/[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 231 by ph10, Tue Sep 11 11:15:33 2007 UTC revision 327 by ph10, Sat Mar 8 19:38:30 2008 UTC
# Line 6  Line 6 
6  and semantics are as close as possible to those of the Perl 5 language.  and semantics are as close as possible to those of the Perl 5 language.
7    
8                         Written by Philip Hazel                         Written by Philip Hazel
9             Copyright (c) 1997-2007 University of Cambridge             Copyright (c) 1997-2008 University of Cambridge
10    
11  -----------------------------------------------------------------------------  -----------------------------------------------------------------------------
12  Redistribution and use in source and binary forms, with or without  Redistribution and use in source and binary forms, with or without
# 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 140  static const short int escapes[] = { Line 140  static const short int escapes[] = {
140  #endif  #endif
141    
142    
143  /* Table of special "verbs" like (*PRUNE) */  /* Table of special "verbs" like (*PRUNE). This is a short table, so it is
144    searched linearly. Put all the names into a single string, in order to reduce
145    the number of relocations when a shared library is dynamically linked. */
146    
147  typedef struct verbitem {  typedef struct verbitem {
   const char *name;  
148    int   len;    int   len;
149    int   op;    int   op;
150  } verbitem;  } verbitem;
151    
152  static verbitem verbs[] = {  static const char verbnames[] =
153    { "ACCEPT", 6, OP_ACCEPT },    "ACCEPT\0"
154    { "COMMIT", 6, OP_COMMIT },    "COMMIT\0"
155    { "F",      1, OP_FAIL },    "F\0"
156    { "FAIL",   4, OP_FAIL },    "FAIL\0"
157    { "PRUNE",  5, OP_PRUNE },    "PRUNE\0"
158    { "SKIP",   4, OP_SKIP  },    "SKIP\0"
159    { "THEN",   4, OP_THEN  }    "THEN";
160    
161    static const verbitem verbs[] = {
162      { 6, OP_ACCEPT },
163      { 6, OP_COMMIT },
164      { 1, OP_FAIL },
165      { 4, OP_FAIL },
166      { 5, OP_PRUNE },
167      { 4, OP_SKIP  },
168      { 4, OP_THEN  }
169  };  };
170    
171  static int verbcount = sizeof(verbs)/sizeof(verbitem);  static const int verbcount = sizeof(verbs)/sizeof(verbitem);
172    
173    
174  /* Tables of names of POSIX character classes and their lengths. The list is  /* Tables of names of POSIX character classes and their lengths. The names are
175  terminated by a zero length entry. The first three must be alpha, lower, upper,  now all in a single string, to reduce the number of relocations when a shared
176  as this is assumed for handling case independence. */  library is dynamically loaded. The list of lengths is terminated by a zero
177    length entry. The first three must be alpha, lower, upper, as this is assumed
178  static const char *const posix_names[] = {  for handling case independence. */
179    "alpha", "lower", "upper",  
180    "alnum", "ascii", "blank", "cntrl", "digit", "graph",  static const char posix_names[] =
181    "print", "punct", "space", "word",  "xdigit" };    "alpha\0"  "lower\0"  "upper\0"  "alnum\0"  "ascii\0"  "blank\0"
182      "cntrl\0"  "digit\0"  "graph\0"  "print\0"  "punct\0"  "space\0"
183      "word\0"   "xdigit";
184    
185  static const uschar posix_name_lengths[] = {  static const uschar posix_name_lengths[] = {
186    5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 4, 6, 0 };    5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 4, 6, 0 };
# Line 207  static const int posix_class_maps[] = { Line 219  static const int posix_class_maps[] = {
219  /* The texts of compile-time error messages. These are "char *" because they  /* The texts of compile-time error messages. These are "char *" because they
220  are passed to the outside world. Do not ever re-use any error number, because  are passed to the outside world. Do not ever re-use any error number, because
221  they are documented. Always add a new error instead. Messages marked DEAD below  they are documented. Always add a new error instead. Messages marked DEAD below
222  are no longer used. */  are no longer used. This used to be a table of strings, but in order to reduce
223    the number of relocations needed when a shared library is loaded dynamically,
224  static const char *error_texts[] = {  it is now one long string. We cannot use a table of offsets, because the
225    "no error",  lengths of inserts such as XSTRING(MAX_NAME_SIZE) are not known. Instead, we
226    "\\ at end of pattern",  simply count through to the one we want - this isn't a performance issue
227    "\\c at end of pattern",  because these strings are used only when there is a compilation error. */
228    "unrecognized character follows \\",  
229    "numbers out of order in {} quantifier",  static const char error_texts[] =
230      "no error\0"
231      "\\ at end of pattern\0"
232      "\\c at end of pattern\0"
233      "unrecognized character follows \\\0"
234      "numbers out of order in {} quantifier\0"
235    /* 5 */    /* 5 */
236    "number too big in {} quantifier",    "number too big in {} quantifier\0"
237    "missing terminating ] for character class",    "missing terminating ] for character class\0"
238    "invalid escape sequence in character class",    "invalid escape sequence in character class\0"
239    "range out of order in character class",    "range out of order in character class\0"
240    "nothing to repeat",    "nothing to repeat\0"
241    /* 10 */    /* 10 */
242    "operand of unlimited repeat could match the empty string",  /** DEAD **/    "operand of unlimited repeat could match the empty string\0"  /** DEAD **/
