--- code/trunk/pcretest.c 2007/02/24 21:38:49 27 +++ code/trunk/pcretest.c 2007/02/24 21:39:50 57 @@ -12,7 +12,14 @@ /* Use the internal info for displaying the results of pcre_study(). */ #include "internal.h" + +/* It is possible to compile this test program without including support for +testing the POSIX interface, though this is not available via the standard +Makefile. */ + +#if !defined NOPOSIX #include "pcreposix.h" +#endif #ifndef CLOCKS_PER_SEC #ifdef CLK_TCK @@ -27,6 +34,114 @@ static FILE *outfile; static int log_store = 0; +static size_t gotten_store; + + + +static int utf8_table1[] = { + 0x0000007f, 0x000007ff, 0x0000ffff, 0x001fffff, 0x03ffffff, 0x7fffffff}; + +static int utf8_table2[] = { + 0, 0xc0, 0xe0, 0xf0, 0xf8, 0xfc}; + +static int utf8_table3[] = { + 0xff, 0x1f, 0x0f, 0x07, 0x03, 0x01}; + + +/************************************************* +* Convert character value to UTF-8 * +*************************************************/ + +/* This function takes an integer value in the range 0 - 0x7fffffff +and encodes it as a UTF-8 character in 0 to 6 bytes. + +Arguments: + cvalue the character value + buffer pointer to buffer for result - at least 6 bytes long + +Returns: number of characters placed in the buffer + -1 if input character is negative + 0 if input character is positive but too big (only when + int is longer than 32 bits) +*/ + +static int +ord2utf8(int cvalue, unsigned char *buffer) +{ +register int i, j; +for (i = 0; i < sizeof(utf8_table1)/sizeof(int); i++) + if (cvalue <= utf8_table1[i]) break; +if (i >= sizeof(utf8_table1)/sizeof(int)) return 0; +if (cvalue < 0) return -1; +*buffer++ = utf8_table2[i] | (cvalue & utf8_table3[i]); +cvalue >>= 6 - i; +for (j = 0; j < i; j++) + { + *buffer++ = 0x80 | (cvalue & 0x3f); + cvalue >>= 6; + } +return i + 1; +} + + +/************************************************* +* Convert UTF-8 string to value * +*************************************************/ + +/* This function takes one or more bytes that represents a UTF-8 character, +and returns the value of the character. + +Argument: + buffer a pointer to the byte vector + vptr a pointer to an int to receive the value + +Returns: > 0 => the number of bytes consumed + -6 to 0 => malformed UTF-8 character at offset = (-return) +*/ + +int +utf82ord(unsigned char *buffer, int *vptr) +{ +int c = *buffer++; +int d = c; +int i, j, s; + +for (i = -1; i < 6; i++) /* i is number of additional bytes */ + { + if ((d & 0x80) == 0) break; + d <<= 1; + } + +if (i == -1) { *vptr = c; return 1; } /* ascii character */ +if (i == 0 || i == 6) return 0; /* invalid UTF-8 */ + +/* i now has a value in the range 1-5 */ + +d = c & utf8_table3[i]; +s = 6 - i; + +for (j = 0; j < i; j++) + { + c = *buffer++; + if ((c & 0xc0) != 0x80) return -(j+1); + d |= (c & 0x3f) << s; + s += 6; + } + +/* Check that encoding was the correct unique one */ + +for (j = 0; j < sizeof(utf8_table1)/sizeof(int); j++) + if (d <= utf8_table1[j]) break; +if (j != i) return -(i+1); + +/* Valid value */ + +*vptr = d; +return i+1; +} + + + @@ -41,14 +156,14 @@ "*", "*?", "+", "+?", "?", "??", "{", "{", "{", "*", "*?", "+", "+?", "?", "??", "{", "{", "{", "*", "*?", "+", "+?", "?", "??", "{", "{", - "class", "Ref", + "class", "Ref", "Recurse", "Alt", "Ket", "KetRmax", "KetRmin", "Assert", "Assert