/[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 216 by ph10, Wed Aug 15 14:35:57 2007 UTC revision 285 by ph10, Wed Dec 12 17:03:50 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 524  else Line 564  else
564      c = 0;      c = 0;
565      while ((digitab[ptr[1]] & ctype_digit) != 0)      while ((digitab[ptr[1]] & ctype_digit) != 0)
566        c = c * 10 + *(++ptr) - '0';        c = c * 10 + *(++ptr) - '0';
567    
568      if (c < 0)      if (c < 0)
569        {        {
570        *errorcodeptr = ERR61;        *errorcodeptr = ERR61;
571        break;        break;
572        }        }
573    
574      if (c == 0 || (braced && *(++ptr) != '}'))      if (c == 0 || (braced && *(++ptr) != '}'))
575        {        {
# Line 574  else Line 614  else
614        if (c < 0)        if (c < 0)
615          {          {
616          *errorcodeptr = ERR61;          *errorcodeptr = ERR61;
617          break;          break;
618          }          }
619        if (c < 10 || c <= bracount)        if (c < 10 || c <= bracount)
620          {          {
621          c = -(ESC_REF + c);          c = -(ESC_REF + c);
# 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 950  for (; *ptr != 0; ptr++) Line 990  for (; *ptr != 0; ptr++)
990      {      {
991      while (*(++ptr) != ']')      while (*(++ptr) != ']')
992        {        {
993        if (*ptr == 0) return -1;        if (*ptr == 0) return -1;
994        if (*ptr == '\\')        if (*ptr == '\\')
995          {          {
996          if (*(++ptr) == 0) return -1;          if (*(++ptr) == 0) return -1;
# Line 1106  for (;;) Line 1146  for (;;)
1146    {    {
1147    int d;    int d;
1148    register int op = *cc;    register int op = *cc;
   
1149    switch (op)    switch (op)
1150      {      {
1151      case OP_CBRA:      case OP_CBRA:
# Line 1195  for (;;) Line 1234  for (;;)
1234    
1235      case OP_TYPEEXACT:      case OP_TYPEEXACT:
1236      branchlength += GET2(cc,1);      branchlength += GET2(cc,1);
1237        if (cc[3] == OP_PROP || cc[3] == OP_NOTPROP) cc += 2;
1238      cc += 4;      cc += 4;
1239      break;      break;
1240    
# Line 1303  for (;;) Line 1343  for (;;)
1343      code += _pcre_OP_lengths[c];      code += _pcre_OP_lengths[c];
1344      }      }
1345    
1346    /* In UTF-8 mode, opcodes that are followed by a character may be followed by    /* Otherwise, we can get the item's length from the table, except that for
1347    a multi-byte character. The length in the table is a minimum, so we have to    repeated character types, we have to test for \p and \P, which have an extra
1348    arrange to skip the extra bytes. */    two bytes of parameters. */
1349    
1350    else    else
1351      {      {
1352        switch(c)
1353          {
1354          case OP_TYPESTAR:
1355          case OP_TYPEMINSTAR:
1356          case OP_TYPEPLUS:
1357          case OP_TYPEMINPLUS:
1358          case OP_TYPEQUERY:
1359          case OP_TYPEMINQUERY:
1360          case OP_TYPEPOSSTAR:
1361          case OP_TYPEPOSPLUS:
1362          case OP_TYPEPOSQUERY:
1363          if (code[1] == OP_PROP || code[1] == OP_NOTPROP) code += 2;
1364          break;
1365    
1366          case OP_TYPEUPTO:
1367          case OP_TYPEMINUPTO:
1368          case OP_TYPEEXACT:
1369          case OP_TYPEPOSUPTO:
1370          if (code[3] == OP_PROP || code[3] == OP_NOTPROP) code += 2;
1371          break;
1372          }
1373    
1374        /* Add in the fixed length from the table */
1375    
1376      code += _pcre_OP_lengths[c];      code += _pcre_OP_lengths[c];
1377    
1378      /* In UTF-8 mode, opcodes that are followed by a character may be followed by
1379      a multi-byte character. The length in the table is a minimum, so we have to
1380      arrange to skip the extra bytes. */
1381    
1382  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
1383      if (utf8) switch(c)      if (utf8) switch(c)
1384        {        {
# Line 1367  for (;;) Line 1436  for (;;)
1436    
1437    if (c == OP_XCLASS) code += GET(code, 1);    if (c == OP_XCLASS) code += GET(code, 1);
1438    
1439    /* Otherwise, we get the item's length from the table. In UTF-8 mode, opcodes    /* Otherwise, we can get the item's length from the table, except that for
1440    that are followed by a character may be followed by a multi-byte character.    repeated character types, we have to test for \p and \P, which have an extra
