--- code/trunk/pcreposix.c	2007/12/17 14:46:11	286
+++ code/trunk/pcreposix.c	2010/03/03 20:09:39	499
@@ -6,7 +6,7 @@
 and semantics are as close as possible to those of the Perl 5 language.
 
                        Written by Philip Hazel
-           Copyright (c) 1997-2007 University of Cambridge
+           Copyright (c) 1997-2009 University of Cambridge
 
 -----------------------------------------------------------------------------
 Redistribution and use in source and binary forms, with or without
@@ -70,63 +70,80 @@
   REG_EESCAPE, /* \c at end of pattern */
   REG_EESCAPE, /* unrecognized character follows \ */
   REG_BADBR,   /* numbers out of order in {} quantifier */
+  /* 5 */
   REG_BADBR,   /* number too big in {} quantifier */
   REG_EBRACK,  /* missing terminating ] for character class */
   REG_ECTYPE,  /* invalid escape sequence in character class */
   REG_ERANGE,  /* range out of order in character class */
   REG_BADRPT,  /* nothing to repeat */
+  /* 10 */
   REG_BADRPT,  /* operand of unlimited repeat could match the empty string */
   REG_ASSERT,  /* internal error: unexpected repeat */
   REG_BADPAT,  /* unrecognized character after (? */
   REG_BADPAT,  /* POSIX named classes are supported only within a class */
   REG_EPAREN,  /* missing ) */
+  /* 15 */
   REG_ESUBREG, /* reference to non-existent subpattern */
   REG_INVARG,  /* erroffset passed as NULL */
   REG_INVARG,  /* unknown option bit(s) set */
   REG_EPAREN,  /* missing ) after comment */
   REG_ESIZE,   /* parentheses nested too deeply */
+  /* 20 */
   REG_ESIZE,   /* regular expression too large */
   REG_ESPACE,  /* failed to get memory */
-  REG_EPAREN,  /* unmatched brackets */
+  REG_EPAREN,  /* unmatched parentheses */
   REG_ASSERT,  /* internal error: code overflow */
   REG_BADPAT,  /* unrecognized character after (?< */
+  /* 25 */
   REG_BADPAT,  /* lookbehind assertion is not fixed length */
   REG_BADPAT,  /* malformed number or name after (?( */
   REG_BADPAT,  /* conditional group contains more than two branches */
   REG_BADPAT,  /* assertion expected after (?( */
   REG_BADPAT,  /* (?R or (?[+-]digits must be followed by ) */
+  /* 30 */
   REG_ECTYPE,  /* unknown POSIX class name */
   REG_BADPAT,  /* POSIX collating elements are not supported */
   REG_INVARG,  /* this version of PCRE is not compiled with PCRE_UTF8 support */
   REG_BADPAT,  /* spare error */
   REG_BADPAT,  /* character value in \x{...} sequence is too large */
+  /* 35 */
   REG_BADPAT,  /* invalid condition (?(0) */
   REG_BADPAT,  /* \C not allowed in lookbehind assertion */
   REG_EESCAPE, /* PCRE does not support \L, \l, \N, \U, or \u */
   REG_BADPAT,  /* number after (?C is > 255 */
   REG_BADPAT,  /* closing ) for (?C expected */
+  /* 40 */
   REG_BADPAT,  /* recursive call could loop indefinitely */
   REG_BADPAT,  /* unrecognized character after (?P */
   REG_BADPAT,  /* syntax error in subpattern name (missing terminator) */
   REG_BADPAT,  /* two named subpatterns have the same name */
   REG_BADPAT,  /* invalid UTF-8 string */
+  /* 45 */
   REG_BADPAT,  /* support for \P, \p, and \X has not been compiled */
   REG_BADPAT,  /* malformed \P or \p sequence */
   REG_BADPAT,  /* unknown property name after \P or \p */
   REG_BADPAT,  /* subpattern name is too long (maximum 32 characters) */
   REG_BADPAT,  /* too many named subpatterns (maximum 10,000) */
+  /* 50 */
   REG_BADPAT,  /* repeated subpattern is too long */
   REG_BADPAT,  /* octal value is greater than \377 (not in UTF-8 mode) */
   REG_BADPAT,  /* internal error: overran compiling workspace */
   REG_BADPAT,  /* internal error: previously-checked referenced subpattern not found */
   REG_BADPAT,  /* DEFINE group contains more than one branch */
+  /* 55 */
   REG_BADPAT,  /* repeating a DEFINE group is not allowed */
   REG_INVARG,  /* inconsistent NEWLINE options */
   REG_BADPAT,  /* \g is not followed followed by an (optionally braced) non-zero number */
-  REG_BADPAT,  /* (?+ or (?- must be followed by a non-zero number */
+  REG_BADPAT,  /* a numbered reference must not be zero */
+  REG_BADPAT,  /* (*VERB) with an argument is not supported */
+  /* 60 */
+  REG_BADPAT,  /* (*VERB) not recognized */
   REG_BADPAT,  /* number is too big */
   REG_BADPAT,  /* subpattern name expected */
-  REG_BADPAT   /* digit expected after (?+ */
+  REG_BADPAT,  /* digit expected after (?+ */
+  REG_BADPAT,  /* ] is an invalid data character in JavaScript compatibility mode */
+  /* 65 */
+  REG_BADPAT   /* different names for subpatterns of the same number are not allowed */
 };
 
