| 44 |
#include <locale.h> |
#include <locale.h> |
| 45 |
#include <errno.h> |
#include <errno.h> |
| 46 |
|
|
| 47 |
#ifndef _WIN32 |
|
| 48 |
#include <sys/resource.h> |
/* A number of things vary for Windows builds. Originally, pcretest opened its |
| 49 |
|
input and output without "b"; then I was told that "b" was needed in some |
| 50 |
|
environments, so it was added for release 5.0 to both the input and output. (It |
| 51 |
|
makes no difference on Unix-like systems.) Later I was told that it is wrong |
| 52 |
|
for the input on Windows. I've now abstracted the modes into two macros that |
| 53 |
|
are set here, to make it easier to fiddle with them, and removed "b" from the |
| 54 |
|
input mode under Windows. */ |
| 55 |
|
|
| 56 |
|
#if defined(_WIN32) || defined(WIN32) |
| 57 |
|
#include <io.h> /* For _setmode() */ |
| 58 |
|
#include <fcntl.h> /* For _O_BINARY */ |
| 59 |
|
#define INPUT_MODE "r" |
| 60 |
|
#define OUTPUT_MODE "wb" |
| 61 |
|
|
| 62 |
|
#else |
| 63 |
|
#include <sys/time.h> /* These two includes are needed */ |
| 64 |
|
#include <sys/resource.h> /* for setrlimit(). */ |
| 65 |
|
#define INPUT_MODE "rb" |
| 66 |
|
#define OUTPUT_MODE "wb" |
| 67 |
#endif |
#endif |
| 68 |
|
|
| 69 |
|
|
| 70 |
#define PCRE_SPY /* For Win32 build, import data, not export */ |
#define PCRE_SPY /* For Win32 build, import data, not export */ |
| 71 |
|
|
| 72 |
/* We include pcre_internal.h because we need the internal info for displaying |
/* We include pcre_internal.h because we need the internal info for displaying |
| 93 |
|
|
| 94 |
/* We also need the pcre_printint() function for printing out compiled |
/* We also need the pcre_printint() function for printing out compiled |
| 95 |
patterns. This function is in a separate file so that it can be included in |
patterns. This function is in a separate file so that it can be included in |
| 96 |
pcre_compile.c when that module is compiled with debugging enabled. */ |
pcre_compile.c when that module is compiled with debugging enabled. |
| 97 |
|
|
| 98 |
|
The definition of the macro PRINTABLE, which determines whether to print an |
| 99 |
|
output character as-is or as a hex value when showing compiled patterns, is |
| 100 |
|
contained in this file. We uses it here also, in cases when the locale has not |
| 101 |
|
been explicitly changed, so as to get consistent output from systems that |
| 102 |
|
differ in their output from isprint() even in the "C" locale. */ |
| 103 |
|
|
| 104 |
#include "pcre_printint.src" |
#include "pcre_printint.src" |
| 105 |
|
|
| 106 |
|
#define PRINTHEX(c) (locale_set? isprint(c) : PRINTABLE(c)) |
| 107 |
|
|
| 108 |
|
|
| 109 |
/* It is possible to compile this test program without including support for |
/* It is possible to compile this test program without including support for |
| 110 |
testing the POSIX interface, though this is not available via the standard |
testing the POSIX interface, though this is not available via the standard |
| 130 |
#endif |
#endif |
| 131 |
#endif |
#endif |
| 132 |
|
|
| 133 |
|
/* This is the default loop count for timing. */ |
| 134 |
|
|
| 135 |
#define LOOPREPEAT 500000 |
#define LOOPREPEAT 500000 |
| 136 |
|
|
| 137 |
/* Static variables */ |
/* Static variables */ |
| 143 |
static int callout_fail_count; |
static int callout_fail_count; |
| 144 |
static int callout_fail_id; |
