/[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 223 by ph10, Mon Aug 20 11:07:53 2007 UTC revision 282 by ph10, Fri Dec 7 19:32:32 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 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 1490  for (code = first_significant_code(code Line 1531  for (code = first_significant_code(code
1531    const uschar *ccode;    const uschar *ccode;
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    
# Line 1597  for (code = first_significant_code(code Line 1648  for (code = first_significant_code(code
1648      case OP_TYPEEXACT:      case OP_TYPEEXACT:
1649      return FALSE;      return FALSE;
1650    
1651        /* These are going to continue, as they may be empty, but we have to
1652        fudge the length for the \p and \P cases. */
1653    
1654        case OP_TYPESTAR:
1655        case OP_TYPEMINSTAR:
1656        case OP_TYPEPOSSTAR:
1657        case OP_TYPEQUERY:
1658        case OP_TYPEMINQUERY:
1659        case OP_TYPEPOSQUERY:
1660        if (code[1] == OP_PROP || code[1] == OP_NOTPROP) code += 2;
1661        break;
1662    
1663        /* Same for these */
1664    
1665        case OP_TYPEUPTO:
1666        case OP_TYPEMINUPTO:
1667        case OP_TYPEPOSUPTO:
1668        if (code[3] == OP_PROP || code[3] == OP_NOTPROP) code += 2;
1669        break;
1670    
1671      /* End of branch */      /* End of branch */
1672    
1673      case OP_KET:      case OP_KET:
# Line 1713  Returns: a value representing the na Line 1784  Returns: a value representing the na
1784  static int  static int
1785  check_posix_name(const uschar *ptr, int len)  check_posix_name(const uschar *ptr, int len)
1786  {  {
1787    const char *pn = posix_names;
1788  register int yield = 0;  register int yield = 0;
1789  while (posix_name_lengths[yield] != 0)  while (posix_name_lengths[yield] != 0)
1790    {    {
1791    if (len == posix_name_lengths[yield] &&    if (len == posix_name_lengths[yield] &&
1792      strncmp((const char *)ptr, posix_names[yield], len) == 0) return yield;      strncmp((const char *)ptr, pn, len) == 0) return yield;
1793      pn += posix_name_lengths[yield] + 1;
1794    yield++;    yield++;
1795    }    }
1796  return -1;  return -1;
# Line 1759  adjust_recurse(uschar *group, int adjust Line 1832  adjust_recurse(uschar *group, int adjust
1832    uschar *save_hwm)    uschar *save_hwm)
1833  {  {
1834  uschar *ptr = group;  uschar *ptr = group;
1835    
1836  while ((ptr = (uschar *)find_recurse(ptr, utf8)) != NULL)  while ((ptr = (uschar *)find_recurse(ptr, utf8)) != NULL)
1837    {    {
1838    int offset;    int offset;
# Line 2322  req_caseopt = ((options & PCRE_CASELESS) Line 2396  req_caseopt = ((options & PCRE_CASELESS)
2396  for (;; ptr++)  for (;; ptr++)
2397    {    {
2398    BOOL negate_class;    BOOL negate_class;
2399      BOOL should_flip_negation;
2400    BOOL possessive_quantifier;    BOOL possessive_quantifier;
2401    BOOL is_quantifier;    BOOL is_quantifier;
2402    BOOL is_recurse;    BOOL is_recurse;
# Line 2570  for (;; ptr++) Line 2645  for (;; ptr++)
2645        else break;        else break;
2646        }        }
2647    
2648        /* If a class contains a negative special such as \S, we need to flip the
2649        negation flag at the end, so that support for characters > 255 works
2650        correctly (they are all included in the class). */
2651    
2652        should_flip_negation = FALSE;
2653    
2654      /* 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
2655      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
2656      valued UTF-8 characters, we don't yet do any optimization. */      valued UTF-8 characters, we don't yet do any optimization. */
# Line 2642  for (;; ptr++) Line 2723  for (;; ptr++)
2723          if (*ptr == '^')          if (*ptr == '^')
2724            {            {
2725            local_negate = TRUE;            local_negate = TRUE;
2726              should_flip_negation = TRUE;  /* Note negative special */
2727            ptr++;            ptr++;
2728            }            }
2729    
# Line 2716  for (;; ptr++) Line 2798  for (;; ptr++)
2798          c = check_escape(&ptr, errorcodeptr, cd->bracount, options, TRUE);          c = check_escape(&ptr, errorcodeptr, cd->bracount, options, TRUE);
2799          if (*errorcodeptr != 0) goto FAILED;          if (*errorcodeptr != 0) goto FAILED;
2800    
2801          if (-c == ESC_b) c = '\b';       /* \b is backslash in a class */          if (-c == ESC_b) c = '\b';       /* \b is backspace in a class */
2802          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 */
2803          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 */
2804          else if (-c == ESC_Q)            /* Handle start of quoted string */          else if (-c == ESC_Q)            /* Handle start of quoted string */
