/[pcre]/code/trunk/pcre_compile.c
ViewVC logotype

Diff of /code/trunk/pcre_compile.c

Parent Directory Parent Directory | Revision Log Revision Log | View Patch Patch

revision 216 by ph10, Wed Aug 15 14:35:57 2007 UTC revision 230 by ph10, Mon Sep 10 13:23:56 2007 UTC
# Line 283  static const char *error_texts[] = { Line 283  static const char *error_texts[] = {
283    "(*VERB) with an argument is not supported",    "(*VERB) with an argument is not supported",
284    /* 60 */    /* 60 */
285    "(*VERB) not recognized",    "(*VERB) not recognized",
286    "number is too big"    "number is too big"
287  };  };
288    
289    
# Line 524  else Line 524  else
524      c = 0;      c = 0;
525      while ((digitab[ptr[1]] & ctype_digit) != 0)      while ((digitab[ptr[1]] & ctype_digit) != 0)
526        c = c * 10 + *(++ptr) - '0';        c = c * 10 + *(++ptr) - '0';
527    
528      if (c < 0)      if (c < 0)
529        {        {
530        *errorcodeptr = ERR61;        *errorcodeptr = ERR61;
531        break;        break;
532        }        }
533    
534      if (c == 0 || (braced && *(++ptr) != '}'))      if (c == 0 || (braced && *(++ptr) != '}'))
535        {        {
# Line 574  else Line 574  else
574        if (c < 0)        if (c < 0)
575          {          {
576          *errorcodeptr = ERR61;          *errorcodeptr = ERR61;
577          break;          break;
578          }          }
579        if (c < 10 || c <= bracount)        if (c < 10 || c <= bracount)
580          {          {
581          c = -(ESC_REF + c);          c = -(ESC_REF + c);
# Line 950  for (; *ptr != 0; ptr++) Line 950  for (; *ptr != 0; ptr++)
950      {      {
951      while (*(++ptr) != ']')      while (*(++ptr) != ']')
952        {        {
953        if (*ptr == 0) return -1;        if (*ptr == 0) return -1;
954        if (*ptr == '\\')        if (*ptr == '\\')
955          {          {
956          if (*(++ptr) == 0) return -1;          if (*(++ptr) == 0) return -1;
# Line 1106  for (;;) Line 1106  for (;;)
1106    {    {
1107    int d;    int d;
1108    register int op = *cc;    register int op = *cc;
   
1109    switch (op)    switch (op)
1110      {      {
1111      case OP_CBRA:      case OP_CBRA:
# Line 1195  for (;;) Line 1194  for (;;)
1194    
1195      case OP_TYPEEXACT:      case OP_TYPEEXACT:
1196      branchlength += GET2(cc,1);      branchlength += GET2(cc,1);
1197        if (cc[3] == OP_PROP || cc[3] == OP_NOTPROP) cc += 2;
1198      cc += 4;      cc += 4;
1199      break;      break;
1200    
# Line 1303  for (;;) Line 1303  for (;;)
1303      code += _pcre_OP_lengths[c];      code += _pcre_OP_lengths[c];
1304      }      }
1305    
1306    /* In UTF-8 mode, opcodes that are followed by a character may be followed by    /* Otherwise, we can get the item's length from the table, except that for
1307    a multi-byte character. The length in the table is a minimum, so we have to    repeated character types, we have to test for \p and \P, which have an extra
1308    arrange to skip the extra bytes. */    two bytes of parameters. */
1309    
1310    else    else
1311      {      {
1312        switch(c)
1313          {
1314          case OP_TYPESTAR:
1315          case OP_TYPEMINSTAR:
1316          case OP_TYPEPLUS:
1317          case OP_TYPEMINPLUS:
1318          case OP_TYPEQUERY:
1319          case OP_TYPEMINQUERY:
1320          case OP_TYPEPOSSTAR:
1321          case OP_TYPEPOSPLUS:
1322          case OP_TYPEPOSQUERY:
1323          if (code[1] == OP_PROP || code[1] == OP_NOTPROP) code += 2;
1324          break;
1325    
1326          case OP_TYPEUPTO:
1327          case OP_TYPEMINUPTO:
1328          case OP_TYPEEXACT:
1329          case OP_TYPEPOSUPTO:
1330          if (code[3] == OP_PROP || code[3] == OP_NOTPROP) code += 2;
1331          break;
1332          }
1333    
1334        /* Add in the fixed length from the table */
1335    
1336      code += _pcre_OP_lengths[c];      code += _pcre_OP_lengths[c];