static int callout_fail_id; |
| 145 |
static int first_callout; |
static int first_callout; |
| 146 |
|
static int locale_set = 0; |
| 147 |
static int show_malloc; |
static int show_malloc; |
| 148 |
static int use_utf8; |
static int use_utf8; |
| 149 |
static size_t gotten_store; |
static size_t gotten_store; |
| 187 |
for (;;) |
for (;;) |
| 188 |
{ |
{ |
| 189 |
int rlen = buffer_size - (here - buffer); |
int rlen = buffer_size - (here - buffer); |
| 190 |
|
|
| 191 |
if (rlen > 1000) |
if (rlen > 1000) |
| 192 |
{ |
{ |
| 193 |
int dlen; |
int dlen; |
| 244 |
|
|
| 245 |
/* We don't use strtoul() because SunOS4 doesn't have it. Rather than mess |
/* We don't use strtoul() because SunOS4 doesn't have it. Rather than mess |
| 246 |
around with conditional compilation, just do the job by hand. It is only used |
around with conditional compilation, just do the job by hand. It is only used |
| 247 |
for unpicking the -o argument, so just keep it simple. |
for unpicking arguments, so just keep it simple. |
| 248 |
|
|
| 249 |
Arguments: |
Arguments: |
| 250 |
str string to be converted |
str string to be converted |
| 342 |
Returns: number of characters placed in the buffer |
Returns: number of characters placed in the buffer |
| 343 |
*/ |
*/ |
| 344 |
|
|
| 345 |
|
#if !defined NOUTF8 |
| 346 |
|
|
| 347 |
static int |
static int |
| 348 |
ord2utf8(int cvalue, uschar *utf8bytes) |
ord2utf8(int cvalue, uschar *utf8bytes) |
| 349 |
{ |
{ |
| 360 |
return i + 1; |
return i + 1; |
| 361 |
} |
} |
| 362 |
|
|
| 363 |
|
#endif |
| 364 |
|
|
| 365 |
|
|
| 366 |
|
|
| 367 |
/************************************************* |
/************************************************* |
| 388 |
{ |
{ |
| 389 |
length -= rc - 1; |
length -= rc - 1; |
| 390 |
p += rc; |
p += rc; |
| 391 |
if (c < 256 && isprint(c)) |
if (PRINTHEX(c)) |
| 392 |
{ |
{ |
| 393 |
if (f != NULL) fprintf(f, "%c", c); |
if (f != NULL) fprintf(f, "%c", c); |
| 394 |
yield++; |
yield++; |
| 395 |
} |
} |
| 396 |
else |
else |
| 397 |
{ |
{ |
| 398 |
int n; |
int n = 4; |
| 399 |
if (f != NULL) fprintf(f, "\\x{%02x}%n", c, &n); |
if (f != NULL) fprintf(f, "\\x{%02x}", c); |
| 400 |
yield += n; |
yield += (n <= 0x000000ff)? 2 : |
| 401 |
|
(n <= 0x00000fff)? 3 : |
| 402 |
|
(n <= 0x0000ffff)? 4 : |
| 403 |
|
(n <= 0x000fffff)? 5 : 6; |
| 404 |
} |
} |
| 405 |
continue; |
continue; |
| 406 |
} |
} |
| 409 |
|
|
| 410 |
/* Not UTF-8, or malformed UTF-8 */ |
/* Not UTF-8, or malformed UTF-8 */ |
| 411 |
|
|
| 412 |
if (isprint(c = *(p++))) |
c = *p++; |
| 413 |
|
if (PRINTHEX(c)) |
| 414 |
{ |
{ |
| 415 |
if (f != NULL) fprintf(f, "%c", c); |
if (f != NULL) fprintf(f, "%c", c); |
| 416 |
yield++; |
yield++; |
| 653 |
*************************************************/ |
*************************************************/ |
| 654 |
|
|
| 655 |
/* This is used both at compile and run-time to check for <xxx> escapes, where |
/* This is used both at compile and run-time to check for <xxx> escapes, where |
| 656 |
xxx is LF, CR, or CRLF. Print a message and return 0 if there is no match. |
xxx is LF, CR, CRLF, or ANY. Print a message and return 0 if there is no match. |
| 657 |
|
|
| 658 |
Arguments: |
Arguments: |
| 659 |
p points after the leading '<' |
p points after the leading '<' |
| 668 |
if (strncmp((char *)p, "cr>", 3) == 0) return PCRE_NEWLINE_CR; |