1441    The length in the table is a minimum, so we have to arrange to skip the extra    two bytes of parameters. */
   bytes. */  
1442    
1443    else    else
1444      {      {
1445        switch(c)
1446          {
1447          case OP_TYPESTAR:
1448          case OP_TYPEMINSTAR:
1449          case OP_TYPEPLUS:
1450          case OP_TYPEMINPLUS:
1451          case OP_TYPEQUERY:
1452          case OP_TYPEMINQUERY:
1453          case OP_TYPEPOSSTAR:
1454          case OP_TYPEPOSPLUS:
1455          case OP_TYPEPOSQUERY:
1456          if (code[1] == OP_PROP || code[1] == OP_NOTPROP) code += 2;
1457          break;
1458    
1459          case OP_TYPEPOSUPTO:
1460          case OP_TYPEUPTO:
1461          case OP_TYPEMINUPTO:
1462          case OP_TYPEEXACT:
1463          if (code[3] == OP_PROP || code[3] == OP_NOTPROP) code += 2;
1464          break;
1465          }
1466    
1467        /* Add in the fixed length from the table */
1468    
1469      code += _pcre_OP_lengths[c];      code += _pcre_OP_lengths[c];
1470    
1471        /* In UTF-8 mode, opcodes that are followed by a character may be followed
1472        by a multi-byte character. The length in the table is a minimum, so we have
1473        to arrange to skip the extra bytes. */
1474    
1475  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
1476      if (utf8) switch(c)      if (utf8) switch(c)
1477        {        {
# Line 1411  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 1433  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 1473  for (code = first_significant_code(code Line 1581  for (code = first_significant_code(code
1581      /* Check for quantifiers after a class. XCLASS is used for classes that      /* Check for quantifiers after a class. XCLASS is used for classes that
1582      cannot be represented just by a bit map. This includes negated single      cannot be represented just by a bit map. This includes negated single
1583      high-valued characters. The length in _pcre_OP_lengths[] is zero; the      high-valued characters. The length in _pcre_OP_lengths[] is zero; the
1584      actual length is stored in the compiled code, so we must update "code"      actual length is stored in the compiled code, so we must update "code"
1585      here. */      here. */
1586    
1587  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
# Line 1540  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 1656  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 1702  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 2265  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 2513  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 2585  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 2659  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 2671  for (;; ptr++) Line 2810  for (;; ptr++)
2810            else inescq = TRUE;            else inescq = TRUE;
2811            continue;            continue;
2812            }            }
2813            else if (-c == ESC_E) continue;  /* Ignore orphan \E */
2814    
2815          if (c < 0)          if (c < 0)
2816            {            {
# Line 2686  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 2694  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 2703  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 2897  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 2938  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 2964  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 3117  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. In non-UTF-8 mode we can always optimize. In UTF-8 mode, we      less than 256. As long as there were no characters >= 128 and there was no
3291      can optimize the negative case only if there were no characters >= 128      use of \p or \P, in other words, no use of any XCLASS features, we can
3292      because OP_NOT and the related opcodes like OP_NOTSTAR operate on      optimize.
3293      single-bytes only. This is an historical hangover. Maybe one day we can  
3294      tidy these opcodes to handle multi-byte characters.      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
3296        operate on single-bytes only. This is an historical hangover. Maybe one day
3297        we can tidy these opcodes to handle multi-byte characters.