243    "internal error: unexpected repeat",    "internal error: unexpected repeat\0"
244    "unrecognized character after (?",    "unrecognized character after (? or (?-\0"
245    "POSIX named classes are supported only within a class",    "POSIX named classes are supported only within a class\0"
246    "missing )",    "missing )\0"
247    /* 15 */    /* 15 */
248    "reference to non-existent subpattern",    "reference to non-existent subpattern\0"
249    "erroffset passed as NULL",    "erroffset passed as NULL\0"
250    "unknown option bit(s) set",    "unknown option bit(s) set\0"
251    "missing ) after comment",    "missing ) after comment\0"
252    "parentheses nested too deeply",  /** DEAD **/    "parentheses nested too deeply\0"  /** DEAD **/
253    /* 20 */    /* 20 */
254    "regular expression is too large",    "regular expression is too large\0"
255    "failed to get memory",    "failed to get memory\0"
256    "unmatched parentheses",    "unmatched parentheses\0"
257    "internal error: code overflow",    "internal error: code overflow\0"
258    "unrecognized character after (?<",    "unrecognized character after (?<\0"
259    /* 25 */    /* 25 */
260    "lookbehind assertion is not fixed length",    "lookbehind assertion is not fixed length\0"
261    "malformed number or name after (?(",    "malformed number or name after (?(\0"
262    "conditional group contains more than two branches",    "conditional group contains more than two branches\0"
263    "assertion expected after (?(",    "assertion expected after (?(\0"
264    "(?R or (?[+-]digits must be followed by )",    "(?R or (?[+-]digits must be followed by )\0"
265    /* 30 */    /* 30 */
266    "unknown POSIX class name",    "unknown POSIX class name\0"
267    "POSIX collating elements are not supported",    "POSIX collating elements are not supported\0"
268    "this version of PCRE is not compiled with PCRE_UTF8 support",    "this version of PCRE is not compiled with PCRE_UTF8 support\0"
269    "spare error",  /** DEAD **/    "spare error\0"  /** DEAD **/
270    "character value in \\x{...} sequence is too large",    "character value in \\x{...} sequence is too large\0"
271    /* 35 */    /* 35 */
272    "invalid condition (?(0)",    "invalid condition (?(0)\0"
273    "\\C not allowed in lookbehind assertion",    "\\C not allowed in lookbehind assertion\0"
274    "PCRE does not support \\L, \\l, \\N, \\U, or \\u",    "PCRE does not support \\L, \\l, \\N, \\U, or \\u\0"
275    "number after (?C is > 255",    "number after (?C is > 255\0"
276    "closing ) for (?C expected",    "closing ) for (?C expected\0"
277    /* 40 */    /* 40 */
278    "recursive call could loop indefinitely",    "recursive call could loop indefinitely\0"
279    "unrecognized character after (?P",    "unrecognized character after (?P\0"
280    "syntax error in subpattern name (missing terminator)",    "syntax error in subpattern name (missing terminator)\0"
281    "two named subpatterns have the same name",    "two named subpatterns have the same name\0"
282    "invalid UTF-8 string",    "invalid UTF-8 string\0"
283    /* 45 */    /* 45 */
284    "support for \\P, \\p, and \\X has not been compiled",    "support for \\P, \\p, and \\X has not been compiled\0"
285    "malformed \\P or \\p sequence",    "malformed \\P or \\p sequence\0"
286    "unknown property name after \\P or \\p",    "unknown property name after \\P or \\p\0"
287    "subpattern name is too long (maximum " XSTRING(MAX_NAME_SIZE) " characters)",    "subpattern name is too long (maximum " XSTRING(MAX_NAME_SIZE) " characters)\0"
288    "too many named subpatterns (maximum " XSTRING(MAX_NAME_COUNT) ")",    "too many named subpatterns (maximum " XSTRING(MAX_NAME_COUNT) ")\0"
289    /* 50 */    /* 50 */
290    "repeated subpattern is too long",    /** DEAD **/    "repeated subpattern is too long\0"    /** DEAD **/
291    "octal value is greater than \\377 (not in UTF-8 mode)",    "octal value is greater than \\377 (not in UTF-8 mode)\0"
292    "internal error: overran compiling workspace",    "internal error: overran compiling workspace\0"
293    "internal error: previously-checked referenced subpattern not found",    "internal error: previously-checked referenced subpattern not found\0"
294    "DEFINE group contains more than one branch",    "DEFINE group contains more than one branch\0"
295    /* 55 */    /* 55 */
296    "repeating a DEFINE group is not allowed",    "repeating a DEFINE group is not allowed\0"
297    "inconsistent NEWLINE options",    "inconsistent NEWLINE options\0"
298    "\\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\0"
299    "(?+ or (?- or (?(+ or (?(- must be followed by a non-zero number",    "(?+ or (?- or (?(+ or (?(- must be followed by a non-zero number\0"
300    "(*VERB) with an argument is not supported",    "(*VERB) with an argument is not supported\0"
301    /* 60 */    /* 60 */
302    "(*VERB) not recognized",    "(*VERB) not recognized\0"
303    "number is too big"    "number is too big\0"
304  };    "subpattern name expected\0"
305      "digit expected after (?+";
306    
307    
308  /* Table to identify digits and hex digits. This is used when compiling  /* Table to identify digits and hex digits. This is used when compiling
# Line 420  static BOOL Line 438  static BOOL
438    
439    
440  /*************************************************  /*************************************************
441    *            Find an error text                  *
442    *************************************************/
443    
444    /* The error texts are now all in one long string, to save on relocations. As
445    some of the text is of unknown length, we can't use a table of offsets.