if (strncmp((char *)p, "cr>", 3) == 0) return PCRE_NEWLINE_CR; |
| 669 |
if (strncmp((char *)p, "lf>", 3) == 0) return PCRE_NEWLINE_LF; |
if (strncmp((char *)p, "lf>", 3) == 0) return PCRE_NEWLINE_LF; |
| 670 |
if (strncmp((char *)p, "crlf>", 5) == 0) return PCRE_NEWLINE_CRLF; |
if (strncmp((char *)p, "crlf>", 5) == 0) return PCRE_NEWLINE_CRLF; |
| 671 |
|
if (strncmp((char *)p, "any>", 4) == 0) return PCRE_NEWLINE_ANY; |
| 672 |
fprintf(f, "Unknown newline type at: <%s\n", p); |
fprintf(f, "Unknown newline type at: <%s\n", p); |
| 673 |
return 0; |
return 0; |
| 674 |
} |
} |
| 676 |
|
|
| 677 |
|
|
| 678 |
/************************************************* |
/************************************************* |
| 679 |
|
* Usage function * |
| 680 |
|
*************************************************/ |
| 681 |
|
|
| 682 |
|
static void |
| 683 |
|
usage(void) |
| 684 |
|
{ |
| 685 |
|
printf("Usage: pcretest [options] [<input> [<output>]]\n"); |
| 686 |
|
printf(" -b show compiled code (bytecode)\n"); |
| 687 |
|
printf(" -C show PCRE compile-time options and exit\n"); |
| 688 |
|
printf(" -d debug: show compiled code and information (-b and -i)\n"); |
| 689 |
|
#if !defined NODFA |
| 690 |
|
printf(" -dfa force DFA matching for all subjects\n"); |
| 691 |
|
#endif |
| 692 |
|
printf(" -help show usage information\n"); |
| 693 |
|
printf(" -i show information about compiled patterns\n" |
| 694 |
|
" -m output memory used information\n" |
| 695 |
|
" -o <n> set size of offsets vector to <n>\n"); |
| 696 |
|
#if !defined NOPOSIX |
| 697 |
|
printf(" -p use POSIX interface\n"); |
| 698 |
|
#endif |
| 699 |
|
printf(" -q quiet: do not output PCRE version number at start\n"); |
| 700 |
|
printf(" -S <n> set stack size to <n> megabytes\n"); |
| 701 |
|
printf(" -s output store (memory) used information\n" |
| 702 |
|
" -t time compilation and execution\n"); |
| 703 |
|
printf(" -t <n> time compilation and execution, repeating <n> times\n"); |
| 704 |
|
printf(" -tm time execution (matching) only\n"); |
| 705 |
|
printf(" -tm <n> time execution (matching) only, repeating <n> times\n"); |
| 706 |
|
} |
| 707 |
|
|
| 708 |
|
|
| 709 |
|
|
| 710 |
|
/************************************************* |
| 711 |
* Main Program * |
* Main Program * |
| 712 |
*************************************************/ |
*************************************************/ |
| 713 |
|
|
| 722 |
int study_options = 0; |
int study_options = 0; |
| 723 |
int op = 1; |
int op = 1; |
| 724 |
int timeit = 0; |
int timeit = 0; |
| 725 |
|
int timeitm = 0; |
| 726 |
int showinfo = 0; |
int showinfo = 0; |
| 727 |
int showstore = 0; |
int showstore = 0; |
| 728 |
int quiet = 0; |
int quiet = 0; |
| 754 |
dbuffer = (unsigned char *)malloc(buffer_size); |
dbuffer = (unsigned char *)malloc(buffer_size); |
| 755 |
pbuffer = (unsigned char *)malloc(buffer_size); |
pbuffer = (unsigned char *)malloc(buffer_size); |
| 756 |
|
|
| 757 |
/* The outfile variable is static so that new_malloc can use it. The _setmode() |
/* The outfile variable is static so that new_malloc can use it. */ |
|
stuff is some magic that I don't understand, but which apparently does good |
|
|
things in Windows. It's related to line terminations. */ |
|
|
|
|
|
#if defined(_WIN32) || defined(WIN32) |
|
|
_setmode( _fileno( stdout ), 0x8000 ); |
|
|
#endif /* defined(_WIN32) || defined(WIN32) */ |
|
| 758 |
