xref: /PHP-7.0/ext/pcre/pcrelib/pcreposix.c (revision 5f9df47e)
1 /*************************************************
2 *      Perl-Compatible Regular Expressions       *
3 *************************************************/
4 
5 /* PCRE is a library of functions to support regular expressions whose syntax
6 and semantics are as close as possible to those of the Perl 5 language.
7 
8                        Written by Philip Hazel
9            Copyright (c) 1997-2014 University of Cambridge
10 
11 -----------------------------------------------------------------------------
12 Redistribution and use in source and binary forms, with or without
13 modification, are permitted provided that the following conditions are met:
14 
15     * Redistributions of source code must retain the above copyright notice,
16       this list of conditions and the following disclaimer.
17 
18     * Redistributions in binary form must reproduce the above copyright
19       notice, this list of conditions and the following disclaimer in the
20       documentation and/or other materials provided with the distribution.
21 
22     * Neither the name of the University of Cambridge nor the names of its
23       contributors may be used to endorse or promote products derived from
24       this software without specific prior written permission.
25 
26 THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
27 AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
28 IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
29 ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE
30 LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
31 CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
32 SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
33 INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
34 CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
35 ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
36 POSSIBILITY OF SUCH DAMAGE.
37 -----------------------------------------------------------------------------
38 */
39 
40 
41 /* This module is a wrapper that provides a POSIX API to the underlying PCRE
42 functions. */
43 
44 
45 #include "config.h"
46 
47 
48 /* Ensure that the PCREPOSIX_EXP_xxx macros are set appropriately for
49 compiling these functions. This must come before including pcreposix.h, where
50 they are set for an application (using these functions) if they have not
51 previously been set. */
52 
53 #if defined(_WIN32) && !defined(PCRE_STATIC)
54 #  define PCREPOSIX_EXP_DECL extern __declspec(dllexport)
55 #  define PCREPOSIX_EXP_DEFN __declspec(dllexport)
56 #endif
57 
58 /* We include pcre.h before pcre_internal.h so that the PCRE library functions
59 are declared as "import" for Windows by defining PCRE_EXP_DECL as "import".
60 This is needed even though pcre_internal.h itself includes pcre.h, because it
61 does so after it has set PCRE_EXP_DECL to "export" if it is not already set. */
62 
63 #include "pcre.h"
64 #include "pcre_internal.h"
65 #include "pcreposix.h"
66 
67 
68 /* Table to translate PCRE compile time error codes into POSIX error codes. */
69 
70 static const int eint[] = {
71   0,           /* no error */
72   REG_EESCAPE, /* \ at end of pattern */
73   REG_EESCAPE, /* \c at end of pattern */
74   REG_EESCAPE, /* unrecognized character follows \ */
75   REG_BADBR,   /* numbers out of order in {} quantifier */
76   /* 5 */
77   REG_BADBR,   /* number too big in {} quantifier */
78   REG_EBRACK,  /* missing terminating ] for character class */
79   REG_ECTYPE,  /* invalid escape sequence in character class */
80   REG_ERANGE,  /* range out of order in character class */
81   REG_BADRPT,  /* nothing to repeat */
82   /* 10 */
83   REG_BADRPT,  /* operand of unlimited repeat could match the empty string */
