/[pcre]/code/trunk/pcreposix.c
ViewVC logotype

Diff of /code/trunk/pcreposix.c

Parent Directory Parent Directory | Revision Log Revision Log | View Patch Patch

revision 220 by ph10, Thu Aug 16 13:29:39 2007 UTC revision 760 by ph10, Tue Nov 22 11:23:43 2011 UTC
# Line 6  Line 6 
6  and semantics are as close as possible to those of the Perl 5 language.  and semantics are as close as possible to those of the Perl 5 language.
7    
8                         Written by Philip Hazel                         Written by Philip Hazel
9             Copyright (c) 1997-2007 University of Cambridge             Copyright (c) 1997-2010 University of Cambridge
10    
11  -----------------------------------------------------------------------------  -----------------------------------------------------------------------------
12  Redistribution and use in source and binary forms, with or without  Redistribution and use in source and binary forms, with or without
# Line 43  functions. */ Line 43  functions. */
43    
44    
45  #ifdef HAVE_CONFIG_H  #ifdef HAVE_CONFIG_H
46  #include <config.h>  #include "config.h"
47  #endif  #endif
48    
49    
# Line 57  previously been set. */ Line 57  previously been set. */
57  #  define PCREPOSIX_EXP_DEFN __declspec(dllexport)  #  define PCREPOSIX_EXP_DEFN __declspec(dllexport)
58  #endif  #endif
59    
60  #include <pcre.h>  /* We include pcre.h before pcre_internal.h so that the PCRE library functions
61    are declared as "import" for Windows by defining PCRE_EXP_DECL as "import".
62    This is needed even though pcre_internal.h itself includes pcre.h, because it
63    does so after it has set PCRE_EXP_DECL to "export" if it is not already set. */
64    
65    #include "pcre.h"
66  #include "pcre_internal.h"  #include "pcre_internal.h"
67  #include "pcreposix.h"  #include "pcreposix.h"
68    
69    
   
70  /* Table to translate PCRE compile time error codes into POSIX error codes. */  /* Table to translate PCRE compile time error codes into POSIX error codes. */
71    
72  static const int eint[] = {  static const int eint[] = {
# Line 71  static const int eint[] = { Line 75  static const int eint[] = {
75    REG_EESCAPE, /* \c at end of pattern */    REG_EESCAPE, /* \c at end of pattern */
76    REG_EESCAPE, /* unrecognized character follows \ */    REG_EESCAPE, /* unrecognized character follows \ */
77    REG_BADBR,   /* numbers out of order in {} quantifier */    REG_BADBR,   /* numbers out of order in {} quantifier */
78      /* 5 */
79    REG_BADBR,   /* number too big in {} quantifier */    REG_BADBR,   /* number too big in {} quantifier */
80    REG_EBRACK,  /* missing terminating ] for character class */    REG_EBRACK,  /* missing terminating ] for character class */
81    REG_ECTYPE,  /* invalid escape sequence in character class */    REG_ECTYPE,  /* invalid escape sequence in character class */
82    REG_ERANGE,  /* range out of order in character class */    REG_ERANGE,  /* range out of order in character class */
83    REG_BADRPT,  /* nothing to repeat */    REG_BADRPT,  /* nothing to repeat */
84      /* 10 */
85    REG_BADRPT,  /* operand of unlimited repeat could match the empty string */    REG_BADRPT,  /* operand of unlimited repeat could match the empty string */
86    REG_ASSERT,  /* internal error: unexpected repeat */    REG_ASSERT,  /* internal error: unexpected repeat */
87    REG_BADPAT,  /* unrecognized character after (? */    REG_BADPAT,  /* unrecognized character after (? */
88    REG_BADPAT,  /* POSIX named classes are supported only within a class */    REG_BADPAT,  /* POSIX named classes are supported only within a class */
89    REG_EPAREN,  /* missing ) */    REG_EPAREN,  /* missing ) */
90      /* 15 */
91    REG_ESUBREG, /* reference to non-existent subpattern */    REG_ESUBREG, /* reference to non-existent subpattern */
92    REG_INVARG,  /* erroffset passed as NULL */    REG_INVARG,  /* erroffset passed as NULL */