|
|
| 759 |
outfile = stdout; |
outfile = stdout; |
| 760 |
|
|
| 761 |
|
/* The following _setmode() stuff is some Windows magic that tells its runtime |
| 762 |
|
library to translate CRLF into a single LF character. At least, that's what |
| 763 |
|
I've been told: never having used Windows I take this all on trust. Originally |
| 764 |
|
it set 0x8000, but then I was advised that _O_BINARY was better. */ |
| 765 |
|
|
| 766 |
|
#if defined(_WIN32) || defined(WIN32) |
| 767 |
|
_setmode( _fileno( stdout ), _O_BINARY ); |
| 768 |
|
#endif |
| 769 |
|
|
| 770 |
/* Scan options */ |
/* Scan options */ |
| 771 |
|
|
| 772 |
while (argc > 1 && argv[op][0] == '-') |
while (argc > 1 && argv[op][0] == '-') |
| 775 |
|
|
| 776 |
if (strcmp(argv[op], "-s") == 0 || strcmp(argv[op], "-m") == 0) |
if (strcmp(argv[op], "-s") == 0 || strcmp(argv[op], "-m") == 0) |
| 777 |
showstore = 1; |
showstore = 1; |
|
else if (strcmp(argv[op], "-t") == 0) timeit = 1; |
|
| 778 |
else if (strcmp(argv[op], "-q") == 0) quiet = 1; |
else if (strcmp(argv[op], "-q") == 0) quiet = 1; |
| 779 |
|
else if (strcmp(argv[op], "-b") == 0) debug = 1; |
| 780 |
else if (strcmp(argv[op], "-i") == 0) showinfo = 1; |
else if (strcmp(argv[op], "-i") == 0) showinfo = 1; |
| 781 |
else if (strcmp(argv[op], "-d") == 0) showinfo = debug = 1; |
else if (strcmp(argv[op], "-d") == 0) showinfo = debug = 1; |
| 782 |
#if !defined NODFA |
#if !defined NODFA |
| 789 |
op++; |
op++; |
| 790 |
argc--; |
argc--; |
| 791 |
} |
} |
| 792 |
|
else if (strcmp(argv[op], "-t") == 0 || strcmp(argv[op], "-tm") == 0) |
| 793 |
|
{ |
| 794 |
|
int both = argv[op][2] == 0; |
| 795 |
|
int temp; |
| 796 |
|
if (argc > 2 && (temp = get_value((unsigned char *)argv[op+1], &endptr), |
| 797 |
|
*endptr == 0)) |
| 798 |
|
{ |
| 799 |
|
timeitm = temp; |
| 800 |
|
op++; |
| 801 |
|
argc--; |
| 802 |
|
} |
| 803 |
|
else timeitm = LOOPREPEAT; |
| 804 |
|
if (both) timeit = timeitm; |
| 805 |
|
} |
| 806 |
else if (strcmp(argv[op], "-S") == 0 && argc > 2 && |
else if (strcmp(argv[op], "-S") == 0 && argc > 2 && |
| 807 |
((stack_size = get_value((unsigned char *)argv[op+1], &endptr)), |
((stack_size = get_value((unsigned char *)argv[op+1], &endptr)), |
| 808 |
*endptr == 0)) |
*endptr == 0)) |
| 809 |
{ |
{ |
| 810 |
#ifdef _WIN32 |
#if defined(_WIN32) || defined(WIN32) |
| 811 |
printf("PCRE: -S not supported on this OS\n"); |
printf("PCRE: -S not supported on this OS\n"); |
| 812 |
exit(1); |
exit(1); |
| 813 |
#else |
#else |
| 839 |
printf(" %sUnicode properties support\n", rc? "" : "No "); |
printf(" %sUnicode properties support\n", rc? "" : "No "); |
| 840 |
(void)pcre_config(PCRE_CONFIG_NEWLINE, &rc); |
(void)pcre_config(PCRE_CONFIG_NEWLINE, &rc); |
| 841 |
printf(" Newline sequence is %s\n", (rc == '\r')? "CR" : |
printf(" Newline sequence is %s\n", (rc == '\r')? "CR" : |
| 842 |
(rc == '\n')? "LF" : "CRLF"); |
(rc == '\n')? "LF" : (rc == ('\r'<<8 | '\n'))? "CRLF" : |
| 843 |
|
(rc == -1)? "ANY" : "???"); |
| 844 |
(void)pcre_config(PCRE_CONFIG_LINK_SIZE, &rc); |
(void)pcre_config(PCRE_CONFIG_LINK_SIZE, &rc); |
| 845 |
printf(" Internal link size = %d\n", rc); |
printf(" Internal link size = %d\n", rc); |
| 846 |
(void)pcre_config(PCRE_CONFIG_POSIX_MALLOC_THRESHOLD, &rc); |
(void)pcre_config(PCRE_CONFIG_POSIX_MALLOC_THRESHOLD, &rc); |
| 853 |
printf(" Match recursion uses %s\n", rc? "stack" : "heap"); |