not", "AssertB", "AssertB not", "Reverse", "Once", "Cond", "Cref", - "Brazero", "Braminzero", "Bra" + "Brazero", "Braminzero", "Branumber", "Bra" }; -static void print_internals(pcre *re, FILE *outfile) +static void print_internals(pcre *re) { unsigned char *code = ((real_pcre *)re)->code; @@ -63,7 +178,10 @@ if (*code >= OP_BRA) { - fprintf(outfile, "%3d Bra %d", (code[1] << 8) + code[2], *code - OP_BRA); + if (*code - OP_BRA > EXTRACT_BASIC_MAX) + fprintf(outfile, "%3d Bra extra", (code[1] << 8) + code[2]); + else + fprintf(outfile, "%3d Bra %d", (code[1] << 8) + code[2], *code - OP_BRA); code += 2; } @@ -79,16 +197,6 @@ code++; break; - case OP_COND: - fprintf(outfile, "%3d Cond", (code[1] << 8) + code[2]); - code += 2; - break; - - case OP_CREF: - fprintf(outfile, " %.2d %s", code[1], OP_names[*code]); - code++; - break; - case OP_CHARS: charlength = *(++code); fprintf(outfile, "%3d ", charlength); @@ -106,11 +214,10 @@ case OP_ASSERTBACK: case OP_ASSERTBACK_NOT: case OP_ONCE: - fprintf(outfile, "%3d %s", (code[1] << 8) + code[2], OP_names[*code]); - code += 2; - break; - + case OP_COND: + case OP_BRANUMBER: case OP_REVERSE: + case OP_CREF: fprintf(outfile, "%3d %s", (code[1] << 8) + code[2], OP_names[*code]); code += 2; break; @@ -183,8 +290,8 @@ break; case OP_REF: - fprintf(outfile, " \\%d", *(++code)); - code++; + fprintf(outfile, " \\%d", (code[1] << 8) | code[2]); + code += 3; goto CLASS_REF_REPEAT; case OP_CLASS: @@ -257,14 +364,31 @@ -/* Character string printing function. */ +/* Character string printing function. A "normal" and a UTF-8 version. */ -static void pchars(unsigned char *p, int length) +static void pchars(unsigned char *p, int length, int utf8) { int c; while (length-- > 0) + { + if (utf8) + { + int rc = utf82ord(p, &c); + if (rc > 0) + { + length -= rc - 1; + p += rc; + if (c < 256 && isprint(c)) fprintf(outfile, "%c", c); + else fprintf(outfile, "\\x{%02x}", c); + continue; + } + } + + /* Not UTF-8, or malformed UTF-8 */ + if (isprint(c = *(p++))) fprintf(outfile, "%c", c); else fprintf(outfile, "\\x%02x", c); + } } @@ -274,12 +398,28 @@ static void *new_malloc(size_t size) { -if (log_store) fprintf(outfile, "Store size request: %d\n", (int)size); +gotten_store = size; +if (log_store) + fprintf(outfile, "Memory allocation (code space): %d\n", + (int)((int)size - offsetof(real_pcre, code[0]))); return malloc(size); } + +/* Get one piece of information from the pcre_fullinfo() function */ + +static void new_info(pcre *re, pcre_extra *study, int option, void *ptr) +{ +int rc; +if ((rc = pcre_fullinfo(re, study, option, ptr)) < 0) + fprintf(outfile, "Error %d from pcre_fullinfo(%d)\n", rc, option); +} + + + + /* Read lines from named file or stdin and write to named file or stdout; lines consist of a regular expression, in delimiters and optionally followed by options, followed by a set of test data, terminated by an empty line. */ @@ -292,7 +432,13 @@ int op = 1; int timeit = 0; int showinfo = 0; +int showstore = 0; +int size_offsets = 45; +int size_offsets_max; +int *offsets; +#if !defined NOPOSIX int posix = 0; +#endif int debug = 0; int done = 0; unsigned char buffer[30000]; @@ -306,26 +452,51 @@ while (argc > 1 && argv[op][0] == '-') { - if (strcmp(argv[op], "-s") == 0) log_store = 1; + char *endptr; + + if (strcmp(argv[op], "-s") == 