/[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 213 by ph10, Wed Aug 15 11:34:14 2007 UTC revision 223 by ph10, Mon Aug 20 11:07:53 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;
954        if (*ptr == '\\')        if (*ptr == '\\')
955          {          {
956          if (*(++ptr) == 0) return -1;          if (*(++ptr) == 0) return -1;
# Line 1105  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 1194  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 1302  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 1366  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 1469  for (code = first_significant_code(code Line 1527  for (code = first_significant_code(code
1527    
1528    switch (c)    switch (c)
1529      {      {
1530      /* Check for quantifiers after a class */      /* 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
1532        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"
1534        here. */
1535    
1536  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
1537      case OP_XCLASS:      case OP_XCLASS:
1538      ccode = code + GET(code, 1);      ccode = code += GET(code, 1);
1539      goto CHECK_CLASS_REPEAT;      goto CHECK_CLASS_REPEAT;
1540  #endif  #endif
1541    
# Line 2666  for (;; ptr++) Line 2728  for (;; ptr++)
2728            else inescq = TRUE;            else inescq = TRUE;
2729            continue;            continue;
2730            }            }
2731            else if (-c == ESC_E) continue;  /* Ignore orphan \E */
2732    
2733          if (c < 0)          if (c < 0)
2734            {            {
# Line 3113  for (;; ptr++) Line 3176  for (;; ptr++)
3176        }        }
3177    
3178      /* 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
3179      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
3180      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
3181      because OP_NOT and the related opcodes like OP_NOTSTAR operate on      optimize.
3182      single-bytes only. This is an historical hangover. Maybe one day we can  
3183      tidy these opcodes to handle multi-byte characters.      In UTF-8 mode, we can optimize the negative case only if there were no
3184        characters >= 128 because OP_NOT and the related opcodes like OP_NOTSTAR
3185        operate on single-bytes only. This is an historical hangover. Maybe one day
3186        we can tidy these opcodes to handle multi-byte characters.
3187    
3188      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
3189      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 3127  for (;; ptr++) Line 3193  for (;; ptr++)
3193      reqbyte, save the previous value for reinstating. */      reqbyte, save the previous value for reinstating. */
3194    
3195  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
3196      if (class_charcount == 1 &&      if (class_charcount == 1 && !class_utf8 &&
3197            (!utf8 ||        (!utf8 || !negate_class || class_lastchar < 128))
           (!class_utf8 && (!negate_class || class_lastchar < 128))))  
   
3198  #else  #else
3199      if (class_charcount == 1)      if (class_charcount == 1)
3200  #endif  #endif
# Line 4740  for (;; ptr++) Line 4804  for (;; ptr++)
4804        goto FAILED;        goto FAILED;
4805        }        }
4806    
4807      /* 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,
4808      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
4809      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
4810      a quantifier. */      duplicated by a quantifier.*/
4811    
4812      if (lengthptr != NULL)      if (lengthptr != NULL)
4813        {        {
# Line 4753  for (;; ptr++) Line 4817  for (;; ptr++)
4817          goto FAILED;          goto FAILED;
4818          }          }
4819        *lengthptr += length_prevgroup - 2 - 2*LINK_SIZE;        *lengthptr += length_prevgroup - 2 - 2*LINK_SIZE;
4820        code++;        *code++ = OP_BRA;
4821        PUTINC(code, 0, 1 + LINK_SIZE);        PUTINC(code, 0, 1 + LINK_SIZE);
4822        *code++ = OP_KET;        *code++ = OP_KET;
4823        PUTINC(code, 0, 1 + LINK_SIZE);        PUTINC(code, 0, 1 + LINK_SIZE);
4824          break;    /* No need to waste time with special character handling */
4825        }        }
4826    
4827      /* Otherwise update the main code pointer to the end of the group. */      /* Otherwise update the main code pointer to the end of the group. */
4828    
4829      else code = tempcode;      code = tempcode;
4830    
4831      /* For a DEFINE group, required and first character settings are not      /* For a DEFINE group, required and first character settings are not
4832      relevant. */      relevant. */

Legend:
Removed from v.213  
changed lines
  Added in v.223

  ViewVC Help
Powered by ViewVC 1.1.5