# Line 2744  for (;; ptr++) Line 2826  for (;; ptr++)
2826              continue;              continue;
2827    
2828              case ESC_D:              case ESC_D:
2829                should_flip_negation = TRUE;
2830              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_digit];              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_digit];
2831              continue;              continue;
2832    
# Line 2752  for (;; ptr++) Line 2835  for (;; ptr++)
2835              continue;              continue;
2836    
2837              case ESC_W:              case ESC_W:
2838                should_flip_negation = TRUE;
2839              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_word];              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_word];
2840              continue;              continue;
2841    
# Line 2761  for (;; ptr++) Line 2845  for (;; ptr++)
2845              continue;              continue;
2846    
2847              case ESC_S:              case ESC_S:
2848                should_flip_negation = TRUE;
2849              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_space];              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_space];
2850              classbits[1] |= 0x08;    /* Perl 5.004 onwards omits VT from \s */              classbits[1] |= 0x08;    /* Perl 5.004 onwards omits VT from \s */
2851              continue;              continue;
2852    
             case ESC_E: /* Perl ignores an orphan \E */  
             continue;  
   
2853              default:    /* Not recognized; fall through */              default:    /* Not recognized; fall through */
2854              break;      /* Need "default" setting to stop compiler warning. */              break;      /* Need "default" setting to stop compiler warning. */
2855              }              }
# Line 2955  for (;; ptr++) Line 3037  for (;; ptr++)
3037    
3038        oldptr = ptr;        oldptr = ptr;
3039    
3040          /* Remember \r or \n */
3041    
3042          if (c == '\r' || c == '\n') cd->external_flags |= PCRE_HASCRORLF;
3043    
3044          /* Check for range */
3045    
3046        if (!inescq && ptr[1] == '-')        if (!inescq && ptr[1] == '-')
3047          {          {
3048          int d;          int d;
# Line 2996  for (;; ptr++) Line 3084  for (;; ptr++)
3084            d = check_escape(&ptr, errorcodeptr, cd->bracount, options, TRUE);            d = check_escape(&ptr, errorcodeptr, cd->bracount, options, TRUE);
3085            if (*errorcodeptr != 0) goto FAILED;            if (*errorcodeptr != 0) goto FAILED;
3086    
3087            /* \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
3088            special means the '-' was literal */            special means the '-' was literal */
3089    
3090            if (d < 0)            if (d < 0)
# Line 3022  for (;; ptr++) Line 3110  for (;; ptr++)
3110    
3111          if (d == c) goto LONE_SINGLE_CHARACTER;  /* A few lines below */          if (d == c) goto LONE_SINGLE_CHARACTER;  /* A few lines below */
3112    
3113            /* Remember \r or \n */
3114    
3115            if (d == '\r' || d == '\n') cd->external_flags |= PCRE_HASCRORLF;
3116    
3117          /* 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
3118          matching, we have to use an XCLASS with extra data items. Caseless          matching, we have to use an XCLASS with extra data items. Caseless
3119          matching for characters > 127 is available only if UCP support is          matching for characters > 127 is available only if UCP support is
# Line 3175  for (;; ptr++) Line 3267  for (;; ptr++)
3267        goto FAILED;        goto FAILED;
3268        }        }
3269    
3270    
3271    /* This code has been disabled because it would mean that \s counts as
3272    an explicit \r or \n reference, and that's not really what is wanted. Now
3273    we set the flag only if there is a literal "\r" or "\n" in the class. */
3274    
3275    #if 0
3276        /* Remember whether \r or \n are in this class */
3277    
3278        if (negate_class)
3279          {
3280          if ((classbits[1] & 0x24) != 0x24) cd->external_flags |= PCRE_HASCRORLF;
3281          }
3282        else
3283          {
3284          if ((classbits[1] & 0x24) != 0) cd->external_flags |= PCRE_HASCRORLF;
3285          }
3286    #endif
3287    
3288    
3289      /* 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
3290      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
3291      use of \p or \P, in other words, no use of any XCLASS features, we can      use of \p or \P, in other words, no use of any XCLASS features, we can
3292      optimize.      optimize.