1337    
1338      /* In UTF-8 mode, opcodes that are followed by a character may be followed by
1339      a multi-byte character. The length in the table is a minimum, so we have to
1340      arrange to skip the extra bytes. */
1341    
1342  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
1343      if (utf8) switch(c)      if (utf8) switch(c)
1344        {        {
# Line 1367  for (;;) Line 1396  for (;;)
1396    
1397    if (c == OP_XCLASS) code += GET(code, 1);    if (c == OP_XCLASS) code += GET(code, 1);
1398    
1399    /* Otherwise, we get the item's length from the table. In UTF-8 mode, opcodes    /* Otherwise, we can get the item's length from the table, except that for
1400    that are followed by a character may be followed by a multi-byte character.    repeated character types, we have to test for \p and \P, which have an extra
1401    The length in the table is a minimum, so we have to arrange to skip the extra    two bytes of parameters. */
   bytes. */  
1402    
1403    else    else
1404      {      {
1405        switch(c)
1406          {
1407          case OP_TYPESTAR:
1408          case OP_TYPEMINSTAR:
1409          case OP_TYPEPLUS:
1410          case OP_TYPEMINPLUS:
1411          case OP_TYPEQUERY:
1412          case OP_TYPEMINQUERY:
1413          case OP_TYPEPOSSTAR:
1414          case OP_TYPEPOSPLUS:
1415          case OP_TYPEPOSQUERY:
1416          if (code[1] == OP_PROP || code[1] == OP_NOTPROP) code += 2;
1417          break;
1418    
1419          case OP_TYPEPOSUPTO:
1420          case OP_TYPEUPTO:
1421          case OP_TYPEMINUPTO:
1422          case OP_TYPEEXACT:
1423          if (code[3] == OP_PROP || code[3] == OP_NOTPROP) code += 2;
1424          break;
1425          }
1426    
1427        /* Add in the fixed length from the table */
1428    
1429      code += _pcre_OP_lengths[c];      code += _pcre_OP_lengths[c];
1430    
1431        /* In UTF-8 mode, opcodes that are followed by a character may be followed
1432        by a multi-byte character. The length in the table is a minimum, so we have
1433        to arrange to skip the extra bytes. */
1434    
1435  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
1436      if (utf8) switch(c)      if (utf8) switch(c)
1437        {        {
# Line 1473  for (code = first_significant_code(code Line 1530  for (code = first_significant_code(code
1530      /* Check for quantifiers after a class. XCLASS is used for classes that      /* Check for quantifiers after a class. XCLASS is used for classes that
1531      cannot be represented just by a bit map. This includes negated single      cannot be represented just by a bit map. This includes negated single
1532      high-valued characters. The length in _pcre_OP_lengths[] is zero; the      high-valued characters. The length in _pcre_OP_lengths[] is zero; the
1533      actual length is stored in the compiled code, so we must update "code"      actual length is stored in the compiled code, so we must update "code"
1534      here. */      here. */
1535    
1536  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
# Line 1540  for (code = first_significant_code(code Line 1597  for (code = first_significant_code(code
1597      case OP_TYPEEXACT:      case OP_TYPEEXACT:
1598      return FALSE;      return FALSE;
1599    
1600        /* These are going to continue, as they may be empty, but we have to
1601        fudge the length for the \p and \P cases. */
1602    
1603        case OP_TYPESTAR:
1604        case OP_TYPEMINSTAR:
1605        case OP_TYPEPOSSTAR:
1606        case OP_TYPEQUERY:
1607        case OP_TYPEMINQUERY:
1608        case OP_TYPEPOSQUERY:
1609        if (code[1] == OP_PROP || code[1] == OP_NOTPROP) code += 2;
1610        break;
1611    
1612        /* Same for these */
1613    
1614        case OP_TYPEUPTO:
1615        case OP_TYPEMINUPTO:
1616        case OP_TYPEPOSUPTO:
1617        if (code[3] == OP_PROP || code[3] == OP_NOTPROP) code += 2;
1618        break;
