/[pcre]/code/branches/pcre16/pcre_compile.c
ViewVC logotype

Diff of /code/branches/pcre16/pcre_compile.c

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

revision 224 by ph10, Mon Aug 20 12:58:29 2007 UTC revision 236 by ph10, Tue Sep 11 12:57:06 2007 UTC
# Line 43  supporting internal functions that are n Line 43  supporting internal functions that are n
43    
44    
45  #ifdef HAVE_CONFIG_H  #ifdef HAVE_CONFIG_H
46  #include <config.h>  #include "config.h"
47  #endif  #endif
48    
49  #define NLBLOCK cd             /* Block containing newline information */  #define NLBLOCK cd             /* Block containing newline information */
# Line 1596  for (code = first_significant_code(code Line 1596  for (code = first_significant_code(code
1596      case OP_TYPEPOSPLUS:      case OP_TYPEPOSPLUS:
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      /* These are going to continue, as they may be empty, but we have to
1601      fudge the length for the \p and \P cases. */      fudge the length for the \p and \P cases. */
1602    
1603      case OP_TYPESTAR:      case OP_TYPESTAR:
1604      case OP_TYPEMINSTAR:      case OP_TYPEMINSTAR:
1605      case OP_TYPEPOSSTAR:      case OP_TYPEPOSSTAR:
# Line 1607  for (code = first_significant_code(code Line 1607  for (code = first_significant_code(code
1607      case OP_TYPEMINQUERY:      case OP_TYPEMINQUERY:
1608      case OP_TYPEPOSQUERY:      case OP_TYPEPOSQUERY:
1609      if (code[1] == OP_PROP || code[1] == OP_NOTPROP) code += 2;      if (code[1] == OP_PROP || code[1] == OP_NOTPROP) code += 2;
1610      break;      break;
1611    
1612      /* Same for these */      /* Same for these */
1613    
1614      case OP_TYPEUPTO:      case OP_TYPEUPTO:
1615      case OP_TYPEMINUPTO:      case OP_TYPEMINUPTO:
1616      case OP_TYPEPOSUPTO:      case OP_TYPEPOSUPTO:
# Line 2976  for (;; ptr++) Line 2976  for (;; ptr++)
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          {          {
2987          int d;          int d;
# Line 3043  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 3196  for (;; ptr++) Line 3206  for (;; ptr++)
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. As long as there were no characters >= 128 and there was no      less than 256. As long as there were no characters >= 128 and there was no
3230      use of \p or \P, in other words, no use of any XCLASS features, we can      use of \p or \P, in other words, no use of any XCLASS features, we can
3231      optimize.      optimize.
3232    
3233      In UTF-8 mode, we can optimize the negative case only if there were no      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      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      operate on single-bytes only. This is an historical hangover. Maybe one day
# Line 3214  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 && !class_utf8 &&      if (class_charcount == 1 && !class_utf8 &&
3247        (!utf8 || !negate_class || class_lastchar < 128))        (!utf8 || !negate_class || class_lastchar < 128))
3248  #else  #else
3249      if (class_charcount == 1)      if (class_charcount == 1)
# Line 3487  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 3637  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 4634  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 5051  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 5650  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 5728  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 global one-time settings at the start of the pattern, and remember
5767    the offset for later. */
5768    
5769    while (ptr[skipatstart] == '(' && ptr[skipatstart+1] == '*')
5770      {
5771      int newnl = 0;
5772      int newbsr = 0;
5773    
5774      if (strncmp((char *)(ptr+skipatstart+2), "CR)", 3) == 0)
5775        { skipatstart += 5; newnl = PCRE_NEWLINE_CR; }
5776      else if (strncmp((char *)(ptr+skipatstart+2), "LF)", 3)  == 0)
5777        { skipatstart += 5; newnl = PCRE_NEWLINE_LF; }
5778      else if (strncmp((char *)(ptr+skipatstart+2), "CRLF)", 5)  == 0)
5779        { skipatstart += 7; newnl = PCRE_NEWLINE_CR + PCRE_NEWLINE_LF; }