printf(" Match recursion uses %s\n", rc? "stack" : "heap"); |
| 854 |
exit(0); |
exit(0); |
| 855 |
} |
} |
| 856 |
|
else if (strcmp(argv[op], "-help") == 0 || |
| 857 |
|
strcmp(argv[op], "--help") == 0) |
| 858 |
|
{ |
| 859 |
|
usage(); |
| 860 |
|
goto EXIT; |
| 861 |
|
} |
| 862 |
else |
else |
| 863 |
{ |
{ |
| 864 |
printf("** Unknown or malformed option %s\n", argv[op]); |
printf("** Unknown or malformed option %s\n", argv[op]); |
| 865 |
printf("Usage: pcretest [options] [<input> [<output>]]\n"); |
usage(); |
|
printf(" -C show PCRE compile-time options and exit\n"); |
|
|
printf(" -d debug: show compiled code; implies -i\n"); |
|
|
#if !defined NODFA |
|
|
printf(" -dfa force DFA matching for all subjects\n"); |
|
|
#endif |
|
|
printf(" -i show information about compiled pattern\n" |
|
|
" -m output memory used information\n" |
|
|
" -o <n> set size of offsets vector to <n>\n"); |
|
|
#if !defined NOPOSIX |
|
|
printf(" -p use POSIX interface\n"); |
|
|
#endif |
|
|
printf(" -S <n> set stack size to <n> megabytes\n"); |
|
|
printf(" -s output store (memory) used information\n" |
|
|
" -t time compilation and execution\n"); |
|
| 866 |
yield = 1; |
yield = 1; |
| 867 |
goto EXIT; |
goto EXIT; |
| 868 |
} |
} |
| 886 |
|
|
| 887 |
if (argc > 1) |
if (argc > 1) |
| 888 |
{ |
{ |
| 889 |
infile = fopen(argv[op], "rb"); |
infile = fopen(argv[op], INPUT_MODE); |
| 890 |
if (infile == NULL) |
if (infile == NULL) |
| 891 |
{ |
{ |
| 892 |
printf("** Failed to open %s\n", argv[op]); |
printf("** Failed to open %s\n", argv[op]); |
| 897 |
|
|
| 898 |
if (argc > 2) |
if (argc > 2) |
| 899 |
{ |
{ |
| 900 |
outfile = fopen(argv[op+1], "wb"); |
outfile = fopen(argv[op+1], OUTPUT_MODE); |
| 901 |
if (outfile == NULL) |
if (outfile == NULL) |
| 902 |
{ |
{ |
| 903 |
printf("** Failed to open %s\n", argv[op+1]); |
printf("** Failed to open %s\n", argv[op+1]); |
| 942 |
int do_showinfo = showinfo; |
int do_showinfo = showinfo; |
| 943 |
int do_showrest = 0; |
int do_showrest = 0; |
| 944 |
int do_flip = 0; |
int do_flip = 0; |
| 945 |
int erroroffset, len, delimiter; |
int erroroffset, len, delimiter, poffset; |
| 946 |
|
|
| 947 |
use_utf8 = 0; |
use_utf8 = 0; |
| 948 |
|
|
| 1052 |
} |
} |
| 1053 |
|
|
| 1054 |
pp = p; |
pp = p; |
| 1055 |
|
poffset = p - buffer; |
| 1056 |
|
|
| 1057 |
for(;;) |
for(;;) |
| 1058 |
{ |
{ |
| 1073 |
if (infile != stdin) fprintf(outfile, "%s", (char *)pp); |
if (infile != stdin) fprintf(outfile, "%s", (char *)pp); |
| 1074 |
} |
} |
| 1075 |
|
|
| 1076 |
|
/* The buffer may have moved while being extended; reset the start of data |
| 1077 |
|
pointer to the correct relative point in the buffer. */ |
| 1078 |
|
|
| 1079 |
|
p = buffer + poffset; |
| 1080 |
|
|
| 1081 |
/* If the first character after the delimiter is backslash, make |
/* If the first character after the delimiter is backslash, make |
| 1082 |
the pattern end with backslash. This is purely to provide a way |
the pattern end with backslash. This is purely to provide a way |
| 1083 |
of testing for the error message when a pattern ends with backslash. */ |
of testing for the error message when a pattern ends with backslash. */ |
| 1109 |
|
|
| 1110 |
case '+': do_showrest = 1; break; |
case '+': do_showrest = 1; break; |
| 1111 |
case 'A': options |= PCRE_ANCHORED; break; |
case 'A': options |= PCRE_ANCHORED; break; |