3293    
3294      In UTF-8 mode, we can optimize the negative case only if there were no      In UTF-8 mode, we can optimize the negative case only if there were no
3295      characters >= 128 because OP_NOT and the related opcodes like OP_NOTSTAR      characters >= 128 because OP_NOT and the related opcodes like OP_NOTSTAR
3296      operate on single-bytes only. This is an historical hangover. Maybe one day      operate on single-bytes only. This is an historical hangover. Maybe one day
# Line 3193  for (;; ptr++) Line 3304  for (;; ptr++)
3304      reqbyte, save the previous value for reinstating. */      reqbyte, save the previous value for reinstating. */
3305    
3306  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
3307      if (class_charcount == 1 && !class_utf8 &&      if (class_charcount == 1 && !class_utf8 &&
3308        (!utf8 || !negate_class || class_lastchar < 128))        (!utf8 || !negate_class || class_lastchar < 128))
3309  #else  #else
3310      if (class_charcount == 1)      if (class_charcount == 1)
# Line 3237  for (;; ptr++) Line 3348  for (;; ptr++)
3348      zeroreqbyte = reqbyte;      zeroreqbyte = reqbyte;
3349    
3350      /* If there are characters with values > 255, we have to compile an      /* If there are characters with values > 255, we have to compile an
3351      extended class, with its own opcode. If there are no characters < 256,      extended class, with its own opcode, unless there was a negated special
3352      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
3353        the class, so any that were explicitly given as well can be ignored. If
3354        (when there are explicit characters > 255 that must be listed) there are no
3355        characters < 256, we can omit the bitmap in the actual compiled code. */
3356    
3357  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
3358      if (class_utf8)      if (class_utf8 && !should_flip_negation)
3359        {        {
3360        *class_utf8data++ = XCL_END;    /* Marks the end of extra data */        *class_utf8data++ = XCL_END;    /* Marks the end of extra data */
3361        *code++ = OP_XCLASS;        *code++ = OP_XCLASS;
# Line 3267  for (;; ptr++) Line 3381  for (;; ptr++)
3381        }        }
3382  #endif  #endif
3383    
3384      /* 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
3385      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
3386      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
3387      setting must remain unchanged after any kind of repeat. */      map into the code vector, negating it if necessary. */
3388    
3389        *code++ = (negate_class == should_flip_negation) ? OP_CLASS : OP_NCLASS;
3390      if (negate_class)      if (negate_class)
3391        {        {
       *code++ = OP_NCLASS;  
3392        if (lengthptr == NULL)    /* Save time in the pre-compile phase */        if (lengthptr == NULL)    /* Save time in the pre-compile phase */
3393          for (c = 0; c < 32; c++) code[c] = ~classbits[c];          for (c = 0; c < 32; c++) code[c] = ~classbits[c];
3394        }        }
3395      else      else
3396        {        {
       *code++ = OP_CLASS;  
3397        memcpy(code, classbits, 32);        memcpy(code, classbits, 32);
3398        }        }
3399      code += 32;      code += 32;
# Line 3466  for (;; ptr++) Line 3579  for (;; ptr++)
3579        /* All real repeats make it impossible to handle partial matching (maybe        /* All real repeats make it impossible to handle partial matching (maybe
3580        one day we will be able to remove this restriction). */        one day we will be able to remove this restriction). */
3581    
3582        if (repeat_max != 1) cd->nopartial = TRUE;        if (repeat_max != 1) cd->external_flags |= PCRE_NOPARTIAL;
3583    
3584        /* Combine the op_type with the repeat_type */        /* Combine the op_type with the repeat_type */
3585    
# Line 3616  for (;; ptr++) Line 3729  for (;; ptr++)
3729        /* All real repeats make it impossible to handle partial matching (maybe        /* All real repeats make it impossible to handle partial matching (maybe
3730        one day we will be able to remove this restriction). */        one day we will be able to remove this restriction). */
3731    
3732        if (repeat_max != 1) cd->nopartial = TRUE;        if (repeat_max != 1) cd->external_flags |= PCRE_NOPARTIAL;
3733    
3734        if (repeat_min == 0 && repeat_max == -1)        if (repeat_min == 0 && repeat_max == -1)
3735          *code++ = OP_CRSTAR + repeat_type;          *code++ = OP_CRSTAR + repeat_type;
# Line 3974  for (;; ptr++) Line 4087  for (;; ptr++)
4087      if (*(++ptr) == '*' && (cd->ctypes[ptr[1]] & ctype_letter) != 0)      if (*(++ptr) == '*' && (cd->ctypes[ptr[1]] & ctype_letter) != 0)
4088        {        {
