/[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 230 by ph10, Mon Sep 10 13:23:56 2007 UTC revision 275 by ph10, Wed Nov 21 15:35:09 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 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 const char verbnames[] =
153      "ACCEPT\0"
154      "COMMIT\0"
155      "F\0"
156      "FAIL\0"
157      "PRUNE\0"
158      "SKIP\0"
159      "THEN";
160    
161  static verbitem verbs[] = {  static verbitem verbs[] = {
162    { "ACCEPT", 6, OP_ACCEPT },    { 6, OP_ACCEPT },
163    { "COMMIT", 6, OP_COMMIT },    { 6, OP_COMMIT },
164    { "F",      1, OP_FAIL },    { 1, OP_FAIL },
165    { "FAIL",   4, OP_FAIL },    { 4, OP_FAIL },
166    { "PRUNE",  5, OP_PRUNE },    { 5, OP_PRUNE },
167    { "SKIP",   4, OP_SKIP  },    { 4, OP_SKIP  },
168    { "THEN",   4, OP_THEN  }    { 4, OP_THEN  }
169  };  };
170    
171  static int verbcount = sizeof(verbs)/sizeof(verbitem);  static 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 1733  Returns: a value representing the na Line 1773  Returns: a value representing the na
1773  static int  static int
1774  check_posix_name(const uschar *ptr, int len)  check_posix_name(const uschar *ptr, int len)
1775  {  {
1776    const char *pn = posix_names;
1777  register int yield = 0;  register int yield = 0;
1778  while (posix_name_lengths[yield] != 0)  while (posix_name_lengths[yield] != 0)
1779    {    {
1780    if (len == posix_name_lengths[yield] &&    if (len == posix_name_lengths[yield] &&
1781      strncmp((const char *)ptr, posix_names[yield], len) == 0) return yield;      strncmp((const char *)ptr, pn, len) == 0) return yield;
1782      pn += posix_name_lengths[yield] + 1;
1783    yield++;    yield++;
1784    }    }
1785  return -1;  return -1;
# Line 2343  req_caseopt = ((options & PCRE_CASELESS) Line 2385  req_caseopt = ((options & PCRE_CASELESS)
2385  for (;; ptr++)  for (;; ptr++)
2386    {    {
2387    BOOL negate_class;    BOOL negate_class;
2388      BOOL should_flip_negation;
2389    BOOL possessive_quantifier;    BOOL possessive_quantifier;
2390    BOOL is_quantifier;    BOOL is_quantifier;
2391    BOOL is_recurse;    BOOL is_recurse;
# Line 2591  for (;; ptr++) Line 2634  for (;; ptr++)
2634        else break;        else break;
2635        }        }
2636    
2637        /* If a class contains a negative special such as \S, we need to flip the
2638        negation flag at the end, so that support for characters > 255 works
2639        correctly (they are all included in the class). */
2640    
2641        should_flip_negation = FALSE;
2642    
2643      /* 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
2644      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
2645      valued UTF-8 characters, we don't yet do any optimization. */      valued UTF-8 characters, we don't yet do any optimization. */
# Line 2663  for (;; ptr++) Line 2712  for (;; ptr++)
2712          if (*ptr == '^')          if (*ptr == '^')
2713            {            {
2714            local_negate = TRUE;            local_negate = TRUE;
2715              should_flip_negation = TRUE;  /* Note negative special */
2716            ptr++;            ptr++;
2717            }            }
2718    
# Line 2737  for (;; ptr++) Line 2787  for (;; ptr++)
2787          c = check_escape(&ptr, errorcodeptr, cd->bracount, options, TRUE);          c = check_escape(&ptr, errorcodeptr, cd->bracount, options, TRUE);
2788          if (*errorcodeptr != 0) goto FAILED;          if (*errorcodeptr != 0) goto FAILED;
2789    
2790          if (-c == ESC_b) c = '\b';       /* \b is backslash in a class */          if (-c == ESC_b) c = '\b';       /* \b is backspace in a class */
2791          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 */
2792          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 */
2793          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 2815  for (;; ptr++)
2815              continue;              continue;
2816    
2817              case ESC_D:              case ESC_D:
2818                should_flip_negation = TRUE;
2819              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_digit];              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_digit];
2820              continue;              continue;
2821    
