/[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 77 by nigel, Sat Feb 24 21:40:45 2007 UTC revision 1404 by ph10, Tue Nov 19 15:36:57 2013 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) 1997-2005 University of Cambridge             Copyright (c) 1997-2012 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 42  POSSIBILITY OF SUCH DAMAGE. Line 42  POSSIBILITY OF SUCH DAMAGE.
42  functions. */  functions. */
43    
44    
45    #ifdef HAVE_CONFIG_H
46    #include "config.h"
47    #endif
48    
49    
50    /* Ensure that the PCREPOSIX_EXP_xxx macros are set appropriately for
51    compiling these functions. This must come before including pcreposix.h, where
52    they are set for an application (using these functions) if they have not
53    previously been set. */
54    
55    #if defined(_WIN32) && !defined(PCRE_STATIC)
56    #  define PCREPOSIX_EXP_DECL extern __declspec(dllexport)
57    #  define PCREPOSIX_EXP_DEFN __declspec(dllexport)
58    #endif
59    
60    /* 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"
 #include "stdlib.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. */
# Line 56  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 after (?( */    REG_BADPAT,  /* malformed number or name after (?( */
105    REG_BADPAT,  /* conditional group containe 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{} or \o{} 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 after (?P */    REG_BADPAT,  /* syntax error in subpattern name (missing terminator) */
124    REG_BADPAT,  /* two named groups 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) */
131      REG_BADPAT,  /* too many named subpatterns (maximum 10,000) */
132      /* 50 */
133      REG_BADPAT,  /* repeated subpattern is too long */
134      REG_BADPAT,  /* octal value is greater than \377 (not in UTF-8 mode) */
135      REG_BADPAT,  /* internal error: overran compiling workspace */
136      REG_BADPAT,  /* internal error: previously-checked referenced subpattern not found */
137      REG_BADPAT,  /* DEFINE group contains more than one branch */
138      /* 55 */
139      REG_BADPAT,  /* repeating a DEFINE group is not allowed */
140      REG_INVARG,  /* inconsistent NEWLINE options */
141      REG_BADPAT,  /* \g is not followed followed by an (optionally braced) non-zero number */
142      REG_BADPAT,  /* a numbered reference must not be zero */
143      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      REG_BADPAT,  /* disallowed UTF-8/16/32 code point (>= 0xd800 && <= 0xdfff) */
161      REG_BADPAT,  /* invalid UTF-16 string (should not occur) */
162      /* 75 */
163      REG_BADPAT,  /* overlong MARK name */
164      REG_BADPAT,  /* character value in \u.... sequence is too large */
165      REG_BADPAT,  /* invalid UTF-32 string (should not occur) */
166      REG_BADPAT,  /* setting UTF is disabled by the application */
167      REG_BADPAT,  /* non-hex character in \\x{} (closing brace missing?) */
168      /* 80 */
169      REG_BADPAT,  /* non-octal character in \o{} (closing brace missing?) */
170      REG_BADPAT,  /* missing opening brace after \o */
171      REG_BADPAT,  /* parentheses too deeply nested */
172      REG_BADPAT,  /* invalid range in character class */
173      REG_BADPAT   /* group name must start with a non-digit */
174  };  };
175    
176  /* Table of texts corresponding to POSIX error codes */  /* Table of texts corresponding to POSIX error codes */
# Line 131  static const char *const pstring[] = { Line 203  static const char *const pstring[] = {
203  *          Translate error code to string        *  *          Translate error code to string        *
204  *************************************************/  *************************************************/
205    
206  EXPORT size_t  PCREPOSIX_EXP_DEFN size_t PCRE_CALL_CONVENTION
207  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)
208  {  {
209  const char *message, *addmessage;  const char *message, *addmessage;
# Line 166  return length + addlength; Line 238  return length + addlength;
238  *           Free store held by a regex           *  *           Free store held by a regex           *
239  *************************************************/  *************************************************/
240    
241  EXPORT void  PCREPOSIX_EXP_DEFN void PCRE_CALL_CONVENTION
242  regfree(regex_t *preg)  regfree(regex_t *preg)
243  {  {
244  (pcre_free)(preg->re_pcre);  (PUBL(free))(preg->re_pcre);
245  }  }
246    
247    
# Line 189  Returns:      0 on success Line 261  Returns:      0 on success
261                various non-zero codes on failure                various non-zero codes on failure
262  */  */
263    
264  EXPORT int  PCREPOSIX_EXP_DEFN int PCRE_CALL_CONVENTION
265  regcomp(regex_t *preg, const char *pattern, int cflags)  regcomp(regex_t *preg, const char *pattern, int cflags)