93    REG_INVARG,  /* unknown option bit(s) set */    REG_INVARG,  /* unknown option bit(s) set */
94    REG_EPAREN,  /* missing ) after comment */    REG_EPAREN,  /* missing ) after comment */
95    REG_ESIZE,   /* parentheses nested too deeply */    REG_ESIZE,   /* parentheses nested too deeply */
96      /* 20 */
97    REG_ESIZE,   /* regular expression too large */    REG_ESIZE,   /* regular expression too large */
98    REG_ESPACE,  /* failed to get memory */    REG_ESPACE,  /* failed to get memory */
99    REG_EPAREN,  /* unmatched brackets */    REG_EPAREN,  /* unmatched parentheses */
100    REG_ASSERT,  /* internal error: code overflow */    REG_ASSERT,  /* internal error: code overflow */
101    REG_BADPAT,  /* unrecognized character after (?< */    REG_BADPAT,  /* unrecognized character after (?< */
102      /* 25 */
103    REG_BADPAT,  /* lookbehind assertion is not fixed length */    REG_BADPAT,  /* lookbehind assertion is not fixed length */
104    REG_BADPAT,  /* malformed number or name after (?( */    REG_BADPAT,  /* malformed number or name after (?( */
105    REG_BADPAT,  /* conditional group contains more than two branches */    REG_BADPAT,  /* conditional group contains more than two branches */
106    REG_BADPAT,  /* assertion expected after (?( */    REG_BADPAT,  /* assertion expected after (?( */
107    REG_BADPAT,  /* (?R or (?[+-]digits must be followed by ) */    REG_BADPAT,  /* (?R or (?[+-]digits must be followed by ) */
108      /* 30 */
109    REG_ECTYPE,  /* unknown POSIX class name */    REG_ECTYPE,  /* unknown POSIX class name */
110    REG_BADPAT,  /* POSIX collating elements are not supported */    REG_BADPAT,  /* POSIX collating elements are not supported */
111    REG_INVARG,  /* this version of PCRE is not compiled with PCRE_UTF8 support */    REG_INVARG,  /* this version of PCRE is not compiled with PCRE_UTF8 support */
112    REG_BADPAT,  /* spare error */    REG_BADPAT,  /* spare error */
113    REG_BADPAT,  /* character value in \x{...} sequence is too large */    REG_BADPAT,  /* character value in \x{...} sequence is too large */
114      /* 35 */
115    REG_BADPAT,  /* invalid condition (?(0) */    REG_BADPAT,  /* invalid condition (?(0) */
116    REG_BADPAT,  /* \C not allowed in lookbehind assertion */    REG_BADPAT,  /* \C not allowed in lookbehind assertion */
117    REG_EESCAPE, /* PCRE does not support \L, \l, \N, \U, or \u */    REG_EESCAPE, /* PCRE does not support \L, \l, \N, \U, or \u */
118    REG_BADPAT,  /* number after (?C is > 255 */    REG_BADPAT,  /* number after (?C is > 255 */
119    REG_BADPAT,  /* closing ) for (?C expected */    REG_BADPAT,  /* closing ) for (?C expected */
120      /* 40 */
121    REG_BADPAT,  /* recursive call could loop indefinitely */    REG_BADPAT,  /* recursive call could loop indefinitely */
122    REG_BADPAT,  /* unrecognized character after (?P */    REG_BADPAT,  /* unrecognized character after (?P */
123    REG_BADPAT,  /* syntax error in subpattern name (missing terminator) */    REG_BADPAT,  /* syntax error in subpattern name (missing terminator) */
124    REG_BADPAT,  /* two named subpatterns have the same name */    REG_BADPAT,  /* two named subpatterns have the same name */
125    REG_BADPAT,  /* invalid UTF-8 string */    REG_BADPAT,  /* invalid UTF-8 string */
126      /* 45 */
127    REG_BADPAT,  /* support for \P, \p, and \X has not been compiled */    REG_BADPAT,  /* support for \P, \p, and \X has not been compiled */
128    REG_BADPAT,  /* malformed \P or \p sequence */    REG_BADPAT,  /* malformed \P or \p sequence */
129    REG_BADPAT,  /* unknown property name after \P or \p */    REG_BADPAT,  /* unknown property name after \P or \p */
130    REG_BADPAT,  /* subpattern name is too long (maximum 32 characters) */    REG_BADPAT,  /* subpattern name is too long (maximum 32 characters) */
131    REG_BADPAT,  /* too many named subpatterns (maximum 10,000) */    REG_BADPAT,  /* too many named subpatterns (maximum 10,000) */