# Line 2773  for (;; ptr++) Line 2824  for (;; ptr++)
2824              continue;              continue;
2825    
2826              case ESC_W:              case ESC_W:
2827                should_flip_negation = TRUE;
2828              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_word];              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_word];
2829              continue;              continue;
2830    
# Line 2782  for (;; ptr++) Line 2834  for (;; ptr++)
2834              continue;              continue;
2835    
2836              case ESC_S:              case ESC_S:
2837                should_flip_negation = TRUE;
2838              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_space];              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_space];
2839              classbits[1] |= 0x08;    /* Perl 5.004 onwards omits VT from \s */              classbits[1] |= 0x08;    /* Perl 5.004 onwards omits VT from \s */
2840              continue;              continue;
# Line 2975  for (;; ptr++) Line 3028  for (;; ptr++)
3028          }          }
3029    
3030        oldptr = ptr;        oldptr = ptr;
3031    
3032        /* Remember \r or \n */        /* Remember \r or \n */
3033    
3034        if (c == '\r' || c == '\n') cd->external_flags |= PCRE_HASCRORLF;        if (c == '\r' || c == '\n') cd->external_flags |= PCRE_HASCRORLF;
3035    
3036        /* Check for range */        /* Check for range */
3037    
3038        if (!inescq && ptr[1] == '-')        if (!inescq && ptr[1] == '-')
# Line 3023  for (;; ptr++) Line 3076  for (;; ptr++)
3076            d = check_escape(&ptr, errorcodeptr, cd->bracount, options, TRUE);            d = check_escape(&ptr, errorcodeptr, cd->bracount, options, TRUE);
3077            if (*errorcodeptr != 0) goto FAILED;            if (*errorcodeptr != 0) goto FAILED;
3078    
3079            /* \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
3080            special means the '-' was literal */            special means the '-' was literal */
3081    
3082            if (d < 0)            if (d < 0)
# Line 3050  for (;; ptr++) Line 3103  for (;; ptr++)
3103          if (d == c) goto LONE_SINGLE_CHARACTER;  /* A few lines below */          if (d == c) goto LONE_SINGLE_CHARACTER;  /* A few lines below */
3104    
3105          /* Remember \r or \n */          /* Remember \r or \n */
3106    
3107          if (d == '\r' || d == '\n') cd->external_flags |= PCRE_HASCRORLF;          if (d == '\r' || d == '\n') cd->external_flags |= PCRE_HASCRORLF;
3108    
3109          /* 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
3110          matching, we have to use an XCLASS with extra data items. Caseless          matching, we have to use an XCLASS with extra data items. Caseless
3111          matching for characters > 127 is available only if UCP support is          matching for characters > 127 is available only if UCP support is
# Line 3157  for (;; ptr++) Line 3210  for (;; ptr++)
3210        apparent range that isn't. */        apparent range that isn't. */
3211    
3212        LONE_SINGLE_CHARACTER:        LONE_SINGLE_CHARACTER:
3213    
3214        /* Handle a character that cannot go in the bit map */        /* Handle a character that cannot go in the bit map */
3215    
3216  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
# Line 3205  for (;; ptr++) Line 3258  for (;; ptr++)
3258        *errorcodeptr = ERR6;        *errorcodeptr = ERR6;
3259        goto FAILED;        goto FAILED;
3260        }        }
3261    
3262    
3263  /* This code has been disabled because it would mean that \s counts as  /* This code has been disabled because it would mean that \s counts as
3264  an explicit \r or \n reference, and that's not really what is wanted. Now  an explicit \r or \n reference, and that's not really what is wanted. Now
3265  we set the flag only if there is a literal "\r" or "\n" in the class. */  we set the flag only if there is a literal "\r" or "\n" in the class. */
# Line 3223  we set the flag only if there is a liter Line 3276  we set the flag only if there is a liter
3276        if ((classbits[1] & 0x24) != 0) cd->external_flags |= PCRE_HASCRORLF;        if ((classbits[1] & 0x24) != 0) cd->external_flags |= PCRE_HASCRORLF;
3277        }        }
3278  #endif  #endif