446    Instead, just count through the strings. This is not a performance issue
447    because it happens only when there has been a compilation error.
448    
449    Argument:   the error number
450    Returns:    pointer to the error string
451    */
452    
453    static const char *
454    find_error_text(int n)
455    {
456    const char *s = error_texts;
457    for (; n > 0; n--) while (*s++ != 0);
458    return s;
459    }
460    
461    
462    /*************************************************
463  *            Handle escapes                      *  *            Handle escapes                      *
464  *************************************************/  *************************************************/
465    
# Line 458  ptr--;                            /* Set Line 498  ptr--;                            /* Set
498    
499  if (c == 0) *errorcodeptr = ERR1;  if (c == 0) *errorcodeptr = ERR1;
500    
501  /* Non-alphamerics are literals. For digits or letters, do an initial lookup in  /* Non-alphanumerics are literals. For digits or letters, do an initial lookup
502  a table. A non-zero result is something that can be returned immediately.  in a table. A non-zero result is something that can be returned immediately.
503  Otherwise further processing may be required. */  Otherwise further processing may be required. */
504    
505  #ifndef EBCDIC  /* ASCII coding */  #ifndef EBCDIC  /* ASCII coding */
506  else if (c < '0' || c > 'z') {}                           /* Not alphameric */  else if (c < '0' || c > 'z') {}                           /* Not alphanumeric */
507  else if ((i = escapes[c - '0']) != 0) c = i;  else if ((i = escapes[c - '0']) != 0) c = i;
508    
509  #else           /* EBCDIC coding */  #else           /* EBCDIC coding */
510  else if (c < 'a' || (ebcdic_chartab[c] & 0x0E) == 0) {}   /* Not alphameric */  else if (c < 'a' || (ebcdic_chartab[c] & 0x0E) == 0) {}   /* Not alphanumeric */
511  else if ((i = escapes[c - 0x48]) != 0)  c = i;  else if ((i = escapes[c - 0x48]) != 0)  c = i;
512  #endif  #endif
513    
# Line 684  else Line 724  else
724      break;      break;
725    
726      /* PCRE_EXTRA enables extensions to Perl in the matter of escapes. Any      /* PCRE_EXTRA enables extensions to Perl in the matter of escapes. Any
727      other alphameric following \ is an error if PCRE_EXTRA was set; otherwise,      other alphanumeric following \ is an error if PCRE_EXTRA was set;
728      for Perl compatibility, it is a literal. This code looks a bit odd, but      otherwise, for Perl compatibility, it is a literal. This code looks a bit
729      there used to be some cases other than the default, and there may be again      odd, but there used to be some cases other than the default, and there may
730      in future, so I haven't "optimized" it. */      be again in future, so I haven't "optimized" it. */
731    
732      default:      default:
733      if ((options & PCRE_EXTRA) != 0) switch(c)      if ((options & PCRE_EXTRA) != 0) switch(c)
# Line 776  top = _pcre_utt_size; Line 816  top = _pcre_utt_size;
816  while (bot < top)  while (bot < top)
817    {    {
818    i = (bot + top) >> 1;    i = (bot + top) >> 1;
819    c = strcmp(name, _pcre_utt[i].name);    c = strcmp(name, _pcre_utt_names + _pcre_utt[i].name_offset);
820    if (c == 0)    if (c == 0)
821      {      {
822      *dptr = _pcre_utt[i].value;      *dptr = _pcre_utt[i].value;
# Line 1468  for (;;) Line 1508  for (;;)
1508  can match the empty string or not. It is called from could_be_empty()  can match the empty string or not. It is called from could_be_empty()
1509  below and from compile_branch() when checking for an unlimited repeat of a  below and from compile_branch() when checking for an unlimited repeat of a
1510  group that can match nothing. Note that first_significant_code() skips over  group that can match nothing. Note that first_significant_code() skips over
1511  assertions. If we hit an unclosed bracket, we return "empty" - this means we've  backward and negative forward assertions when its final argument is TRUE. If we
1512  struck an inner bracket whose current branch will already have been scanned.  hit an unclosed bracket, we return "empty" - this means we've struck an inner
1513    bracket whose current branch will already have been scanned.