5780      else if (strncmp((char *)(ptr+skipatstart+2), "ANY)", 4) == 0)
5781        { skipatstart += 6; newnl = PCRE_NEWLINE_ANY; }
5782      else if (strncmp((char *)(ptr+skipatstart+2), "ANYCRLF)", 8)  == 0)
5783        { skipatstart += 10; newnl = PCRE_NEWLINE_ANYCRLF; }
5784    
5785      else if (strncmp((char *)(ptr+skipatstart+2), "BSR_ANYCRLF)", 12) == 0)
5786        { skipatstart += 14; newbsr = PCRE_BSR_ANYCRLF; }
5787      else if (strncmp((char *)(ptr+skipatstart+2), "BSR_UNICODE)", 12) == 0)
5788        { skipatstart += 14; newbsr = PCRE_BSR_UNICODE; }
5789    
5790      if (newnl != 0)
5791        options = (options & ~PCRE_NEWLINE_BITS) | newnl;
5792      else if (newbsr != 0)
5793        options = (options & ~(PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE)) | newbsr;
5794      else break;
5795      }
5796    
5797    /* Check validity of \R options. */
5798    
5799    switch (options & (PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE))
5800      {
5801      case 0:
5802      case PCRE_BSR_ANYCRLF:
5803      case PCRE_BSR_UNICODE:
5804      break;
5805      default: errorcode = ERR56; goto PCRE_EARLY_ERROR_RETURN;
5806      }
5807    
5808  /* Handle different types of newline. The three bits give seven cases. The  /* Handle different types of newline. The three bits give seven cases. The
5809  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
5810  "anycrlf". */  "anycrlf". */
5811    
5812  switch (options & (PCRE_NEWLINE_CRLF | PCRE_NEWLINE_ANY))  switch (options & PCRE_NEWLINE_BITS)
5813    {    {
5814    case 0: newline = NEWLINE; break;   /* Compile-time default */    case 0: newline = NEWLINE; break;   /* Build-time default */
5815    case PCRE_NEWLINE_CR: newline = '\r'; break;    case PCRE_NEWLINE_CR: newline = '\r'; break;
5816    case PCRE_NEWLINE_LF: newline = '\n'; break;    case PCRE_NEWLINE_LF: newline = '\n'; break;
5817    case PCRE_NEWLINE_CR+    case PCRE_NEWLINE_CR+
# Line 5797  cd->hwm = cworkspace; Line 5874  cd->hwm = cworkspace;
5874  cd->start_pattern = (const uschar *)pattern;  cd->start_pattern = (const uschar *)pattern;
5875  cd->end_pattern = (const uschar *)(pattern + strlen(pattern));  cd->end_pattern = (const uschar *)(pattern + strlen(pattern));
5876  cd->req_varyopt = 0;  cd->req_varyopt = 0;
 cd->nopartial = FALSE;  
5877  cd->external_options = options;  cd->external_options = options;
5878    cd->external_flags = 0;
5879    
5880  /* 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
5881  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 5806  been put into the cd block so that they Line 5883  been put into the cd block so that they
5883  found within the regex right at the beginning. Bringing initial option settings  found within the regex right at the beginning. Bringing initial option settings
5884  outside can help speed up starting point checks. */  outside can help speed up starting point checks. */
5885    
5886    ptr += skipatstart;
5887  code = cworkspace;  code = cworkspace;
5888  *code = OP_BRA;  *code = OP_BRA;
5889  (void)compile_regex(cd->external_options, cd->external_options & PCRE_IMS,  (void)compile_regex(cd->external_options, cd->external_options & PCRE_IMS,
# Line 5836  if (re == NULL) Line 5914  if (re == NULL)
5914    goto PCRE_EARLY_ERROR_RETURN;    goto PCRE_EARLY_ERROR_RETURN;
5915    }    }
5916    
5917  /* 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
5918  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
5919  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
5920  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
5921    pointers. */
5922    
5923  re->magic_number = MAGIC_NUMBER;  re->magic_number = MAGIC_NUMBER;
5924  re->size = size;  re->size = size;
5925  re->options = cd->external_options;  re->options = cd->external_options;
5926    re->flags = cd->external_flags;
5927  re->dummy1 = 0;  re->dummy1 = 0;
5928  re->first_byte = 0;  re->first_byte = 0;
5929  re->req_byte = 0;  re->req_byte = 0;
# Line 5868  codestart = cd->name_table + re->name_en Line 5948  codestart = cd->name_table + re->name_en
5948  cd->start_code = codestart;  cd->start_code = codestart;