1619    
1620      /* End of branch */      /* End of branch */
1621    
1622      case OP_KET:      case OP_KET:
# Line 1702  adjust_recurse(uschar *group, int adjust Line 1779  adjust_recurse(uschar *group, int adjust
1779    uschar *save_hwm)    uschar *save_hwm)
1780  {  {
1781  uschar *ptr = group;  uschar *ptr = group;
1782    
1783  while ((ptr = (uschar *)find_recurse(ptr, utf8)) != NULL)  while ((ptr = (uschar *)find_recurse(ptr, utf8)) != NULL)
1784    {    {
1785    int offset;    int offset;
# Line 2671  for (;; ptr++) Line 2749  for (;; ptr++)
2749            else inescq = TRUE;            else inescq = TRUE;
2750            continue;            continue;
2751            }            }
2752            else if (-c == ESC_E) continue;  /* Ignore orphan \E */
2753    
2754          if (c < 0)          if (c < 0)
2755            {            {
# Line 2896  for (;; ptr++) Line 2975  for (;; ptr++)
2975          }          }
2976    
2977        oldptr = ptr;        oldptr = ptr;
2978    
2979          /* Remember \r or \n */
2980    
2981          if (c == '\r' || c == '\n') cd->external_flags |= PCRE_HASCRORLF;
2982    
2983          /* Check for range */
2984    
2985        if (!inescq && ptr[1] == '-')        if (!inescq && ptr[1] == '-')
2986          {          {
# Line 2964  for (;; ptr++) Line 3049  for (;; ptr++)
3049    
3050          if (d == c) goto LONE_SINGLE_CHARACTER;  /* A few lines below */          if (d == c) goto LONE_SINGLE_CHARACTER;  /* A few lines below */
3051    
3052            /* Remember \r or \n */
3053    
3054            if (d == '\r' || d == '\n') cd->external_flags |= PCRE_HASCRORLF;
3055    
3056          /* In UTF-8 mode, if the upper limit is > 255, or > 127 for caseless          /* In UTF-8 mode, if the upper limit is > 255, or > 127 for caseless
3057          matching, we have to use an XCLASS with extra data items. Caseless          matching, we have to use an XCLASS with extra data items. Caseless
3058          matching for characters > 127 is available only if UCP support is          matching for characters > 127 is available only if UCP support is
# Line 3068  for (;; ptr++) Line 3157  for (;; ptr++)
3157        apparent range that isn't. */        apparent range that isn't. */
3158    
3159        LONE_SINGLE_CHARACTER:        LONE_SINGLE_CHARACTER:
3160    
3161        /* Handle a character that cannot go in the bit map */        /* Handle a character that cannot go in the bit map */
3162    
3163  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
# Line 3116  for (;; ptr++) Line 3205  for (;; ptr++)
3205        *errorcodeptr = ERR6;        *errorcodeptr = ERR6;
3206        goto FAILED;        goto FAILED;
3207        }        }
3208    
3209    
3210    /* This code has been disabled because it would mean that \s counts as
3211    an explicit \r or \n reference, and that's not really what is wanted. Now
3212    we set the flag only if there is a literal "\r" or "\n" in the class. */
3213    
3214    #if 0
3215        /* Remember whether \r or \n are in this class */
3216    
3217        if (negate_class)
3218          {
3219          if ((classbits[1] & 0x24) != 0x24) cd->external_flags |= PCRE_HASCRORLF;
3220          }
3221        else
3222          {
3223          if ((classbits[1] & 0x24) != 0) cd->external_flags |= PCRE_HASCRORLF;
3224          }
3225    #endif
3226    
3227    
3228      /* If class_charcount is 1, we saw precisely one character whose value is      /* If class_charcount is 1, we saw precisely one character whose value is
3229      less than 256. In non-UTF-8 mode we can always optimize. In UTF-8 mode, we      less than 256. As long as there were no characters >= 128 and there was no
3230      can optimize the negative case only if there were no characters >= 128      use of \p or \P, in other words, no use of any XCLASS features, we can
3231      because OP_NOT and the related opcodes like OP_NOTSTAR operate on      optimize.