| 1112 |
|
case 'B': do_debug = 1; break; |
| 1113 |
case 'C': options |= PCRE_AUTO_CALLOUT; break; |
case 'C': options |= PCRE_AUTO_CALLOUT; break; |
| 1114 |
case 'D': do_debug = do_showinfo = 1; break; |
case 'D': do_debug = do_showinfo = 1; break; |
| 1115 |
case 'E': options |= PCRE_DOLLAR_ENDONLY; break; |
case 'E': options |= PCRE_DOLLAR_ENDONLY; break; |
| 1132 |
|
|
| 1133 |
case 'L': |
case 'L': |
| 1134 |
ppp = pp; |
ppp = pp; |
| 1135 |
/* The '\r' test here is so that it works on Windows */ |
/* The '\r' test here is so that it works on Windows. */ |
| 1136 |
while (*ppp != '\n' && *ppp != '\r' && *ppp != ' ') ppp++; |
/* The '0' test is just in case this is an unterminated line. */ |
| 1137 |
|
while (*ppp != 0 && *ppp != '\n' && *ppp != '\r' && *ppp != ' ') ppp++; |
| 1138 |
*ppp = 0; |
*ppp = 0; |
| 1139 |
if (setlocale(LC_CTYPE, (const char *)pp) == NULL) |
if (setlocale(LC_CTYPE, (const char *)pp) == NULL) |
| 1140 |
{ |
{ |
| 1141 |
fprintf(outfile, "** Failed to set locale \"%s\"\n", pp); |
fprintf(outfile, "** Failed to set locale \"%s\"\n", pp); |
| 1142 |
goto SKIP_DATA; |
goto SKIP_DATA; |
| 1143 |
} |
} |
| 1144 |
|
locale_set = 1; |
| 1145 |
tables = pcre_maketables(); |
tables = pcre_maketables(); |
| 1146 |
pp = ppp; |
pp = ppp; |
| 1147 |
break; |
break; |
| 1208 |
#endif /* !defined NOPOSIX */ |
#endif /* !defined NOPOSIX */ |
| 1209 |
|
|
| 1210 |
{ |
{ |
| 1211 |
if (timeit) |
if (timeit > 0) |
| 1212 |
{ |
{ |
| 1213 |
register int i; |
register int i; |
| 1214 |
clock_t time_taken; |
clock_t time_taken; |
| 1215 |
clock_t start_time = clock(); |
clock_t start_time = clock(); |
| 1216 |
for (i = 0; i < LOOPREPEAT; i++) |
for (i = 0; i < timeit; i++) |
| 1217 |
{ |
{ |
| 1218 |
re = pcre_compile((char *)p, options, &error, &erroroffset, tables); |
re = pcre_compile((char *)p, options, &error, &erroroffset, tables); |
| 1219 |
if (re != NULL) free(re); |
if (re != NULL) free(re); |
| 1220 |
} |
} |
| 1221 |
time_taken = clock() - start_time; |
time_taken = clock() - start_time; |
| 1222 |
fprintf(outfile, "Compile time %.3f milliseconds\n", |
fprintf(outfile, "Compile time %.4f milliseconds\n", |
| 1223 |
(((double)time_taken * 1000.0) / (double)LOOPREPEAT) / |
(((double)time_taken * 1000.0) / (double)timeit) / |
| 1224 |
(double)CLOCKS_PER_SEC); |
(double)CLOCKS_PER_SEC); |
| 1225 |
} |
} |
| 1226 |
|
|
| 1272 |
|
|
| 1273 |
if (do_study) |
if (do_study) |
| 1274 |
{ |
{ |
| 1275 |
if (timeit) |
if (timeit > 0) |
| 1276 |
{ |
{ |
| 1277 |
register int i; |
register int i; |
| 1278 |
clock_t time_taken; |
clock_t time_taken; |
| 1279 |
clock_t start_time = clock(); |
clock_t start_time = clock(); |
| 1280 |
for (i = 0; i < LOOPREPEAT; i++) |
for (i = 0; i < timeit; i++) |
| 1281 |
extra = pcre_study(re, study_options, &error); |
extra = pcre_study(re, study_options, &error); |
| 1282 |
time_taken = clock() - start_time; |
time_taken = clock() - start_time; |
| 1283 |
if (extra != NULL) free(extra); |
if (extra != NULL) free(extra); |
| 1284 |
fprintf(outfile, " Study time %.3f milliseconds\n", |
fprintf(outfile, " Study time %.4f milliseconds\n", |
| 1285 |
(((double)time_taken * 1000.0) / (double)LOOPREPEAT) / |
(((double)time_taken * 1000.0) / (double)timeit) / |
| 1286 |
(double)CLOCKS_PER_SEC); |
(double)CLOCKS_PER_SEC); |
| 1287 |
} |
} |
| 1288 |