1514    
1515  Arguments:  Arguments:
1516    code        points to start of search    code        points to start of search
# Line 1491  for (code = first_significant_code(code Line 1532  for (code = first_significant_code(code
1532    
1533    c = *code;    c = *code;
1534    
1535      /* Skip over forward assertions; the other assertions are skipped by
1536      first_significant_code() with a TRUE final argument. */
1537    
1538      if (c == OP_ASSERT)
1539        {
1540        do code += GET(code, 1); while (*code == OP_ALT);
1541        c = *code;
1542        continue;
1543        }
1544    
1545    /* Groups with zero repeats can of course be empty; skip them. */    /* Groups with zero repeats can of course be empty; skip them. */
1546    
1547    if (c == OP_BRAZERO || c == OP_BRAMINZERO)    if (c == OP_BRAZERO || c == OP_BRAMINZERO)
# Line 1686  return TRUE; Line 1737  return TRUE;
1737  *************************************************/  *************************************************/
1738    
1739  /* This function is called when the sequence "[:" or "[." or "[=" is  /* This function is called when the sequence "[:" or "[." or "[=" is
1740  encountered in a character class. It checks whether this is followed by an  encountered in a character class. It checks whether this is followed by a
1741  optional ^ and then a sequence of letters, terminated by a matching ":]" or  sequence of characters terminated by a matching ":]" or ".]" or "=]". If we
1742  ".]" or "=]".  reach an unescaped ']' without the special preceding character, return FALSE.
1743    
1744    Originally, this function only recognized a sequence of letters between the
1745    terminators, but it seems that Perl recognizes any sequence of characters,
1746    though of course unknown POSIX names are subsequently rejected. Perl gives an
1747    "Unknown POSIX class" error for [:f\oo:] for example, where previously PCRE
1748    didn't consider this to be a POSIX class. Likewise for [:1234:].
1749    
1750    The problem in trying to be exactly like Perl is in the handling of escapes. We
1751    have to be sure that [abc[:x\]pqr] is *not* treated as containing a POSIX
1752    class, but [abc[:x\]pqr:]] is (so that an error can be generated). The code
1753    below handles the special case of \], but does not try to do any other escape
1754    processing. This makes it different from Perl for cases such as [:l\ower:]
1755    where Perl recognizes it as the POSIX class "lower" but PCRE does not recognize
1756    "l\ower". This is a lesser evil that not diagnosing bad classes when Perl does,
1757    I think.
1758    
1759  Argument:  Arguments:
1760    ptr      pointer to the initial [    ptr      pointer to the initial [
1761    endptr   where to return the end pointer    endptr   where to return the end pointer
   cd       pointer to compile data  
1762    
1763  Returns:   TRUE or FALSE  Returns:   TRUE or FALSE
1764  */  */
1765    
1766  static BOOL  static BOOL
1767  check_posix_syntax(const uschar *ptr, const uschar **endptr, compile_data *cd)  check_posix_syntax(const uschar *ptr, const uschar **endptr)
1768  {  {
1769  int terminator;          /* Don't combine these lines; the Solaris cc */  int terminator;          /* Don't combine these lines; the Solaris cc */
1770  terminator = *(++ptr);   /* compiler warns about "non-constant" initializer. */  terminator = *(++ptr);   /* compiler warns about "non-constant" initializer. */
1771  if (*(++ptr) == '^') ptr++;  for (++ptr; *ptr != 0; ptr++)
 while ((cd->ctypes[*ptr] & ctype_letter) != 0) ptr++;  
 if (*ptr == terminator && ptr[1] == ']')  
1772    {    {
1773    *endptr = ptr;    if (*ptr == '\\' && ptr[1] == ']') ptr++; else
1774    return TRUE;      {
1775        if (*ptr == ']') return FALSE;
1776        if (*ptr == terminator && ptr[1] == ']')
1777          {
1778          *endptr = ptr;
1779          return TRUE;
1780          }
1781        }
1782    }    }
1783  return FALSE;  return FALSE;
1784  }  }
# Line 1733  Returns:     a value representing the na Line 1803  Returns:     a value representing the na
1803  static int  static int
1804  check_posix_name(const uschar *ptr, int len)  check_posix_name(const uschar *ptr, int len)
1805  {  {
1806    const char *pn = posix_names;
1807  register int yield = 0;  register int yield = 0;
1808  while (posix_name_lengths[yield] != 0)  while (posix_name_lengths[yield] != 0)
1809    {    {
1810    if (len == posix_name_lengths[yield] &&    if (len == posix_name_lengths[yield] &&
1811      strncmp((const char *)ptr, posix_names[yield], len) == 0) return yield;      strncmp((const char *)ptr, pn, len) == 0) return yield;
1812      pn += posix_name_lengths[yield] + 1;
1813    yield++;    yield++;
1814    }    }
1815  return -1;  return -1;
# Line 2041  if (next >= 0) switch(op_code) Line 2113  if (next >= 0) switch(op_code)
2113    /* For OP_NOT, "item" must be a single-byte character. */    /* For OP_NOT, "item" must be a single-byte character. */
2114    
2115    case OP_NOT:    case OP_NOT:
   if (next < 0) return FALSE;  /* Not a character */  
2116    if (item == next) return TRUE;    if (item == next) return TRUE;
2117    if ((options & PCRE_CASELESS) == 0) return FALSE;    if ((options & PCRE_CASELESS) == 0) return FALSE;