3232      single-bytes only. This is an historical hangover. Maybe one day we can  
3233      tidy these opcodes to handle multi-byte characters.      In UTF-8 mode, we can optimize the negative case only if there were no
3234        characters >= 128 because OP_NOT and the related opcodes like OP_NOTSTAR
3235        operate on single-bytes only. This is an historical hangover. Maybe one day
3236        we can tidy these opcodes to handle multi-byte characters.
3237    
3238      The optimization throws away the bit map. We turn the item into a      The optimization throws away the bit map. We turn the item into a
3239      1-character OP_CHAR[NC] if it's positive, or OP_NOT if it's negative. Note      1-character OP_CHAR[NC] if it's positive, or OP_NOT if it's negative. Note
# Line 3132  for (;; ptr++) Line 3243  for (;; ptr++)
3243      reqbyte, save the previous value for reinstating. */      reqbyte, save the previous value for reinstating. */
3244    
3245  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
3246      if (class_charcount == 1 &&      if (class_charcount == 1 && !class_utf8 &&
3247            (!utf8 ||        (!utf8 || !negate_class || class_lastchar < 128))
           (!class_utf8 && (!negate_class || class_lastchar < 128))))  
   
3248  #else  #else
3249      if (class_charcount == 1)      if (class_charcount == 1)
3250  #endif  #endif
# Line 3407  for (;; ptr++) Line 3516  for (;; ptr++)
3516        /* All real repeats make it impossible to handle partial matching (maybe        /* All real repeats make it impossible to handle partial matching (maybe
3517        one day we will be able to remove this restriction). */        one day we will be able to remove this restriction). */
3518    
3519        if (repeat_max != 1) cd->nopartial = TRUE;        if (repeat_max != 1) cd->external_flags |= PCRE_NOPARTIAL;
3520    
3521        /* Combine the op_type with the repeat_type */        /* Combine the op_type with the repeat_type */
3522    
# Line 3557  for (;; ptr++) Line 3666  for (;; ptr++)
3666        /* All real repeats make it impossible to handle partial matching (maybe        /* All real repeats make it impossible to handle partial matching (maybe
3667        one day we will be able to remove this restriction). */        one day we will be able to remove this restriction). */
3668    
3669        if (repeat_max != 1) cd->nopartial = TRUE;        if (repeat_max != 1) cd->external_flags |= PCRE_NOPARTIAL;
3670    
3671        if (repeat_min == 0 && repeat_max == -1)        if (repeat_min == 0 && repeat_max == -1)
3672          *code++ = OP_CRSTAR + repeat_type;          *code++ = OP_CRSTAR + repeat_type;
# Line 4554  for (;; ptr++) Line 4663  for (;; ptr++)
4663    
4664              case 'J':    /* Record that it changed in the external options */              case 'J':    /* Record that it changed in the external options */
4665              *optset |= PCRE_DUPNAMES;              *optset |= PCRE_DUPNAMES;
