/[pcre]/code/trunk/pcretest.c

Diff of /code/trunk/pcretest.c

Parent Directory | Revision Log | View Patch Patch

-revision 71 by nigel,
Sat Feb 24 21:40:24 2007 UTC
+revision 93 by nigel,
Sat Feb 24 21:41:42 2007 UTC
 Line 4
  /* This program was hacked up as a tester for PCRE. I really should have
  written it more tidily in the first place. Will I ever learn? It has grown and
- been extended and consequently is now rather untidy in places. */
+ been extended and consequently is now rather, er, *very* untidy in places.
+ -----------------------------------------------------------------------------
+ Redistribution and use in source and binary forms, with or without
+ modification, are permitted provided that the following conditions are met:
+     * Redistributions of source code must retain the above copyright notice,
+       this list of conditions and the following disclaimer.
+     * Redistributions in binary form must reproduce the above copyright
+       notice, this list of conditions and the following disclaimer in the
+       documentation and/or other materials provided with the distribution.
+     * Neither the name of the University of Cambridge nor the names of its
+       contributors may be used to endorse or promote products derived from
+       this software without specific prior written permission.
+ THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+ AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+ IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
+ ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE
+ LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
+ CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
+ SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
+ INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
+ CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
+ ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
+ POSSIBILITY OF SUCH DAMAGE.
+ -----------------------------------------------------------------------------
+ */
  #include <ctype.h>
  #include <stdio.h>
-Line 12 
 been extended and consequently is now ra
+Line 42 
 been extended and consequently is now ra
  #include <stdlib.h>
  #include <time.h>
  #include <locale.h>
+ #include <errno.h>
+ /* A number of things vary for Windows builds. Originally, pcretest opened its
+ input and output without "b"; then I was told that "b" was needed in some
+ environments, so it was added for release 5.0 to both the input and output. (It
+ makes no difference on Unix-like systems.) Later I was told that it is wrong
+ for the input on Windows. I've now abstracted the modes into two macros that
+ are set here, to make it easier to fiddle with them, and removed "b" from the
+ input mode under Windows. */
+ #if defined(_WIN32) || defined(WIN32)
+ #include <io.h>                /* For _setmode() */
+ #include <fcntl.h>             /* For _O_BINARY */
+ #define INPUT_MODE   "r"
+ #define OUTPUT_MODE  "wb"
+ #else
+ #include <sys/time.h>          /* These two includes are needed */
+ #include <sys/resource.h>      /* for setrlimit(). */
+ #define INPUT_MODE   "rb"
+ #define OUTPUT_MODE  "wb"
+ #endif
- /* We need the internal info for displaying the results of pcre_study(). Also
- for getting the opcodes for showing compiled code. */
  #define PCRE_SPY        /* For Win32 build, import data, not export */
- #include "internal.h"
+ /* We include pcre_internal.h because we need the internal info for displaying
+ the results of pcre_study() and we also need to know about the internal
+ macros, structures, and other internal data values; pcretest has "inside
+ information" compared to a program that strictly follows the PCRE API. */
+ #include "pcre_internal.h"
+ /* We need access to the data tables that PCRE uses. So as not to have to keep
+ two copies, we include the source file here, changing the names of the external
+ symbols to prevent clashes. */
+ #define _pcre_utf8_table1      utf8_table1
+ #define _pcre_utf8_table1_size utf8_table1_size
+ #define _pcre_utf8_table2      utf8_table2
+ #define _pcre_utf8_table3      utf8_table3
+ #define _pcre_utf8_table4      utf8_table4
+ #define _pcre_utt              utt
+ #define _pcre_utt_size         utt_size
+ #define _pcre_OP_lengths       OP_lengths
+ #include "pcre_tables.c"
+ /* We also need the pcre_printint() function for printing out compiled
+ patterns. This function is in a separate file so that it can be included in
+ pcre_compile.c when that module is compiled with debugging enabled.
+ The definition of the macro PRINTABLE, which determines whether to print an
+ output character as-is or as a hex value when showing compiled patterns, is
+ contained in this file. We uses it here also, in cases when the locale has not
+ been explicitly changed, so as to get consistent output from systems that
+ differ in their output from isprint() even in the "C" locale. */
+ #include "pcre_printint.src"
+ #define PRINTHEX(c) (locale_set? isprint(c) : PRINTABLE(c))
  /* It is possible to compile this test program without including support for
  testing the POSIX interface, though this is not available via the standard
-Line 27 
 Makefile. */
+Line 114 
 Makefile. */
  #include "pcreposix.h"
  #endif
+ /* It is also possible, for the benefit of the version imported into Exim, to
+ build pcretest without support for UTF8 (define NOUTF8), without the interface
+ to the DFA matcher (NODFA), and without the doublecheck of the old "info"
+ function (define NOINFOCHECK). */
+ /* Other parameters */
  #ifndef CLOCKS_PER_SEC
  #ifdef CLK_TCK
  #define CLOCKS_PER_SEC CLK_TCK
-Line 35 
 Makefile. */
+Line 130 
 Makefile. */
  #endif
  #endif
- #define LOOPREPEAT 50000
+ /* This is the default loop count for timing. */
- #define BUFFER_SIZE 30000
+ #define LOOPREPEAT 500000
- #define DBUFFER_SIZE 1024
+ /* Static variables */
  static FILE *outfile;
  static int log_store = 0;
-Line 48 
 static int callout_extra;
+Line 143 
 static int callout_extra;
  static int callout_fail_count;
  static int callout_fail_id;
  static int first_callout;
+ static int locale_set = 0;
+ static int show_malloc;
  static int use_utf8;
  static size_t gotten_store;
+ /* The buffers grow automatically if very long input lines are encountered. */
- static const int utf8_table1[] = {
+ static int buffer_size = 50000;
-x0000007f, 0x000007ff, 0x0000ffff, 0x001fffff, 0x03ffffff, 0x7fffffff};
+ static uschar *buffer = NULL;
+ static uschar *dbuffer = NULL;
- static const int utf8_table2[] = {
+ static uschar *pbuffer = NULL;
-, 0xc0, 0xe0, 0xf0, 0xf8, 0xfc};
- static const int utf8_table3[] = {
-xff, 0x1f, 0x0f, 0x07, 0x03, 0x01};
  /*************************************************
- *         Print compiled regex                   *
+ *        Read or extend an input line            *
  *************************************************/
- /* The code for doing this is held in a separate file that is also included in
+ /* Input lines are read into buffer, but both patterns and data lines can be
- pcre.c when it is compiled with the debug switch. It defines a function called
+ continued over multiple input lines. In addition, if the buffer fills up, we
- print_internals(), which uses a table of opcode lengths defined by the macro
+ want to automatically expand it so as to be able to handle extremely large
- OP_LENGTHS, whose name must be OP_lengths. */
+ lines that are needed for certain stress tests. When the input buffer is
+ expanded, the other two buffers must also be expanded likewise, and the
+ contents of pbuffer, which are a copy of the input for callouts, must be
+ preserved (for when expansion happens for a data line). This is not the most
+ optimal way of handling this, but hey, this is just a test program!
+ Arguments:
+   f            the file to read
+   start        where in buffer to start (this *must* be within buffer)
+ Returns:       pointer to the start of new data
+                could be a copy of start, or could be moved
+                NULL if no data read and EOF reached
+ */
+ static uschar *
+ extend_inputline(FILE *f, uschar *start)
+ {
+ uschar *here = start;
+ for (;;)
+   {
+   int rlen = buffer_size - (here - buffer);
+   if (rlen > 1000)
+     {
+     int dlen;
+     if (fgets((char *)here, rlen,  f) == NULL)
+       return (here == start)? NULL : start;
+     dlen = (int)strlen((char *)here);
+     if (dlen > 0 && here[dlen - 1] == '\n') return start;
+     here += dlen;
+     }
+   else
+     {
+     int new_buffer_size = 2*buffer_size;
+     uschar *new_buffer = (unsigned char *)malloc(new_buffer_size);
+     uschar *new_dbuffer = (unsigned char *)malloc(new_buffer_size);
+     uschar *new_pbuffer = (unsigned char *)malloc(new_buffer_size);
+     if (new_buffer == NULL || new_dbuffer == NULL || new_pbuffer == NULL)
+       {
+       fprintf(stderr, "pcretest: malloc(%d) failed\n", new_buffer_size);
+       exit(1);
+       }
+     memcpy(new_buffer, buffer, buffer_size);
+     memcpy(new_pbuffer, pbuffer, buffer_size);
+     buffer_size = new_buffer_size;
+     start = new_buffer + (start - buffer);
+     here = new_buffer + (here - buffer);
+     free(buffer);
+     free(dbuffer);
+     free(pbuffer);
+     buffer = new_buffer;
+     dbuffer = new_dbuffer;
+     pbuffer = new_pbuffer;
+     }
+   }
+ return NULL;  /* Control never gets here */
+ }
- static uschar OP_lengths[] = { OP_LENGTHS };
- #include "printint.c"
-Line 84 
 static uschar OP_lengths[] = { OP_LENGTH
+Line 244 
 static uschar OP_lengths[] = { OP_LENGTH
  /* We don't use strtoul() because SunOS4 doesn't have it. Rather than mess
  around with conditional compilation, just do the job by hand. It is only used
- for unpicking the -o argument, so just keep it simple.