2118  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
# Line 2304  uschar classbits[32]; Line 2375  uschar classbits[32];
2375  BOOL class_utf8;  BOOL class_utf8;
2376  BOOL utf8 = (options & PCRE_UTF8) != 0;  BOOL utf8 = (options & PCRE_UTF8) != 0;
2377  uschar *class_utf8data;  uschar *class_utf8data;
2378    uschar *class_utf8data_base;
2379  uschar utf8_char[6];  uschar utf8_char[6];
2380  #else  #else
2381  BOOL utf8 = FALSE;  BOOL utf8 = FALSE;
# Line 2343  req_caseopt = ((options & PCRE_CASELESS) Line 2415  req_caseopt = ((options & PCRE_CASELESS)
2415  for (;; ptr++)  for (;; ptr++)
2416    {    {
2417    BOOL negate_class;    BOOL negate_class;
2418      BOOL should_flip_negation;
2419    BOOL possessive_quantifier;    BOOL possessive_quantifier;
2420    BOOL is_quantifier;    BOOL is_quantifier;
2421    BOOL is_recurse;    BOOL is_recurse;
# Line 2566  for (;; ptr++) Line 2639  for (;; ptr++)
2639      they are encountered at the top level, so we'll do that too. */      they are encountered at the top level, so we'll do that too. */
2640    
2641      if ((ptr[1] == ':' || ptr[1] == '.' || ptr[1] == '=') &&      if ((ptr[1] == ':' || ptr[1] == '.' || ptr[1] == '=') &&
2642          check_posix_syntax(ptr, &tempptr, cd))          check_posix_syntax(ptr, &tempptr))
2643        {        {
2644        *errorcodeptr = (ptr[1] == ':')? ERR13 : ERR31;        *errorcodeptr = (ptr[1] == ':')? ERR13 : ERR31;
2645        goto FAILED;        goto FAILED;
# Line 2591  for (;; ptr++) Line 2664  for (;; ptr++)
2664        else break;        else break;
2665        }        }
2666    
2667        /* If a class contains a negative special such as \S, we need to flip the
2668        negation flag at the end, so that support for characters > 255 works
2669        correctly (they are all included in the class). */
2670    
2671        should_flip_negation = FALSE;
2672    
2673      /* 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
2674      of just a single character (as long as it's < 256). However, For higher      of just a single character (as long as it's < 256). However, For higher
2675      valued UTF-8 characters, we don't yet do any optimization. */      valued UTF-8 characters, we don't yet do any optimization. */
# Line 2608  for (;; ptr++) Line 2687  for (;; ptr++)
2687  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
2688      class_utf8 = FALSE;                       /* No chars >= 256 */      class_utf8 = FALSE;                       /* No chars >= 256 */
2689      class_utf8data = code + LINK_SIZE + 2;    /* For UTF-8 items */      class_utf8data = code + LINK_SIZE + 2;    /* For UTF-8 items */
2690        class_utf8data_base = class_utf8data;     /* For resetting in pass 1 */
2691  #endif  #endif
2692    
2693      /* Process characters until ] is reached. By writing this as a "do" it      /* Process characters until ] is reached. By writing this as a "do" it
# Line 2623  for (;; ptr++) Line 2703  for (;; ptr++)
2703          {                           /* Braces are required because the */          {                           /* Braces are required because the */
2704          GETCHARLEN(c, ptr, ptr);    /* macro generates multiple statements */          GETCHARLEN(c, ptr, ptr);    /* macro generates multiple statements */
2705          }          }
2706    
2707          /* In the pre-compile phase, accumulate the length of any UTF-8 extra
2708          data and reset the pointer. This is so that very large classes that
2709          contain a zillion UTF-8 characters no longer overwrite the work space
2710          (which is on the stack). */
2711    
2712          if (lengthptr != NULL)
2713            {
2714            *lengthptr += class_utf8data - class_utf8data_base;
2715            class_utf8data = class_utf8data_base;
2716            }
2717    
2718  #endif  #endif
2719    
2720        /* Inside \Q...\E everything is literal except \E */        /* Inside \Q...\E everything is literal except \E */
# Line 2646  for (;; ptr++) Line 2738  for (;; ptr++)
2738    
2739        if (c == '[' &&        if (c == '[' &&
2740            (ptr[1] == ':' || ptr[1] == '.' || ptr[1] == '=') &&            (ptr[1] == ':' || ptr[1] == '.' || ptr[1] == '=') &&
2741            check_posix_syntax(ptr, &tempptr, cd))            check_posix_syntax(ptr, &tempptr))
2742          {          {
2743          BOOL local_negate = FALSE;          BOOL local_negate = FALSE;
2744          int posix_class, taboffset, tabopt;          int posix_class, taboffset, tabopt;
# Line 2663  for (;; ptr++) Line 2755  for (;; ptr++)
2755          if (*ptr == '^')          if (*ptr == '^')
2756            {            {
2757            local_negate = TRUE;            local_negate = TRUE;
2758              should_flip_negation = TRUE;  /* Note negative special */
2759            ptr++;            ptr++;
2760            }            }
2761    
# Line 2737  for (;; ptr++) Line 2830  for (;; ptr++)
2830          c = check_escape(&ptr, errorcodeptr, cd->bracount, options, TRUE);          c = check_escape(&ptr, errorcodeptr, cd->bracount, options, TRUE);
2831          if (*errorcodeptr != 0) goto FAILED;          if (*errorcodeptr != 0) goto FAILED;
2832    
2833          if (-c == ESC_b) c = '\b';       /* \b is backslash in a class */          if (-c == ESC_b) c = '\b';       /* \b is backspace in a class */