4089        int i, namelen;        int i, namelen;
4090          const char *vn = verbnames;
4091        const uschar *name = ++ptr;        const uschar *name = ++ptr;
4092        previous = NULL;        previous = NULL;
4093        while ((cd->ctypes[*++ptr] & ctype_letter) != 0);        while ((cd->ctypes[*++ptr] & ctype_letter) != 0);
# Line 3991  for (;; ptr++) Line 4105  for (;; ptr++)
4105        for (i = 0; i < verbcount; i++)        for (i = 0; i < verbcount; i++)
4106          {          {
4107          if (namelen == verbs[i].len &&          if (namelen == verbs[i].len &&
4108              strncmp((char *)name, verbs[i].name, namelen) == 0)              strncmp((char *)name, vn, namelen) == 0)
4109            {            {
4110            *code = verbs[i].op;            *code = verbs[i].op;
4111            if (*code++ == OP_ACCEPT) cd->had_accept = TRUE;            if (*code++ == OP_ACCEPT) cd->had_accept = TRUE;
4112            break;            break;
4113            }            }
4114            vn += verbs[i].len + 1;
4115          }          }
4116        if (i < verbcount) continue;        if (i < verbcount) continue;
4117        *errorcodeptr = ERR60;        *errorcodeptr = ERR60;
# Line 4141  for (;; ptr++) Line 4256  for (;; ptr++)
4256              *errorcodeptr = ERR58;              *errorcodeptr = ERR58;
4257              goto FAILED;              goto FAILED;
4258              }              }
4259            if (refsign == '-')            recno = (refsign == '-')?
4260                cd->bracount - recno + 1 : recno +cd->bracount;
4261              if (recno <= 0 || recno > cd->final_bracount)
4262              {              {
4263              recno = cd->bracount - recno + 1;              *errorcodeptr = ERR15;
4264              if (recno <= 0)              goto FAILED;
               {  
               *errorcodeptr = ERR15;  
               goto FAILED;  
               }  
4265              }              }
           else recno += cd->bracount;  
4266            PUT2(code, 2+LINK_SIZE, recno);            PUT2(code, 2+LINK_SIZE, recno);
4267            break;            break;
4268            }            }
# Line 4222  for (;; ptr++) Line 4334  for (;; ptr++)
4334            skipbytes = 1;            skipbytes = 1;
4335            }            }
4336    
4337          /* Check for the "name" actually being a subpattern number. */          /* Check for the "name" actually being a subpattern number. We are
4338            in the second pass here, so final_bracount is set. */
4339    
4340          else if (recno > 0)          else if (recno > 0 && recno <= cd->final_bracount)
4341            {            {
4342            PUT2(code, 2+LINK_SIZE, recno);            PUT2(code, 2+LINK_SIZE, recno);
4343            }            }
# Line 4418  for (;; ptr++) Line 4531  for (;; ptr++)
4531    
4532          /* We come here from the Python syntax above that handles both          /* We come here from the Python syntax above that handles both
4533          references (?P=name) and recursion (?P>name), as well as falling          references (?P=name) and recursion (?P>name), as well as falling
4534          through from the Perl recursion syntax (?&name). */          through from the Perl recursion syntax (?&name). We also come here from
4535            the Perl \k<name> or \k'name' back reference syntax and the \k{name}
4536            .NET syntax. */
4537    
4538          NAMED_REF_OR_RECURSE:          NAMED_REF_OR_RECURSE:
4539          name = ++ptr;          name = ++ptr;
# Line 4430  for (;; ptr++) Line 4545  for (;; ptr++)
4545    
4546          if (lengthptr != NULL)          if (lengthptr != NULL)
4547            {            {
4548              if (namelen == 0)
4549                {
4550                *errorcodeptr = ERR62;
4551                goto FAILED;
4552                }
4553            if (*ptr != terminator)            if (*ptr != terminator)
4554              {              {
4555              *errorcodeptr = ERR42;              *errorcodeptr = ERR42;
# Line 4443  for (;; ptr++) Line 4563  for (;; ptr++)
4563            recno = 0;            recno = 0;
4564            }            }
4565    
4566          /* In the real compile, seek the name in the table */          /* In the real compile, seek the name in the table. We check the name
4567            first, and then check that we have reached the end of the name in the
4568            table. That way, if the name that is longer than any in the table,
4569            the comparison will fail without reading beyond the table entry. */
4570    
4571          else          else
4572            {            {
4573            slot = cd->name_table;            slot = cd->name_table;
4574            for (i = 0; i < cd->names_found; i++)            for (i = 0; i < cd->names_found; i++)