0 || strcmp(argv[op], "-m") == 0) + showstore = 1; else if (strcmp(argv[op], "-t") == 0) timeit = 1; else if (strcmp(argv[op], "-i") == 0) showinfo = 1; else if (strcmp(argv[op], "-d") == 0) showinfo = debug = 1; + else if (strcmp(argv[op], "-o") == 0 && argc > 2 && + ((size_offsets = strtoul(argv[op+1], &endptr, 10)), *endptr == 0)) + { + op++; + argc--; + } +#if !defined NOPOSIX else if (strcmp(argv[op], "-p") == 0) posix = 1; +#endif else { - printf("*** Unknown option %s\n", argv[op]); - printf("Usage: pcretest [-d] [-i] [-p] [-s] [-t] [ []]\n"); - printf(" -d debug: show compiled code; implies -i\n" - " -i show information about compiled pattern\n" - " -p use POSIX interface\n" - " -s output store information\n" - " -t time compilation and execution\n"); + printf("** Unknown or malformed option %s\n", argv[op]); + printf("Usage: pcretest [-d] [-i] [-o ] [-p] [-s] [-t] [ []]\n"); + printf(" -d debug: show compiled code; implies -i\n" + " -i show information about compiled pattern\n" + " -o set size of offsets vector to \n"); +#if !defined NOPOSIX + printf(" -p use POSIX interface\n"); +#endif + printf(" -s output store information\n" + " -t time compilation and execution\n"); return 1; } op++; argc--; } +/* Get the store for the offsets vector, and remember what it was */ + +size_offsets_max = size_offsets; +offsets = malloc(size_offsets_max * sizeof(int)); +if (offsets == NULL) + { + printf("** Failed to get %d bytes of memory for offsets vector\n", + size_offsets_max * sizeof(int)); + return 1; + } + /* Sort out the input and output files */ if (argc > 1) @@ -362,14 +533,22 @@ { pcre *re = NULL; pcre_extra *extra = NULL; + +#if !defined NOPOSIX /* There are still compilers that require no indent */ regex_t preg; + int do_posix = 0; +#endif + const char *error; unsigned char *p, *pp, *ppp; - unsigned const char *tables = NULL; + const unsigned char *tables = NULL; int do_study = 0; int do_debug = debug; + int do_G = 0; + int do_g = 0; int do_showinfo = showinfo; - int do_posix = 0; + int do_showrest = 0; + int utf8 = 0; int erroroffset, len, delimiter; if (infile == stdin) printf(" re> "); @@ -385,9 +564,9 @@ delimiter = *p++; - if (isalnum(delimiter)) + if (isalnum(delimiter) || delimiter == '\\') { - fprintf(outfile, "** Delimiter must not be alphameric\n"); + fprintf(outfile, "** Delimiter must not be alphameric or \\\n"); goto SKIP_DATA; } @@ -395,7 +574,12 @@ for(;;) { - while (*pp != 0 && *pp != delimiter) pp++; + while (*pp != 0) + { + if (*pp == '\\' && pp[1] != 0) pp++; + else if (*pp == delimiter) break; + pp++; + } if (*pp != 0) break; len = sizeof(buffer) - (pp - buffer); @@ -415,6 +599,12 @@ if (infile != stdin) fprintf(outfile, "%s", (char *)pp); } + /* If the first character after the delimiter is backslash, make + the pattern end with backslash. This is purely to provide a way + of testing for the error message when a pattern ends with backslash. */ + + if (pp[1] == '\\') *pp++ = '\\'; + /* Terminate the pattern at the delimiter */ *pp++ = 0; @@ -423,23 +613,34 @@ options = 0; study_options = 0; + log_store = showstore; /* default from command line */ + while (*pp != 0) { switch (*pp++) { + case 'g': do_g = 1; break; case 'i': options |= PCRE_CASELESS; break; case 'm': options |= PCRE_MULTILINE; break; case 's': options |= PCRE_DOTALL; break; case 'x': options |= PCRE_EXTENDED; break; + case '+': do_showrest = 1; break; case 'A': options |= PCRE_ANCHORED; break; case 'D': do_debug = do_showinfo = 1; break; case 'E': options |= PCRE_DOLLAR_ENDONLY; break; + case 'G': do_G = 1; break; case 'I': do_showinfo = 1; break; + case 'M': log_store = 1; break; + +#if !defined NOPOSIX case 'P': do_posix = 1; break; +#endif + case 'S': do_study = 1; break; case 'U': options |= PCRE_UNGREEDY; break; case 'X': options |= PCRE_EXTRA; break; + case '8': options |= PCRE_UTF8; utf8 = 1; break; case 'L': ppp = pp; @@ -465,6 +666,7 @@ timing, showing, or debugging options, nor the ability to pass over local character tables. */ +#if !defined NOPOSIX if (posix || do_posix) { int rc; @@ -487,6 +689,8 @@ /* Handle compiling via the native interface */ else +#endif /* !defined NOPOSIX */ + { if (timeit) { @@ -531,45 +735,92 @@ goto CONTINUE; } - /* Compilation succeeded; print data if required */ + /* Compilation succeeded; print data if required. There are now two + info-returning functions. The old one has a limited interface and + returns only limited data. Check that it agrees with the newer one. */ if (do_showinfo) { - int first_char, count; + unsigned long int get_options; + int old_first_char, old_options, old_count; + int count, backrefmax, first_char, need_char; + size_t size; + + if (do_debug) print_internals(re); + + new_info(re, NULL, PCRE_INFO_OPTIONS, &get_options); + new_info(re, NULL, PCRE_INFO_SIZE, &size); + new_info(re, NULL, PCRE_INFO_CAPTURECOUNT, &count); + new_info(re, NULL, PCRE_INFO_BACKREFMAX, &backrefmax); + new_info(re, NULL, PCRE_INFO_FIRSTCHAR, &first_char); + new_info(re, NULL, PCRE_INFO_LASTLITERAL, &need_char); - if (do_debug) print_internals(re, outfile); - - count = pcre_info(re, &options, &first_char); + old_count = pcre_info(re, &old_options, &old_first_char); if (count < 0) fprintf(outfile, - "Error %d while reading info\n", count); + "Error %d from pcre_info()\n", count); else { - fprintf(outfile, "Identifying subpattern count = %d\n", count); - if (options == 0) fprintf(outfile, "No options\n"); - else fprintf(outfile, "Options:%s%s%s%s%s%s%s%s\n", - ((options & PCRE_ANCHORED) != 0)? " anchored" : "", - ((options & PCRE_CASELESS) != 0)? " caseless" : "", - ((options & PCRE_EXTENDED) != 0)? " extended" : "", - ((options & PCRE_MULTILINE) != 0)? " multiline" : "", - ((options & PCRE_DOTALL) != 0)? " dotall" : "", - ((options & PCRE_DOLLAR_ENDONLY) != 0)? " dollar_endonly" : "", - ((options & PCRE_EXTRA) != 0)? " extra" : "", - ((options & PCRE_UNGREEDY) != 0)? " ungreedy" : ""); - if (first_char == -1) - { - fprintf(outfile, "First char at start or follows \\n\n"); - } - else if (first_char < 0) - { - fprintf(outfile, "No first char\n"); - } + if (old_count != count) fprintf(outfile, + "Count disagreement: pcre_fullinfo=%d pcre_info=%d\n", count, + old_count); + + if (old_first_char != first_char) fprintf(outfile, + "First char disagreement: pcre_fullinfo=%d pcre_info=%d\n", + first_char, old_first_char); + + if (old_options != (int)get_options) fprintf(outfile, + "Options disagreement: pcre_fullinfo=%ld pcre_info=%d\n", + get_options, old_options); + } + + if (size != gotten_store) fprintf(outfile, + "Size disagreement: pcre_fullinfo=%d call to malloc for %d\n", + size, gotten_store); + + fprintf(outfile, "Capturing subpattern count = %d\n", count); + if (backrefmax > 0) + fprintf(outfile, "Max back reference = %d\n", backrefmax); + if (get_options == 0) fprintf(outfile, "No options\n"); + else fprintf(outfile, "Options:%s%s%s%s%s%s%s%s%s\n", + ((get_options & PCRE_ANCHORED) != 0)? " anchored" : "", + ((get_options & PCRE_CASELESS) != 0)? " caseless" : "", + ((get_options & PCRE_EXTENDED) != 0)? " extended" : "", + ((get_options & PCRE_MULTILINE) != 0)? " multiline" : "", + ((get_options & PCRE_DOTALL) != 0)? " dotall" : "", + ((get_options & PCRE_DOLLAR_ENDONLY) != 0)? " dollar_endonly" : "", + ((get_options & PCRE_EXTRA) != 0)? " extra" : "", + ((get_options & PCRE_UNGREEDY) != 0)? " ungreedy" : "", + ((get_options & PCRE_UTF8) != 0)? " utf8" : ""); + + if (((((real_pcre *)re)->options) & PCRE_ICHANGED) != 0) + fprintf(outfile, "Case state changes\n"); + + if (first_char == -1) + { + fprintf(outfile, "First char at start or follows \\n\n"); + } + else if (first_char < 0) + { + fprintf(outfile, "No first char\n"); + } + else + { + if (isprint(first_char)) + fprintf(outfile, "First char = \'%c\'\n", first_char); else - { - if (isprint(first_char)) - fprintf(outfile, "First char = \'%c\'\n", first_char); - else - fprintf(outfile, "First char = %d\n", first_char); - } + fprintf(outfile, "First char = %d\n", first_char); + } + + if (need_char < 0) + { + fprintf(outfile, "No need char\n"); + } + else + { + if (isprint(need_char)) + fprintf(outfile, "Need char = \'%c\'\n", need_char); + else + fprintf(outfile, "Need char = %d\n", need_char); } } @@ -598,13 +849,11 @@ else if (extra == NULL) fprintf(outfile, "Study returned NULL\n"); - /* This looks at internal information. A bit kludgy to do it this - way, but it is useful for testing. */ - else if (do_showinfo) { - real_pcre_extra *xx = (real_pcre_extra *)extra; - if ((xx->options & PCRE_STUDY_MAPPED) == 0) + uschar *start_bits = NULL; + new_info(re, extra, PCRE_INFO_FIRSTTABLE, &start_bits); + if (start_bits == NULL) fprintf(outfile, "No starting character set\n"); else { @@ -613,7 +862,7 @@ fprintf(outfile, "Starting character set: "); for (i = 0; i < 256; i++) { - if ((xx->start_bits[i/8] & (1<<(i%8))) != 0) + if ((start_bits[i/8] & (1<<(i%8))) != 0) { if (c > 75) { @@ -643,13 +892,20 @@ for (;;) { unsigned char *q; + unsigned char *bptr = dbuffer; + int *use_offsets = offsets; + int use_size_offsets = size_offsets; int count, c; - int offsets[45]; - int size_offsets = sizeof(offsets)/sizeof(int); + int copystrings = 0; + int getstrings = 0; + int getlist = 0; + int gmatched = 0; + int start_offset = 0; + int g_notempty = 0; options = 0; - if (infile == stdin) printf(" data> "); + if (infile == stdin) printf("data> "); if (fgets((char *)buffer, sizeof(buffer), infile) == NULL) { done = 1; @@ -689,6 +945,30 @@ break; case 'x': + + /* Handle \x{..