3279    
3280    
3281      /* 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
3282      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
# Line 3287  we set the flag only if there is a liter Line 3340  we set the flag only if there is a liter
3340      zeroreqbyte = reqbyte;      zeroreqbyte = reqbyte;
3341    
3342      /* If there are characters with values > 255, we have to compile an      /* If there are characters with values > 255, we have to compile an
3343      extended class, with its own opcode. If there are no characters < 256,      extended class, with its own opcode, unless there was a negated special
3344      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
3345        the class, so any that were explicitly given as well can be ignored. If
3346        (when there are explicit characters > 255 that must be listed) there are no
3347        characters < 256, we can omit the bitmap in the actual compiled code. */
3348    
3349  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
3350      if (class_utf8)      if (class_utf8 && !should_flip_negation)
3351        {        {
3352        *class_utf8data++ = XCL_END;    /* Marks the end of extra data */        *class_utf8data++ = XCL_END;    /* Marks the end of extra data */
3353        *code++ = OP_XCLASS;        *code++ = OP_XCLASS;
# Line 3317  we set the flag only if there is a liter Line 3373  we set the flag only if there is a liter
3373        }        }
3374  #endif  #endif
3375    
3376      /* 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
3377      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
3378      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
3379      setting must remain unchanged after any kind of repeat. */      map into the code vector, negating it if necessary. */
3380    
3381        *code++ = (negate_class == should_flip_negation) ? OP_CLASS : OP_NCLASS;
3382      if (negate_class)      if (negate_class)
3383        {        {
       *code++ = OP_NCLASS;  
3384        if (lengthptr == NULL)    /* Save time in the pre-compile phase */        if (lengthptr == NULL)    /* Save time in the pre-compile phase */
3385          for (c = 0; c < 32; c++) code[c] = ~classbits[c];          for (c = 0; c < 32; c++) code[c] = ~classbits[c];
3386        }        }
3387      else      else
3388        {        {
       *code++ = OP_CLASS;  
3389        memcpy(code, classbits, 32);        memcpy(code, classbits, 32);
3390        }        }
3391      code += 32;      code += 32;
# Line 4024  we set the flag only if there is a liter Line 4079  we set the flag only if there is a liter
4079      if (*(++ptr) == '*' && (cd->ctypes[ptr[1]] & ctype_letter) != 0)      if (*(++ptr) == '*' && (cd->ctypes[ptr[1]] & ctype_letter) != 0)
4080        {        {
4081        int i, namelen;        int i, namelen;
4082          const char *vn = verbnames;
4083        const uschar *name = ++ptr;        const uschar *name = ++ptr;
4084        previous = NULL;        previous = NULL;
4085        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 4097  we set the flag only if there is a liter
4097        for (i = 0; i < verbcount; i++)        for (i = 0; i < verbcount; i++)
4098          {          {
4099          if (namelen == verbs[i].len &&          if (namelen == verbs[i].len &&
4100              strncmp((char *)name, verbs[i].name, namelen) == 0)              strncmp((char *)name, vn, namelen) == 0)
4101            {            {
4102            *code = verbs[i].op;            *code = verbs[i].op;
4103            if (*code++ == OP_ACCEPT) cd->had_accept = TRUE;            if (*code++ == OP_ACCEPT) cd->had_accept = TRUE;
4104            break;            break;
4105            }            }
4106            vn += verbs[i].len + 1;
4107          }          }
4108        if (i < verbcount) continue;        if (i < verbcount) continue;
4109        *errorcodeptr = ERR60;        *errorcodeptr = ERR60;
# Line 4191  we set the flag only if there is a liter Line 4248  we set the flag only if there is a liter
4248              *errorcodeptr = ERR58;              *errorcodeptr = ERR58;
4249              goto FAILED;              goto FAILED;
4250              }              }
4251            if (refsign == '-')            recno = (refsign == '-')?