+ for unpicking arguments, so just keep it simple.
  Arguments:
    str           string to be converted
-Line 105 
 return(result);
+Line 265 
 return(result);
- /*************************************************
- *       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
-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 += i;
- for (j = i; j > 0; j--)
-  {
-  *buffer-- = 0x80 | (cvalue & 0x3f);
-  cvalue >>= 6;
-  }
- *buffer = utf8_table2[i] | cvalue;
- return i + 1;
- }
  /*************************************************
  *            Convert UTF-8 string to value       *
-Line 150 
 return i + 1;
+Line 274 
 return i + 1;
  and returns the value of the character.
  Argument:
-   buffer   a pointer to the byte vector
+   utf8bytes   a pointer to the byte vector
-   vptr     a pointer to an int to receive the value
+   vptr        a pointer to an int to receive the value
- Returns:   >  0 => the number of bytes consumed
+ Returns:      >  0 => the number of bytes consumed
-            -6 to 0 => malformed UTF-8 character at offset = (-return)
+               -6 to 0 => malformed UTF-8 character at offset = (-return)
  */
+ #if !defined NOUTF8
  static int
- utf82ord(unsigned char *buffer, int *vptr)
+ utf82ord(unsigned char *utf8bytes, int *vptr)
  {
- int c = *buffer++;
+ int c = *utf8bytes++;
  int d = c;
  int i, j, s;
-Line 180 
 d = (c & utf8_table3[i]) << s;
+Line 306 
 d = (c & utf8_table3[i]) << s;
  for (j = 0; j < i; j++)
    {
-   c = *buffer++;
+   c = *utf8bytes++;
    if ((c & 0xc0) != 0x80) return -(j+1);
    s -= 6;
    d |= (c & 0x3f) << s;
-Line 188 
 for (j = 0; j < i; j++)
+Line 314 
 for (j = 0; j < i; j++)
  /* Check that encoding was the correct unique one */
- for (j = 0; j < sizeof(utf8_table1)/sizeof(int); j++)
+ for (j = 0; j < utf8_table1_size; j++)
    if (d <= utf8_table1[j]) break;
  if (j != i) return -(i+1);
-Line 198 
 if (j != i) return -(i+1);
+Line 324 
 if (j != i) return -(i+1);
  return i+1;
  }
+ #endif
+ /*************************************************
+ *       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
+   utf8bytes  pointer to buffer for result - at least 6 bytes long
+ Returns:     number of characters placed in the buffer
+ */
+ #if !defined NOUTF8
+ static int
+ ord2utf8(int cvalue, uschar *utf8bytes)
+ {
+ register int i, j;
+ for (i = 0; i < utf8_table1_size; i++)
+   if (cvalue <= utf8_table1[i]) break;
+ utf8bytes += i;
+ for (j = i; j > 0; j--)
+  {
+  *utf8bytes-- = 0x80 | (cvalue & 0x3f);
+  cvalue >>= 6;
+  }
+ *utf8bytes = utf8_table2[i] | cvalue;
+ return i + 1;
+ }
+ #endif
  /*************************************************
-Line 210 
 chars without printing. */
+Line 374 
 chars without printing. */
  static int pchars(unsigned char *p, int length, FILE *f)
  {
- int c;
+ int c = 0;
  int yield = 0;
  while (length-- > 0)
    {
+ #if !defined NOUTF8
    if (use_utf8)
      {
      int rc = utf82ord(p, &c);
-Line 223 
 while (length-- > 0)
+Line 388 
 while (length-- > 0)
        {
        length -= rc - 1;
        p += rc;
-       if (c < 256 && isprint(c))
+       if (PRINTHEX(c))
          {
          if (f != NULL) fprintf(f, "%c", c);
          yield++;
          }
        else
          {
-         int n;
+         int n = 4;
-         if (f != NULL) fprintf(f, "\\x{%02x}%n", c, &n);
+         if (f != NULL) fprintf(f, "\\x{%02x}", c);
-         yield += n;
+         yield += (n <= 0x000000ff)? 2 :
+                  (n <= 0x00000fff)? 3 :
+                  (n <= 0x0000ffff)? 4 :
+                  (n <= 0x000fffff)? 5 : 6;
          }
        continue;
        }
      }
+ #endif
     /* Not UTF-8, or malformed UTF-8  */
-   if (isprint(c = *(p++)))
+   c = *p++;
+   if (PRINTHEX(c))
      {
      if (f != NULL) fprintf(f, "%c", c);
      yield++;
-Line 268 
 data is not zero. */
+Line 438 
 data is not zero. */
  static int callout(pcre_callout_block *cb)
  {
  FILE *f = (first_callout | callout_extra)? outfile : NULL;
- int i, pre_start, post_start;
+ int i, pre_start, post_start, subject_length;
  if (callout_extra)
    {
-Line 299 
 pre_start = pchars((unsigned char *)cb->
+Line 469 
 pre_start = pchars((unsigned char *)cb->
  post_start = pchars((unsigned char *)(cb->subject + cb->start_match),
    cb->current_position - cb->start_match, f);
+ subject_length = pchars((unsigned char *)cb->subject, cb->subject_length, NULL);
  (void)pchars((unsigned char *)(cb->subject + cb->current_position),
    cb->subject_length - cb->current_position, f);
  if (f != NULL) fprintf(f, "\n");
  /* Always print appropriate indicators, with callout number if not already
- shown */
+ shown. For automatic callouts, show the pattern offset. */
- if (callout_extra) fprintf(outfile, "    ");
+ if (cb->callout_number == 255)
-   else fprintf(outfile, "%3d ", cb->callout_number);
+   {
+   fprintf(outfile, "%+3d ", cb->pattern_position);
+   if (cb->pattern_position > 99) fprintf(outfile, "\n    ");
+   }
+ else
+   {
+   if (callout_extra) fprintf(outfile, "    ");
+     else fprintf(outfile, "%3d ", cb->callout_number);
+   }
  for (i = 0; i < pre_start; i++) fprintf(outfile, " ");
  fprintf(outfile, "^");
-Line 319 
 if (post_start > 0)
+Line 499 
 if (post_start > 0)
    fprintf(outfile, "^");
    }
+ for (i = 0; i < subject_length - pre_start - post_start + 4; i++)
+   fprintf(outfile, " ");
+ fprintf(outfile, "%.*s", (cb->next_item_length == 0)? 1 : cb->next_item_length,
+   pbuffer + cb->pattern_position);
  fprintf(outfile, "\n");
  first_callout = 0;
-Line 338 
 return (cb->callout_number != callout_fa
+Line 524 
 return (cb->callout_number != callout_fa
  /*************************************************
- *            Local malloc function               *
+ *            Local malloc functions              *
  *************************************************/
  /* Alternative malloc function, to test functionality and show the size of the
-Line 346 
 compiled re. */
+Line 532 
 compiled re. */
  static void *new_malloc(size_t size)
  {
+ void *block = malloc(size);
  gotten_store = size;
- return malloc(size);
+ if (show_malloc)
+   fprintf(outfile, "malloc       %3d %p\n", (int)size, block);
+ return block;
+ }
+ static void new_free(void *block)
+ {
+ if (show_malloc)
+   fprintf(outfile, "free             %p\n", block);
+ free(block);
  }
+ /* For recursion malloc/free, to test stacking calls */
+ static void *stack_malloc(size_t size)
+ {
+ void *block = malloc(size);
+ if (show_malloc)
+   fprintf(outfile, "stack_malloc %3d %p\n", (int)size, block);
+ return block;
+ }
+ static void stack_free(void *block)
+ {
+ if (show_malloc)
+   fprintf(outfile, "stack_free       %p\n", block);
+ free(block);
+ }
  /*************************************************
  *          Call pcre_fullinfo()                  *
-Line 368 
 if ((rc = pcre_fullinfo(re, study, optio
+Line 581 
 if ((rc = pcre_fullinfo(re, study, optio
  /*************************************************
+ *         Byte flipping function                 *
+ *************************************************/
+ static unsigned long int
+ byteflip(unsigned long int value, int n)
+ {
+ if (n == 2) return ((value & 0x00ff) << 8) | ((value & 0xff00) >> 8);
+ return ((value & 0x000000ff) << 24) |
+        ((value & 0x0000ff00) <<  8) |
+        ((value & 0x00ff0000) >>  8) |
+        ((value & 0xff000000) >> 24);
+ }
+ /*************************************************
+ *        Check match or recursion limit          *
+ *************************************************/
+ static int
+ check_match_limit(pcre *re, pcre_extra *extra, uschar *bptr, int len,
+   int start_offset, int options, int *use_offsets, int use_size_offsets,
+   int flag, unsigned long int *limit, int errnumber, const char *msg)
+ {
+ int count;
+ int min = 0;
+ int mid = 64;
+ int max = -1;
+ extra->flags |= flag;
+ for (;;)
+   {
+   *limit = mid;
+   count = pcre_exec(re, extra, (char *)bptr, len, start_offset, options,
+     use_offsets, use_size_offsets);
+   if (count == errnumber)
+     {
+     /* fprintf(outfile, "Testing %s limit = %d\n", msg, mid); */
+     min = mid;
+     mid = (mid == max - 1)? max : (max > 0)? (min + max)/2 : mid*2;
+     }
+   else if (count >= 0 || count == PCRE_ERROR_NOMATCH ||
+                          count == PCRE_ERROR_PARTIAL)
+     {
+     if (mid == min + 1)
+       {
+       fprintf(outfile, "Minimum %s limit = %d\n", msg, mid);
+       break;
+       }
+     /* fprintf(outfile, "Testing %s limit = %d\n", msg, mid); */
+     max = mid;
+     mid = (min + mid)/2;
+     }
+   else break;    /* Some other error */
+   }
+ extra->flags &= ~flag;
+ return count;
+ }
+ /*************************************************
+ *         Check newline indicator                *
+ *************************************************/
+ /* This is used both at compile and run-time to check for <xxx> escapes, where
+ xxx is LF, CR, CRLF, or ANY. Print a message and return 0 if there is no match.