} specially - new Perl thing for utf8 */ + + if (*p == '{') + { + unsigned char *pt = p; + c = 0; + while (isxdigit(*(++pt))) + c = c * 16 + tolower(*pt) - ((isdigit(*pt))? '0' : 'W'); + if (*pt == '}') + { + unsigned char buffer[8]; + int ii, utn; + utn = ord2utf8(c, buffer); + for (ii = 0; ii < utn - 1; ii++) *q++ = buffer[ii]; + c = buffer[ii]; /* Last byte */ + p = pt + 1; + break; + } + /* Not correct form; fall through */ + } + + /* Ordinary \x */ + c = 0; while (i++ < 2 && isxdigit(*p)) { @@ -709,9 +989,40 @@ options |= PCRE_NOTBOL; continue; + case 'C': + while(isdigit(*p)) n = n * 10 + *p++ - '0'; + copystrings |= 1 << n; + continue; + + case 'G': + while(isdigit(*p)) n = n * 10 + *p++ - '0'; + getstrings |= 1 << n; + continue; + + case 'L': + getlist = 1; + continue; + + case 'N': + options |= PCRE_NOTEMPTY; + continue; + case 'O': while(isdigit(*p)) n = n * 10 + *p++ - '0'; - if (n <= (int)(sizeof(offsets)/sizeof(int))) size_offsets = n; + if (n > size_offsets_max) + { + size_offsets_max = n; + free(offsets); + use_offsets = offsets = malloc(size_offsets_max * sizeof(int)); + if (offsets == NULL) + { + printf("** Failed to get %d bytes of memory for offsets vector\n", + size_offsets_max * sizeof(int)); + return 1; + } + } + use_size_offsets = n; + if (n == 0) use_offsets = NULL; continue; case 'Z': @@ -726,16 +1037,16 @@ /* Handle matching via the POSIX interface, which does not support timing. */ +#if !defined NOPOSIX if (posix || do_posix) { int rc; int eflags = 0; - regmatch_t pmatch[30]; + regmatch_t *pmatch = malloc(sizeof(regmatch_t) * use_size_offsets); if ((options & PCRE_NOTBOL) != 0) eflags |= REG_NOTBOL; if ((options & PCRE_NOTEOL) != 0) eflags |= REG_NOTEOL; - rc = regexec(&preg, (char *)dbuffer, sizeof(pmatch)/sizeof(regmatch_t), - pmatch, eflags); + rc = regexec(&preg, (const char *)bptr, use_size_offsets, pmatch, eflags); if (rc != 0) { @@ -745,22 +1056,32 @@ else { size_t i; - for (i = 0; i < sizeof(pmatch)/sizeof(regmatch_t); i++) + for (i = 0; i < use_size_offsets; i++) { if (pmatch[i].rm_so >= 0) { fprintf(outfile, "%2d: ", (int)i); pchars(dbuffer + pmatch[i].rm_so, - pmatch[i].rm_eo - pmatch[i].rm_so); + pmatch[i].rm_eo - pmatch[i].rm_so, utf8); fprintf(outfile, "\n"); + if (i == 0 && do_showrest) + { + fprintf(outfile, " 0+ "); + pchars(dbuffer + pmatch[i].rm_eo, len - pmatch[i].rm_eo, utf8); + fprintf(outfile, "\n"); + } } } } + free(pmatch); } - /* Handle matching via the native interface */ + /* Handle matching via the native interface - repeats for /g and /G */ else +#endif /* !defined NOPOSIX */ + + for (;; gmatched++) /* Loop for /g or /G */ { if (timeit) { @@ -768,49 +1089,162 @@ clock_t time_taken; clock_t start_time = clock(); for (i = 0; i < LOOPREPEAT; i++) - count = pcre_exec(re, extra, (char *)dbuffer, len, options, offsets, - size_offsets); + count = pcre_exec(re, extra, (char *)bptr, len, + start_offset, options | g_notempty, use_offsets, use_size_offsets); time_taken = clock() - start_time; fprintf(outfile, "Execute time %.3f milliseconds\n", ((double)time_taken * 1000.0)/ ((double)LOOPREPEAT * (double)CLOCKS_PER_SEC)); } - count = pcre_exec(re, extra, (char *)dbuffer, len, options, offsets, - size_offsets); + count = pcre_exec(re, extra, (char *)bptr, len, + start_offset, options | g_notempty, use_offsets, use_size_offsets); if (count == 0) { fprintf(outfile, "Matched, but too many substrings\n"); - count = size_offsets/3; + count = use_size_offsets/3; } + /* Matched */ + if (count >= 0) { int i; - count *= 2; - for (i = 0; i < count; i += 2) + for (i = 0; i < count * 2; i += 2) { - if (offsets[i] < 0) + if (use_offsets[i] < 0) fprintf(outfile, "%2d: \n", i/2); else { fprintf(outfile, "%2d: ", i/2); - pchars(dbuffer + offsets[i], offsets[i+1] - offsets[i]); + pchars(bptr + use_offsets[i], use_offsets[i+1] - use_offsets[i], utf8); fprintf(outfile, "\n"); + if (i == 0) + { + if (do_showrest) + { + fprintf(outfile, " 0+ "); + pchars(bptr + use_offsets[i+1], len - use_offsets[i+1], utf8); + fprintf(outfile, "\n"); + } + } + } + } + + for (i = 0; i < 32; i++) + { + if ((copystrings & (1 << i)) != 0) + { + char copybuffer[16]; + int rc = pcre_copy_substring((char *)bptr, use_offsets, count, + i, copybuffer, sizeof(copybuffer)); + if (rc < 0) + fprintf(outfile, "copy substring %d failed %d\n", i, rc); + else + fprintf(outfile, "%2dC %s (%d)\n", i, copybuffer, rc); + } + } + + for (i = 0; i < 32; i++) + { + if ((getstrings & (1 << i)) != 0) + { + const char *substring; + int rc = pcre_get_substring((char *)bptr, use_offsets, count, + i, &substring); + if (rc < 0) + fprintf(outfile, "get substring %d failed %d\n", i, rc); + else + { + fprintf(outfile, "%2dG %s (%d)\n", i, substring, rc); + /* free((void *)substring); */ + pcre_free_substring(substring); + } + } + } + + if (getlist) + { + const char **stringlist; + int rc = pcre_get_substring_list((char *)bptr, use_offsets, count, + &stringlist); + if (rc < 0) + fprintf(outfile, "get substring list failed %d\n", rc); + else + { + for (i = 0; i < count; i++) + fprintf(outfile, "%2dL %s\n", i, stringlist[i]); + if (stringlist[i] != NULL) + fprintf(outfile, "string list not terminated by NULL\n"); + /* free((void *)stringlist); */ + pcre_free_substring_list(stringlist); } } } + + /* Failed to match. If this is a /g or /G loop and we previously set + g_notempty after a null match, this is not necessarily the end. + We want to advance the start offset, and continue. Fudge the offset + values to achieve this. We won't be at the end of the string - that + was checked before setting g_notempty. */ + else { - if (count == -1) fprintf(outfile, "No match\n"); - else fprintf(outfile, "Error %d\n", count); + if (g_notempty != 0) + { + use_offsets[0] = start_offset; + use_offsets[1] = start_offset + 1; + } + else + { + if (gmatched == 0) /* Error if no previous matches */ + { + if (count == -1) fprintf(outfile, "No match\n"); + else fprintf(outfile, "Error %d\n", count); + } + break; /* Out of the /g loop */ + } } - } - } + + /* If not /g or /G we are done */ + + if (!do_g && !do_G) break; + + /* If we have matched an empty string, first check to see if we are at + the end of the subject. If so, the /g loop is over. Otherwise, mimic + what Perl's /g options does. This turns out to be rather cunning. First + we set PCRE_NOTEMPTY and PCRE_ANCHORED and try the match again at the + same point. If this fails (picked up above) we advance to the next + character. */ + + g_notempty = 0; + if (use_offsets[0] == use_offsets[1]) + { + if (use_offsets[0] == len) break; + g_notempty = PCRE_NOTEMPTY | PCRE_ANCHORED; + } + + /* For /g, update the start offset, leaving the rest alone */ + + if (do_g) start_offset = use_offsets[1]; + + /* For /G, update the pointer and length */ + + else + { + bptr += use_offsets[1]; + len -= use_offsets[1]; + } + } /* End of loop for /g and /G */ + } /* End of loop for data lines */ CONTINUE: + +#if !defined NOPOSIX if (posix || do_posix) regfree(&preg); +#endif + if (re != NULL) free(re); if (extra != NULL) free(extra); if (tables != NULL)