4666              cd->external_options |= PCRE_JCHANGED;              cd->external_flags |= PCRE_JCHANGED;
4667              break;              break;
4668    
4669              case 'i': *optset |= PCRE_CASELESS; break;              case 'i': *optset |= PCRE_CASELESS; break;
# Line 4745  for (;; ptr++) Line 4854  for (;; ptr++)
4854        goto FAILED;        goto FAILED;
4855        }        }
4856    
4857      /* In the pre-compile phase, update the length by the length of the nested      /* In the pre-compile phase, update the length by the length of the group,
4858      group, less the brackets at either end. Then reduce the compiled code to      less the brackets at either end. Then reduce the compiled code to just a
4859      just the brackets so that it doesn't use much memory if it is duplicated by      set of non-capturing brackets so that it doesn't use much memory if it is
4860      a quantifier. */      duplicated by a quantifier.*/
4861    
4862      if (lengthptr != NULL)      if (lengthptr != NULL)
4863        {        {
# Line 4758  for (;; ptr++) Line 4867  for (;; ptr++)
4867          goto FAILED;          goto FAILED;
4868          }          }
4869        *lengthptr += length_prevgroup - 2 - 2*LINK_SIZE;        *lengthptr += length_prevgroup - 2 - 2*LINK_SIZE;
4870        code++;        *code++ = OP_BRA;
4871        PUTINC(code, 0, 1 + LINK_SIZE);        PUTINC(code, 0, 1 + LINK_SIZE);
4872        *code++ = OP_KET;        *code++ = OP_KET;
4873        PUTINC(code, 0, 1 + LINK_SIZE);        PUTINC(code, 0, 1 + LINK_SIZE);
4874          break;    /* No need to waste time with special character handling */
4875        }        }
4876    
4877      /* Otherwise update the main code pointer to the end of the group. */      /* Otherwise update the main code pointer to the end of the group. */
4878    
4879      else code = tempcode;      code = tempcode;
4880    
4881      /* For a DEFINE group, required and first character settings are not      /* For a DEFINE group, required and first character settings are not
4882      relevant. */      relevant. */
# Line 4970  for (;; ptr++) Line 5080  for (;; ptr++)
5080      *code++ = ((options & PCRE_CASELESS) != 0)? OP_CHARNC : OP_CHAR;      *code++ = ((options & PCRE_CASELESS) != 0)? OP_CHARNC : OP_CHAR;
5081      for (c = 0; c < mclength; c++) *code++ = mcbuffer[c];      for (c = 0; c < mclength; c++) *code++ = mcbuffer[c];
5082    
5083        /* Remember if \r or \n were seen */
5084    
5085        if (mcbuffer[0] == '\r' || mcbuffer[0] == '\n')
5086          cd->external_flags |= PCRE_HASCRORLF;
5087    
5088      /* Set the first and required bytes appropriately. If no previous first      /* Set the first and required bytes appropriately. If no previous first
5089      byte, set it from this character, but revert to none on a zero repeat.      byte, set it from this character, but revert to none on a zero repeat.