2834          else if (-c == ESC_X) c = 'X';   /* \X is literal X in a class */          else if (-c == ESC_X) c = 'X';   /* \X is literal X in a class */
2835          else if (-c == ESC_R) c = 'R';   /* \R is literal R in a class */          else if (-c == ESC_R) c = 'R';   /* \R is literal R in a class */
2836          else if (-c == ESC_Q)            /* Handle start of quoted string */          else if (-c == ESC_Q)            /* Handle start of quoted string */
# Line 2765  for (;; ptr++) Line 2858  for (;; ptr++)
2858              continue;              continue;
2859    
2860              case ESC_D:              case ESC_D:
2861                should_flip_negation = TRUE;
2862              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_digit];              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_digit];
2863              continue;              continue;
2864    
# Line 2773  for (;; ptr++) Line 2867  for (;; ptr++)
2867              continue;              continue;
2868    
2869              case ESC_W:              case ESC_W:
2870                should_flip_negation = TRUE;
2871              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_word];              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_word];
2872              continue;              continue;
2873    
# Line 2782  for (;; ptr++) Line 2877  for (;; ptr++)
2877              continue;              continue;
2878    
2879              case ESC_S:              case ESC_S:
2880                should_flip_negation = TRUE;
2881              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_space];              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_space];
2882              classbits[1] |= 0x08;    /* Perl 5.004 onwards omits VT from \s */              classbits[1] |= 0x08;    /* Perl 5.004 onwards omits VT from \s */
2883              continue;              continue;
2884    
             case ESC_E: /* Perl ignores an orphan \E */  
             continue;  
   
2885              default:    /* Not recognized; fall through */              default:    /* Not recognized; fall through */
2886              break;      /* Need "default" setting to stop compiler warning. */              break;      /* Need "default" setting to stop compiler warning. */
2887              }              }
# Line 3023  for (;; ptr++) Line 3116  for (;; ptr++)
3116            d = check_escape(&ptr, errorcodeptr, cd->bracount, options, TRUE);            d = check_escape(&ptr, errorcodeptr, cd->bracount, options, TRUE);
3117            if (*errorcodeptr != 0) goto FAILED;            if (*errorcodeptr != 0) goto FAILED;
3118    
3119            /* \b is backslash; \X is literal X; \R is literal R; any other            /* \b is backspace; \X is literal X; \R is literal R; any other
3120            special means the '-' was literal */            special means the '-' was literal */
3121    
3122            if (d < 0)            if (d < 0)
# Line 3287  we set the flag only if there is a liter Line 3380  we set the flag only if there is a liter
3380      zeroreqbyte = reqbyte;      zeroreqbyte = reqbyte;
3381    
3382      /* If there are characters with values > 255, we have to compile an      /* If there are characters with values > 255, we have to compile an
3383      extended class, with its own opcode. If there are no characters < 256,      extended class, with its own opcode, unless there was a negated special
3384      we can omit the bitmap in the actual compiled code. */      such as \S in the class, because in that case all characters > 255 are in
3385        the class, so any that were explicitly given as well can be ignored. If
3386        (when there are explicit characters > 255 that must be listed) there are no
3387        characters < 256, we can omit the bitmap in the actual compiled code. */
3388    
3389  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
3390      if (class_utf8)      if (class_utf8 && !should_flip_negation)
3391        {        {
3392        *class_utf8data++ = XCL_END;    /* Marks the end of extra data */        *class_utf8data++ = XCL_END;    /* Marks the end of extra data */
3393        *code++ = OP_XCLASS;        *code++ = OP_XCLASS;
# Line 3317  we set the flag only if there is a liter Line 3413  we set the flag only if there is a liter
3413        }        }
3414  #endif  #endif
3415    
3416      /* If there are no characters > 255, negate the 32-byte map if necessary,      /* If there are no characters > 255, set the opcode to OP_CLASS or
3417      and copy it into the code vector. If this is the first thing in the branch,      OP_NCLASS, depending on whether the whole class was negated and whether
3418      there can be no first char setting, whatever the repeat count. Any reqbyte      there were negative specials such as \S in the class. Then copy the 32-byte
3419      setting must remain unchanged after any kind of repeat. */      map into the code vector, negating it if necessary. */
3420    
3421        *code++ = (negate_class == should_flip_negation) ? OP_CLASS : OP_NCLASS;
3422      if (negate_class)      if (negate_class)
3423        {        {
       *code++ = OP_NCLASS;  