5949  cd->hwm = cworkspace;  cd->hwm = cworkspace;
5950  cd->req_varyopt = 0;  cd->req_varyopt = 0;
 cd->nopartial = FALSE;  
5951  cd->had_accept = FALSE;  cd->had_accept = FALSE;
5952    
5953  /* Set up a starting, non-extracting bracket, then compile the expression. On  /* Set up a starting, non-extracting bracket, then compile the expression. On
5954  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
5955  of the function here. */  of the function here. */
5956    
5957  ptr = (const uschar *)pattern;  ptr = (const uschar *)pattern + skipatstart;
5958  code = (uschar *)codestart;  code = (uschar *)codestart;
5959  *code = OP_BRA;  *code = OP_BRA;
5960  (void)compile_regex(re->options, re->options & PCRE_IMS, &code, &ptr,  (void)compile_regex(re->options, re->options & PCRE_IMS, &code, &ptr,
5961    &errorcode, FALSE, FALSE, 0, &firstbyte, &reqbyte, NULL, cd, NULL);    &errorcode, FALSE, FALSE, 0, &firstbyte, &reqbyte, NULL, cd, NULL);
5962  re->top_bracket = cd->bracount;  re->top_bracket = cd->bracount;
5963  re->top_backref = cd->top_backref;  re->top_backref = cd->top_backref;
5964    re->flags = cd->external_flags;
5965    
 if (cd->nopartial) re->options |= PCRE_NOPARTIAL;  
5966  if (cd->had_accept) reqbyte = -1;   /* Must disable after (*ACCEPT) */  if (cd->had_accept) reqbyte = -1;   /* Must disable after (*ACCEPT) */
5967    
5968  /* 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 5955  if ((re->options & PCRE_ANCHORED) == 0) Line 6034  if ((re->options & PCRE_ANCHORED) == 0)
6034        int ch = firstbyte & 255;        int ch = firstbyte & 255;
6035        re->first_byte = ((firstbyte & REQ_CASELESS) != 0 &&        re->first_byte = ((firstbyte & REQ_CASELESS) != 0 &&
6036           cd->fcc[ch] == ch)? ch : firstbyte;           cd->fcc[ch] == ch)? ch : firstbyte;
6037        re->options |= PCRE_FIRSTSET;        re->flags |= PCRE_FIRSTSET;
6038        }        }
6039      else if (is_startline(codestart, 0, cd->backref_map))      else if (is_startline(codestart, 0, cd->backref_map))
6040        re->options |= PCRE_STARTLINE;        re->flags |= PCRE_STARTLINE;
6041      }      }
6042    }    }
6043    
# Line 5972  if (reqbyte >= 0 && Line 6051  if (reqbyte >= 0 &&
6051    int ch = reqbyte & 255;    int ch = reqbyte & 255;
6052    re->req_byte = ((reqbyte & REQ_CASELESS) != 0 &&    re->req_byte = ((reqbyte & REQ_CASELESS) != 0 &&
6053      cd->fcc[ch] == ch)? (reqbyte & ~REQ_CASELESS) : reqbyte;      cd->fcc[ch] == ch)? (reqbyte & ~REQ_CASELESS) : reqbyte;
6054    re->options |= PCRE_REQCHSET;    re->flags |= PCRE_REQCHSET;
6055    }    }
6056    
6057  /* 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 5983  case when building a production library. Line 6062  case when building a production library.
6062  printf("Length = %d top_bracket = %d top_backref = %d\n",  printf("Length = %d top_bracket = %d top_backref = %d\n",
6063    length, re->top_bracket, re->top_backref);    length, re->top_bracket, re->top_backref);
6064    
6065  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 " : "");  
   }  
6066    
6067  if ((re->options & PCRE_FIRSTSET) != 0)  if ((re->flags & PCRE_FIRSTSET) != 0)
6068    {    {
6069    int ch = re->first_byte & 255;    int ch = re->first_byte & 255;
6070    const char *caseless = ((re->first_byte & REQ_CASELESS) == 0)?    const char *caseless = ((re->first_byte & REQ_CASELESS) == 0)?
# Line 6006  if ((re->options & PCRE_FIRSTSET) != 0) Line 6073  if ((re->options & PCRE_FIRSTSET) != 0)
6073      else printf("First char = \\x%02x%s\n", ch, caseless);      else printf("First char = \\x%02x%s\n", ch, caseless);
6074    }    }
6075    
6076  if ((re->options & PCRE_REQCHSET) != 0)  if ((re->flags & PCRE_REQCHSET) != 0)
6077    {    {
6078    int ch = re->req_byte & 255;    int ch = re->req_byte & 255;
6079    const char *caseless = ((re->req_byte & REQ_CASELESS) == 0)?    const char *caseless = ((re->req_byte & REQ_CASELESS) == 0)?

Legend:
Removed from v.224  
changed lines
  Added in v.236

  ViewVC Help
Powered by ViewVC 1.1.5