+ Arguments:
+   p           points after the leading '<'
+   f           file for error message
+ Returns:      appropriate PCRE_NEWLINE_xxx flags, or 0
+ */
+ static int
+ check_newline(uschar *p, FILE *f)
+ {
+ if (strncmp((char *)p, "cr>", 3) == 0) return PCRE_NEWLINE_CR;
+ if (strncmp((char *)p, "lf>", 3) == 0) return PCRE_NEWLINE_LF;
+ if (strncmp((char *)p, "crlf>", 5) == 0) return PCRE_NEWLINE_CRLF;
+ if (strncmp((char *)p, "any>", 4) == 0) return PCRE_NEWLINE_ANY;
+ fprintf(f, "Unknown newline type at: <%s\n", p);
+ return 0;
+ }
+ /*************************************************
+ *             Usage function                     *
+ *************************************************/
+ static void
+ usage(void)
+ {
+ printf("Usage:     pcretest [options] [<input> [<output>]]\n");
+ printf("  -b       show compiled code (bytecode)\n");
+ printf("  -C       show PCRE compile-time options and exit\n");
+ printf("  -d       debug: show compiled code and information (-b and -i)\n");
+ #if !defined NODFA
+ printf("  -dfa     force DFA matching for all subjects\n");
+ #endif
+ printf("  -help    show usage information\n");
+ printf("  -i       show information about compiled patterns\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("  -q       quiet: do not output PCRE version number at start\n");
+ printf("  -S <n>   set stack size to <n> megabytes\n");
+ printf("  -s       output store (memory) used information\n"
+        "  -t       time compilation and execution\n");
+ printf("  -t <n>   time compilation and execution, repeating <n> times\n");
+ printf("  -tm      time execution (matching) only\n");
+ printf("  -tm <n>  time execution (matching) only, repeating <n> times\n");
+ }
+ /*************************************************
  *                Main Program                    *
  *************************************************/
-Line 382 
 int options = 0;
+Line 722 
 int options = 0;
  int study_options = 0;
  int op = 1;
  int timeit = 0;
+ int timeitm = 0;
  int showinfo = 0;
  int showstore = 0;
+ int quiet = 0;
  int size_offsets = 45;
  int size_offsets_max;
- int *offsets;
+ int *offsets = NULL;
  #if !defined NOPOSIX
  int posix = 0;
  #endif
  int debug = 0;
  int done = 0;
+ int all_use_dfa = 0;
+ int yield = 0;
+ int stack_size;
+ /* These vectors store, end-to-end, a list of captured substring names. Assume
+ that 1024 is plenty long enough for the few names we'll be testing. */
- unsigned char *buffer;
+ uschar copynames[1024];
- unsigned char *dbuffer;
+ uschar getnames[1024];
+ uschar *copynamesptr;
+ uschar *getnamesptr;
  /* Get buffers from malloc() so that Electric Fence will check their misuse
- when I am debugging. */
+ when I am debugging. They grow automatically when very long lines are read. */
- buffer = (unsigned char *)malloc(BUFFER_SIZE);
+ buffer = (unsigned char *)malloc(buffer_size);
- dbuffer = (unsigned char *)malloc(DBUFFER_SIZE);
+ dbuffer = (unsigned char *)malloc(buffer_size);
+ pbuffer = (unsigned char *)malloc(buffer_size);
- /* Static so that new_malloc can use it. */
+ /* The outfile variable is static so that new_malloc can use it. */
  outfile = stdout;
+ /* The following  _setmode() stuff is some Windows magic that tells its runtime
+ library to translate CRLF into a single LF character. At least, that's what
+ I've been told: never having used Windows I take this all on trust. Originally
+ it set 0x8000, but then I was advised that _O_BINARY was better. */
+ #if defined(_WIN32) || defined(WIN32)
+ _setmode( _fileno( stdout ), _O_BINARY );
+ #endif
  /* Scan options */
  while (argc > 1 && argv[op][0] == '-')
-Line 414 
 while (argc > 1 && argv[op][0] == '-')
+Line 775 
 while (argc > 1 && argv[op][0] == '-')
    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], "-q") == 0) quiet = 1;
+   else if (strcmp(argv[op], "-b") == 0) debug = 1;
    else if (strcmp(argv[op], "-i") == 0) showinfo = 1;
    else if (strcmp(argv[op], "-d") == 0) showinfo = debug = 1;
+ #if !defined NODFA
+   else if (strcmp(argv[op], "-dfa") == 0) all_use_dfa = 1;
+ #endif
    else if (strcmp(argv[op], "-o") == 0 && argc > 2 &&
        ((size_offsets = get_value((unsigned char *)argv[op+1], &endptr)),
          *endptr == 0))
-Line 424 
 while (argc > 1 && argv[op][0] == '-')
+Line 789 
 while (argc > 1 && argv[op][0] == '-')
      op++;
      argc--;
      }
+   else if (strcmp(argv[op], "-t") == 0 || strcmp(argv[op], "-tm") == 0)
+     {
+     int both = argv[op][2] == 0;
+     int temp;
+     if (argc > 2 && (temp = get_value((unsigned char *)argv[op+1], &endptr),
+                      *endptr == 0))
+       {
+       timeitm = temp;
+       op++;
+       argc--;
+       }
+     else timeitm = LOOPREPEAT;
+     if (both) timeit = timeitm;
+     }
+   else if (strcmp(argv[op], "-S") == 0 && argc > 2 &&
+       ((stack_size = get_value((unsigned char *)argv[op+1], &endptr)),
+         *endptr == 0))
+     {
+ #if defined(_WIN32) || defined(WIN32)
+     printf("PCRE: -S not supported on this OS\n");
+     exit(1);
+ #else
+     int rc;
+     struct rlimit rlim;
+     getrlimit(RLIMIT_STACK, &rlim);
+     rlim.rlim_cur = stack_size * 1024 * 1024;
+     rc = setrlimit(RLIMIT_STACK, &rlim);
+     if (rc != 0)
+       {
+     printf("PCRE: setrlimit() failed with error %d\n", rc);
+     exit(1);
+       }
+     op++;
+     argc--;
+ #endif
+     }
  #if !defined NOPOSIX
    else if (strcmp(argv[op], "-p") == 0) posix = 1;
  #endif
-Line 434 
 while (argc > 1 && argv[op][0] == '-')
+Line 835 
 while (argc > 1 && argv[op][0] == '-')
      printf("Compiled with\n");
      (void)pcre_config(PCRE_CONFIG_UTF8, &rc);
      printf("  %sUTF-8 support\n", rc? "" : "No ");
+     (void)pcre_config(PCRE_CONFIG_UNICODE_PROPERTIES, &rc);
+     printf("  %sUnicode properties support\n", rc? "" : "No ");
      (void)pcre_config(PCRE_CONFIG_NEWLINE, &rc);
-     printf("  Newline character is %s\n", (rc == '\r')? "CR" : "LF");
+     printf("  Newline sequence is %s\n", (rc == '\r')? "CR" :
+       (rc == '\n')? "LF" : (rc == ('\r'<<8 | '\n'))? "CRLF" :
+       (rc == -1)? "ANY" : "???");
      (void)pcre_config(PCRE_CONFIG_LINK_SIZE, &rc);
      printf("  Internal link size = %d\n", rc);
      (void)pcre_config(PCRE_CONFIG_POSIX_MALLOC_THRESHOLD, &rc);
      printf("  POSIX malloc threshold = %d\n", rc);
      (void)pcre_config(PCRE_CONFIG_MATCH_LIMIT, &rc);
      printf("  Default match limit = %d\n", rc);
+     (void)pcre_config(PCRE_CONFIG_MATCH_LIMIT_RECURSION, &rc);
+     printf("  Default recursion depth limit = %d\n", rc);
+     (void)pcre_config(PCRE_CONFIG_STACKRECURSE, &rc);
+     printf("  Match recursion uses %s\n", rc? "stack" : "heap");
      exit(0);
      }
+   else if (strcmp(argv[op], "-help") == 0 ||
+            strcmp(argv[op], "--help") == 0)
+     {
+     usage();
+     goto EXIT;
+     }
    else
      {
      printf("** Unknown or malformed option %s\n", argv[op]);
-     printf("Usage:   pcretest [-d] [-i] [-o <n>] [-p] [-s] [-t] [<input> [<output>]]\n");
+     usage();
-     printf("  -C     show PCRE compile-time options and exit\n");
+     yield = 1;
-     printf("  -d     debug: show compiled code; implies -i\n"
+     goto EXIT;
-            "  -i     show information about compiled pattern\n"
-            "  -o <n> set size of offsets vector to <n>\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--;
-Line 471 
 if (offsets == NULL)
+Line 878 
 if (offsets == NULL)
    {
    printf("** Failed to get %d bytes of memory for offsets vector\n",
      size_offsets_max * sizeof(int));
-   return 1;
+   yield = 1;
+   goto EXIT;
    }
  /* Sort out the input and output files */
  if (argc > 1)
    {
-   infile = fopen(argv[op], "r");
+   infile = fopen(argv[op], INPUT_MODE);
    if (infile == NULL)
      {
      printf("** Failed to open %s\n", argv[op]);
-     return 1;
+     yield = 1;
+     goto EXIT;
      }
    }
  if (argc > 2)
    {
-   outfile = fopen(argv[op+1], "w");
+   outfile = fopen(argv[op+1], OUTPUT_MODE);
    if (outfile == NULL)
      {