3298    
3299      The optimization throws away the bit map. We turn the item into a      The optimization throws away the bit map. We turn the item into a
3300      1-character OP_CHAR[NC] if it's positive, or OP_NOT if it's negative. Note      1-character OP_CHAR[NC] if it's positive, or OP_NOT if it's negative. Note
# Line 3132  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 &&      if (class_charcount == 1 && !class_utf8 &&
3308            (!utf8 ||        (!utf8 || !negate_class || class_lastchar < 128))
           (!class_utf8 && (!negate_class || class_lastchar < 128))))  
   
3309  #else  #else
3310      if (class_charcount == 1)      if (class_charcount == 1)
3311  #endif  #endif
# Line 3178  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 3208  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 3407  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 3557  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 3857  for (;; ptr++) Line 4029  for (;; ptr++)
4029        int len;        int len;
4030        if (*tempcode == OP_EXACT || *tempcode == OP_TYPEEXACT ||        if (*tempcode == OP_EXACT || *tempcode == OP_TYPEEXACT ||
4031            *tempcode == OP_NOTEXACT)            *tempcode == OP_NOTEXACT)
4032          tempcode += _pcre_OP_lengths[*tempcode];          tempcode += _pcre_OP_lengths[*tempcode] +
4033              ((*tempcode == OP_TYPEEXACT &&
4034                 (tempcode[3] == OP_PROP || tempcode[3] == OP_NOTPROP))? 2:0);
4035        len = code - tempcode;        len = code - tempcode;
4036        if (len > 0) switch (*tempcode)        if (len > 0) switch (*tempcode)
4037          {          {
# Line 3915  for (;; ptr++) Line 4089  for (;; ptr++)
4089      if (*(++ptr) == '*' && (cd->ctypes[ptr[1]] & ctype_letter) != 0)      if (*(++ptr) == '*' && (cd->ctypes[ptr[1]] & ctype_letter) != 0)
4090        {        {
4091        int i, namelen;        int i, namelen;
4092          const char *vn = verbnames;
4093        const uschar *name = ++ptr;        const uschar *name = ++ptr;
4094        previous = NULL;        previous = NULL;
4095        while ((cd->ctypes[*++ptr] & ctype_letter) != 0);        while ((cd->ctypes[*++ptr] & ctype_letter) != 0);
# Line 3932  for (;; ptr++) Line 4107  for (;; ptr++)
4107        for (i = 0; i < verbcount; i++)        for (i = 0; i < verbcount; i++)
4108          {          {
4109          if (namelen == verbs[i].len &&          if (namelen == verbs[i].len &&
4110              strncmp((char *)name, verbs[i].name, namelen) == 0)              strncmp((char *)name, vn, namelen) == 0)
4111            {            {
4112            *code = verbs[i].op;            *code = verbs[i].op;
4113            if (*code++ == OP_ACCEPT) cd->had_accept = TRUE;            if (*code++ == OP_ACCEPT) cd->had_accept = TRUE;
4114            break;            break;
4115            }            }
4116            vn += verbs[i].len + 1;
4117          }          }
4118        if (i < verbcount) continue;        if (i < verbcount) continue;
4119        *errorcodeptr = ERR60;        *errorcodeptr = ERR60;
# Line 4082  for (;; ptr++) Line 4258  for (;; ptr++)
4258              *errorcodeptr = ERR58;              *errorcodeptr = ERR58;
4259              goto FAILED;              goto FAILED;
4260              }              }
4261            if (refsign == '-')            recno = (refsign == '-')?