4252                cd->bracount - recno + 1 : recno +cd->bracount;
4253              if (recno <= 0 || recno > cd->final_bracount)
4254              {              {
4255              recno = cd->bracount - recno + 1;              *errorcodeptr = ERR15;
4256              if (recno <= 0)              goto FAILED;
               {  
               *errorcodeptr = ERR15;  
               goto FAILED;  
               }  
4257              }              }
           else recno += cd->bracount;  
4258            PUT2(code, 2+LINK_SIZE, recno);            PUT2(code, 2+LINK_SIZE, recno);
4259            break;            break;
4260            }            }
# Line 4272  we set the flag only if there is a liter Line 4326  we set the flag only if there is a liter
4326            skipbytes = 1;            skipbytes = 1;
4327            }            }
4328    
4329          /* Check for the "name" actually being a subpattern number. */          /* Check for the "name" actually being a subpattern number. We are
4330            in the second pass here, so final_bracount is set. */
4331    
4332          else if (recno > 0)          else if (recno > 0 && recno <= cd->final_bracount)
4333            {            {
4334            PUT2(code, 2+LINK_SIZE, recno);            PUT2(code, 2+LINK_SIZE, recno);
4335            }            }
# Line 4468  we set the flag only if there is a liter Line 4523  we set the flag only if there is a liter
4523    
4524          /* We come here from the Python syntax above that handles both          /* We come here from the Python syntax above that handles both
4525          references (?P=name) and recursion (?P>name), as well as falling          references (?P=name) and recursion (?P>name), as well as falling
4526          through from the Perl recursion syntax (?&name). */          through from the Perl recursion syntax (?&name). We also come here from
4527            the Perl \k<name> or \k'name' back reference syntax and the \k{name}
4528            .NET syntax. */
4529    
4530          NAMED_REF_OR_RECURSE:          NAMED_REF_OR_RECURSE:
4531          name = ++ptr;          name = ++ptr;
# Line 4480  we set the flag only if there is a liter Line 4537  we set the flag only if there is a liter
4537    
4538          if (lengthptr != NULL)          if (lengthptr != NULL)
4539            {            {
4540              if (namelen == 0)
4541                {
4542                *errorcodeptr = ERR62;
4543                goto FAILED;
4544                }
4545            if (*ptr != terminator)            if (*ptr != terminator)
4546              {              {
4547              *errorcodeptr = ERR42;              *errorcodeptr = ERR42;
# Line 4493  we set the flag only if there is a liter Line 4555  we set the flag only if there is a liter
4555            recno = 0;            recno = 0;
4556            }            }
4557    
4558          /* In the real compile, seek the name in the table */          /* In the real compile, seek the name in the table. We check the name
4559            first, and then check that we have reached the end of the name in the
4560            table. That way, if the name that is longer than any in the table,
4561            the comparison will fail without reading beyond the table entry. */
4562    
4563          else          else
4564            {            {
4565            slot = cd->name_table;            slot = cd->name_table;
4566            for (i = 0; i < cd->names_found; i++)            for (i = 0; i < cd->names_found; i++)
4567              {              {