      printf("** Failed to open %s\n", argv[op+1]);
-     return 1;
+     yield = 1;
+     goto EXIT;
      }
    }
  /* Set alternative malloc function */
  pcre_malloc = new_malloc;
+ pcre_free = new_free;
+ pcre_stack_malloc = stack_malloc;
+ pcre_stack_free = stack_free;
- /* Heading line, then prompt for first regex if stdin */
+ /* Heading line unless quiet, then prompt for first regex if stdin */
- fprintf(outfile, "PCRE version %s\n\n", pcre_version());
+ if (!quiet) fprintf(outfile, "PCRE version %s\n\n", pcre_version());
  /* Main loop */
-Line 518 
 while (!done)
+Line 931 
 while (!done)
    const char *error;
    unsigned char *p, *pp, *ppp;
+   unsigned char *to_file = NULL;
    const unsigned char *tables = NULL;
+   unsigned long int true_size, true_study_size = 0;
+   size_t size, regex_gotten_store;
    int do_study = 0;
    int do_debug = debug;
    int do_G = 0;
    int do_g = 0;
    int do_showinfo = showinfo;
    int do_showrest = 0;
-   int erroroffset, len, delimiter;
+   int do_flip = 0;
+   int erroroffset, len, delimiter, poffset;
    use_utf8 = 0;
    if (infile == stdin) printf("  re> ");
-   if (fgets((char *)buffer, BUFFER_SIZE, infile) == NULL) break;
+   if (extend_inputline(infile, buffer) == NULL) break;
    if (infile != stdin) fprintf(outfile, "%s", (char *)buffer);
    fflush(outfile);
-Line 538 
 while (!done)
+Line 955 
 while (!done)
    while (isspace(*p)) p++;
    if (*p == 0) continue;
-   /* Get the delimiter and seek the end of the pattern; if is isn't
+   /* See if the pattern is to be loaded pre-compiled from a file. */
-   complete, read more. */
+   if (*p == '<' && strchr((char *)(p+1), '<') == NULL)
+     {
+     unsigned long int magic, get_options;
+     uschar sbuf[8];
+     FILE *f;
+     p++;
+     pp = p + (int)strlen((char *)p);
+     while (isspace(pp[-1])) pp--;
+     *pp = 0;
+     f = fopen((char *)p, "rb");
+     if (f == NULL)
+       {
+       fprintf(outfile, "Failed to open %s: %s\n", p, strerror(errno));
+       continue;
+       }
+     if (fread(sbuf, 1, 8, f) != 8) goto FAIL_READ;
+     true_size =
+       (sbuf[0] << 24) | (sbuf[1] << 16) | (sbuf[2] << 8) | sbuf[3];
+     true_study_size =
+       (sbuf[4] << 24) | (sbuf[5] << 16) | (sbuf[6] << 8) | sbuf[7];
+     re = (real_pcre *)new_malloc(true_size);
+     regex_gotten_store = gotten_store;
+     if (fread(re, 1, true_size, f) != true_size) goto FAIL_READ;
+     magic = ((real_pcre *)re)->magic_number;
+     if (magic != MAGIC_NUMBER)
+       {
+       if (byteflip(magic, sizeof(magic)) == MAGIC_NUMBER)
+         {
+         do_flip = 1;
+         }
+       else
+         {
+         fprintf(outfile, "Data in %s is not a compiled PCRE regex\n", p);
+         fclose(f);
+         continue;
+         }
+       }
+     fprintf(outfile, "Compiled regex%s loaded from %s\n",
+       do_flip? " (byte-inverted)" : "", p);
+     /* Need to know if UTF-8 for printing data strings */
+     new_info(re, NULL, PCRE_INFO_OPTIONS, &get_options);
+     use_utf8 = (get_options & PCRE_UTF8) != 0;
+     /* Now see if there is any following study data */
+     if (true_study_size != 0)
+       {
+       pcre_study_data *psd;
+       extra = (pcre_extra *)new_malloc(sizeof(pcre_extra) + true_study_size);
+       extra->flags = PCRE_EXTRA_STUDY_DATA;
+       psd = (pcre_study_data *)(((char *)extra) + sizeof(pcre_extra));
+       extra->study_data = psd;
+       if (fread(psd, 1, true_study_size, f) != true_study_size)
+         {
+         FAIL_READ:
+         fprintf(outfile, "Failed to read data from %s\n", p);
+         if (extra != NULL) new_free(extra);
+         if (re != NULL) new_free(re);
+         fclose(f);
+         continue;
+         }
+       fprintf(outfile, "Study data loaded from %s\n", p);
+       do_study = 1;     /* To get the data output if requested */
+       }
+     else fprintf(outfile, "No study data\n");
+     fclose(f);
+     goto SHOW_INFO;
+     }
+   /* In-line pattern (the usual case). Get the delimiter and seek the end of
+   the pattern; if is isn't complete, read more. */
    delimiter = *p++;
-Line 550 
 while (!done)
+Line 1052 
 while (!done)
      }
    pp = p;
+   poffset = p - buffer;
    for(;;)
      {
-Line 560 
 while (!done)
+Line 1063 
 while (!done)
        pp++;
        }
      if (*pp != 0) break;
-     len = BUFFER_SIZE - (pp - buffer);
-     if (len < 256)
-       {
-       fprintf(outfile, "** Expression too long - missing delimiter?\n");
-       goto SKIP_DATA;
-       }
      if (infile == stdin) printf("    > ");
-     if (fgets((char *)pp, len, infile) == NULL)
+     if ((pp = extend_inputline(infile, pp)) == NULL)
        {
        fprintf(outfile, "** Unexpected EOF\n");
        done = 1;
-Line 578 
 while (!done)
+Line 1073 
 while (!done)
      if (infile != stdin) fprintf(outfile, "%s", (char *)pp);
      }
+   /* The buffer may have moved while being extended; reset the start of data
+   pointer to the correct relative point in the buffer. */
+   p = buffer + poffset;
    /* 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 */
+   /* Terminate the pattern at the delimiter, and save a copy of the pattern
+   for callouts. */
    *pp++ = 0;
+   strcpy((char *)pbuffer, (char *)p);
    /* Look for options after final delimiter */
-Line 598 
 while (!done)
+Line 1100 
 while (!done)
      {
      switch (*pp++)
        {
+       case 'f': options |= PCRE_FIRSTLINE; break;
        case 'g': do_g = 1; break;
        case 'i': options |= PCRE_CASELESS; break;
        case 'm': options |= PCRE_MULTILINE; break;
-Line 606 
 while (!done)
+Line 1109 
 while (!done)
        case '+': do_showrest = 1; break;
        case 'A': options |= PCRE_ANCHORED; break;
+       case 'B': do_debug = 1; break;
+       case 'C': options |= PCRE_AUTO_CALLOUT; break;
        case 'D': do_debug = do_showinfo = 1; break;
        case 'E': options |= PCRE_DOLLAR_ENDONLY; break;
+       case 'F': do_flip = 1; break;
        case 'G': do_G = 1; break;
        case 'I': do_showinfo = 1; break;
+       case 'J': options |= PCRE_DUPNAMES; break;
        case 'M': log_store = 1; break;
        case 'N': options |= PCRE_NO_AUTO_CAPTURE; break;
-Line 625 
 while (!done)
+Line 1132 
 while (!done)
        case 'L':
        ppp = pp;
-       while (*ppp != '\n' && *ppp != ' ') ppp++;
+       /* The '\r' test here is so that it works on Windows. */
+       /* The '0' test is just in case this is an unterminated line. */
+       while (*ppp != 0 && *ppp != '\n' && *ppp != '\r' && *ppp != ' ') ppp++;
        *ppp = 0;
        if (setlocale(LC_CTYPE, (const char *)pp) == NULL)
          {
          fprintf(outfile, "** Failed to set locale \"%s\"\n", pp);
          goto SKIP_DATA;
          }
+       locale_set = 1;
        tables = pcre_maketables();
        pp = ppp;
        break;
-       case '\n': case ' ': break;
+       case '>':
+       to_file = pp;
+       while (*pp != 0) pp++;
+       while (isspace(pp[-1])) pp--;
+       *pp = 0;
+       break;
+       case '<':
+         {
+         int x = check_newline(pp, outfile);
+         if (x == 0) goto SKIP_DATA;
+         options |= x;
+         while (*pp++ != '>');
+         }
+       break;
+       case '\r':                      /* So that it works in Windows */
+       case '\n':
+       case ' ':
+       break;
        default:
        fprintf(outfile, "** Unknown option '%c'\n", pp[-1]);
        goto SKIP_DATA;
-Line 652 
 while (!done)
+Line 1182 
 while (!done)
      {
      int rc;
      int cflags = 0;
      if ((options & PCRE_CASELESS) != 0) cflags |= REG_ICASE;
      if ((options & PCRE_MULTILINE) != 0) cflags |= REG_NEWLINE;
+     if ((options & PCRE_DOTALL) != 0) cflags |= REG_DOTALL;
+     if ((options & PCRE_NO_AUTO_CAPTURE) != 0) cflags |= REG_NOSUB;
+     if ((options & PCRE_UTF8) != 0) cflags |= REG_UTF8;
      rc = regcomp(&preg, (char *)p, cflags);
      /* Compilation failed; go back for another re, skipping to blank line
-Line 661 
 while (!done)
+Line 1196 
 while (!done)
      if (rc != 0)
        {