266  {  {
267  const char *errorptr;  const char *errorptr;
268  int erroffset;  int erroffset;
269  int errorcode;  int errorcode;
270  int options = 0;  int options = 0;
271    int re_nsub = 0;
272    
273  if ((cflags & REG_ICASE) != 0) options |= PCRE_CASELESS;  if ((cflags & REG_ICASE) != 0)    options |= PCRE_CASELESS;
274  if ((cflags & REG_NEWLINE) != 0) options |= PCRE_MULTILINE;  if ((cflags & REG_NEWLINE) != 0)  options |= PCRE_MULTILINE;
275  if ((cflags & REG_DOTALL) != 0) options |= PCRE_DOTALL;  if ((cflags & REG_DOTALL) != 0)   options |= PCRE_DOTALL;
276    if ((cflags & REG_NOSUB) != 0)    options |= PCRE_NO_AUTO_CAPTURE;
277    if ((cflags & REG_UTF8) != 0)     options |= PCRE_UTF8;
278    if ((cflags & REG_UCP) != 0)      options |= PCRE_UCP;
279    if ((cflags & REG_UNGREEDY) != 0) options |= PCRE_UNGREEDY;
280    
281  preg->re_pcre = pcre_compile2(pattern, options, &errorcode, &errorptr,  preg->re_pcre = pcre_compile2(pattern, options, &errorcode, &errorptr,
282    &erroffset, NULL);    &erroffset, NULL);
283  preg->re_erroffset = erroffset;  preg->re_erroffset = erroffset;
284    
285  if (preg->re_pcre == NULL) return eint[errorcode];  /* Safety: if the error code is too big for the translation vector (which
286    should not happen, but we all make mistakes), return REG_BADPAT. */
287    
288    if (preg->re_pcre == NULL)
289      {
290      return (errorcode < (int)(sizeof(eint)/sizeof(const int)))?
291        eint[errorcode] : REG_BADPAT;
292      }
293    
294  preg->re_nsub = pcre_info((const pcre *)preg->re_pcre, NULL, NULL);  (void)pcre_fullinfo((const pcre *)preg->re_pcre, NULL, PCRE_INFO_CAPTURECOUNT,
295      &re_nsub);
296    preg->re_nsub = (size_t)re_nsub;
297  return 0;  return 0;
298  }  }
299    
# Line 223  substring, so we have to get and release Line 309  substring, so we have to get and release
309  the POSIX structures as was done in earlier releases when PCRE needed only 2  the POSIX structures as was done in earlier releases when PCRE needed only 2
310  ints. However, if the number of possible capturing brackets is small, use a  ints. However, if the number of possible capturing brackets is small, use a
311  block of store on the stack, to reduce the use of malloc/free. The threshold is  block of store on the stack, to reduce the use of malloc/free. The threshold is
312  in a macro that can be changed at configure time. */  in a macro that can be changed at configure time.
313    
314  EXPORT int  If REG_NOSUB was specified at compile time, the PCRE_NO_AUTO_CAPTURE flag will
315    be set. When this is the case, the nmatch and pmatch arguments are ignored, and
316    the only result is yes/no/error. */
317    
318    PCREPOSIX_EXP_DEFN int PCRE_CALL_CONVENTION
319  regexec(const regex_t *preg, const char *string, size_t nmatch,  regexec(const regex_t *preg, const char *string, size_t nmatch,
320    regmatch_t pmatch[], int eflags)    regmatch_t pmatch[], int eflags)
321  {  {
322  int rc;  int rc, so, eo;
323  int options = 0;  int options = 0;
324  int *ovector = NULL;  int *ovector = NULL;
325  int small_ovector[POSIX_MALLOC_THRESHOLD * 3];  int small_ovector[POSIX_MALLOC_THRESHOLD * 3];
326  BOOL allocated_ovector = FALSE;  BOOL allocated_ovector = FALSE;
327    BOOL nosub =
328      (REAL_PCRE_OPTIONS((const pcre *)preg->re_pcre) & PCRE_NO_AUTO_CAPTURE) != 0;
329    
330  if ((eflags & REG_NOTBOL) != 0) options |= PCRE_NOTBOL;  if ((eflags & REG_NOTBOL) != 0) options |= PCRE_NOTBOL;
331  if ((eflags & REG_NOTEOL) != 0) options |= PCRE_NOTEOL;  if ((eflags & REG_NOTEOL) != 0) options |= PCRE_NOTEOL;
332    if ((eflags & REG_NOTEMPTY) != 0) options |= PCRE_NOTEMPTY;
333    
334  ((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 */
335    
336  if (nmatch > 0)  /* When no string data is being returned, or no vector has been passed in which
337    to put it, ensure that nmatch is zero. Otherwise, ensure the vector for holding
338    the return data is large enough. */
339    
340    if (nosub || pmatch == NULL) nmatch = 0;
341    
342    else if (nmatch > 0)
343    {    {
344    if (nmatch <= POSIX_MALLOC_THRESHOLD)    if (nmatch <= POSIX_MALLOC_THRESHOLD)
345      {      {
# Line 248  if (nmatch > 0) Line 347  if (nmatch > 0)
347      }      }
348    else    else
349      {      {
350        if (nmatch > INT_MAX/(sizeof(int) * 3)) return REG_ESPACE;
351      ovector = (int *)malloc(sizeof(int) * nmatch * 3);      ovector = (int *)malloc(sizeof(int) * nmatch * 3);
352      if (ovector == NULL) return REG_ESPACE;      if (ovector == NULL) return REG_ESPACE;
353      allocated_ovector = TRUE;      allocated_ovector = TRUE;
354      }      }
355    }    }
356    
357  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.