3424        if (lengthptr == NULL)    /* Save time in the pre-compile phase */        if (lengthptr == NULL)    /* Save time in the pre-compile phase */
3425          for (c = 0; c < 32; c++) code[c] = ~classbits[c];          for (c = 0; c < 32; c++) code[c] = ~classbits[c];
3426        }        }
3427      else      else
3428        {        {
       *code++ = OP_CLASS;  
3429        memcpy(code, classbits, 32);        memcpy(code, classbits, 32);
3430        }        }
3431      code += 32;      code += 32;
# Line 3966  we set the flag only if there is a liter Line 4061  we set the flag only if there is a liter
4061        int len;        int len;
4062        if (*tempcode == OP_EXACT || *tempcode == OP_TYPEEXACT ||        if (*tempcode == OP_EXACT || *tempcode == OP_TYPEEXACT ||
4063            *tempcode == OP_NOTEXACT)            *tempcode == OP_NOTEXACT)
4064          tempcode += _pcre_OP_lengths[*tempcode];          tempcode += _pcre_OP_lengths[*tempcode] +
4065              ((*tempcode == OP_TYPEEXACT &&
4066                 (tempcode[3] == OP_PROP || tempcode[3] == OP_NOTPROP))? 2:0);
4067        len = code - tempcode;        len = code - tempcode;
4068        if (len > 0) switch (*tempcode)        if (len > 0) switch (*tempcode)
4069          {          {
# Line 4024  we set the flag only if there is a liter Line 4121  we set the flag only if there is a liter
4121      if (*(++ptr) == '*' && (cd->ctypes[ptr[1]] & ctype_letter) != 0)      if (*(++ptr) == '*' && (cd->ctypes[ptr[1]] & ctype_letter) != 0)
4122        {        {
4123        int i, namelen;        int i, namelen;
4124          const char *vn = verbnames;
4125        const uschar *name = ++ptr;        const uschar *name = ++ptr;
4126        previous = NULL;        previous = NULL;
4127        while ((cd->ctypes[*++ptr] & ctype_letter) != 0);        while ((cd->ctypes[*++ptr] & ctype_letter) != 0);
# Line 4041  we set the flag only if there is a liter Line 4139  we set the flag only if there is a liter
4139        for (i = 0; i < verbcount; i++)        for (i = 0; i < verbcount; i++)
4140          {          {
4141          if (namelen == verbs[i].len &&          if (namelen == verbs[i].len &&
4142              strncmp((char *)name, verbs[i].name, namelen) == 0)              strncmp((char *)name, vn, namelen) == 0)
4143            {            {
4144            *code = verbs[i].op;            *code = verbs[i].op;
4145            if (*code++ == OP_ACCEPT) cd->had_accept = TRUE;            if (*code++ == OP_ACCEPT) cd->had_accept = TRUE;
4146            break;            break;
4147            }            }
4148            vn += verbs[i].len + 1;
4149          }          }
4150        if (i < verbcount) continue;        if (i < verbcount) continue;
4151        *errorcodeptr = ERR60;        *errorcodeptr = ERR60;
# Line 4191  we set the flag only if there is a liter Line 4290  we set the flag only if there is a liter
4290              *errorcodeptr = ERR58;              *errorcodeptr = ERR58;
4291              goto FAILED;              goto FAILED;
4292              }              }
4293            if (refsign == '-')            recno = (refsign == '-')?
4294                cd->bracount - recno + 1 : recno +cd->bracount;
4295              if (recno <= 0 || recno > cd->final_bracount)
4296              {              {
4297              recno = cd->bracount - recno + 1;              *errorcodeptr = ERR15;
4298              if (recno <= 0)              goto FAILED;
               {  
               *errorcodeptr = ERR15;  
               goto FAILED;  
               }  
4299              }              }
           else recno += cd->bracount;  
4300            PUT2(code, 2+LINK_SIZE, recno);            PUT2(code, 2+LINK_SIZE, recno);
4301            break;            break;
4302            }            }
# Line 4272  we set the flag only if there is a liter Line 4368  we set the flag only if there is a liter
4368            skipbytes = 1;            skipbytes = 1;
4369            }            }
4370    
4371          /* Check for the "name" actually being a subpattern number. */          /* Check for the "name" actually being a subpattern number. We are
4372            in the second pass here, so final_bracount is set. */
4373    
4374          else if (recno > 0)          else if (recno > 0 && recno <= cd->final_bracount)
4375            {            {
4376            PUT2(code, 2+LINK_SIZE, recno);            PUT2(code, 2+LINK_SIZE, recno);
4377            }            }
# Line 4468  we set the flag only if there is a liter Line 4565  we set the flag only if there is a liter
4565    
4566          /* We come here from the Python syntax above that handles both          /* We come here from the Python syntax above that handles both
4567          references (?P=name) and recursion (?P>name), as well as falling          references (?P=name) and recursion (?P>name), as well as falling
4568          through from the Perl recursion syntax (?&name). */          through from the Perl recursion syntax (?&name). We also come here from
4569            the Perl \k<name> or \k'name' back reference syntax and the \k{name}
4570            .NET syntax. */
4571    
4572          NAMED_REF_OR_RECURSE:          NAMED_REF_OR_RECURSE:
4573          name = ++ptr;          name = ++ptr;