4575              {              {
4576              if (strncmp((char *)name, (char *)slot+2, namelen) == 0) break;              if (strncmp((char *)name, (char *)slot+2, namelen) == 0 &&
4577                    slot[2+namelen] == 0)
4578                  break;
4579              slot += cd->name_entry_size;              slot += cd->name_entry_size;
4580              }              }
4581    
# Line 4487  for (;; ptr++) Line 4612  for (;; ptr++)
4612            {            {
4613            const uschar *called;            const uschar *called;
4614    
4615            if ((refsign = *ptr) == '+') ptr++;            if ((refsign = *ptr) == '+')
4616                {
4617                ptr++;
4618                if ((digitab[*ptr] & ctype_digit) == 0)
4619                  {
4620                  *errorcodeptr = ERR63;
4621                  goto FAILED;
4622                  }
4623                }
4624            else if (refsign == '-')            else if (refsign == '-')
4625              {              {
4626              if ((digitab[ptr[1]] & ctype_digit) == 0)              if ((digitab[ptr[1]] & ctype_digit) == 0)
# Line 4613  for (;; ptr++) Line 4746  for (;; ptr++)
4746    
4747              case 'J':    /* Record that it changed in the external options */              case 'J':    /* Record that it changed in the external options */
4748              *optset |= PCRE_DUPNAMES;              *optset |= PCRE_DUPNAMES;
4749              cd->external_options |= PCRE_JCHANGED;              cd->external_flags |= PCRE_JCHANGED;
4750              break;              break;
4751    
4752              case 'i': *optset |= PCRE_CASELESS; break;              case 'i': *optset |= PCRE_CASELESS; break;
# Line 5030  for (;; ptr++) Line 5163  for (;; ptr++)
5163      *code++ = ((options & PCRE_CASELESS) != 0)? OP_CHARNC : OP_CHAR;      *code++ = ((options & PCRE_CASELESS) != 0)? OP_CHARNC : OP_CHAR;
5164      for (c = 0; c < mclength; c++) *code++ = mcbuffer[c];      for (c = 0; c < mclength; c++) *code++ = mcbuffer[c];
5165    
5166        /* Remember if \r or \n were seen */
5167    
5168        if (mcbuffer[0] == '\r' || mcbuffer[0] == '\n')
5169          cd->external_flags |= PCRE_HASCRORLF;
5170    
5171      /* Set the first and required bytes appropriately. If no previous first      /* Set the first and required bytes appropriately. If no previous first
5172      byte, set it from this character, but revert to none on a zero repeat.      byte, set it from this character, but revert to none on a zero repeat.
5173      Otherwise, leave the firstbyte value alone, and don't change it on a zero      Otherwise, leave the firstbyte value alone, and don't change it on a zero
# Line 5629  real_pcre *re; Line 5767  real_pcre *re;
5767  int length = 1;  /* For final END opcode */  int length = 1;  /* For final END opcode */
5768  int firstbyte, reqbyte, newline;  int firstbyte, reqbyte, newline;
5769  int errorcode = 0;  int errorcode = 0;
5770    int skipatstart = 0;
5771  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
5772  BOOL utf8;  BOOL utf8;
5773  #endif  #endif
# Line 5707  cd->fcc = tables + fcc_offset; Line 5846  cd->fcc = tables + fcc_offset;
5846  cd->cbits = tables + cbits_offset;  cd->cbits = tables + cbits_offset;
5847  cd->ctypes = tables + ctypes_offset;  cd->ctypes = tables + ctypes_offset;
5848    
5849    /* Check for global one-time settings at the start of the pattern, and remember
5850    the offset for later. */
5851    
5852    while (ptr[skipatstart] == '(' && ptr[skipatstart+1] == '*')
5853      {
5854      int newnl = 0;
5855      int newbsr = 0;
5856    
5857      if (strncmp((char *)(ptr+skipatstart+2), "CR)", 3) == 0)
5858        { skipatstart += 5; newnl = PCRE_NEWLINE_CR; }
5859      else if (strncmp((char *)(ptr+skipatstart+2), "LF)", 3)  == 0)
5860        { skipatstart += 5; newnl = PCRE_NEWLINE_LF; }
5861      else if (strncmp((char *)(ptr+skipatstart+2), "CRLF)", 5)  == 0)
5862        { skipatstart += 7; newnl = PCRE_NEWLINE_CR + PCRE_NEWLINE_LF; }
5863      else if (strncmp((char *)(ptr+skipatstart+2), "ANY)", 4) == 0)
5864        { skipatstart += 6; newnl = PCRE_NEWLINE_ANY; }
5865      else if (strncmp((char *)(ptr+skipatstart+2), "ANYCRLF)", 8)  == 0)
5866        { skipatstart += 10; newnl = PCRE_NEWLINE_ANYCRLF; }
5867    
5868      else if (strncmp((char *)(ptr+skipatstart+2), "BSR_ANYCRLF)", 12) == 0)
5869        { skipatstart += 14; newbsr = PCRE_BSR_ANYCRLF; }
5870      else if (strncmp((char *)(ptr+skipatstart+2), "BSR_UNICODE)", 12) == 0)
5871        { skipatstart += 14; newbsr = PCRE_BSR_UNICODE; }
5872    
5873      if (newnl != 0)
5874        options = (options & ~PCRE_NEWLINE_BITS) | newnl;
5875      else if (newbsr != 0)