4262                cd->bracount - recno + 1 : recno +cd->bracount;
4263              if (recno <= 0 || recno > cd->final_bracount)
4264              {              {
4265              recno = cd->bracount - recno + 1;              *errorcodeptr = ERR15;
4266              if (recno <= 0)              goto FAILED;
               {  
               *errorcodeptr = ERR15;  
               goto FAILED;  
               }  
4267              }              }
           else recno += cd->bracount;  
4268            PUT2(code, 2+LINK_SIZE, recno);            PUT2(code, 2+LINK_SIZE, recno);
4269            break;            break;
4270            }            }
# Line 4163  for (;; ptr++) Line 4336  for (;; ptr++)
4336            skipbytes = 1;            skipbytes = 1;
4337            }            }
4338    
4339          /* Check for the "name" actually being a subpattern number. */          /* Check for the "name" actually being a subpattern number. We are
4340            in the second pass here, so final_bracount is set. */
4341    
4342          else if (recno > 0)          else if (recno > 0 && recno <= cd->final_bracount)
4343            {            {
4344            PUT2(code, 2+LINK_SIZE, recno);            PUT2(code, 2+LINK_SIZE, recno);
4345            }            }
# Line 4359  for (;; ptr++) Line 4533  for (;; ptr++)
4533    
4534          /* We come here from the Python syntax above that handles both          /* We come here from the Python syntax above that handles both
4535          references (?P=name) and recursion (?P>name), as well as falling          references (?P=name) and recursion (?P>name), as well as falling
4536          through from the Perl recursion syntax (?&name). */          through from the Perl recursion syntax (?&name). We also come here from
4537            the Perl \k<name> or \k'name' back reference syntax and the \k{name}
4538            .NET syntax. */
4539    
4540          NAMED_REF_OR_RECURSE:          NAMED_REF_OR_RECURSE:
4541          name = ++ptr;          name = ++ptr;
# Line 4371  for (;; ptr++) Line 4547  for (;; ptr++)
4547    
4548          if (lengthptr != NULL)          if (lengthptr != NULL)
4549            {            {
4550              if (namelen == 0)
4551                {
4552                *errorcodeptr = ERR62;
4553                goto FAILED;
4554                }
4555            if (*ptr != terminator)            if (*ptr != terminator)
4556              {              {
4557              *errorcodeptr = ERR42;              *errorcodeptr = ERR42;
# Line 4384  for (;; ptr++) Line 4565  for (;; ptr++)
4565            recno = 0;            recno = 0;
4566            }            }
4567    
4568          /* In the real compile, seek the name in the table */          /* In the real compile, seek the name in the table. We check the name
4569            first, and then check that we have reached the end of the name in the
4570            table. That way, if the name that is longer than any in the table,
4571            the comparison will fail without reading beyond the table entry. */
4572    
4573          else          else
4574            {            {
4575            slot = cd->name_table;            slot = cd->name_table;
4576            for (i = 0; i < cd->names_found; i++)            for (i = 0; i < cd->names_found; i++)
4577              {              {
4578              if (strncmp((char *)name, (char *)slot+2, namelen) == 0) break;              if (strncmp((char *)name, (char *)slot+2, namelen) == 0 &&
4579                    slot[2+namelen] == 0)
4580                  break;
4581              slot += cd->name_entry_size;              slot += cd->name_entry_size;
4582              }              }
4583    
# Line 4428  for (;; ptr++) Line 4614  for (;; ptr++)
4614            {            {
4615            const uschar *called;            const uschar *called;
4616    