5090      Otherwise, leave the firstbyte value alone, and don't change it on a zero      Otherwise, leave the firstbyte value alone, and don't change it on a zero
# Line 5569  real_pcre *re; Line 5684  real_pcre *re;
5684  int length = 1;  /* For final END opcode */  int length = 1;  /* For final END opcode */
5685  int firstbyte, reqbyte, newline;  int firstbyte, reqbyte, newline;
5686  int errorcode = 0;  int errorcode = 0;
5687    int skipatstart = 0;
5688  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
5689  BOOL utf8;  BOOL utf8;
5690  #endif  #endif
# Line 5647  cd->fcc = tables + fcc_offset; Line 5763  cd->fcc = tables + fcc_offset;
5763  cd->cbits = tables + cbits_offset;  cd->cbits = tables + cbits_offset;
5764  cd->ctypes = tables + ctypes_offset;  cd->ctypes = tables + ctypes_offset;
5765    
5766    /* Check for newline settings at the start of the pattern, and remember the
5767    offset for later. */
5768    
5769    if (ptr[0] == '(' && ptr[1] == '*')
5770      {
5771      int newnl = 0;
5772      if (strncmp((char *)(ptr+2), "CR)", 3) == 0)
5773        { skipatstart = 5; newnl = PCRE_NEWLINE_CR; }
5774      else if (strncmp((char *)(ptr+2), "LF)", 3)  == 0)
5775        { skipatstart = 5; newnl = PCRE_NEWLINE_LF; }
5776      else if (strncmp((char *)(ptr+2), "CRLF)", 5)  == 0)
5777        { skipatstart = 7; newnl = PCRE_NEWLINE_CR + PCRE_NEWLINE_LF; }
5778      else if (strncmp((char *)(ptr+2), "ANY)", 4) == 0)
5779        { skipatstart = 6; newnl = PCRE_NEWLINE_ANY; }
5780      else if (strncmp((char *)(ptr+2), "ANYCRLF)", 8)  == 0)
5781        { skipatstart = 10; newnl = PCRE_NEWLINE_ANYCRLF; }
5782      if (skipatstart > 0)
5783        options = (options & ~PCRE_NEWLINE_BITS) | newnl;
5784      }
5785    
5786  /* Handle different types of newline. The three bits give seven cases. The  /* Handle different types of newline. The three bits give seven cases. The
5787  current code allows for fixed one- or two-byte sequences, plus "any" and  current code allows for fixed one- or two-byte sequences, plus "any" and
5788  "anycrlf". */  "anycrlf". */
5789    
5790  switch (options & (PCRE_NEWLINE_CRLF | PCRE_NEWLINE_ANY))  switch (options & PCRE_NEWLINE_BITS)
5791    {    {
5792    case 0: newline = NEWLINE; break;   /* Compile-time default */    case 0: newline = NEWLINE; break;   /* Build-time default */
5793    case PCRE_NEWLINE_CR: newline = '\r'; break;    case PCRE_NEWLINE_CR: newline = '\r'; break;
5794    case PCRE_NEWLINE_LF: newline = '\n'; break;    case PCRE_NEWLINE_LF: newline = '\n'; break;
5795    case PCRE_NEWLINE_CR+    case PCRE_NEWLINE_CR+
# Line 5716  cd->hwm = cworkspace; Line 5852  cd->hwm = cworkspace;
5852  cd->start_pattern = (const uschar *)pattern;  cd->start_pattern = (const uschar *)pattern;
5853  cd->end_pattern = (const uschar *)(pattern + strlen(pattern));  cd->end_pattern = (const uschar *)(pattern + strlen(pattern));
5854  cd->req_varyopt = 0;  cd->req_varyopt = 0;
 cd->nopartial = FALSE;  
5855  cd->external_options = options;  cd->external_options = options;
5856    cd->external_flags = 0;
5857    
5858  /* Now do the pre-compile. On error, errorcode will be set non-zero, so we  /* Now do the pre-compile. On error, errorcode will be set non-zero, so we
5859  don't need to look at the result of the function here. The initial options have  don't need to look at the result of the function here. The initial options have