-       (void)regerror(rc, &preg, (char *)buffer, BUFFER_SIZE);
+       (void)regerror(rc, &preg, (char *)buffer, buffer_size);
        fprintf(outfile, "Failed: POSIX code %d: %s\n", rc, buffer);
        goto SKIP_DATA;
        }
-Line 673 
 while (!done)
+Line 1208 
 while (!done)
  #endif  /* !defined NOPOSIX */
      {
-     if (timeit)
+     if (timeit > 0)
        {
        register int i;
        clock_t time_taken;
        clock_t start_time = clock();
-       for (i = 0; i < LOOPREPEAT; i++)
+       for (i = 0; i < timeit; i++)
          {
          re = pcre_compile((char *)p, options, &error, &erroroffset, tables);
          if (re != NULL) free(re);
          }
        time_taken = clock() - start_time;
-       fprintf(outfile, "Compile time %.3f milliseconds\n",
+       fprintf(outfile, "Compile time %.4f milliseconds\n",
-         (((double)time_taken * 1000.0) / (double)LOOPREPEAT) /
+         (((double)time_taken * 1000.0) / (double)timeit) /
            (double)CLOCKS_PER_SEC);
        }
-Line 702 
 while (!done)
+Line 1237 
 while (!done)
          {
          for (;;)
            {
-           if (fgets((char *)buffer, BUFFER_SIZE, infile) == NULL)
+           if (extend_inputline(infile, buffer) == NULL)
              {
              done = 1;
              goto CONTINUE;
-Line 726 
 while (!done)
+Line 1261 
 while (!done)
                sizeof(real_pcre) -
                ((real_pcre *)re)->name_count * ((real_pcre *)re)->name_entry_size));
+     /* Extract the size for possible writing before possibly flipping it,
+     and remember the store that was got. */
+     true_size = ((real_pcre *)re)->size;
+     regex_gotten_store = gotten_store;
+     /* If /S was present, study the regexp to generate additional info to
+     help with the matching. */
+     if (do_study)
+       {
+       if (timeit > 0)
+         {
+         register int i;
+         clock_t time_taken;
+         clock_t start_time = clock();
+         for (i = 0; i < timeit; i++)
+           extra = pcre_study(re, study_options, &error);
+         time_taken = clock() - start_time;
+         if (extra != NULL) free(extra);
+         fprintf(outfile, "  Study time %.4f milliseconds\n",
+           (((double)time_taken * 1000.0) / (double)timeit) /
+             (double)CLOCKS_PER_SEC);
+         }
+       extra = pcre_study(re, study_options, &error);
+       if (error != NULL)
+         fprintf(outfile, "Failed to study: %s\n", error);
+       else if (extra != NULL)
+         true_study_size = ((pcre_study_data *)(extra->study_data))->size;
+       }
+     /* If the 'F' option was present, we flip the bytes of all the integer
+     fields in the regex data block and the study block. This is to make it
+     possible to test PCRE's handling of byte-flipped patterns, e.g. those
+     compiled on a different architecture. */
+     if (do_flip)
+       {
+       real_pcre *rre = (real_pcre *)re;
+       rre->magic_number = byteflip(rre->magic_number, sizeof(rre->magic_number));
+       rre->size = byteflip(rre->size, sizeof(rre->size));
+       rre->options = byteflip(rre->options, sizeof(rre->options));
+       rre->top_bracket = byteflip(rre->top_bracket, sizeof(rre->top_bracket));
+       rre->top_backref = byteflip(rre->top_backref, sizeof(rre->top_backref));
+       rre->first_byte = byteflip(rre->first_byte, sizeof(rre->first_byte));
+       rre->req_byte = byteflip(rre->req_byte, sizeof(rre->req_byte));
+       rre->name_table_offset = byteflip(rre->name_table_offset,
+         sizeof(rre->name_table_offset));
+       rre->name_entry_size = byteflip(rre->name_entry_size,
+         sizeof(rre->name_entry_size));
+       rre->name_count = byteflip(rre->name_count, sizeof(rre->name_count));
+       if (extra != NULL)
+         {
+         pcre_study_data *rsd = (pcre_study_data *)(extra->study_data);
+         rsd->size = byteflip(rsd->size, sizeof(rsd->size));
+         rsd->options = byteflip(rsd->options, sizeof(rsd->options));
+         }
+       }
+     /* Extract information from the compiled data if required */
+     SHOW_INFO:
+     if (do_debug)
+       {
+       fprintf(outfile, "------------------------------------------------------------------\n");
+       pcre_printint(re, outfile);
+       }
      if (do_showinfo)
        {
-       unsigned long int get_options;
+       unsigned long int get_options, all_options;
+ #if !defined NOINFOCHECK
        int old_first_char, old_options, old_count;
+ #endif
        int count, backrefmax, first_char, need_char;
        int nameentrysize, namecount;
        const uschar *nametable;
-       size_t size;
-       if (do_debug)
-         {
-         fprintf(outfile, "------------------------------------------------------------------\n");
-         print_internals(re, outfile);
-         }
        new_info(re, NULL, PCRE_INFO_OPTIONS, &get_options);
        new_info(re, NULL, PCRE_INFO_SIZE, &size);
-Line 751 
 while (!done)
+Line 1351 
 while (!done)
        new_info(re, NULL, PCRE_INFO_NAMECOUNT, &namecount);
        new_info(re, NULL, PCRE_INFO_NAMETABLE, (void *)&nametable);
+ #if !defined NOINFOCHECK
        old_count = pcre_info(re, &old_options, &old_first_char);
        if (count < 0) fprintf(outfile,
          "Error %d from pcre_info()\n", count);
-Line 768 
 while (!done)
+Line 1369 
 while (!done)
            "Options disagreement: pcre_fullinfo=%ld pcre_info=%d\n",
              get_options, old_options);
          }
+ #endif
-       if (size != gotten_store) fprintf(outfile,
+       if (size != regex_gotten_store) fprintf(outfile,
          "Size disagreement: pcre_fullinfo=%d call to malloc for %d\n",
-         size, gotten_store);
+         (int)size, (int)regex_gotten_store);
        fprintf(outfile, "Capturing subpattern count = %d\n", count);
        if (backrefmax > 0)
-Line 789 
 while (!done)
+Line 1391 
 while (!done)
            }
          }
+       /* The NOPARTIAL bit is a private bit in the options, so we have
+       to fish it out via out back door */
+       all_options = ((real_pcre *)re)->options;
+       if (do_flip)
+         {
+         all_options = byteflip(all_options, sizeof(all_options));
+          }
+       if ((all_options & PCRE_NOPARTIAL) != 0)
+         fprintf(outfile, "Partial matching not supported\n");
        if (get_options == 0) fprintf(outfile, "No options\n");
-         else fprintf(outfile, "Options:%s%s%s%s%s%s%s%s%s%s\n",
+         else fprintf(outfile, "Options:%s%s%s%s%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_FIRSTLINE) != 0)? " firstline" : "",
            ((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_NO_AUTO_CAPTURE) != 0)? " no_auto_capture" : "",
            ((get_options & PCRE_UTF8) != 0)? " utf8" : "",
-           ((get_options & PCRE_NO_UTF8_CHECK) != 0)? " no_utf8_check" : "");
+           ((get_options & PCRE_NO_UTF8_CHECK) != 0)? " no_utf8_check" : "",
+           ((get_options & PCRE_DUPNAMES) != 0)? " dupnames" : "");
-       if (((((real_pcre *)re)->options) & PCRE_ICHANGED) != 0)
+       switch (get_options & PCRE_NEWLINE_BITS)
-         fprintf(outfile, "Case state changes\n");
+         {
+         case PCRE_NEWLINE_CR:
+         fprintf(outfile, "Forced newline sequence: CR\n");
+         break;
+         case PCRE_NEWLINE_LF:
+         fprintf(outfile, "Forced newline sequence: LF\n");
+         break;
+         case PCRE_NEWLINE_CRLF:
+         fprintf(outfile, "Forced newline sequence: CRLF\n");
+         break;
+         case PCRE_NEWLINE_ANY:
+         fprintf(outfile, "Forced newline sequence: ANY\n");
+         break;
+         default:
+         break;
+         }
        if (first_char == -1)
          {
-         fprintf(outfile, "First char at start or follows \\n\n");
+         fprintf(outfile, "First char at start or follows newline\n");
          }
        else if (first_char < 0)
          {
-Line 818 
 while (!done)
+Line 1454 
 while (!done)
          int ch = first_char & 255;
          const char *caseless = ((first_char & REQ_CASELESS) == 0)?