4617            if ((refsign = *ptr) == '+') ptr++;            if ((refsign = *ptr) == '+')
4618                {
4619                ptr++;
4620                if ((digitab[*ptr] & ctype_digit) == 0)
4621                  {
4622                  *errorcodeptr = ERR63;
4623                  goto FAILED;
4624                  }
4625                }
4626            else if (refsign == '-')            else if (refsign == '-')
4627              {              {
4628              if ((digitab[ptr[1]] & ctype_digit) == 0)              if ((digitab[ptr[1]] & ctype_digit) == 0)
# Line 4554  for (;; ptr++) Line 4748  for (;; ptr++)
4748    
4749              case 'J':    /* Record that it changed in the external options */              case 'J':    /* Record that it changed in the external options */
4750              *optset |= PCRE_DUPNAMES;              *optset |= PCRE_DUPNAMES;
4751              cd->external_options |= PCRE_JCHANGED;              cd->external_flags |= PCRE_JCHANGED;
4752              break;              break;
4753    
4754              case 'i': *optset |= PCRE_CASELESS; break;              case 'i': *optset |= PCRE_CASELESS; break;
# Line 4745  for (;; ptr++) Line 4939  for (;; ptr++)
4939        goto FAILED;        goto FAILED;
4940        }        }
4941    
4942      /* In the pre-compile phase, update the length by the length of the nested      /* In the pre-compile phase, update the length by the length of the group,
4943      group, less the brackets at either end. Then reduce the compiled code to      less the brackets at either end. Then reduce the compiled code to just a
4944      just the brackets so that it doesn't use much memory if it is duplicated by      set of non-capturing brackets so that it doesn't use much memory if it is
4945      a quantifier. */      duplicated by a quantifier.*/
4946    
4947      if (lengthptr != NULL)      if (lengthptr != NULL)
4948        {        {
# Line 4758  for (;; ptr++) Line 4952  for (;; ptr++)
4952          goto FAILED;          goto FAILED;
4953          }          }
4954        *lengthptr += length_prevgroup - 2 - 2*LINK_SIZE;        *lengthptr += length_prevgroup - 2 - 2*LINK_SIZE;
4955        code++;        *code++ = OP_BRA;
4956        PUTINC(code, 0, 1 + LINK_SIZE);        PUTINC(code, 0, 1 + LINK_SIZE);
4957        *code++ = OP_KET;        *code++ = OP_KET;
4958        PUTINC(code, 0, 1 + LINK_SIZE);        PUTINC(code, 0, 1 + LINK_SIZE);
4959          break;    /* No need to waste time with special character handling */
4960        }        }
4961    
4962      /* Otherwise update the main code pointer to the end of the group. */      /* Otherwise update the main code pointer to the end of the group. */
4963    
4964      else code = tempcode;      code = tempcode;
4965    
4966      /* For a DEFINE group, required and first character settings are not      /* For a DEFINE group, required and first character settings are not
4967      relevant. */      relevant. */
# Line 4970  for (;; ptr++) Line 5165  for (;; ptr++)
5165      *code++ = ((options & PCRE_CASELESS) != 0)? OP_CHARNC : OP_CHAR;      *code++ = ((options & PCRE_CASELESS) != 0)? OP_CHARNC : OP_CHAR;
5166      for (c = 0; c < mclength; c++) *code++ = mcbuffer[c];      for (c = 0; c < mclength; c++) *code++ = mcbuffer[c];
5167    
5168        /* Remember if \r or \n were seen */
5169    
5170        if (mcbuffer[0] == '\r' || mcbuffer[0] == '\n')
5171          cd->external_flags |= PCRE_HASCRORLF;
5172    
5173      /* Set the first and required bytes appropriately. If no previous first      /* Set the first and required bytes appropriately. If no previous first
5174      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.
5175      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 5569  real_pcre *re; Line 5769  real_pcre *re;
5769  int length = 1;  /* For final END opcode */  int length = 1;  /* For final END opcode */