# Line 5725  been put into the cd block so that they Line 5861  been put into the cd block so that they
5861  found within the regex right at the beginning. Bringing initial option settings  found within the regex right at the beginning. Bringing initial option settings
5862  outside can help speed up starting point checks. */  outside can help speed up starting point checks. */
5863    
5864    ptr += skipatstart;
5865  code = cworkspace;  code = cworkspace;
5866  *code = OP_BRA;  *code = OP_BRA;
5867  (void)compile_regex(cd->external_options, cd->external_options & PCRE_IMS,  (void)compile_regex(cd->external_options, cd->external_options & PCRE_IMS,
# Line 5755  if (re == NULL) Line 5892  if (re == NULL)
5892    goto PCRE_EARLY_ERROR_RETURN;    goto PCRE_EARLY_ERROR_RETURN;
5893    }    }
5894    
5895  /* Put in the magic number, and save the sizes, initial options, and character  /* Put in the magic number, and save the sizes, initial options, internal
5896  table pointer. NULL is used for the default character tables. The nullpad field  flags, and character table pointer. NULL is used for the default character
5897  is at the end; it's there to help in the case when a regex compiled on a system  tables. The nullpad field is at the end; it's there to help in the case when a
5898  with 4-byte pointers is run on another with 8-byte pointers. */  regex compiled on a system with 4-byte pointers is run on another with 8-byte
5899    pointers. */
5900    
5901  re->magic_number = MAGIC_NUMBER;  re->magic_number = MAGIC_NUMBER;
5902  re->size = size;  re->size = size;
5903  re->options = cd->external_options;  re->options = cd->external_options;
5904    re->flags = cd->external_flags;
5905  re->dummy1 = 0;  re->dummy1 = 0;
5906  re->first_byte = 0;  re->first_byte = 0;
5907  re->req_byte = 0;  re->req_byte = 0;
# Line 5787  codestart = cd->name_table + re->name_en Line 5926  codestart = cd->name_table + re->name_en
5926  cd->start_code = codestart;  cd->start_code = codestart;
5927  cd->hwm = cworkspace;  cd->hwm = cworkspace;
5928  cd->req_varyopt = 0;  cd->req_varyopt = 0;
 cd->nopartial = FALSE;  
5929  cd->had_accept = FALSE;  cd->had_accept = FALSE;
5930    
5931  /* Set up a starting, non-extracting bracket, then compile the expression. On  /* Set up a starting, non-extracting bracket, then compile the expression. On
5932  error, errorcode will be set non-zero, so we don't need to look at the result  error, errorcode will be set non-zero, so we don't need to look at the result
5933  of the function here. */  of the function here. */
5934    
5935  ptr = (const uschar *)pattern;  ptr = (const uschar *)pattern + skipatstart;
5936  code = (uschar *)codestart;  code = (uschar *)codestart;
5937  *code = OP_BRA;  *code = OP_BRA;
5938  (void)compile_regex(re->options, re->options & PCRE_IMS, &code, &ptr,  (void)compile_regex(re->options, re->options & PCRE_IMS, &code, &ptr,
5939    &errorcode, FALSE, FALSE, 0, &firstbyte, &reqbyte, NULL, cd, NULL);    &errorcode, FALSE, FALSE, 0, &firstbyte, &reqbyte, NULL, cd, NULL);
5940  re->top_bracket = cd->bracount;  re->top_bracket = cd->bracount;
5941  re->top_backref = cd->top_backref;  re->top_backref = cd->top_backref;
5942    re->flags = cd->external_flags;
5943    
 if (cd->nopartial) re->options |= PCRE_NOPARTIAL;  
5944  if (cd->had_accept) reqbyte = -1;   /* Must disable after (*ACCEPT) */  if (cd->had_accept) reqbyte = -1;   /* Must disable after (*ACCEPT) */
5945    