            "" : " (caseless)";
-         if (isprint(ch))
+         if (PRINTHEX(ch))
            fprintf(outfile, "First char = \'%c\'%s\n", ch, caseless);
          else
            fprintf(outfile, "First char = %d%s\n", ch, caseless);
-Line 833 
 while (!done)
+Line 1469 
 while (!done)
          int ch = need_char & 255;
          const char *caseless = ((need_char & REQ_CASELESS) == 0)?
            "" : " (caseless)";
-         if (isprint(ch))
+         if (PRINTHEX(ch))
            fprintf(outfile, "Need char = \'%c\'%s\n", ch, caseless);
          else
            fprintf(outfile, "Need char = %d%s\n", ch, caseless);
          }
-       }
-     /* If /S was present, study the regexp to generate additional info to
-     help with the matching. */
-     if (do_study)
-       {
-       if (timeit)
-         {
-         register int i;
-         clock_t time_taken;
-         clock_t start_time = clock();
-         for (i = 0; i < LOOPREPEAT; i++)
-           extra = pcre_study(re, study_options, &error);
-         time_taken = clock() - start_time;
-         if (extra != NULL) free(extra);
-         fprintf(outfile, "  Study time %.3f milliseconds\n",
-           (((double)time_taken * 1000.0) / (double)LOOPREPEAT) /
-             (double)CLOCKS_PER_SEC);
-         }
-       extra = pcre_study(re, study_options, &error);
-       if (error != NULL)
-         fprintf(outfile, "Failed to study: %s\n", error);
-       else if (extra == NULL)
-         fprintf(outfile, "Study returned NULL\n");
        /* Don't output study size; at present it is in any case a fixed
        value, but it varies, depending on the computer architecture, and
-       so messes up the test suite. */
+       so messes up the test suite. (And with the /F option, it might be
+       flipped.) */
-       else if (do_showinfo)
+       if (do_study)
          {
-         size_t size;
+         if (extra == NULL)
-         uschar *start_bits = NULL;
+           fprintf(outfile, "Study returned NULL\n");
-         new_info(re, extra, PCRE_INFO_STUDYSIZE, &size);
-         new_info(re, extra, PCRE_INFO_FIRSTTABLE, &start_bits);
-         /* fprintf(outfile, "Study size = %d\n", size); */
-         if (start_bits == NULL)
-           fprintf(outfile, "No starting character set\n");
          else
            {
-           int i;
+           uschar *start_bits = NULL;
-           int c = 24;
+           new_info(re, extra, PCRE_INFO_FIRSTTABLE, &start_bits);
-           fprintf(outfile, "Starting character set: ");
-           for (i = 0; i < 256; i++)
+           if (start_bits == NULL)
+             fprintf(outfile, "No starting byte set\n");
+           else
              {
-             if ((start_bits[i/8] & (1<<(i%8))) != 0)
+             int i;
+             int c = 24;
+             fprintf(outfile, "Starting byte set: ");
+             for (i = 0; i < 256; i++)
                {
-               if (c > 75)
+               if ((start_bits[i/8] & (1<<(i&7))) != 0)
-                 {
-                 fprintf(outfile, "\n  ");
-                 c = 2;
-                 }
-               if (isprint(i) && i != ' ')
-                 {
-                 fprintf(outfile, "%c ", i);
-                 c += 2;
-                 }
-               else
                  {
-                 fprintf(outfile, "\\x%02x ", i);
+                 if (c > 75)
-                 c += 5;
+                   {
+                   fprintf(outfile, "\n  ");
+                   c = 2;
+                   }
+                 if (PRINTHEX(i) && i != ' ')
+                   {
+                   fprintf(outfile, "%c ", i);
+                   c += 2;
+                   }
+                 else
+                   {
+                   fprintf(outfile, "\\x%02x ", i);
+                   c += 5;
+                   }
                  }
                }
+             fprintf(outfile, "\n");
              }
-           fprintf(outfile, "\n");
            }
          }
        }
-     }
+     /* If the '>' option was present, we write out the regex to a file, and
+     that is all. The first 8 bytes of the file are the regex length and then
+     the study length, in big-endian order. */
+     if (to_file != NULL)
+       {
+       FILE *f = fopen((char *)to_file, "wb");
+       if (f == NULL)
+         {
+         fprintf(outfile, "Unable to open %s: %s\n", to_file, strerror(errno));
+         }
+       else
+         {
+         uschar sbuf[8];
+         sbuf[0] = (true_size >> 24)  & 255;
+         sbuf[1] = (true_size >> 16)  & 255;
+         sbuf[2] = (true_size >>  8)  & 255;
+         sbuf[3] = (true_size)  & 255;
+         sbuf[4] = (true_study_size >> 24)  & 255;
+         sbuf[5] = (true_study_size >> 16)  & 255;
+         sbuf[6] = (true_study_size >>  8)  & 255;
+         sbuf[7] = (true_study_size)  & 255;
+         if (fwrite(sbuf, 1, 8, f) < 8 ||
+             fwrite(re, 1, true_size, f) < true_size)
+           {
+           fprintf(outfile, "Write error on %s: %s\n", to_file, strerror(errno));
+           }
+         else
+           {
+           fprintf(outfile, "Compiled regex written to %s\n", to_file);
+           if (extra != NULL)
+             {
+             if (fwrite(extra->study_data, 1, true_study_size, f) <
+                 true_study_size)
+               {
+               fprintf(outfile, "Write error on %s: %s\n", to_file,
+                 strerror(errno));
+               }
+             else fprintf(outfile, "Study data written to %s\n", to_file);
+             }
+           }
+         fclose(f);
+         }
+       new_free(re);
+       if (extra != NULL) new_free(extra);
+       if (tables != NULL) new_free((void *)tables);
+       continue;  /* With next regex */
+       }
+     }        /* End of non-POSIX compile */
    /* Read data lines and test them */
    for (;;)
      {
-     unsigned char *q;
+     uschar *q;
-     unsigned char *bptr = dbuffer;
+     uschar *bptr = dbuffer;
      int *use_offsets = offsets;
      int use_size_offsets = size_offsets;
      int callout_data = 0;
-Line 928 
 while (!done)
+Line 1595 
 while (!done)
      int gmatched = 0;
      int start_offset = 0;
      int g_notempty = 0;
+     int use_dfa = 0;
      options = 0;
+     *copynames = 0;
+     *getnames = 0;
+     copynamesptr = copynames;
+     getnamesptr = getnames;
      pcre_callout = callout;
      first_callout = 1;
      callout_extra = 0;
      callout_count = 0;
      callout_fail_count = 999999;
      callout_fail_id = -1;
+     show_malloc = 0;
+     if (extra != NULL) extra->flags &=
+       ~(PCRE_EXTRA_MATCH_LIMIT|PCRE_EXTRA_MATCH_LIMIT_RECURSION);
-     if (infile == stdin) printf("data> ");
+     len = 0;
-     if (fgets((char *)buffer, BUFFER_SIZE, infile) == NULL)
+     for (;;)
        {
-       done = 1;
+       if (infile == stdin) printf("data> ");
-       goto CONTINUE;
+       if (extend_inputline(infile, buffer + len) == NULL)
+         {
+         if (len > 0) break;
+         done = 1;
+         goto CONTINUE;
+         }
+       if (infile != stdin) fprintf(outfile, "%s", (char *)buffer);
+       len = (int)strlen((char *)buffer);
+       if (buffer[len-1] == '\n') break;
        }
-     if (infile != stdin) fprintf(outfile, "%s", (char *)buffer);
-     len = (int)strlen((char *)buffer);
      while (len > 0 && isspace(buffer[len-1])) len--;
      buffer[len] = 0;
      if (len == 0) break;
-Line 976 
 while (!done)
+Line 1660 
 while (!done)
          c -= '0';
          while (i++ < 2 && isdigit(*p) && *p != '8' && *p != '9')
            c = c * 8 + *p++ - '0';
+ #if !defined NOUTF8
+         if (use_utf8 && c > 255)
+           {
+           unsigned char buff8[8];
+           int ii, utn;
+           utn = ord2utf8(c, buff8);
+           for (ii = 0; ii < utn - 1; ii++) *q++ = buff8[ii];
+           c = buff8[ii];   /* Last byte */
+           }
+ #endif
          break;
          case 'x':
          /* Handle \x{..} specially - new Perl thing for utf8 */