 /* Table of texts corresponding to POSIX error codes */
@@ -159,7 +176,7 @@
 *          Translate error code to string        *
 *************************************************/
 
-PCREPOSIX_EXP_DEFN size_t
+PCREPOSIX_EXP_DEFN size_t PCRE_CALL_CONVENTION
 regerror(int errcode, const regex_t *preg, char *errbuf, size_t errbuf_size)
 {
 const char *message, *addmessage;
@@ -194,7 +211,7 @@
 *           Free store held by a regex           *
 *************************************************/
 
-PCREPOSIX_EXP_DEFN void
+PCREPOSIX_EXP_DEFN void PCRE_CALL_CONVENTION
 regfree(regex_t *preg)
 {
 (pcre_free)(preg->re_pcre);
@@ -217,7 +234,7 @@
               various non-zero codes on failure
 */
 
-PCREPOSIX_EXP_DEFN int
+PCREPOSIX_EXP_DEFN int PCRE_CALL_CONVENTION
 regcomp(regex_t *preg, const char *pattern, int cflags)
 {
 const char *errorptr;
@@ -225,17 +242,25 @@
 int errorcode;
 int options = 0;
 
-if ((cflags & REG_ICASE) != 0)   options |= PCRE_CASELESS;
-if ((cflags & REG_NEWLINE) != 0) options |= PCRE_MULTILINE;
-if ((cflags & REG_DOTALL) != 0)  options |= PCRE_DOTALL;
-if ((cflags & REG_NOSUB) != 0)   options |= PCRE_NO_AUTO_CAPTURE;
-if ((cflags & REG_UTF8) != 0)    options |= PCRE_UTF8;
+if ((cflags & REG_ICASE) != 0)    options |= PCRE_CASELESS;
+if ((cflags & REG_NEWLINE) != 0)  options |= PCRE_MULTILINE;
+if ((cflags & REG_DOTALL) != 0)   options |= PCRE_DOTALL;
+if ((cflags & REG_NOSUB) != 0)    options |= PCRE_NO_AUTO_CAPTURE;
+if ((cflags & REG_UTF8) != 0)     options |= PCRE_UTF8;
+if ((cflags & REG_UNGREEDY) != 0) options |= PCRE_UNGREEDY;
 
 preg->re_pcre = pcre_compile2(pattern, options, &errorcode, &errorptr,
   &erroffset, NULL);
 preg->re_erroffset = erroffset;
 
-if (preg->re_pcre == NULL) return eint[errorcode];
+/* Safety: if the error code is too big for the translation vector (which
+should not happen, but we all make mistakes), return REG_BADPAT. */
+
+if (preg->re_pcre == NULL)
+  {
+  return (errorcode < sizeof(eint)/sizeof(const int))?
+    eint[errorcode] : REG_BADPAT;
+  }
 
 preg->re_nsub = pcre_info((const pcre *)preg->re_pcre, NULL, NULL);
 return 0;
@@ -259,11 +284,11 @@
 be set. When this is the case, the nmatch and pmatch arguments are ignored, and
 the only result is yes/no/error. */
 