5770  int firstbyte, reqbyte, newline;  int firstbyte, reqbyte, newline;
5771  int errorcode = 0;  int errorcode = 0;
5772    int skipatstart = 0;
5773  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
5774  BOOL utf8;  BOOL utf8;
5775  #endif  #endif
# Line 5647  cd->fcc = tables + fcc_offset; Line 5848  cd->fcc = tables + fcc_offset;
5848  cd->cbits = tables + cbits_offset;  cd->cbits = tables + cbits_offset;
5849  cd->ctypes = tables + ctypes_offset;  cd->ctypes = tables + ctypes_offset;
5850    
5851    /* Check for global one-time settings at the start of the pattern, and remember
5852    the offset for later. */
5853    
5854    while (ptr[skipatstart] == '(' && ptr[skipatstart+1] == '*')
5855      {
5856      int newnl = 0;
5857      int newbsr = 0;
5858    
5859      if (strncmp((char *)(ptr+skipatstart+2), "CR)", 3) == 0)
5860        { skipatstart += 5; newnl = PCRE_NEWLINE_CR; }
5861      else if (strncmp((char *)(ptr+skipatstart+2), "LF)", 3)  == 0)
5862        { skipatstart += 5; newnl = PCRE_NEWLINE_LF; }
5863      else if (strncmp((char *)(ptr+skipatstart+2), "CRLF)", 5)  == 0)
5864        { skipatstart += 7; newnl = PCRE_NEWLINE_CR + PCRE_NEWLINE_LF; }
5865      else if (strncmp((char *)(ptr+skipatstart+2), "ANY)", 4) == 0)
5866        { skipatstart += 6; newnl = PCRE_NEWLINE_ANY; }
5867      else if (strncmp((char *)(ptr+skipatstart+2), "ANYCRLF)", 8)  == 0)
5868        { skipatstart += 10; newnl = PCRE_NEWLINE_ANYCRLF; }
5869    
5870      else if (strncmp((char *)(ptr+skipatstart+2), "BSR_ANYCRLF)", 12) == 0)
5871        { skipatstart += 14; newbsr = PCRE_BSR_ANYCRLF; }
5872      else if (strncmp((char *)(ptr+skipatstart+2), "BSR_UNICODE)", 12) == 0)
5873        { skipatstart += 14; newbsr = PCRE_BSR_UNICODE; }
5874    
5875      if (newnl != 0)
5876        options = (options & ~PCRE_NEWLINE_BITS) | newnl;
5877      else if (newbsr != 0)
5878        options = (options & ~(PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE)) | newbsr;
5879      else break;
5880      }
5881    
5882    /* Check validity of \R options. */
5883    
5884    switch (options & (PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE))
5885      {
5886      case 0:
5887      case PCRE_BSR_ANYCRLF:
5888      case PCRE_BSR_UNICODE:
5889      break;
5890      default: errorcode = ERR56; goto PCRE_EARLY_ERROR_RETURN;
5891      }
5892    
5893  /* Handle different types of newline. The three bits give seven cases. The  /* Handle different types of newline. The three bits give seven cases. The
5894  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
5895  "anycrlf". */  "anycrlf". */
5896    
5897  switch (options & (PCRE_NEWLINE_CRLF | PCRE_NEWLINE_ANY))  switch (options & PCRE_NEWLINE_BITS)
5898    {    {
5899    case 0: newline = NEWLINE; break;   /* Compile-time default */    case 0: newline = NEWLINE; break;   /* Build-time default */
5900    case PCRE_NEWLINE_CR: newline = '\r'; break;    case PCRE_NEWLINE_CR: newline = '\r'; break;
5901    case PCRE_NEWLINE_LF: newline = '\n'; break;    case PCRE_NEWLINE_LF: newline = '\n'; break;
5902    case PCRE_NEWLINE_CR+    case PCRE_NEWLINE_CR+
# Line 5706  to compile parts of the pattern into; th Line 5949  to compile parts of the pattern into; th
5949  no longer needed, so hopefully this workspace will never overflow, though there  no longer needed, so hopefully this workspace will never overflow, though there
5950  is a test for its doing so. */  is a test for its doing so. */
5951    
5952  cd->bracount = 0;  cd->bracount = cd->final_bracount = 0;
5953  cd->names_found = 0;  cd->names_found = 0;
5954  cd->name_entry_size = 0;  cd->name_entry_size = 0;
5955  cd->name_table = NULL;  cd->name_table = NULL;
# Line 5716  cd->hwm = cworkspace; Line 5959  cd->hwm = cworkspace;
5959  cd->start_pattern = (const uschar *)pattern;  cd->start_pattern = (const uschar *)pattern;