132      /* 50 */
133    REG_BADPAT,  /* repeated subpattern is too long */    REG_BADPAT,  /* repeated subpattern is too long */
134    REG_BADPAT,  /* octal value is greater than \377 (not in UTF-8 mode) */    REG_BADPAT,  /* octal value is greater than \377 (not in UTF-8 mode) */
135    REG_BADPAT,  /* internal error: overran compiling workspace */    REG_BADPAT,  /* internal error: overran compiling workspace */
136    REG_BADPAT,  /* internal error: previously-checked referenced subpattern not found */    REG_BADPAT,  /* internal error: previously-checked referenced subpattern not found */
137    REG_BADPAT,  /* DEFINE group contains more than one branch */    REG_BADPAT,  /* DEFINE group contains more than one branch */
138      /* 55 */
139    REG_BADPAT,  /* repeating a DEFINE group is not allowed */    REG_BADPAT,  /* repeating a DEFINE group is not allowed */
140    REG_INVARG,  /* inconsistent NEWLINE options */    REG_INVARG,  /* inconsistent NEWLINE options */
141    REG_BADPAT,  /* \g is not followed followed by an (optionally braced) non-zero number */    REG_BADPAT,  /* \g is not followed followed by an (optionally braced) non-zero number */
142    REG_BADPAT,  /* (?+ or (?- must be followed by a non-zero number */    REG_BADPAT,  /* a numbered reference must not be zero */
143    REG_BADPAT   /* number is too big */    REG_BADPAT,  /* an argument is not allowed for (*ACCEPT), (*FAIL), or (*COMMIT) */
144      /* 60 */
145      REG_BADPAT,  /* (*VERB) not recognized */
146      REG_BADPAT,  /* number is too big */
147      REG_BADPAT,  /* subpattern name expected */
148      REG_BADPAT,  /* digit expected after (?+ */
149      REG_BADPAT,  /* ] is an invalid data character in JavaScript compatibility mode */
150      /* 65 */
151      REG_BADPAT,  /* different names for subpatterns of the same number are not allowed */
152      REG_BADPAT,  /* (*MARK) must have an argument */
153      REG_INVARG,  /* this version of PCRE is not compiled with PCRE_UCP support */
154      REG_BADPAT,  /* \c must be followed by an ASCII character */
155      REG_BADPAT,  /* \k is not followed by a braced, angle-bracketed, or quoted name */
156      /* 70 */
157      REG_BADPAT,  /* internal error: unknown opcode in find_fixedlength() */
158      REG_BADPAT,  /* \N is not supported in a class */
159      REG_BADPAT,  /* too many forward references */
160  };  };
161    
162  /* Table of texts corresponding to POSIX error codes */  /* Table of texts corresponding to POSIX error codes */
# Line 158  static const char *const pstring[] = { Line 189  static const char *const pstring[] = {
189  *          Translate error code to string        *  *          Translate error code to string        *
190  *************************************************/  *************************************************/
191    
192  PCREPOSIX_EXP_DEFN size_t  PCREPOSIX_EXP_DEFN size_t PCRE_CALL_CONVENTION
193  regerror(int errcode, const regex_t *preg, char *errbuf, size_t errbuf_size)  regerror(int errcode, const regex_t *preg, char *errbuf, size_t errbuf_size)
194  {  {
195  const char *message, *addmessage;  const char *message, *addmessage;
# Line 193  return length + addlength; Line 224  return length + addlength;
224  *           Free store held by a regex           *  *           Free store held by a regex           *
225  *************************************************/  *************************************************/
226    
227  PCREPOSIX_EXP_DEFN void  PCREPOSIX_EXP_DEFN void PCRE_CALL_CONVENTION
228  regfree(regex_t *preg)  regfree(regex_t *preg)
229  {  {
230  (pcre_free)(preg->re_pcre);  (pcre_free)(preg->re_pcre);
# Line 216  Returns:      0 on success Line 247  Returns:      0 on success
247                various non-zero codes on failure                various non-zero codes on failure
248  */  */
249    
250  PCREPOSIX_EXP_DEFN int  PCREPOSIX_EXP_DEFN int PCRE_CALL_CONVENTION
