/[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 195 by ph10, Mon Jul 30 13:23:28 2007 UTC revision 216 by ph10, Wed Aug 15 14:35:57 2007 UTC
# Line 42  POSSIBILITY OF SUCH DAMAGE. Line 42  POSSIBILITY OF SUCH DAMAGE.
42  supporting internal functions that are not used by other modules. */  supporting internal functions that are not used by other modules. */
43    
44    
45    #ifdef HAVE_CONFIG_H
46    #include <config.h>
47    #endif
48    
49  #define NLBLOCK cd             /* Block containing newline information */  #define NLBLOCK cd             /* Block containing newline information */
50  #define PSSTART start_pattern  /* Field containing processed string start */  #define PSSTART start_pattern  /* Field containing processed string start */
51  #define PSEND   end_pattern    /* Field containing processed string end */  #define PSEND   end_pattern    /* Field containing processed string end */
52    
   
53  #include "pcre_internal.h"  #include "pcre_internal.h"
54    
55    
# Line 62  used by pcretest. DEBUG is not defined w Line 65  used by pcretest. DEBUG is not defined w
65    
66  #define SETBIT(a,b) a[b/8] |= (1 << (b%8))  #define SETBIT(a,b) a[b/8] |= (1 << (b%8))
67    
68    /* Maximum length value to check against when making sure that the integer that
69    holds the compiled pattern length does not overflow. We make it a bit less than
70    INT_MAX to allow for adding in group terminating bytes, so that we don't have
71    to check them every time. */
72    
73    #define OFLOW_MAX (INT_MAX - 20)
74    
75    
76  /*************************************************  /*************************************************
77  *      Code parameters and static tables         *  *      Code parameters and static tables         *
# Line 130  static const short int escapes[] = { Line 140  static const short int escapes[] = {
140  #endif  #endif
141    
142    
143    /* Table of special "verbs" like (*PRUNE) */
144    
145    typedef struct verbitem {
146      const char *name;
147      int   len;
148      int   op;
149    } verbitem;
150    
151    static verbitem verbs[] = {
152      { "ACCEPT", 6, OP_ACCEPT },
153      { "COMMIT", 6, OP_COMMIT },
154      { "F",      1, OP_FAIL },
155      { "FAIL",   4, OP_FAIL },
156      { "PRUNE",  5, OP_PRUNE },
157      { "SKIP",   4, OP_SKIP  },
158      { "THEN",   4, OP_THEN  }
159    };
160    
161    static int verbcount = sizeof(verbs)/sizeof(verbitem);
162    
163    
164  /* Tables of names of POSIX character classes and their lengths. The list is  /* Tables of names of POSIX character classes and their lengths. The list is
165  terminated by a zero length entry. The first three must be alpha, lower, upper,  terminated by a zero length entry. The first three must be alpha, lower, upper,
166  as this is assumed for handling case independence. */  as this is assumed for handling case independence. */
# Line 203  static const char *error_texts[] = { Line 234  static const char *error_texts[] = {
234    "missing ) after comment",    "missing ) after comment",
235    "parentheses nested too deeply",  /** DEAD **/    "parentheses nested too deeply",  /** DEAD **/
236    /* 20 */    /* 20 */
237    "regular expression too large",    "regular expression is too large",
238    "failed to get memory",    "failed to get memory",
239    "unmatched parentheses",    "unmatched parentheses",
240    "internal error: code overflow",    "internal error: code overflow",
# Line 239  static const char *error_texts[] = { Line 270  static const char *error_texts[] = {
270    "subpattern name is too long (maximum " XSTRING(MAX_NAME_SIZE) " characters)",    "subpattern name is too long (maximum " XSTRING(MAX_NAME_SIZE) " characters)",
271    "too many named subpatterns (maximum " XSTRING(MAX_NAME_COUNT) ")",    "too many named subpatterns (maximum " XSTRING(MAX_NAME_COUNT) ")",
272    /* 50 */    /* 50 */
273    "repeated subpattern is too long",    "repeated subpattern is too long",    /** DEAD **/
274    "octal value is greater than \\377 (not in UTF-8 mode)",    "octal value is greater than \\377 (not in UTF-8 mode)",