5876        options = (options & ~(PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE)) | newbsr;
5877      else break;
5878      }
5879    
5880    /* Check validity of \R options. */
5881    
5882    switch (options & (PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE))
5883      {
5884      case 0:
5885      case PCRE_BSR_ANYCRLF:
5886      case PCRE_BSR_UNICODE:
5887      break;
5888      default: errorcode = ERR56; goto PCRE_EARLY_ERROR_RETURN;
5889      }
5890    
5891  /* Handle different types of newline. The three bits give seven cases. The  /* Handle different types of newline. The three bits give seven cases. The
5892  current code allows for fixed one- or two-byte sequences, plus "any" and  current code allows for fixed one- or two-byte sequences, plus "any" and
5893  "anycrlf". */  "anycrlf". */
5894    
5895  switch (options & (PCRE_NEWLINE_CRLF | PCRE_NEWLINE_ANY))  switch (options & PCRE_NEWLINE_BITS)
5896    {    {
5897    case 0: newline = NEWLINE; break;   /* Compile-time default */    case 0: newline = NEWLINE; break;   /* Build-time default */
5898    case PCRE_NEWLINE_CR: newline = '\r'; break;    case PCRE_NEWLINE_CR: newline = '\r'; break;
5899    case PCRE_NEWLINE_LF: newline = '\n'; break;    case PCRE_NEWLINE_LF: newline = '\n'; break;
5900    case PCRE_NEWLINE_CR+    case PCRE_NEWLINE_CR+
# Line 5766  to compile parts of the pattern into; th Line 5947  to compile parts of the pattern into; th
5947  no longer needed, so hopefully this workspace will never overflow, though there  no longer needed, so hopefully this workspace will never overflow, though there
5948  is a test for its doing so. */  is a test for its doing so. */
5949    
5950  cd->bracount = 0;  cd->bracount = cd->final_bracount = 0;
5951  cd->names_found = 0;  cd->names_found = 0;
5952  cd->name_entry_size = 0;  cd->name_entry_size = 0;
5953  cd->name_table = NULL;  cd->name_table = NULL;
# Line 5776  cd->hwm = cworkspace; Line 5957  cd->hwm = cworkspace;
5957  cd->start_pattern = (const uschar *)pattern;  cd->start_pattern = (const uschar *)pattern;
5958  cd->end_pattern = (const uschar *)(pattern + strlen(pattern));  cd->end_pattern = (const uschar *)(pattern + strlen(pattern));
5959  cd->req_varyopt = 0;  cd->req_varyopt = 0;
 cd->nopartial = FALSE;  
5960  cd->external_options = options;  cd->external_options = options;
5961    cd->external_flags = 0;
5962    
5963  /* Now do the pre-compile. On error, errorcode will be set non-zero, so we  /* Now do the pre-compile. On error, errorcode will be set non-zero, so we
5964  don't need to look at the result of the function here. The initial options have  don't need to look at the result of the function here. The initial options have
# Line 5785  been put into the cd block so that they Line 5966  been put into the cd block so that they
5966  found within the regex right at the beginning. Bringing initial option settings  found within the regex right at the beginning. Bringing initial option settings
5967  outside can help speed up starting point checks. */  outside can help speed up starting point checks. */
5968    
5969    ptr += skipatstart;
5970  code = cworkspace;  code = cworkspace;
5971  *code = OP_BRA;  *code = OP_BRA;
5972  (void)compile_regex(cd->external_options, cd->external_options & PCRE_IMS,  (void)compile_regex(cd->external_options, cd->external_options & PCRE_IMS,
# Line 5815  if (re == NULL) Line 5997  if (re == NULL)
5997    goto PCRE_EARLY_ERROR_RETURN;    goto PCRE_EARLY_ERROR_RETURN;
5998    }    }
5999    
6000  /* Put in the magic number, and save the sizes, initial options, and character  /* Put in the magic number, and save the sizes, initial options, internal
6001  table pointer. NULL is used for the default character tables. The nullpad field  flags, and character table pointer. NULL is used for the default character
6002  is at the end; it's there to help in the case when a regex compiled on a system  tables. The nullpad field is at the end; it's there to help in the case when a
6003  with 4-byte pointers is run on another with 8-byte pointers. */  regex compiled on a system with 4-byte pointers is run on another with 8-byte
6004    pointers. */
6005    
6006  re->magic_number = MAGIC_NUMBER;  re->magic_number = MAGIC_NUMBER;
6007  re->size = size;  re->size = size;
6008  re->options = cd->external_options;  re->options = cd->external_options;
6009    re->flags = cd->external_flags;
6010  re->dummy1 = 0;  re->dummy1 = 0;
6011  re->first_byte = 0;  re->first_byte = 0;
6012  re->req_byte = 0;  re->req_byte = 0;
# Line 5840  field. Reset the bracket count and the n Line 6024  field. Reset the bracket count and the n
6024  field; this time it's used for remembering forward references to subpatterns.  field; this time it's used for remembering forward references to subpatterns.