251  regcomp(regex_t *preg, const char *pattern, int cflags)  regcomp(regex_t *preg, const char *pattern, int cflags)
252  {  {
253  const char *errorptr;  const char *errorptr;
# Line 224  int erroffset; Line 255  int erroffset;
255  int errorcode;  int errorcode;
256  int options = 0;  int options = 0;
257    
258  if ((cflags & REG_ICASE) != 0)   options |= PCRE_CASELESS;  if ((cflags & REG_ICASE) != 0)    options |= PCRE_CASELESS;
259  if ((cflags & REG_NEWLINE) != 0) options |= PCRE_MULTILINE;  if ((cflags & REG_NEWLINE) != 0)  options |= PCRE_MULTILINE;
260  if ((cflags & REG_DOTALL) != 0)  options |= PCRE_DOTALL;  if ((cflags & REG_DOTALL) != 0)   options |= PCRE_DOTALL;
261  if ((cflags & REG_NOSUB) != 0)   options |= PCRE_NO_AUTO_CAPTURE;  if ((cflags & REG_NOSUB) != 0)    options |= PCRE_NO_AUTO_CAPTURE;
262  if ((cflags & REG_UTF8) != 0)    options |= PCRE_UTF8;  if ((cflags & REG_UTF8) != 0)     options |= PCRE_UTF8;
263    if ((cflags & REG_UCP) != 0)      options |= PCRE_UCP;
264    if ((cflags & REG_UNGREEDY) != 0) options |= PCRE_UNGREEDY;
265    
266  preg->re_pcre = pcre_compile2(pattern, options, &errorcode, &errorptr,  preg->re_pcre = pcre_compile2(pattern, options, &errorcode, &errorptr,
267    &erroffset, NULL);    &erroffset, NULL);
268  preg->re_erroffset = erroffset;  preg->re_erroffset = erroffset;
269    
270  if (preg->re_pcre == NULL) return eint[errorcode];  /* Safety: if the error code is too big for the translation vector (which
271    should not happen, but we all make mistakes), return REG_BADPAT. */
272    
273    if (preg->re_pcre == NULL)
274      {
275      return (errorcode < sizeof(eint)/sizeof(const int))?
276        eint[errorcode] : REG_BADPAT;
277      }
278    
279  preg->re_nsub = pcre_info((const pcre *)preg->re_pcre, NULL, NULL);  preg->re_nsub = pcre_info((const pcre *)preg->re_pcre, NULL, NULL);
280  return 0;  return 0;
# Line 258  If REG_NOSUB was specified at compile ti Line 298  If REG_NOSUB was specified at compile ti
298  be set. When this is the case, the nmatch and pmatch arguments are ignored, and  be set. When this is the case, the nmatch and pmatch arguments are ignored, and
299  the only result is yes/no/error. */  the only result is yes/no/error. */
300    
301  PCREPOSIX_EXP_DEFN int  PCREPOSIX_EXP_DEFN int PCRE_CALL_CONVENTION
302  regexec(const regex_t *preg, const char *string, size_t nmatch,  regexec(const regex_t *preg, const char *string, size_t nmatch,
303    regmatch_t pmatch[], int eflags)    regmatch_t pmatch[], int eflags)
304  {  {
305  int rc;  int rc, so, eo;
306  int options = 0;  int options = 0;
307  int *ovector = NULL;  int *ovector = NULL;
308  int small_ovector[POSIX_MALLOC_THRESHOLD * 3];  int small_ovector[POSIX_MALLOC_THRESHOLD * 3];
# Line 272  BOOL nosub = Line 312  BOOL nosub =
312    
313  if ((eflags & REG_NOTBOL) != 0) options |= PCRE_NOTBOL;  if ((eflags & REG_NOTBOL) != 0) options |= PCRE_NOTBOL;
314  if ((eflags & REG_NOTEOL) != 0) options |= PCRE_NOTEOL;  if ((eflags & REG_NOTEOL) != 0) options |= PCRE_NOTEOL;
315    if ((eflags & REG_NOTEMPTY) != 0) options |= PCRE_NOTEMPTY;
316    
317  ((regex_t *)preg)->re_erroffset = (size_t)(-1);  /* Only has meaning after compile */  ((regex_t *)preg)->re_erroffset = (size_t)(-1);  /* Only has meaning after compile */
318    
319  /* When no string data is being returned, ensure that nmatch is zero.  /* When no string data is being returned, or no vector has been passed in which
320  Otherwise, ensure the vector for holding the return data is large enough. */  to put it, ensure that nmatch is zero. Otherwise, ensure the vector for holding
321    the return data is large enough. */
322    
323  if (nosub) nmatch = 0;  if (nosub || pmatch == NULL) nmatch = 0;
324    
325  else if (nmatch > 0)  else if (nmatch > 0)
326    {    {
# Line 295  else if (nmatch > 0) Line 337  else if (nmatch > 0)
337      }      }
338    }    }
339    
340  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.
