **stringlist; |
const char **stringlist; |
936 |
int rc = pcre_get_substring_list((char *)dbuffer, offsets, count, |
int rc = pcre_get_substring_list((char *)bptr, offsets, count, |
937 |
&stringlist); |
&stringlist); |
938 |
if (rc < 0) |
if (rc < 0) |
939 |
fprintf(outfile, "get substring list failed %d\n", rc); |
fprintf(outfile, "get substring list failed %d\n", rc); |
950 |
} |
} |
951 |
else |
else |
952 |
{ |
{ |
953 |
if (count == -1) fprintf(outfile, "No match\n"); |
if (start_offset == 0) |
954 |
else fprintf(outfile, "Error %d\n", count); |
{ |
955 |
|
if (count == -1) fprintf(outfile, "No match\n"); |
956 |
|
else fprintf(outfile, "Error %d\n", count); |
957 |
|
} |
958 |
|
start_offset = -1; |
959 |
|
} |
960 |
|
|
961 |
|
if ((!do_g && !do_G) || start_offset <= 0) break; |
962 |
|
if (do_G) |
963 |
|
{ |
964 |
|
bptr += start_offset; |
965 |
|
len -= start_offset; |
966 |
} |
} |
967 |
} |
} |
968 |
} |
} |
969 |
|
|
970 |
CONTINUE: |
CONTINUE: |
971 |
|
|
972 |
|
#if !defined NOPOSIX |
973 |
if (posix || do_posix) regfree(&preg); |
if (posix || do_posix) regfree(&preg); |
974 |
|
#endif |
975 |
|
|
976 |
if (re != NULL) free(re); |
if (re != NULL) free(re); |
977 |
if (extra != NULL) free(extra); |
if (extra != NULL) free(extra); |
978 |
if (tables != NULL) |
if (tables != NULL) |