275    "internal error: overran compiling workspace",    "internal error: overran compiling workspace",
276    "internal error: previously-checked referenced subpattern not found",    "internal error: previously-checked referenced subpattern not found",
# Line 248  static const char *error_texts[] = { Line 279  static const char *error_texts[] = {
279    "repeating a DEFINE group is not allowed",    "repeating a DEFINE group is not allowed",
280    "inconsistent NEWLINE options",    "inconsistent NEWLINE options",
281    "\\g is not followed by a braced name or an optionally braced non-zero number",    "\\g is not followed by a braced name or an optionally braced non-zero number",
282    "(?+ or (?- or (?(+ or (?(- must be followed by a non-zero number"    "(?+ or (?- or (?(+ or (?(- must be followed by a non-zero number",
283      "(*VERB) with an argument is not supported",
284      /* 60 */
285      "(*VERB) not recognized",
286      "number is too big"
287  };  };
288    
289    
# Line 405  Arguments: Line 440  Arguments:
440    
441  Returns:         zero or positive => a data character  Returns:         zero or positive => a data character
442                   negative => a special escape sequence                   negative => a special escape sequence
443                   on error, errorptr is set                   on error, errorcodeptr is set
444  */  */
445    
446  static int  static int
# Line 489  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)
529          {
530          *errorcodeptr = ERR61;
531          break;
532          }
533    
534      if (c == 0 || (braced && *(++ptr) != '}'))      if (c == 0 || (braced && *(++ptr) != '}'))
535        {        {
536        *errorcodeptr = ERR57;        *errorcodeptr = ERR57;
537        return 0;        break;
538        }        }
539    
540      if (negated)      if (negated)
# Line 501  else Line 542  else
542        if (c > bracount)        if (c > bracount)
543          {          {
544          *errorcodeptr = ERR15;          *errorcodeptr = ERR15;
545          return 0;          break;
546          }          }
547        c = bracount - (c - 1);        c = bracount - (c - 1);
548        }        }
# Line 530  else Line 571  else
571        c -= '0';        c -= '0';
572        while ((digitab[ptr[1]] & ctype_digit) != 0)        while ((digitab[ptr[1]] & ctype_digit) != 0)
573          c = c * 10 + *(++ptr) - '0';          c = c * 10 + *(++ptr) - '0';
574          if (c < 0)
575            {
576            *errorcodeptr = ERR61;
577            break;
578            }
579        if (c < 10 || c <= bracount)        if (c < 10 || c <= bracount)
580          {          {
581          c = -(ESC_REF + c);          c = -(ESC_REF + c);
# Line 625  else Line 671  else
671      if (c == 0)      if (c == 0)
672        {        {
673        *errorcodeptr = ERR2;        *errorcodeptr = ERR2;
674        return 0;        break;
675        }        }
676    
677  #ifndef EBCDIC  /* ASCII coding */  #ifndef EBCDIC  /* ASCII coding */
# Line 701  if (c == '{') Line 747  if (c == '{')
747      *negptr = TRUE;      *negptr = TRUE;
748      ptr++;      ptr++;
749      }      }
750    for (i = 0; i < sizeof(name) - 1; i++)    for (i = 0; i < (int)sizeof(name) - 1; i++)
751      {      {
752      c = *(++ptr);      c = *(++ptr);
753      if (c == 0) goto ERROR_RETURN;      if (c == 0) goto ERROR_RETURN;
# Line 904  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 931  for (; *ptr != 0; ptr++) Line 978  for (; *ptr != 0; ptr++)
978    /* An opening parens must now be a real metacharacter */    /* An opening parens must now be a real metacharacter */
979    
980    if (*ptr != '(') continue;    if (*ptr != '(') continue;
981    if (ptr[1] != '?')    if (ptr[1] != '?' && ptr[1] != '*')
982      {      {
983      count++;      count++;
984      if (name == NULL && count == lorn) return count;      if (name == NULL && count == lorn) return count;
# Line 1399  for (code = first_significant_code(code Line 1446  for (code = first_significant_code(code
1446    