extra = pcre_study(re, study_options, &error); |
extra = pcre_study(re, study_options, &error); |
| 1325 |
|
|
| 1326 |
SHOW_INFO: |
SHOW_INFO: |
| 1327 |
|
|
| 1328 |
|
if (do_debug) |
| 1329 |
|
{ |
| 1330 |
|
fprintf(outfile, "------------------------------------------------------------------\n"); |
| 1331 |
|
pcre_printint(re, outfile); |
| 1332 |
|
} |
| 1333 |
|
|
| 1334 |
if (do_showinfo) |
if (do_showinfo) |
| 1335 |
{ |
{ |
| 1336 |
unsigned long int get_options, all_options; |
unsigned long int get_options, all_options; |
| 1341 |
int nameentrysize, namecount; |
int nameentrysize, namecount; |
| 1342 |
const uschar *nametable; |
const uschar *nametable; |
| 1343 |
|
|
|
if (do_debug) |
|
|
{ |
|
|
fprintf(outfile, "------------------------------------------------------------------\n"); |
|
|
pcre_printint(re, outfile); |
|
|
} |
|
|
|
|
| 1344 |
new_info(re, NULL, PCRE_INFO_OPTIONS, &get_options); |
new_info(re, NULL, PCRE_INFO_OPTIONS, &get_options); |
| 1345 |
new_info(re, NULL, PCRE_INFO_SIZE, &size); |
new_info(re, NULL, PCRE_INFO_SIZE, &size); |
| 1346 |
new_info(re, NULL, PCRE_INFO_CAPTURECOUNT, &count); |
new_info(re, NULL, PCRE_INFO_CAPTURECOUNT, &count); |
| 1419 |
((get_options & PCRE_NO_UTF8_CHECK) != 0)? " no_utf8_check" : "", |
((get_options & PCRE_NO_UTF8_CHECK) != 0)? " no_utf8_check" : "", |
| 1420 |
((get_options & PCRE_DUPNAMES) != 0)? " dupnames" : ""); |
((get_options & PCRE_DUPNAMES) != 0)? " dupnames" : ""); |
| 1421 |
|
|
| 1422 |
switch (get_options & PCRE_NEWLINE_CRLF) |
switch (get_options & PCRE_NEWLINE_BITS) |
| 1423 |
{ |
{ |
| 1424 |
case PCRE_NEWLINE_CR: |
case PCRE_NEWLINE_CR: |
| 1425 |
fprintf(outfile, "Forced newline sequence: CR\n"); |
fprintf(outfile, "Forced newline sequence: CR\n"); |
| 1433 |
fprintf(outfile, "Forced newline sequence: CRLF\n"); |
fprintf(outfile, "Forced newline sequence: CRLF\n"); |
| 1434 |
break; |
break; |
| 1435 |
|
|
| 1436 |
|
case PCRE_NEWLINE_ANY: |
| 1437 |
|
fprintf(outfile, "Forced newline sequence: ANY\n"); |
| 1438 |
|
break; |
| 1439 |
|
|
| 1440 |
default: |
default: |
| 1441 |
break; |
break; |
| 1442 |
} |
} |
| 1454 |
int ch = first_char & 255; |
int ch = first_char & 255; |
| 1455 |
const char *caseless = ((first_char & REQ_CASELESS) == 0)? |
const char *caseless = ((first_char & REQ_CASELESS) == 0)? |
| 1456 |
"" : " (caseless)"; |
"" : " (caseless)"; |
| 1457 |
if (isprint(ch)) |
if (PRINTHEX(ch)) |
| 1458 |
fprintf(outfile, "First char = \'%c\'%s\n", ch, caseless); |
fprintf(outfile, "First char = \'%c\'%s\n", ch, caseless); |
| 1459 |
else |
else |
| 1460 |
fprintf(outfile, "First char = %d%s\n", ch, caseless); |
fprintf(outfile, "First char = %d%s\n", ch, caseless); |
| 1469 |
int ch = need_char & 255; |
int ch = need_char & 255; |
| 1470 |
const char *caseless = ((need_char & REQ_CASELESS) == 0)? |
const char *caseless = ((need_char & REQ_CASELESS) == 0)? |
| 1471 |
"" : " (caseless)"; |
"" : " (caseless)"; |
| 1472 |
if (isprint(ch)) |
if (PRINTHEX(ch)) |
| 1473 |
fprintf(outfile, "Need char = \'%c\'%s\n", ch, caseless); |
fprintf(outfile, "Need char = \'%c\'%s\n", ch, caseless); |
| 1474 |
else |
else |
| 1475 |
fprintf(outfile, "Need char = %d%s\n", ch, caseless); |
fprintf(outfile, "Need char = %d%s\n", ch, caseless); |
| 1505 |
fprintf(outfile, "\n "); |
fprintf(outfile, "\n "); |
| 1506 |
c = 2; |
c = 2; |
| 1507 |
} |
} |
| 1508 |