4568              if (strncmp((char *)name, (char *)slot+2, namelen) == 0) break;              if (strncmp((char *)name, (char *)slot+2, namelen) == 0 &&
4569                    slot[2+namelen] == 0)
4570                  break;
4571              slot += cd->name_entry_size;              slot += cd->name_entry_size;
4572              }              }
4573    
# Line 4537  we set the flag only if there is a liter Line 4604  we set the flag only if there is a liter
4604            {            {
4605            const uschar *called;            const uschar *called;
4606    
4607            if ((refsign = *ptr) == '+') ptr++;            if ((refsign = *ptr) == '+')
4608                {
4609                ptr++;
4610                if ((digitab[*ptr] & ctype_digit) == 0)
4611                  {
4612                  *errorcodeptr = ERR63;
4613                  goto FAILED;
4614                  }
4615                }
4616            else if (refsign == '-')            else if (refsign == '-')
4617              {              {
4618              if ((digitab[ptr[1]] & ctype_digit) == 0)              if ((digitab[ptr[1]] & ctype_digit) == 0)
# Line 5763  cd->fcc = tables + fcc_offset; Line 5838  cd->fcc = tables + fcc_offset;
5838  cd->cbits = tables + cbits_offset;  cd->cbits = tables + cbits_offset;
5839  cd->ctypes = tables + ctypes_offset;  cd->ctypes = tables + ctypes_offset;
5840    
5841  /* Check for newline settings at the start of the pattern, and remember the  /* Check for global one-time settings at the start of the pattern, and remember
5842  offset for later. */  the offset for later. */
5843    
5844  if (ptr[0] == '(' && ptr[1] == '*')  while (ptr[skipatstart] == '(' && ptr[skipatstart+1] == '*')
5845    {    {
5846    int newnl = 0;    int newnl = 0;
5847    if (strncmp((char *)(ptr+2), "CR)", 3) == 0)    int newbsr = 0;
5848      { skipatstart = 5; newnl = PCRE_NEWLINE_CR; }  
5849    else if (strncmp((char *)(ptr+2), "LF)", 3)  == 0)    if (strncmp((char *)(ptr+skipatstart+2), "CR)", 3) == 0)
5850      { skipatstart = 5; newnl = PCRE_NEWLINE_LF; }      { skipatstart += 5; newnl = PCRE_NEWLINE_CR; }
5851    else if (strncmp((char *)(ptr+2), "CRLF)", 5)  == 0)    else if (strncmp((char *)(ptr+skipatstart+2), "LF)", 3)  == 0)
5852      { skipatstart = 7; newnl = PCRE_NEWLINE_CR + PCRE_NEWLINE_LF; }      { skipatstart += 5; newnl = PCRE_NEWLINE_LF; }
5853    else if (strncmp((char *)(ptr+2), "ANY)", 4) == 0)    else if (strncmp((char *)(ptr+skipatstart+2), "CRLF)", 5)  == 0)
5854      { skipatstart = 6; newnl = PCRE_NEWLINE_ANY; }      { skipatstart += 7; newnl = PCRE_NEWLINE_CR + PCRE_NEWLINE_LF; }
5855    else if (strncmp((char *)(ptr+2), "ANYCRLF)", 8)  == 0)    else if (strncmp((char *)(ptr+skipatstart+2), "ANY)", 4) == 0)
5856      { skipatstart = 10; newnl = PCRE_NEWLINE_ANYCRLF; }      { skipatstart += 6; newnl = PCRE_NEWLINE_ANY; }
5857    if (skipatstart > 0)    else if (strncmp((char *)(ptr+skipatstart+2), "ANYCRLF)", 8)  == 0)
5858        { skipatstart += 10; newnl = PCRE_NEWLINE_ANYCRLF; }
5859    
5860      else if (strncmp((char *)(ptr+skipatstart+2), "BSR_ANYCRLF)", 12) == 0)
5861        { skipatstart += 14; newbsr = PCRE_BSR_ANYCRLF; }
5862      else if (strncmp((char *)(ptr+skipatstart+2), "BSR_UNICODE)", 12) == 0)
5863        { skipatstart += 14; newbsr = PCRE_BSR_UNICODE; }
5864    
5865      if (newnl != 0)
5866      options = (options & ~PCRE_NEWLINE_BITS) | newnl;      options = (options & ~PCRE_NEWLINE_BITS) | newnl;
5867      else if (newbsr != 0)
5868        options = (options & ~(PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE)) | newbsr;
5869      else break;
5870      }
5871    