6025  */  */
6026    
6027    cd->final_bracount = cd->bracount;  /* Save for checking forward references */
6028  cd->bracount = 0;  cd->bracount = 0;
6029  cd->names_found = 0;  cd->names_found = 0;
6030  cd->name_table = (uschar *)re + re->name_table_offset;  cd->name_table = (uschar *)re + re->name_table_offset;
# Line 5847  codestart = cd->name_table + re->name_en Line 6032  codestart = cd->name_table + re->name_en
6032  cd->start_code = codestart;  cd->start_code = codestart;
6033  cd->hwm = cworkspace;  cd->hwm = cworkspace;
6034  cd->req_varyopt = 0;  cd->req_varyopt = 0;
 cd->nopartial = FALSE;  
6035  cd->had_accept = FALSE;  cd->had_accept = FALSE;
6036    
6037  /* Set up a starting, non-extracting bracket, then compile the expression. On  /* Set up a starting, non-extracting bracket, then compile the expression. On
6038  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
6039  of the function here. */  of the function here. */
6040    
6041  ptr = (const uschar *)pattern;  ptr = (const uschar *)pattern + skipatstart;
6042  code = (uschar *)codestart;  code = (uschar *)codestart;
6043  *code = OP_BRA;  *code = OP_BRA;
6044  (void)compile_regex(re->options, re->options & PCRE_IMS, &code, &ptr,  (void)compile_regex(re->options, re->options & PCRE_IMS, &code, &ptr,
6045    &errorcode, FALSE, FALSE, 0, &firstbyte, &reqbyte, NULL, cd, NULL);    &errorcode, FALSE, FALSE, 0, &firstbyte, &reqbyte, NULL, cd, NULL);
6046  re->top_bracket = cd->bracount;  re->top_bracket = cd->bracount;
6047  re->top_backref = cd->top_backref;  re->top_backref = cd->top_backref;
6048    re->flags = cd->external_flags;
6049    
 if (cd->nopartial) re->options |= PCRE_NOPARTIAL;  
6050  if (cd->had_accept) reqbyte = -1;   /* Must disable after (*ACCEPT) */  if (cd->had_accept) reqbyte = -1;   /* Must disable after (*ACCEPT) */
6051    
6052  /* If not reached end of pattern on success, there's an excess bracket. */  /* If not reached end of pattern on success, there's an excess bracket. */
# Line 5905  if (errorcode != 0) Line 6089  if (errorcode != 0)
6089    PCRE_EARLY_ERROR_RETURN:    PCRE_EARLY_ERROR_RETURN:
6090    *erroroffset = ptr - (const uschar *)pattern;    *erroroffset = ptr - (const uschar *)pattern;
6091    PCRE_EARLY_ERROR_RETURN2:    PCRE_EARLY_ERROR_RETURN2:
6092    *errorptr = error_texts[errorcode];    *errorptr = find_error_text(errorcode);
6093    if (errorcodeptr != NULL) *errorcodeptr = errorcode;    if (errorcodeptr != NULL) *errorcodeptr = errorcode;
6094    return NULL;    return NULL;
6095    }    }
# Line 5934  if ((re->options & PCRE_ANCHORED) == 0) Line 6118  if ((re->options & PCRE_ANCHORED) == 0)
6118        int ch = firstbyte & 255;        int ch = firstbyte & 255;
6119        re->first_byte = ((firstbyte & REQ_CASELESS) != 0 &&        re->first_byte = ((firstbyte & REQ_CASELESS) != 0 &&
6120           cd->fcc[ch] == ch)? ch : firstbyte;           cd->fcc[ch] == ch)? ch : firstbyte;
6121        re->options |= PCRE_FIRSTSET;        re->flags |= PCRE_FIRSTSET;
6122        }        }
6123      else if (is_startline(codestart, 0, cd->backref_map))      else if (is_startline(codestart, 0, cd->backref_map))
6124        re->options |= PCRE_STARTLINE;        re->flags |= PCRE_STARTLINE;