if (isprint(i) && i != ' ') |
if (PRINTHEX(i) && i != ' ') |
| 1509 |
{ |
{ |
| 1510 |
fprintf(outfile, "%c ", i); |
fprintf(outfile, "%c ", i); |
| 1511 |
c += 2; |
c += 2; |
| 1564 |
strerror(errno)); |
strerror(errno)); |
| 1565 |
} |
} |
| 1566 |
else fprintf(outfile, "Study data written to %s\n", to_file); |
else fprintf(outfile, "Study data written to %s\n", to_file); |
| 1567 |
|
|
| 1568 |
} |
} |
| 1569 |
} |
} |
| 1570 |
fclose(f); |
fclose(f); |
| 1963 |
|
|
| 1964 |
for (;; gmatched++) /* Loop for /g or /G */ |
for (;; gmatched++) /* Loop for /g or /G */ |
| 1965 |
{ |
{ |
| 1966 |
if (timeit) |
if (timeitm > 0) |
| 1967 |
{ |
{ |
| 1968 |
register int i; |
register int i; |
| 1969 |
clock_t time_taken; |
clock_t time_taken; |
| 1973 |
if (all_use_dfa || use_dfa) |
if (all_use_dfa || use_dfa) |
| 1974 |
{ |
{ |
| 1975 |
int workspace[1000]; |
int workspace[1000]; |
| 1976 |
for (i = 0; i < LOOPREPEAT; i++) |
for (i = 0; i < timeitm; i++) |
| 1977 |
count = pcre_dfa_exec(re, NULL, (char *)bptr, len, start_offset, |
count = pcre_dfa_exec(re, NULL, (char *)bptr, len, start_offset, |
| 1978 |
options | g_notempty, use_offsets, use_size_offsets, workspace, |
options | g_notempty, use_offsets, use_size_offsets, workspace, |
| 1979 |
sizeof(workspace)/sizeof(int)); |
sizeof(workspace)/sizeof(int)); |
| 1981 |
else |
else |
| 1982 |
#endif |
#endif |
| 1983 |
|
|
| 1984 |
for (i = 0; i < LOOPREPEAT; i++) |
for (i = 0; i < timeitm; i++) |
| 1985 |
count = pcre_exec(re, extra, (char *)bptr, len, |
count = pcre_exec(re, extra, (char *)bptr, len, |
| 1986 |
start_offset, options | g_notempty, use_offsets, use_size_offsets); |
start_offset, options | g_notempty, use_offsets, use_size_offsets); |
| 1987 |
|
|
| 1988 |
time_taken = clock() - start_time; |
time_taken = clock() - start_time; |
| 1989 |
fprintf(outfile, "Execute time %.3f milliseconds\n", |
fprintf(outfile, "Execute time %.4f milliseconds\n", |
| 1990 |
(((double)time_taken * 1000.0) / (double)LOOPREPEAT) / |
(((double)time_taken * 1000.0) / (double)timeitm) / |
| 1991 |
(double)CLOCKS_PER_SEC); |
(double)CLOCKS_PER_SEC); |
| 1992 |
} |
} |
| 1993 |
|
|
| 2063 |
|
|
| 2064 |
if (count >= 0) |
if (count >= 0) |
| 2065 |
{ |
{ |
| 2066 |
int i; |
int i, maxcount; |
| 2067 |
|
|
| 2068 |
|
#if !defined NODFA |
| 2069 |
|
if (all_use_dfa || use_dfa) maxcount = use_size_offsets/2; else |
| 2070 |
|
#endif |
| 2071 |
|
maxcount = use_size_offsets/3; |
| 2072 |
|
|
| 2073 |
|
/* This is a check against a lunatic return value. */ |
| 2074 |
|
|
| 2075 |
|
if (count > maxcount) |
| 2076 |
|
{ |
| 2077 |
|
fprintf(outfile, |
| 2078 |
|
"** PCRE error: returned count %d is too big for offset size %d\n", |
| 2079 |
|
count, use_size_offsets); |
| 2080 |
|
count = use_size_offsets/3; |
| 2081 |
|
if (do_g || do_G) |
| 2082 |
|
{ |
| 2083 |
|
fprintf(outfile, "** /%c loop abandoned\n", do_g? 'g' : 'G'); |
| 2084 |
|
do_g = do_G = FALSE; /* Break g/G loop */ |
| 2085 |
|
} |
| 2086 |
|
} |
| 2087 |
|
|
| 2088 |
for (i = 0; i < count * 2; i += 2) |
for (i = 0; i < count * 2; i += 2) |
| 2089 |
{ |
{ |
| 2090 |
if (use_offsets[i] < 0) |
if (use_offsets[i] < 0) |
| 2283 |
{ |
{ |
| 2284 |
new_free((void *)tables); |
new_free((void *)tables); |
| 2285 |
setlocale(LC_CTYPE, "C"); |
setlocale(LC_CTYPE, "C"); |
| 2286 |
|
locale_set = 0; |
| 2287 |
} |
} |
| 2288 |
} |
} |
| 2289 |
|
|