5960  cd->end_pattern = (const uschar *)(pattern + strlen(pattern));  cd->end_pattern = (const uschar *)(pattern + strlen(pattern));
5961  cd->req_varyopt = 0;  cd->req_varyopt = 0;
 cd->nopartial = FALSE;  
5962  cd->external_options = options;  cd->external_options = options;
5963    cd->external_flags = 0;
5964    
5965  /* 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
5966  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 5725  been put into the cd block so that they Line 5968  been put into the cd block so that they
5968  found within the regex right at the beginning. Bringing initial option settings  found within the regex right at the beginning. Bringing initial option settings
5969  outside can help speed up starting point checks. */  outside can help speed up starting point checks. */
5970    
5971    ptr += skipatstart;
5972  code = cworkspace;  code = cworkspace;
5973  *code = OP_BRA;  *code = OP_BRA;
5974  (void)compile_regex(cd->external_options, cd->external_options & PCRE_IMS,  (void)compile_regex(cd->external_options, cd->external_options & PCRE_IMS,
# Line 5755  if (re == NULL) Line 5999  if (re == NULL)
5999    goto PCRE_EARLY_ERROR_RETURN;    goto PCRE_EARLY_ERROR_RETURN;
6000    }    }
6001    
6002  /* 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
6003  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
6004  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
6005  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
6006    pointers. */
6007    
6008  re->magic_number = MAGIC_NUMBER;  re->magic_number = MAGIC_NUMBER;
6009  re->size = size;  re->size = size;
6010  re->options = cd->external_options;  re->options = cd->external_options;
6011    re->flags = cd->external_flags;
6012  re->dummy1 = 0;  re->dummy1 = 0;
6013  re->first_byte = 0;  re->first_byte = 0;
6014  re->req_byte = 0;  re->req_byte = 0;
# Line 5780  field. Reset the bracket count and the n Line 6026  field. Reset the bracket count and the n
6026  field; this time it's used for remembering forward references to subpatterns.  field; this time it's used for remembering forward references to subpatterns.
6027  */  */
6028    
6029    cd->final_bracount = cd->bracount;  /* Save for checking forward references */
6030  cd->bracount = 0;  cd->bracount = 0;
6031  cd->names_found = 0;  cd->names_found = 0;
6032  cd->name_table = (uschar *)re + re->name_table_offset;  cd->name_table = (uschar *)re + re->name_table_offset;
# Line 5787  codestart = cd->name_table + re->name_en Line 6034  codestart = cd->name_table + re->name_en
6034  cd->start_code = codestart;  cd->start_code = codestart;
6035  cd->hwm = cworkspace;  cd->hwm = cworkspace;
6036  cd->req_varyopt = 0;  cd->req_varyopt = 0;
 cd->nopartial = FALSE;  
6037  cd->had_accept = FALSE;  cd->had_accept = FALSE;
6038    
6039  /* Set up a starting, non-extracting bracket, then compile the expression. On  /* Set up a starting, non-extracting bracket, then compile the expression. On
6040  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
6041  of the function here. */  of the function here. */
6042    
6043  ptr = (const uschar *)pattern;  ptr = (const uschar *)pattern + skipatstart;
6044  code = (uschar *)codestart;  code = (uschar *)codestart;
6045  *code = OP_BRA;  *code = OP_BRA;
6046  (void)compile_regex(re->options, re->options & PCRE_IMS, &code, &ptr,  (void)compile_regex(re->options, re->options & PCRE_IMS, &code, &ptr,
6047    &errorcode, FALSE, FALSE, 0, &firstbyte, &reqbyte, NULL, cd, NULL);    &errorcode, FALSE, FALSE, 0, &firstbyte, &reqbyte, NULL, cd, NULL);
6048  re->top_bracket = cd->bracount;  re->top_bracket = cd->bracount;
6049  re->top_backref = cd->top_backref;  re->top_backref = cd->top_backref;
6050    re->flags = cd->external_flags;
6051    
 if (cd->nopartial) re->options |= PCRE_NOPARTIAL;  