+ #if !defined NOUTF8
          if (*p == '{')
            {
            unsigned char *pt = p;
-Line 1000 
 while (!done)
+Line 1696 
 while (!done)
              }
            /* Not correct form; fall through */
            }
+ #endif
          /* Ordinary \x */
-Line 1011 
 while (!done)
+Line 1708 
 while (!done)
            }
          break;
-         case 0:   /* Allows for an empty line */
+         case 0:   /* \ followed by EOF allows for an empty line */
          p--;
          continue;
+         case '>':
+         while(isdigit(*p)) start_offset = start_offset * 10 + *p++ - '0';
+         continue;
          case 'A':  /* Option setting */
          options |= PCRE_ANCHORED;
          continue;
-Line 1031 
 while (!done)
+Line 1732 
 while (!done)
            }
          else if (isalnum(*p))
            {
-           uschar name[256];
+           uschar *npp = copynamesptr;
-           uschar *npp = name;
            while (isalnum(*p)) *npp++ = *p++;
+           *npp++ = 0;
            *npp = 0;
-           n = pcre_get_stringnumber(re, (char *)name);
+           n = pcre_get_stringnumber(re, (char *)copynamesptr);
            if (n < 0)
-             fprintf(outfile, "no parentheses with name \"%s\"\n", name);
+             fprintf(outfile, "no parentheses with name \"%s\"\n", copynamesptr);
-           else copystrings |= 1 << n;
+           copynamesptr = npp;
            }
          else if (*p == '+')
            {
-Line 1076 
 while (!done)
+Line 1777 
 while (!done)
            }
          continue;
+ #if !defined NODFA
+         case 'D':
+ #if !defined NOPOSIX
+         if (posix || do_posix)
+           printf("** Can't use dfa matching in POSIX mode: \\D ignored\n");
+         else
+ #endif
+           use_dfa = 1;
+         continue;
+         case 'F':
+         options |= PCRE_DFA_SHORTEST;
+         continue;
+ #endif
          case 'G':
          if (isdigit(*p))
            {
-Line 1084 
 while (!done)
+Line 1800 
 while (!done)
            }
          else if (isalnum(*p))
            {
-           uschar name[256];
+           uschar *npp = getnamesptr;
-           uschar *npp = name;
            while (isalnum(*p)) *npp++ = *p++;
+           *npp++ = 0;
            *npp = 0;
-           n = pcre_get_stringnumber(re, (char *)name);
+           n = pcre_get_stringnumber(re, (char *)getnamesptr);
            if (n < 0)
-             fprintf(outfile, "no parentheses with name \"%s\"\n", name);
+             fprintf(outfile, "no parentheses with name \"%s\"\n", getnamesptr);
-           else getstrings |= 1 << n;
+           getnamesptr = npp;
            }
          continue;
-Line 1118 
 while (!done)
+Line 1834 
 while (!done)
              {
              printf("** Failed to get %d bytes of memory for offsets vector\n",
                size_offsets_max * sizeof(int));
-             return 1;
+             yield = 1;
+             goto EXIT;
              }
            }
          use_size_offsets = n;
          if (n == 0) use_offsets = NULL;   /* Ensures it can't write to it */
          continue;
+         case 'P':
+         options |= PCRE_PARTIAL;
+         continue;
+         case 'Q':
+         while(isdigit(*p)) n = n * 10 + *p++ - '0';
+         if (extra == NULL)
+           {
+           extra = (pcre_extra *)malloc(sizeof(pcre_extra));
+           extra->flags = 0;
+           }
+         extra->flags |= PCRE_EXTRA_MATCH_LIMIT_RECURSION;
+         extra->match_limit_recursion = n;
+         continue;
+         case 'q':
+         while(isdigit(*p)) n = n * 10 + *p++ - '0';
+         if (extra == NULL)
+           {
+           extra = (pcre_extra *)malloc(sizeof(pcre_extra));
+           extra->flags = 0;
+           }
+         extra->flags |= PCRE_EXTRA_MATCH_LIMIT;
+         extra->match_limit = n;
+         continue;
+ #if !defined NODFA
+         case 'R':
+         options |= PCRE_DFA_RESTART;
+         continue;
+ #endif
+         case 'S':
+         show_malloc = 1;
+         continue;
          case 'Z':
          options |= PCRE_NOTEOL;
          continue;
-Line 1132 
 while (!done)
+Line 1885 
 while (!done)
          case '?':
          options |= PCRE_NO_UTF8_CHECK;
          continue;
+         case '<':
+           {
+           int x = check_newline(p, outfile);
+           if (x == 0) goto NEXT_DATA;
+           options |= x;
+           while (*p++ != '>');
+           }
+         continue;
          }
        *q++ = c;
        }
      *q = 0;
      len = q - dbuffer;
+     if ((all_use_dfa || use_dfa) && find_match_limit)
+       {
+       printf("**Match limit not relevant for DFA matching: ignored\n");
+       find_match_limit = 0;
+       }
      /* Handle matching via the POSIX interface, which does not
      support timing or playing with the match limit or callout data. */
-Line 1156 
 while (!done)
+Line 1924 
 while (!done)
        if (rc != 0)
          {
-         (void)regerror(rc, &preg, (char *)buffer, BUFFER_SIZE);
+         (void)regerror(rc, &preg, (char *)buffer, buffer_size);
          fprintf(outfile, "No match: POSIX code %d: %s\n", rc, buffer);
          }
+       else if ((((const pcre *)preg.re_pcre)->options & PCRE_NO_AUTO_CAPTURE)
+               != 0)
+         {
+         fprintf(outfile, "Matched with REG_NOSUB\n");
+         }
        else
          {
          size_t i;
-Line 1190 
 while (!done)
+Line 1963 
 while (!done)
      for (;; gmatched++)    /* Loop for /g or /G */
        {
-       if (timeit)
+       if (timeitm > 0)
          {
          register int i;
          clock_t time_taken;
          clock_t start_time = clock();
-         for (i = 0; i < LOOPREPEAT; i++)
+ #if !defined NODFA
+         if (all_use_dfa || use_dfa)
+           {
+           int workspace[1000];
+           for (i = 0; i < timeitm; i++)
+             count = pcre_dfa_exec(re, NULL, (char *)bptr, len, start_offset,
+               options | g_notempty, use_offsets, use_size_offsets, workspace,
+               sizeof(workspace)/sizeof(int));
+           }
+         else
+ #endif
+         for (i = 0; i < timeitm; i++)
            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",
+         fprintf(outfile, "Execute time %.4f milliseconds\n",
-           (((double)time_taken * 1000.0) / (double)LOOPREPEAT) /
+           (((double)time_taken * 1000.0) / (double)timeitm) /
              (double)CLOCKS_PER_SEC);
          }
        /* If find_match_limit is set, we want to do repeated matches with
-       varying limits in order to find the minimum value. */
+       varying limits in order to find the minimum value for the match limit and
+       for the recursion limit. */
        if (find_match_limit)
          {
-         int min = 0;
-         int mid = 64;
-         int max = -1;
          if (extra == NULL)
            {
            extra = (pcre_extra *)malloc(sizeof(pcre_extra));
            extra->flags = 0;
            }
-         extra->flags |= PCRE_EXTRA_MATCH_LIMIT;
-         for (;;)
+         (void)check_match_limit(re, extra, bptr, len, start_offset,
-           {
+           options|g_notempty, use_offsets, use_size_offsets,
-           extra->match_limit = mid;
+           PCRE_EXTRA_MATCH_LIMIT, &(extra->match_limit),
-           count = pcre_exec(re, extra, (char *)bptr, len, start_offset,
+           PCRE_ERROR_MATCHLIMIT, "match()");
-             options | g_notempty, use_offsets, use_size_offsets);
-           if (count == PCRE_ERROR_MATCHLIMIT)
+         count = check_match_limit(re, extra, bptr, len, start_offset,
-             {
+           options|g_notempty, use_offsets, use_size_offsets,
-             /* fprintf(outfile, "Testing match limit = %d\n", mid); */
+           PCRE_EXTRA_MATCH_LIMIT_RECURSION, &(extra->match_limit_recursion),
-             min = mid;
+           PCRE_ERROR_RECURSIONLIMIT, "match() recursion");
-             mid = (mid == max - 1)? max : (max > 0)? (min + max)/2 : mid*2;
-             }