358    0, options, ovector, nmatch * 3);  The man page from OS X says "REG_STARTEND affects only the location of the
359    string, not how it is matched". That is why the "so" value is used to bump the
360    start location rather than being passed as a PCRE "starting offset". */
361    
362    if ((eflags & REG_STARTEND) != 0)
363      {
364      so = pmatch[0].rm_so;
365      eo = pmatch[0].rm_eo;
366      }
367    else
368      {
369      so = 0;
370      eo = (int)strlen(string);
371      }
372    
373    rc = pcre_exec((const pcre *)preg->re_pcre, NULL, string + so, (eo - so),
374      0, options, ovector, (int)(nmatch * 3));
375    
376  if (rc == 0) rc = nmatch;    /* All captured slots were filled in */  if (rc == 0) rc = (int)nmatch;    /* All captured slots were filled in */
377    
378    /* Successful match */
379    
380  if (rc >= 0)  if (rc >= 0)
381    {    {
382    size_t i;    size_t i;
383    for (i = 0; i < (size_t)rc; i++)    if (!nosub)
384      {      {
385      pmatch[i].rm_so = ovector[i*2];      for (i = 0; i < (size_t)rc; i++)
386      pmatch[i].rm_eo = ovector[i*2+1];        {
387          pmatch[i].rm_so = ovector[i*2];
388          pmatch[i].rm_eo = ovector[i*2+1];
389          }
390        if (allocated_ovector) free(ovector);
391        for (; i < nmatch; i++) pmatch[i].rm_so = pmatch[i].rm_eo = -1;
392      }      }
   if (allocated_ovector) free(ovector);  
   for (; i < nmatch; i++) pmatch[i].rm_so = pmatch[i].rm_eo = -1;  
393    return 0;    return 0;
394    }    }
395    
396  else  /* Unsuccessful match */
397    
398    if (allocated_ovector) free(ovector);
399    switch(rc)
400    {    {
401    if (allocated_ovector) free(ovector);  /* ========================================================================== */
402    switch(rc)    /* These cases are never obeyed. This is a fudge that causes a compile-time
403      {    error if the vector eint, which is indexed by compile-time error number, is
404      case PCRE_ERROR_NOMATCH: return REG_NOMATCH;    not the correct length. It seems to be the only way to do such a check at
405      case PCRE_ERROR_NULL: return REG_INVARG;    compile time, as the sizeof() operator does not work in the C preprocessor.
406      case PCRE_ERROR_BADOPTION: return REG_INVARG;    As all the PCRE_ERROR_xxx values are negative, we can use 0 and 1. */
407      case PCRE_ERROR_BADMAGIC: return REG_INVARG;  
408      case PCRE_ERROR_UNKNOWN_NODE: return REG_ASSERT;    case 0:
409      case PCRE_ERROR_NOMEMORY: return REG_ESPACE;    case (sizeof(eint)/sizeof(int) == ERRCOUNT):
410      case PCRE_ERROR_MATCHLIMIT: return REG_ESPACE;    return REG_ASSERT;
411      case PCRE_ERROR_BADUTF8: return REG_INVARG;  /* ========================================================================== */
412      case PCRE_ERROR_BADUTF8_OFFSET: return REG_INVARG;  
413      default: return REG_ASSERT;    case PCRE_ERROR_NOMATCH: return REG_NOMATCH;
414      }    case PCRE_ERROR_NULL: return REG_INVARG;
415      case PCRE_ERROR_BADOPTION: return REG_INVARG;
416      case PCRE_ERROR_BADMAGIC: return REG_INVARG;
417      case PCRE_ERROR_UNKNOWN_NODE: return REG_ASSERT;
418      case PCRE_ERROR_NOMEMORY: return REG_ESPACE;
419      case PCRE_ERROR_MATCHLIMIT: return REG_ESPACE;
420      case PCRE_ERROR_BADUTF8: return REG_INVARG;
421      case PCRE_ERROR_BADUTF8_OFFSET: return REG_INVARG;
422      case PCRE_ERROR_BADMODE: return REG_INVARG;
423      default: return REG_ASSERT;
424    }    }
425  }  }
426    

Legend:
Removed from v.77  
changed lines
  Added in v.1404

  ViewVC Help
Powered by ViewVC 1.1.5