5872    /* Check validity of \R options. */
5873    
5874    switch (options & (PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE))
5875      {
5876      case 0:
5877      case PCRE_BSR_ANYCRLF:
5878      case PCRE_BSR_UNICODE:
5879      break;
5880      default: errorcode = ERR56; goto PCRE_EARLY_ERROR_RETURN;
5881    }    }
5882    
5883  /* Handle different types of newline. The three bits give seven cases. The  /* Handle different types of newline. The three bits give seven cases. The
# Line 5842  to compile parts of the pattern into; th Line 5939  to compile parts of the pattern into; th
5939  no longer needed, so hopefully this workspace will never overflow, though there  no longer needed, so hopefully this workspace will never overflow, though there
5940  is a test for its doing so. */  is a test for its doing so. */
5941    
5942  cd->bracount = 0;  cd->bracount = cd->final_bracount = 0;
5943  cd->names_found = 0;  cd->names_found = 0;
5944  cd->name_entry_size = 0;  cd->name_entry_size = 0;
5945  cd->name_table = NULL;  cd->name_table = NULL;
# Line 5892  if (re == NULL) Line 5989  if (re == NULL)
5989    goto PCRE_EARLY_ERROR_RETURN;    goto PCRE_EARLY_ERROR_RETURN;
5990    }    }
5991    
5992  /* Put in the magic number, and save the sizes, initial options, internal  /* Put in the magic number, and save the sizes, initial options, internal
5993  flags, and character table pointer. NULL is used for the default character  flags, and character table pointer. NULL is used for the default character
5994  tables. The nullpad field is at the end; it's there to help in the case when a  tables. The nullpad field is at the end; it's there to help in the case when a
5995  regex compiled on a system with 4-byte pointers is run on another with 8-byte  regex compiled on a system with 4-byte pointers is run on another with 8-byte
# Line 5919  field. Reset the bracket count and the n Line 6016  field. Reset the bracket count and the n
6016  field; this time it's used for remembering forward references to subpatterns.  field; this time it's used for remembering forward references to subpatterns.
6017  */  */
6018    
6019    cd->final_bracount = cd->bracount;  /* Save for checking forward references */
6020  cd->bracount = 0;  cd->bracount = 0;
6021  cd->names_found = 0;  cd->names_found = 0;
6022  cd->name_table = (uschar *)re + re->name_table_offset;  cd->name_table = (uschar *)re + re->name_table_offset;
# Line 5983  if (errorcode != 0) Line 6081  if (errorcode != 0)
6081    PCRE_EARLY_ERROR_RETURN:    PCRE_EARLY_ERROR_RETURN:
6082    *erroroffset = ptr - (const uschar *)pattern;    *erroroffset = ptr - (const uschar *)pattern;
6083    PCRE_EARLY_ERROR_RETURN2:    PCRE_EARLY_ERROR_RETURN2:
6084    *errorptr = error_texts[errorcode];    *errorptr = find_error_text(errorcode);
6085    if (errorcodeptr != NULL) *errorcodeptr = errorcode;    if (errorcodeptr != NULL) *errorcodeptr = errorcode;
6086    return NULL;    return NULL;
6087    }    }
# Line 6068  was compiled can be seen. */ Line 6166  was compiled can be seen. */
6166  if (code - codestart > length)  if (code - codestart > length)
6167    {    {
6168    (pcre_free)(re);    (pcre_free)(re);
6169    *errorptr = error_texts[ERR23];    *errorptr = find_error_text(ERR23);
6170    *erroroffset = ptr - (uschar *)pattern;    *erroroffset = ptr - (uschar *)pattern;
6171    if (errorcodeptr != NULL) *errorcodeptr = ERR23;    if (errorcodeptr != NULL) *errorcodeptr = ERR23;
6172    return NULL;    return NULL;
Legend:
Removed from v.230  
changed lines
  Added in v.275
webmaster@exim.org
 ViewVC Help
Powered by ViewVC 1.1.12