84   REG_ASSERT,  /* internal error: unexpected repeat */
85   REG_BADPAT,  /* unrecognized character after (? */
86   REG_BADPAT,  /* POSIX named classes are supported only within a class */
87   REG_EPAREN,  /* missing ) */
88   /* 15 */
89   REG_ESUBREG, /* reference to non-existent subpattern */
90   REG_INVARG,  /* erroffset passed as NULL */
91   REG_INVARG,  /* unknown option bit(s) set */
92   REG_EPAREN,  /* missing ) after comment */
93   REG_ESIZE,   /* parentheses nested too deeply */
94   /* 20 */
95   REG_ESIZE,   /* regular expression too large */
96   REG_ESPACE,  /* failed to get memory */
97   REG_EPAREN,  /* unmatched parentheses */
98   REG_ASSERT,  /* internal error: code overflow */
99   REG_BADPAT,  /* unrecognized character after (?< */
100   /* 25 */
101   REG_BADPAT,  /* lookbehind assertion is not fixed length */
102   REG_BADPAT,  /* malformed number or name after (?( */
103   REG_BADPAT,  /* conditional group contains more than two branches */
104   REG_BADPAT,  /* assertion expected after (?( */
105   REG_BADPAT,  /* (?R or (?[+-]digits must be followed by ) */
106   /* 30 */
107   REG_ECTYPE,  /* unknown POSIX class name */
108   REG_BADPAT,  /* POSIX collating elements are not supported */
109   REG_INVARG,  /* this version of PCRE is not compiled with PCRE_UTF8 support */
110   REG_BADPAT,  /* spare error */
111   REG_BADPAT,  /* character value in \x{} or \o{} is too large */
112   /* 35 */
113   REG_BADPAT,  /* invalid condition (?(0) */
114   REG_BADPAT,  /* \C not allowed in lookbehind assertion */
115   REG_EESCAPE, /* PCRE does not support \L, \l, \N, \U, or \u */
116   REG_BADPAT,  /* number after (?C is > 255 */
117   REG_BADPAT,  /* closing ) for (?C expected */
118   /* 40 */
119   REG_BADPAT,  /* recursive call could loop indefinitely */
120   REG_BADPAT,  /* unrecognized character after (?P */
121   REG_BADPAT,  /* syntax error in subpattern name (missing terminator) */
122   REG_BADPAT,  /* two named subpatterns have the same name */
123   REG_BADPAT,  /* invalid UTF-8 string */
124   /* 45 */
125   REG_BADPAT,  /* support for \P, \p, and \X has not been compiled */
126   REG_BADPAT,  /* malformed \P or \p sequence */
127   REG_BADPAT,  /* unknown property name after \P or \p */
128   REG_BADPAT,  /* subpattern name is too long (maximum 32 characters) */
129   REG_BADPAT,  /* too many named subpatterns (maximum 10,000) */
130   /* 50 */
131   REG_BADPAT,  /* repeated subpattern is too long */
132   REG_BADPAT,  /* octal value is greater than \377 (not in UTF-8 mode) */
133   REG_BADPAT,  /* internal error: overran compiling workspace */
134   REG_BADPAT,  /* internal error: previously-checked referenced subpattern not found */
135   REG_BADPAT,  /* DEFINE group contains more than one branch */
136   /* 55 */
137   REG_BADPAT,  /* repeating a DEFINE group is not allowed */
138   REG_INVARG,  /* inconsistent NEWLINE options */
139   REG_BADPAT,  /* \g is not followed followed by an (optionally braced) non-zero number */
140   REG_BADPAT,  /* a numbered reference must not be zero */
141   REG_BADPAT,  /* an argument is not allowed for (*ACCEPT), (*FAIL), or (*COMMIT) */
142   /* 60 */
143   REG_BADPAT,  /* (*VERB) not recognized */
144   REG_BADPAT,  /* number is too big */
145   REG_BADPAT,  /* subpattern name expected */
146   REG_BADPAT,  /* digit expected after (?+ */
147   REG_BADPAT,  /* ] is an invalid data character in JavaScript compatibility mode */
148   /* 65 */
149   REG_BADPAT,  /* different names for subpatterns of the same number are not allowed */
150   REG_BADPAT,  /* (*MARK) must have an argument */
151   REG_INVARG,  /* this version of PCRE is not compiled with PCRE_UCP support */