6052  if (cd->had_accept) reqbyte = -1;   /* Must disable after (*ACCEPT) */  if (cd->had_accept) reqbyte = -1;   /* Must disable after (*ACCEPT) */
6053    
6054  /* 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 5845  if (errorcode != 0) Line 6091  if (errorcode != 0)
6091    PCRE_EARLY_ERROR_RETURN:    PCRE_EARLY_ERROR_RETURN:
6092    *erroroffset = ptr - (const uschar *)pattern;    *erroroffset = ptr - (const uschar *)pattern;
6093    PCRE_EARLY_ERROR_RETURN2:    PCRE_EARLY_ERROR_RETURN2:
6094    *errorptr = error_texts[errorcode];    *errorptr = find_error_text(errorcode);
6095    if (errorcodeptr != NULL) *errorcodeptr = errorcode;    if (errorcodeptr != NULL) *errorcodeptr = errorcode;
6096    return NULL;    return NULL;
6097    }    }
# Line 5874  if ((re->options & PCRE_ANCHORED) == 0) Line 6120  if ((re->options & PCRE_ANCHORED) == 0)
6120        int ch = firstbyte & 255;        int ch = firstbyte & 255;
6121        re->first_byte = ((firstbyte & REQ_CASELESS) != 0 &&        re->first_byte = ((firstbyte & REQ_CASELESS) != 0 &&
6122           cd->fcc[ch] == ch)? ch : firstbyte;           cd->fcc[ch] == ch)? ch : firstbyte;
6123        re->options |= PCRE_FIRSTSET;        re->flags |= PCRE_FIRSTSET;
6124        }        }
6125      else if (is_startline(codestart, 0, cd->backref_map))      else if (is_startline(codestart, 0, cd->backref_map))
6126        re->options |= PCRE_STARTLINE;        re->flags |= PCRE_STARTLINE;
6127      }      }
6128    }    }
6129    
# Line 5891  if (reqbyte >= 0 && Line 6137  if (reqbyte >= 0 &&
6137    int ch = reqbyte & 255;    int ch = reqbyte & 255;
6138    re->req_byte = ((reqbyte & REQ_CASELESS) != 0 &&    re->req_byte = ((reqbyte & REQ_CASELESS) != 0 &&
6139      cd->fcc[ch] == ch)? (reqbyte & ~REQ_CASELESS) : reqbyte;      cd->fcc[ch] == ch)? (reqbyte & ~REQ_CASELESS) : reqbyte;
6140    re->options |= PCRE_REQCHSET;    re->flags |= PCRE_REQCHSET;
6141    }    }
6142    
6143  /* 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 5902  case when building a production library. Line 6148  case when building a production library.
6148  printf("Length = %d top_bracket = %d top_backref = %d\n",  printf("Length = %d top_bracket = %d top_backref = %d\n",
6149    length, re->top_bracket, re->top_backref);    length, re->top_bracket, re->top_backref);
6150    
6151  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 " : "");  
   }  
6152    
6153  if ((re->options & PCRE_FIRSTSET) != 0)  if ((re->flags & PCRE_FIRSTSET) != 0)
6154    {    {
6155    int ch = re->first_byte & 255;    int ch = re->first_byte & 255;
6156    const char *caseless = ((re->first_byte & REQ_CASELESS) == 0)?    const char *caseless = ((re->first_byte & REQ_CASELESS) == 0)?
# Line 5925  if ((re->options & PCRE_FIRSTSET) != 0) Line 6159  if ((re->options & PCRE_FIRSTSET) != 0)
6159      else printf("First char = \\x%02x%s\n", ch, caseless);      else printf("First char = \\x%02x%s\n", ch, caseless);
6160    }    }
6161    
6162  if ((re->options & PCRE_REQCHSET) != 0)  if ((re->flags & PCRE_REQCHSET) != 0)
6163    {    {
6164    int ch = re->req_byte & 255;    int ch = re->req_byte & 255;
6165    const char *caseless = ((re->req_byte & REQ_CASELESS) == 0)?    const char *caseless = ((re->req_byte & REQ_CASELESS) == 0)?
# Line 5942  was compiled can be seen. */ Line 6176  was compiled can be seen. */
6176  if (code - codestart > length)  if (code - codestart > length)
6177    {    {
6178    (pcre_free)(re);    (pcre_free)(re);
6179    *errorptr = error_texts[ERR23];    *errorptr = find_error_text(ERR23);
6180    *erroroffset = ptr - (uschar *)pattern;    *erroroffset = ptr - (uschar *)pattern;
6181    if (errorcodeptr != NULL) *errorcodeptr = ERR23;    if (errorcodeptr != NULL) *errorcodeptr = ERR23;
6182    return NULL;    return NULL;
Legend:
Removed from v.216  
changed lines
  Added in v.285
webmaster@exim.org
 ViewVC Help
Powered by ViewVC 1.1.12