1447    /* For other groups, scan the branches. */    /* For other groups, scan the branches. */
1448    
1449    if (c == OP_BRA || c == OP_CBRA || c == OP_ONCE)    if (c == OP_BRA || c == OP_CBRA || c == OP_ONCE || c == OP_COND)
1450      {      {
1451      BOOL empty_branch;      BOOL empty_branch;
1452      if (GET(code, 1) == 0) return TRUE;    /* Hit unclosed bracket */      if (GET(code, 1) == 0) return TRUE;    /* Hit unclosed bracket */
# Line 1423  for (code = first_significant_code(code Line 1470  for (code = first_significant_code(code
1470    
1471    switch (c)    switch (c)
1472      {      {
1473      /* Check for quantifiers after a class */      /* Check for quantifiers after a class. XCLASS is used for classes that
1474        cannot be represented just by a bit map. This includes negated single
1475        high-valued characters. The length in _pcre_OP_lengths[] is zero; the
1476        actual length is stored in the compiled code, so we must update "code"
1477        here. */
1478    
1479  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
1480      case OP_XCLASS:      case OP_XCLASS:
1481      ccode = code + GET(code, 1);      ccode = code += GET(code, 1);
1482      goto CHECK_CLASS_REPEAT;      goto CHECK_CLASS_REPEAT;
1483  #endif  #endif
1484    
# Line 2255  for (;; ptr++) Line 2306  for (;; ptr++)
2306      */      */
2307    
2308      if (code < last_code) code = last_code;      if (code < last_code) code = last_code;
2309    
2310        /* Paranoid check for integer overflow */
2311    
2312        if (OFLOW_MAX - *lengthptr < code - last_code)
2313          {
2314          *errorcodeptr = ERR20;
2315          goto FAILED;
2316          }
2317    
2318      *lengthptr += code - last_code;      *lengthptr += code - last_code;
2319      DPRINTF(("length=%d added %d c=%c\n", *lengthptr, code - last_code, c));      DPRINTF(("length=%d added %d c=%c\n", *lengthptr, code - last_code, c));
2320    
# Line 2367  for (;; ptr++) Line 2427  for (;; ptr++)
2427      *ptrptr = ptr;      *ptrptr = ptr;
2428      if (lengthptr != NULL)      if (lengthptr != NULL)
2429        {        {
2430          if (OFLOW_MAX - *lengthptr < code - last_code)
2431            {
2432            *errorcodeptr = ERR20;
2433            goto FAILED;
2434            }
2435        *lengthptr += code - last_code;   /* To include callout length */        *lengthptr += code - last_code;   /* To include callout length */
2436        DPRINTF((">> end branch\n"));        DPRINTF((">> end branch\n"));
2437        }        }
# Line 2429  for (;; ptr++) Line 2494  for (;; ptr++)
2494        goto FAILED;        goto FAILED;
2495        }        }
2496    
2497      /* If the first character is '^', set the negation flag and skip it. */      /* If the first character is '^', set the negation flag and skip it. Also,
2498        if the first few characters (either before or after ^) are \Q\E or \E we
2499        skip them too. This makes for compatibility with Perl. */
2500    
2501      if ((c = *(++ptr)) == '^')      negate_class = FALSE;
2502        for (;;)
2503        {        {
       negate_class = TRUE;  
2504        c = *(++ptr);        c = *(++ptr);
2505        }        if (c == '\\')
2506      else          {
2507        {          if (ptr[1] == 'E') ptr++;
2508        negate_class = FALSE;            else if (strncmp((const char *)ptr+1, "Q\\E", 3) == 0) ptr += 3;
2509                else break;
2510            }
2511          else if (!negate_class && c == '^')
2512            negate_class = TRUE;
2513          else break;
2514        }        }
2515    
2516      /* Keep a count of chars with values < 256 so that we can optimize the case      /* Keep a count of chars with values < 256 so that we can optimize the case
# Line 2579  for (;; ptr++) Line 2651  for (;; ptr++)
2651        of the specials, which just set a flag. The sequence \b is a special        of the specials, which just set a flag. The sequence \b is a special
2652        case. Inside a class (and only there) it is treated as backspace.        case. Inside a class (and only there) it is treated as backspace.