341    0, options, ovector, nmatch * 3);  The man page from OS X says "REG_STARTEND affects only the location of the
342    string, not how it is matched". That is why the "so" value is used to bump the
343    start location rather than being passed as a PCRE "starting offset". */
344    
345  if (rc == 0) rc = nmatch;    /* All captured slots were filled in */  if ((eflags & REG_STARTEND) != 0)
346      {
347      so = pmatch[0].rm_so;
348      eo = pmatch[0].rm_eo;
349      }
350    else
351      {
352      so = 0;
353      eo = (int)strlen(string);
354      }
355    
356    rc = pcre_exec((const pcre *)preg->re_pcre, NULL, string + so, (eo - so),
357      0, options, ovector, (int)(nmatch * 3));
358    
359    if (rc == 0) rc = (int)nmatch;    /* All captured slots were filled in */
360    
361    /* Successful match */
362    
363  if (rc >= 0)  if (rc >= 0)
364    {    {
# Line 316  if (rc >= 0) Line 376  if (rc >= 0)
376    return 0;    return 0;
377    }    }
378    
379  else  /* Unsuccessful match */
380    
381    if (allocated_ovector) free(ovector);
382    switch(rc)
383    {    {
384    if (allocated_ovector) free(ovector);  /* ========================================================================== */
385    switch(rc)    /* These cases are never obeyed. This is a fudge that causes a compile-time
386      {    error if the vector eint, which is indexed by compile-time error number, is
387      case PCRE_ERROR_NOMATCH: return REG_NOMATCH;    not the correct length. It seems to be the only way to do such a check at
388      case PCRE_ERROR_NULL: return REG_INVARG;    compile time, as the sizeof() operator does not work in the C preprocessor.
389      case PCRE_ERROR_BADOPTION: return REG_INVARG;    As all the PCRE_ERROR_xxx values are negative, we can use 0 and 1. */
390      case PCRE_ERROR_BADMAGIC: return REG_INVARG;  
391      case PCRE_ERROR_UNKNOWN_NODE: return REG_ASSERT;    case 0:
392      case PCRE_ERROR_NOMEMORY: return REG_ESPACE;    case (sizeof(eint)/sizeof(int) == ERRCOUNT):
393      case PCRE_ERROR_MATCHLIMIT: return REG_ESPACE;    return REG_ASSERT;
394      case PCRE_ERROR_BADUTF8: return REG_INVARG;  /* ========================================================================== */
395      case PCRE_ERROR_BADUTF8_OFFSET: return REG_INVARG;  
396      default: return REG_ASSERT;    case PCRE_ERROR_NOMATCH: return REG_NOMATCH;
397      }    case PCRE_ERROR_NULL: return REG_INVARG;
398      case PCRE_ERROR_BADOPTION: return REG_INVARG;
399      case PCRE_ERROR_BADMAGIC: return REG_INVARG;
400      case PCRE_ERROR_UNKNOWN_NODE: return REG_ASSERT;
401      case PCRE_ERROR_NOMEMORY: return REG_ESPACE;
402      case PCRE_ERROR_MATCHLIMIT: return REG_ESPACE;
403      case PCRE_ERROR_BADUTF8: return REG_INVARG;
404      case PCRE_ERROR_BADUTF8_OFFSET: return REG_INVARG;
405      default: return REG_ASSERT;
406    }    }
407  }  }
408    

Legend:
Removed from v.220  
changed lines
  Added in v.760

  ViewVC Help
Powered by ViewVC 1.1.5