# Line 4480  we set the flag only if there is a liter Line 4579  we set the flag only if there is a liter
4579    
4580          if (lengthptr != NULL)          if (lengthptr != NULL)
4581            {            {
4582              if (namelen == 0)
4583                {
4584                *errorcodeptr = ERR62;
4585                goto FAILED;
4586                }
4587            if (*ptr != terminator)            if (*ptr != terminator)
4588              {              {
4589              *errorcodeptr = ERR42;              *errorcodeptr = ERR42;
# Line 4493  we set the flag only if there is a liter Line 4597  we set the flag only if there is a liter
4597            recno = 0;            recno = 0;
4598            }            }
4599    
4600          /* In the real compile, seek the name in the table */          /* In the real compile, seek the name in the table. We check the name
4601            first, and then check that we have reached the end of the name in the
4602            table. That way, if the name that is longer than any in the table,
4603            the comparison will fail without reading beyond the table entry. */
4604    
4605          else          else
4606            {            {
4607            slot = cd->name_table;            slot = cd->name_table;
4608            for (i = 0; i < cd->names_found; i++)            for (i = 0; i < cd->names_found; i++)
4609              {              {
4610              if (strncmp((char *)name, (char *)slot+2, namelen) == 0) break;              if (strncmp((char *)name, (char *)slot+2, namelen) == 0 &&
4611                    slot[2+namelen] == 0)
4612                  break;
4613              slot += cd->name_entry_size;              slot += cd->name_entry_size;
4614              }              }
4615    
# Line 4537  we set the flag only if there is a liter Line 4646  we set the flag only if there is a liter
4646            {            {
4647            const uschar *called;            const uschar *called;
4648    
4649            if ((refsign = *ptr) == '+') ptr++;            if ((refsign = *ptr) == '+')
4650                {
4651                ptr++;
4652                if ((digitab[*ptr] & ctype_digit) == 0)
4653                  {
4654                  *errorcodeptr = ERR63;
4655                  goto FAILED;
4656                  }
4657                }
4658            else if (refsign == '-')            else if (refsign == '-')
4659              {              {
4660              if ((digitab[ptr[1]] & ctype_digit) == 0)              if ((digitab[ptr[1]] & ctype_digit) == 0)
# Line 5703  to fill in forward references to subpatt Line 5820  to fill in forward references to subpatt
5820    
5821  uschar cworkspace[COMPILE_WORK_SIZE];  uschar cworkspace[COMPILE_WORK_SIZE];
5822    
   
5823  /* Set this early so that early errors get offset 0. */  /* Set this early so that early errors get offset 0. */
5824    
5825  ptr = (const uschar *)pattern;  ptr = (const uschar *)pattern;
# Line 5864  to compile parts of the pattern into; th Line 5980  to compile parts of the pattern into; th
5980  no longer needed, so hopefully this workspace will never overflow, though there  no longer needed, so hopefully this workspace will never overflow, though there
5981  is a test for its doing so. */  is a test for its doing so. */
5982    
5983  cd->bracount = 0;  cd->bracount = cd->final_bracount = 0;
5984  cd->names_found = 0;  cd->names_found = 0;
5985  cd->name_entry_size = 0;  cd->name_entry_size = 0;
5986  cd->name_table = NULL;  cd->name_table = NULL;
# Line 5941  field. Reset the bracket count and the n Line 6057  field. Reset the bracket count and the n
6057  field; this time it's used for remembering forward references to subpatterns.  field; this time it's used for remembering forward references to subpatterns.
6058  */  */
6059    
6060    cd->final_bracount = cd->bracount;  /* Save for checking forward references */
6061  cd->bracount = 0;  cd->bracount = 0;
6062  cd->names_found = 0;  cd->names_found = 0;
6063  cd->name_table = (uschar *)re + re->name_table_offset;  cd->name_table = (uschar *)re + re->name_table_offset;
# Line 6005  if (errorcode != 0) Line 6122  if (errorcode != 0)
6122    PCRE_EARLY_ERROR_RETURN:    PCRE_EARLY_ERROR_RETURN:
6123    *erroroffset = ptr - (const uschar *)pattern;    *erroroffset = ptr - (const uschar *)pattern;
6124    PCRE_EARLY_ERROR_RETURN2:    PCRE_EARLY_ERROR_RETURN2:
6125    *errorptr = error_texts[errorcode];    *errorptr = find_error_text(errorcode);
6126    if (errorcodeptr != NULL) *errorcodeptr = errorcode;    if (errorcodeptr != NULL) *errorcodeptr = errorcode;
6127    return NULL;    return NULL;
6128    }    }
# Line 6090  was compiled can be seen. */ Line 6207  was compiled can be seen. */
6207  if (code - codestart > length)  if (code - codestart > length)
6208    {    {
6209    (pcre_free)(re);    (pcre_free)(re);
6210    *errorptr = error_texts[ERR23];    *errorptr = find_error_text(ERR23);
6211    *erroroffset = ptr - (uschar *)pattern;    *erroroffset = ptr - (uschar *)pattern;
6212    if (errorcodeptr != NULL) *errorcodeptr = ERR23;    if (errorcodeptr != NULL) *errorcodeptr = ERR23;
6213    return NULL;    return NULL;

Legend:
Removed from v.231  
changed lines
  Added in v.327

  ViewVC Help
Powered by ViewVC 1.1.5