-           else if (count >= 0 || count == PCRE_ERROR_NOMATCH)
-             {
-             if (mid == min + 1)
-               {
-               fprintf(outfile, "Minimum match limit = %d\n", mid);
-               break;
-               }
-             /* fprintf(outfile, "Testing match limit = %d\n", mid); */
-             max = mid;
-             mid = (min + mid)/2;
-             }
-           else break;    /* Some other error */
-           }
-         extra->flags &= ~PCRE_EXTRA_MATCH_LIMIT;
          }
        /* If callout_data is set, use the interface with additional data */
-Line 1267 
 while (!done)
+Line 2033 
 while (!done)
        /* The normal case is just to do the match once, with the default
        value of match_limit. */
-       else count = pcre_exec(re, extra, (char *)bptr, len,
+ #if !defined NODFA
-         start_offset, options | g_notempty, use_offsets, use_size_offsets);
+       else if (all_use_dfa || use_dfa)
+         {
+         int workspace[1000];
+         count = pcre_dfa_exec(re, NULL, (char *)bptr, len, start_offset,
+           options | g_notempty, use_offsets, use_size_offsets, workspace,
+           sizeof(workspace)/sizeof(int));
+         if (count == 0)
+           {
+           fprintf(outfile, "Matched, but too many subsidiary matches\n");
+           count = use_size_offsets/2;
+           }
+         }
+ #endif
-       if (count == 0)
+       else
          {
-         fprintf(outfile, "Matched, but too many substrings\n");
+         count = pcre_exec(re, extra, (char *)bptr, len,
-         count = use_size_offsets/3;
+           start_offset, options | g_notempty, use_offsets, use_size_offsets);
+         if (count == 0)
+           {
+           fprintf(outfile, "Matched, but too many substrings\n");
+           count = use_size_offsets/3;
+           }
          }
        /* Matched */
        if (count >= 0)
          {
-         int i;
+         int i, maxcount;
+ #if !defined NODFA
+         if (all_use_dfa || use_dfa) maxcount = use_size_offsets/2; else
+ #endif
+           maxcount = use_size_offsets/3;
+         /* This is a check against a lunatic return value. */
+         if (count > maxcount)
+           {
+           fprintf(outfile,
+             "** PCRE error: returned count %d is too big for offset size %d\n",
+             count, use_size_offsets);
+           count = use_size_offsets/3;
+           if (do_g || do_G)
+             {
+             fprintf(outfile, "** /%c loop abandoned\n", do_g? 'g' : 'G');
+             do_g = do_G = FALSE;        /* Break g/G loop */
+             }
+           }
          for (i = 0; i < count * 2; i += 2)
            {
            if (use_offsets[i] < 0)
-Line 1308 
 while (!done)
+Line 2112 
 while (!done)
            {
            if ((copystrings & (1 << i)) != 0)
              {
-             char copybuffer[16];
+             char copybuffer[256];
              int rc = pcre_copy_substring((char *)bptr, use_offsets, count,
                i, copybuffer, sizeof(copybuffer));
              if (rc < 0)
-Line 1318 
 while (!done)
+Line 2122 
 while (!done)
              }
            }
+         for (copynamesptr = copynames;
+              *copynamesptr != 0;
+              copynamesptr += (int)strlen((char*)copynamesptr) + 1)
+           {
+           char copybuffer[256];
+           int rc = pcre_copy_named_substring(re, (char *)bptr, use_offsets,
+             count, (char *)copynamesptr, copybuffer, sizeof(copybuffer));
+           if (rc < 0)
+             fprintf(outfile, "copy substring %s failed %d\n", copynamesptr, rc);
+           else
+             fprintf(outfile, "  C %s (%d) %s\n", copybuffer, rc, copynamesptr);
+           }
          for (i = 0; i < 32; i++)
            {
            if ((getstrings & (1 << i)) != 0)
-Line 1330 
 while (!done)
+Line 2147 
 while (!done)
              else
                {
                fprintf(outfile, "%2dG %s (%d)\n", i, substring, rc);
-               /* free((void *)substring); */
                pcre_free_substring(substring);
                }
              }
            }
+         for (getnamesptr = getnames;
+              *getnamesptr != 0;
+              getnamesptr += (int)strlen((char*)getnamesptr) + 1)
+           {
+           const char *substring;
+           int rc = pcre_get_named_substring(re, (char *)bptr, use_offsets,
+             count, (char *)getnamesptr, &substring);
+           if (rc < 0)
+             fprintf(outfile, "copy substring %s failed %d\n", getnamesptr, rc);
+           else
+             {
+             fprintf(outfile, "  G %s (%d) %s\n", substring, rc, getnamesptr);
+             pcre_free_substring(substring);
+             }
+           }
          if (getlist)
            {
            const char **stringlist;
-Line 1355 
 while (!done)
+Line 2187 
 while (!done)
            }
          }
+       /* There was a partial match */
+       else if (count == PCRE_ERROR_PARTIAL)
+         {
+         fprintf(outfile, "Partial match");
+ #if !defined NODFA
+         if ((all_use_dfa || use_dfa) && use_size_offsets > 2)
+           fprintf(outfile, ": %.*s", use_offsets[1] - use_offsets[0],
+             bptr + use_offsets[0]);
+ #endif
+         fprintf(outfile, "\n");
+         break;  /* Out of the /g loop */
+         }
        /* 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
+       We want to advance the start offset, and continue. In the case of UTF-8
-       values to achieve this. We won't be at the end of the string - that
+       matching, the advance must be one character, not one byte. Fudge the
-       was checked before setting g_notempty. */
+       offset values to achieve this. We won't be at the end of the string -
+       that was checked before setting g_notempty. */
        else
          {
          if (g_notempty != 0)
            {
+           int onechar = 1;
            use_offsets[0] = start_offset;
-           use_offsets[1] = start_offset + 1;
+           if (use_utf8)
+             {
+             while (start_offset + onechar < len)
+               {
+               int tb = bptr[start_offset+onechar];
+               if (tb <= 127) break;
+               tb &= 0xc0;
+               if (tb != 0 && tb != 0xc0) onechar++;
+               }
+             }
+           use_offsets[1] = start_offset + onechar;
            }
          else
            {
-           if (gmatched == 0)   /* Error if no previous matches */
+           if (count == PCRE_ERROR_NOMATCH)
              {
-             if (count == -1) fprintf(outfile, "No match\n");
+             if (gmatched == 0) fprintf(outfile, "No match\n");
-               else fprintf(outfile, "Error %d\n", count);
              }
+           else fprintf(outfile, "Error %d\n", count);
            break;  /* Out of the /g loop */
            }
          }
-Line 1409 
 while (!done)
+Line 2267 
 while (!done)
          len -= use_offsets[1];
          }
        }  /* End of loop for /g and /G */
+     NEXT_DATA: continue;
      }    /* End of loop for data lines */
    CONTINUE:
-Line 1417 
 while (!done)
+Line 2277 
 while (!done)
    if (posix || do_posix) regfree(&preg);
  #endif
-   if (re != NULL) free(re);
+   if (re != NULL) new_free(re);
-   if (extra != NULL) free(extra);
+   if (extra != NULL) new_free(extra);
    if (tables != NULL)
      {
-     free((void *)tables);
+     new_free((void *)tables);
      setlocale(LC_CTYPE, "C");
+     locale_set = 0;
      }
    }
- fprintf(outfile, "\n");
+ if (infile == stdin) fprintf(outfile, "\n");
- return 0;
+ EXIT:
+ if (infile != NULL && infile != stdin) fclose(infile);
+ if (outfile != NULL && outfile != stdout) fclose(outfile);
+ free(buffer);
+ free(dbuffer);
+ free(pbuffer);
+ free(offsets);
+ return yield;
  }
- /* End */
+ /* End of pcretest.c */

 Legend:



Removed from v.71
 


changed lines


 
Added in v.93
 Legend:



Removed from v.71
 


changed lines


 
Added in v.93
-Removed from v.71
+Added in v.93

	ViewVC Help
Powered by ViewVC 1.1.5