2653        Elsewhere it marks a word boundary. Other escapes have preset maps ready        Elsewhere it marks a word boundary. Other escapes have preset maps ready
2654        to or into the one we are building. We assume they have more than one        to 'or' into the one we are building. We assume they have more than one
2655        character in them, so set class_charcount bigger than one. */        character in them, so set class_charcount bigger than one. */
2656    
2657        if (c == '\\')        if (c == '\\')
# Line 3521  for (;; ptr++) Line 3593  for (;; ptr++)
3593          goto FAILED;          goto FAILED;
3594          }          }
3595    
       /* This is a paranoid check to stop integer overflow later on */  
   
       if (len > MAX_DUPLENGTH)  
         {  
         *errorcodeptr = ERR50;  
         goto FAILED;  
         }  
   
3596        /* If the maximum repeat count is unlimited, find the end of the bracket        /* If the maximum repeat count is unlimited, find the end of the bracket
3597        by scanning through from the start, and compute the offset back to it        by scanning through from the start, and compute the offset back to it
3598        from the current code pointer. There may be an OP_OPT setting following        from the current code pointer. There may be an OP_OPT setting following
# Line 3617  for (;; ptr++) Line 3681  for (;; ptr++)
3681          if (repeat_min > 1)          if (repeat_min > 1)
3682            {            {
3683            /* In the pre-compile phase, we don't actually do the replication. We            /* In the pre-compile phase, we don't actually do the replication. We
3684            just adjust the length as if we had. */            just adjust the length as if we had. Do some paranoid checks for
3685              potential integer overflow. */
3686    
3687            if (lengthptr != NULL)            if (lengthptr != NULL)
3688              *lengthptr += (repeat_min - 1)*length_prevgroup;              {
3689                int delta = (repeat_min - 1)*length_prevgroup;
3690                if ((double)(repeat_min - 1)*(double)length_prevgroup >
3691                                                                (double)INT_MAX ||
3692                    OFLOW_MAX - *lengthptr < delta)
3693                  {
3694                  *errorcodeptr = ERR20;
3695                  goto FAILED;
3696                  }
3697                *lengthptr += delta;
3698                }
3699    
3700            /* This is compiling for real */            /* This is compiling for real */
3701    
# Line 3658  for (;; ptr++) Line 3733  for (;; ptr++)
3733          /* In the pre-compile phase, we don't actually do the replication. We          /* In the pre-compile phase, we don't actually do the replication. We
3734          just adjust the length as if we had. For each repetition we must add 1          just adjust the length as if we had. For each repetition we must add 1
3735          to the length for BRAZERO and for all but the last repetition we must          to the length for BRAZERO and for all but the last repetition we must
3736          add 2 + 2*LINKSIZE to allow for the nesting that occurs. */          add 2 + 2*LINKSIZE to allow for the nesting that occurs. Do some
3737            paranoid checks to avoid integer overflow. */
3738    
3739          if (lengthptr != NULL && repeat_max > 0)          if (lengthptr != NULL && repeat_max > 0)
3740            *lengthptr += repeat_max * (length_prevgroup + 1 + 2 + 2*LINK_SIZE) -            {
3741              2 - 2*LINK_SIZE;  /* Last one doesn't nest */            int delta = repeat_max * (length_prevgroup + 1 + 2 + 2*LINK_SIZE) -
3742                          2 - 2*LINK_SIZE;   /* Last one doesn't nest */
3743              if ((double)repeat_max *
3744                    (double)(length_prevgroup + 1 + 2 + 2*LINK_SIZE)
3745                      > (double)INT_MAX ||
3746                  OFLOW_MAX - *lengthptr < delta)
3747                {
3748                *errorcodeptr = ERR20;
3749                goto FAILED;
3750                }
3751              *lengthptr += delta;
3752              }
3753    
3754          /* This is compiling for real */          /* This is compiling for real */
3755    
# Line 3814  for (;; ptr++) Line 3901  for (;; ptr++)
3901      /* ===================================================================*/      /* ===================================================================*/
3902      /* Start of nested parenthesized sub-expression, or comment or lookahead or      /* Start of nested parenthesized sub-expression, or comment or lookahead or
3903      lookbehind or option setting or condition or all the other extended      lookbehind or option setting or condition or all the other extended
3904      parenthesis forms. First deal with the specials; all are introduced by ?,      parenthesis forms.  */
     and the appearance of any of them means that this is not a capturing  
     group. */  
3905    
3906      case '(':      case '(':
3907      newoptions = options;      newoptions = options;
# Line 3825  for (;; ptr++) Line 3910  for (;; ptr++)
3910      save_hwm = cd->hwm;      save_hwm = cd->hwm;
3911      reset_bracount = FALSE;      reset_bracount = FALSE;
3912    
3913      if (*(++ptr) == '?')      /* First deal with various "verbs" that can be introduced by '*'. */
3914    
3915        if (*(++ptr) == '*' && (cd->ctypes[ptr[1]] & ctype_letter) != 0)
3916          {
3917          int i, namelen;
3918          const uschar *name = ++ptr;
3919          previous = NULL;
3920          while ((cd->ctypes[*++ptr] & ctype_letter) != 0);
3921          if (*ptr == ':')
3922            {
3923            *errorcodeptr = ERR59;   /* Not supported */
3924            goto FAILED;
3925            }
3926          if (*ptr != ')')
3927            {
3928            *errorcodeptr = ERR60;
3929            goto FAILED;
3930            }
3931          namelen = ptr - name;
3932          for (i = 0; i < verbcount; i++)
3933            {
3934            if (namelen == verbs[i].len &&
3935                strncmp((char *)name, verbs[i].name, namelen) == 0)
3936              {
3937              *code = verbs[i].op;
3938              if (*code++ == OP_ACCEPT) cd->had_accept = TRUE;
3939              break;
3940              }
3941            }
3942          if (i < verbcount) continue;
3943          *errorcodeptr = ERR60;
3944          goto FAILED;
3945          }
3946    
3947        /* Deal with the extended parentheses; all are introduced by '?', and the
3948        appearance of any of them means that this is not a capturing group. */
3949    
3950        else if (*ptr == '?')