-PCREPOSIX_EXP_DEFN int
+PCREPOSIX_EXP_DEFN int PCRE_CALL_CONVENTION
 regexec(const regex_t *preg, const char *string, size_t nmatch,
   regmatch_t pmatch[], int eflags)
 {
-int rc;
+int rc, so, eo;
 int options = 0;
 int *ovector = NULL;
 int small_ovector[POSIX_MALLOC_THRESHOLD * 3];
@@ -273,13 +298,15 @@
 
 if ((eflags & REG_NOTBOL) != 0) options |= PCRE_NOTBOL;
 if ((eflags & REG_NOTEOL) != 0) options |= PCRE_NOTEOL;
+if ((eflags & REG_NOTEMPTY) != 0) options |= PCRE_NOTEMPTY;
 
 ((regex_t *)preg)->re_erroffset = (size_t)(-1);  /* Only has meaning after compile */
 
-/* When no string data is being returned, ensure that nmatch is zero.
-Otherwise, ensure the vector for holding the return data is large enough. */
+/* When no string data is being returned, or no vector has been passed in which
+to put it, ensure that nmatch is zero. Otherwise, ensure the vector for holding
+the return data is large enough. */
 
-if (nosub) nmatch = 0;
+if (nosub || pmatch == NULL) nmatch = 0;
 
 else if (nmatch > 0)
   {
@@ -296,11 +323,29 @@
     }
   }
 
-rc = pcre_exec((const pcre *)preg->re_pcre, NULL, string, (int)strlen(string),
+/* REG_STARTEND is a BSD extension, to allow for non-NUL-terminated strings.
+The man page from OS X says "REG_STARTEND affects only the location of the
+string, not how it is matched". That is why the "so" value is used to bump the
+start location rather than being passed as a PCRE "starting offset". */
+
+if ((eflags & REG_STARTEND) != 0)
+  {
+  so = pmatch[0].rm_so;
+  eo = pmatch[0].rm_eo;
+  }
+else
+  {
+  so = 0;
+  eo = strlen(string);
+  }
+
+rc = pcre_exec((const pcre *)preg->re_pcre, NULL, string + so, (eo - so),
   0, options, ovector, nmatch * 3);
 
 if (rc == 0) rc = nmatch;    /* All captured slots were filled in */
 
+/* Successful match */
+
 if (rc >= 0)
   {
   size_t i;
@@ -317,22 +362,33 @@
   return 0;
   }
 
-else
+/* Unsuccessful match */
+
+if (allocated_ovector) free(ovector);
+switch(rc)
   {
-  if (allocated_ovector) free(ovector);
-  switch(rc)
-    {
-    case PCRE_ERROR_NOMATCH: return REG_NOMATCH;
-    case PCRE_ERROR_NULL: return REG_INVARG;
-    case PCRE_ERROR_BADOPTION: return REG_INVARG;
-    case PCRE_ERROR_BADMAGIC: return REG_INVARG;
-    case PCRE_ERROR_UNKNOWN_NODE: return REG_ASSERT;
-    case PCRE_ERROR_NOMEMORY: return REG_ESPACE;
-    case PCRE_ERROR_MATCHLIMIT: return REG_ESPACE;
-    case PCRE_ERROR_BADUTF8: return REG_INVARG;
-    case PCRE_ERROR_BADUTF8_OFFSET: return REG_INVARG;
-    default: return REG_ASSERT;
-    }
+/* ========================================================================== */
+  /* These cases are never obeyed. This is a fudge that causes a compile-time
+  error if the vector eint, which is indexed by compile-time error number, is
+  not the correct length. It seems to be the only way to do such a check at
+  compile time, as the sizeof() operator does not work in the C preprocessor.
+  As all the PCRE_ERROR_xxx values are negative, we can use 0 and 1. */ 
+
+  case 0:
+  case (sizeof(eint)/sizeof(int) == ERRCOUNT):
+  return REG_ASSERT;
+/* ========================================================================== */
+
+  case PCRE_ERROR_NOMATCH: return REG_NOMATCH;
+  case PCRE_ERROR_NULL: return REG_INVARG;
+  case PCRE_ERROR_BADOPTION: return REG_INVARG;
+  case PCRE_ERROR_BADMAGIC: return REG_INVARG;
+  case PCRE_ERROR_UNKNOWN_NODE: return REG_ASSERT;
+  case PCRE_ERROR_NOMEMORY: return REG_ESPACE;
+  case PCRE_ERROR_MATCHLIMIT: return REG_ESPACE;
+  case PCRE_ERROR_BADUTF8: return REG_INVARG;
+  case PCRE_ERROR_BADUTF8_OFFSET: return REG_INVARG;
+  default: return REG_ASSERT;
   }
 }