6125      }      }
6126    }    }
6127    
# Line 5951  if (reqbyte >= 0 && Line 6135  if (reqbyte >= 0 &&
6135    int ch = reqbyte & 255;    int ch = reqbyte & 255;
6136    re->req_byte = ((reqbyte & REQ_CASELESS) != 0 &&    re->req_byte = ((reqbyte & REQ_CASELESS) != 0 &&
6137      cd->fcc[ch] == ch)? (reqbyte & ~REQ_CASELESS) : reqbyte;      cd->fcc[ch] == ch)? (reqbyte & ~REQ_CASELESS) : reqbyte;
6138    re->options |= PCRE_REQCHSET;    re->flags |= PCRE_REQCHSET;
6139    }    }
6140    
6141  /* Print out the compiled data if debugging is enabled. This is never the  /* Print out the compiled data if debugging is enabled. This is never the
# Line 5962  case when building a production library. Line 6146  case when building a production library.
6146  printf("Length = %d top_bracket = %d top_backref = %d\n",  printf("Length = %d top_bracket = %d top_backref = %d\n",
6147    length, re->top_bracket, re->top_backref);    length, re->top_bracket, re->top_backref);
6148    
6149  if (re->options != 0)  printf("Options=%08x\n", re->options);
   {  
   printf("%s%s%s%s%s%s%s%s%s\n",  
     ((re->options & PCRE_NOPARTIAL) != 0)? "nopartial " : "",  
     ((re->options & PCRE_ANCHORED) != 0)? "anchored " : "",  
     ((re->options & PCRE_CASELESS) != 0)? "caseless " : "",  
     ((re->options & PCRE_EXTENDED) != 0)? "extended " : "",  
     ((re->options & PCRE_MULTILINE) != 0)? "multiline " : "",  
     ((re->options & PCRE_DOTALL) != 0)? "dotall " : "",  
     ((re->options & PCRE_DOLLAR_ENDONLY) != 0)? "endonly " : "",  
     ((re->options & PCRE_EXTRA) != 0)? "extra " : "",  
     ((re->options & PCRE_UNGREEDY) != 0)? "ungreedy " : "");  
   }  
6150    
6151  if ((re->options & PCRE_FIRSTSET) != 0)  if ((re->flags & PCRE_FIRSTSET) != 0)
6152    {    {
6153    int ch = re->first_byte & 255;    int ch = re->first_byte & 255;
6154    const char *caseless = ((re->first_byte & REQ_CASELESS) == 0)?    const char *caseless = ((re->first_byte & REQ_CASELESS) == 0)?
# Line 5985  if ((re->options & PCRE_FIRSTSET) != 0) Line 6157  if ((re->options & PCRE_FIRSTSET) != 0)
6157      else printf("First char = \\x%02x%s\n", ch, caseless);      else printf("First char = \\x%02x%s\n", ch, caseless);
6158    }    }
6159    
6160  if ((re->options & PCRE_REQCHSET) != 0)  if ((re->flags & PCRE_REQCHSET) != 0)
6161    {    {
6162    int ch = re->req_byte & 255;    int ch = re->req_byte & 255;
6163    const char *caseless = ((re->req_byte & REQ_CASELESS) == 0)?    const char *caseless = ((re->req_byte & REQ_CASELESS) == 0)?
# Line 6002  was compiled can be seen. */ Line 6174  was compiled can be seen. */
6174  if (code - codestart > length)  if (code - codestart > length)
6175    {    {
6176    (pcre_free)(re);    (pcre_free)(re);
6177    *errorptr = error_texts[ERR23];    *errorptr = find_error_text(ERR23);
6178    *erroroffset = ptr - (uschar *)pattern;    *erroroffset = ptr - (uschar *)pattern;
6179    if (errorcodeptr != NULL) *errorcodeptr = ERR23;    if (errorcodeptr != NULL) *errorcodeptr = ERR23;
6180    return NULL;    return NULL;

Legend:
Removed from v.223  
changed lines
  Added in v.282

webmaster@exim.org
ViewVC Help
Powered by ViewVC 1.1.12