3951        {        {
3952        int i, set, unset, namelen;        int i, set, unset, namelen;
3953        int *optset;        int *optset;
# Line 4067  for (;; ptr++) Line 4189  for (;; ptr++)
4189    
4190          /* ------------------------------------------------------------ */          /* ------------------------------------------------------------ */
4191          case '!':                 /* Negative lookahead */          case '!':                 /* Negative lookahead */
         bravalue = OP_ASSERT_NOT;  
4192          ptr++;          ptr++;
4193            if (*ptr == ')')          /* Optimize (?!) */
4194              {
4195              *code++ = OP_FAIL;
4196              previous = NULL;
4197              continue;
4198              }
4199            bravalue = OP_ASSERT_NOT;
4200          break;          break;
4201    
4202    
# Line 4624  for (;; ptr++) Line 4752  for (;; ptr++)
4752    
4753      if (lengthptr != NULL)      if (lengthptr != NULL)
4754        {        {
4755          if (OFLOW_MAX - *lengthptr < length_prevgroup - 2 - 2*LINK_SIZE)
4756            {
4757            *errorcodeptr = ERR20;
4758            goto FAILED;
4759            }
4760        *lengthptr += length_prevgroup - 2 - 2*LINK_SIZE;        *lengthptr += length_prevgroup - 2 - 2*LINK_SIZE;
4761        code++;        code++;
4762        PUTINC(code, 0, 1 + LINK_SIZE);        PUTINC(code, 0, 1 + LINK_SIZE);
# Line 5119  for (;;) Line 5252  for (;;)
5252      *ptrptr = ptr;      *ptrptr = ptr;
5253      *firstbyteptr = firstbyte;      *firstbyteptr = firstbyte;
5254      *reqbyteptr = reqbyte;      *reqbyteptr = reqbyte;
5255      if (lengthptr != NULL) *lengthptr += length;      if (lengthptr != NULL)
5256          {
5257          if (OFLOW_MAX - *lengthptr < length)
5258            {
5259            *errorcodeptr = ERR20;
5260            return FALSE;
5261            }
5262          *lengthptr += length;
5263          }
5264      return TRUE;      return TRUE;
5265      }      }
5266    
# Line 5647  cd->start_code = codestart; Line 5788  cd->start_code = codestart;
5788  cd->hwm = cworkspace;  cd->hwm = cworkspace;
5789  cd->req_varyopt = 0;  cd->req_varyopt = 0;
5790  cd->nopartial = FALSE;  cd->nopartial = FALSE;
5791    cd->had_accept = FALSE;
5792    
5793  /* Set up a starting, non-extracting bracket, then compile the expression. On  /* Set up a starting, non-extracting bracket, then compile the expression. On
5794  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
# Line 5661  re->top_bracket = cd->bracount; Line 5803  re->top_bracket = cd->bracount;
5803  re->top_backref = cd->top_backref;  re->top_backref = cd->top_backref;
5804    
5805  if (cd->nopartial) re->options |= PCRE_NOPARTIAL;  if (cd->nopartial) re->options |= PCRE_NOPARTIAL;
5806    if (cd->had_accept) reqbyte = -1;   /* Must disable after (*ACCEPT) */
5807    
5808  /* 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. */
5809    

Legend:
Removed from v.195  
changed lines
  Added in v.216

  ViewVC Help
Powered by ViewVC 1.1.5