5946  /* If not reached end of pattern on success, there's an excess bracket. */  /* If not reached end of pattern on success, there's an excess bracket. */
# Line 5874  if ((re->options & PCRE_ANCHORED) == 0) Line 6012  if ((re->options & PCRE_ANCHORED) == 0)
6012        int ch = firstbyte & 255;        int ch = firstbyte & 255;
6013        re->first_byte = ((firstbyte & REQ_CASELESS) != 0 &&        re->first_byte = ((firstbyte & REQ_CASELESS) != 0 &&
6014           cd->fcc[ch] == ch)? ch : firstbyte;           cd->fcc[ch] == ch)? ch : firstbyte;
6015        re->options |= PCRE_FIRSTSET;        re->flags |= PCRE_FIRSTSET;
6016        }        }
6017      else if (is_startline(codestart, 0, cd->backref_map))      else if (is_startline(codestart, 0, cd->backref_map))
6018        re->options |= PCRE_STARTLINE;        re->flags |= PCRE_STARTLINE;
6019      }      }
6020    }    }
6021    
# Line 5891  if (reqbyte >= 0 && Line 6029  if (reqbyte >= 0 &&
6029    int ch = reqbyte & 255;    int ch = reqbyte & 255;
6030    re->req_byte = ((reqbyte & REQ_CASELESS) != 0 &&    re->req_byte = ((reqbyte & REQ_CASELESS) != 0 &&
6031      cd->fcc[ch] == ch)? (reqbyte & ~REQ_CASELESS) : reqbyte;      cd->fcc[ch] == ch)? (reqbyte & ~REQ_CASELESS) : reqbyte;
6032    re->options |= PCRE_REQCHSET;    re->flags |= PCRE_REQCHSET;
6033    }    }
6034    
6035  /* Print out the compiled data if debugging is enabled. This is never the  /* Print out the compiled data if debugging is enabled. This is never the
# Line 5902  case when building a production library. Line 6040  case when building a production library.
6040  printf("Length = %d top_bracket = %d top_backref = %d\n",  printf("Length = %d top_bracket = %d top_backref = %d\n",
6041    length, re->top_bracket, re->top_backref);    length, re->top_bracket, re->top_backref);
6042    
6043  if (re->options != 0)  printf("Options=%08x\n", re->options);
   {  
   printf("%s%s%s%s%s%s%s%s%s\n",  
     ((re->options & PCRE_NOPARTIAL) != 0)? "nopartial " : "",  
     ((re->options & PCRE_ANCHORED) != 0)? "anchored " : "",  
     ((re->options & PCRE_CASELESS) != 0)? "caseless " : "",  
     ((re->options & PCRE_EXTENDED) != 0)? "extended " : "",  
     ((re->options & PCRE_MULTILINE) != 0)? "multiline " : "",  
     ((re->options & PCRE_DOTALL) != 0)? "dotall " : "",  
     ((re->options & PCRE_DOLLAR_ENDONLY) != 0)? "endonly " : "",  
     ((re->options & PCRE_EXTRA) != 0)? "extra " : "",  
     ((re->options & PCRE_UNGREEDY) != 0)? "ungreedy " : "");  
   }  
6044    
6045  if ((re->options & PCRE_FIRSTSET) != 0)  if ((re->flags & PCRE_FIRSTSET) != 0)
6046    {    {
6047    int ch = re->first_byte & 255;    int ch = re->first_byte & 255;
6048    const char *caseless = ((re->first_byte & REQ_CASELESS) == 0)?    const char *caseless = ((re->first_byte & REQ_CASELESS) == 0)?
# Line 5925  if ((re->options & PCRE_FIRSTSET) != 0) Line 6051  if ((re->options & PCRE_FIRSTSET) != 0)
6051      else printf("First char = \\x%02x%s\n", ch, caseless);      else printf("First char = \\x%02x%s\n", ch, caseless);
6052    }    }
6053    
6054  if ((re->options & PCRE_REQCHSET) != 0)  if ((re->flags & PCRE_REQCHSET) != 0)
6055    {    {
6056    int ch = re->req_byte & 255;    int ch = re->req_byte & 255;
6057    const char *caseless = ((re->req_byte & REQ_CASELESS) == 0)?    const char *caseless = ((re->req_byte & REQ_CASELESS) == 0)?
Legend:
Removed from v.216  
changed lines
  Added in v.230
webmaster@exim.org
 ViewVC Help
Powered by ViewVC 1.1.12