152   REG_BADPAT,  /* \c must be followed by an ASCII character */
153   REG_BADPAT,  /* \k is not followed by a braced, angle-bracketed, or quoted name */
154   /* 70 */
155   REG_BADPAT,  /* internal error: unknown opcode in find_fixedlength() */
156   REG_BADPAT,  /* \N is not supported in a class */
157   REG_BADPAT,  /* too many forward references */
158   REG_BADPAT,  /* disallowed UTF-8/16/32 code point (>= 0xd800 && <= 0xdfff) */
159   REG_BADPAT,  /* invalid UTF-16 string (should not occur) */
160   /* 75 */
161   REG_BADPAT,  /* overlong MARK name */
162   REG_BADPAT,  /* character value in \u.... sequence is too large */
163   REG_BADPAT,  /* invalid UTF-32 string (should not occur) */
164   REG_BADPAT,  /* setting UTF is disabled by the application */
165   REG_BADPAT,  /* non-hex character in \\x{} (closing brace missing?) */
166   /* 80 */
167   REG_BADPAT,  /* non-octal character in \o{} (closing brace missing?) */
168   REG_BADPAT,  /* missing opening brace after \o */
169   REG_BADPAT,  /* parentheses too deeply nested */
170   REG_BADPAT,  /* invalid range in character class */
171   REG_BADPAT,  /* group name must start with a non-digit */
172   /* 85 */
173   REG_BADPAT,  /* parentheses too deeply nested (stack check) */
174   REG_BADPAT   /* missing digits in \x{} or \o{} */
175 };
176 
177 /* Table of texts corresponding to POSIX error codes */
178 
179 static const char *const pstring[] = {
180   "",                                /* Dummy for value 0 */
181   "internal error",                  /* REG_ASSERT */
182   "invalid repeat counts in {}",     /* BADBR      */
183   "pattern error",                   /* BADPAT     */
184   "? * + invalid",                   /* BADRPT     */
185   "unbalanced {}",                   /* EBRACE     */
186   "unbalanced []",                   /* EBRACK     */
187   "collation error - not relevant",  /* ECOLLATE   */
188   "bad class",                       /* ECTYPE     */
189   "bad escape sequence",             /* EESCAPE    */
190   "empty expression",                /* EMPTY      */
191   "unbalanced ()",                   /* EPAREN     */
192   "bad range inside []",             /* ERANGE     */
193   "expression too big",              /* ESIZE      */
194   "failed to get memory",            /* ESPACE     */
195   "bad back reference",              /* ESUBREG    */
196   "bad argument",                    /* INVARG     */
197   "match failed"                     /* NOMATCH    */
198 };
199 
200 
201 
202 
203 /*************************************************
204 *          Translate error code to string        *
205 *************************************************/
206 
207 PCREPOSIX_EXP_DEFN size_t PCRE_CALL_CONVENTION
regerror(int errcode,const regex_t * preg,char * errbuf,size_t errbuf_size)208 regerror(int errcode, const regex_t *preg, char *errbuf, size_t errbuf_size)
209 {
210 const char *message, *addmessage;
211 size_t length, addlength;
212 
213 message = (errcode >= (int)(sizeof(pstring)/sizeof(char *)))?
214   "unknown error code" : pstring[errcode];
215 length = strlen(message) + 1;
216 
217 addmessage = " at offset ";
218 addlength = (preg != NULL && (int)preg->re_erroffset != -1)?
219   strlen(addmessage) + 6 : 0;
220 
221 if (errbuf_size > 0)
222   {
223   if (addlength > 0 && errbuf_size >= length + addlength)
224     sprintf(errbuf, "%s%s%-6d", message, addmessage, (int)preg->re_erroffset);
225   else
226     {
227     strncpy(errbuf, message, errbuf_size - 1);
228     errbuf[errbuf_size-1] = 0;
229     }
230   }
231 
232 return length + addlength;
233 }
234 
235 
236 
237 
238 /*************************************************
239 *           Free store held by a regex           *
240 *************************************************/
241 
242 PCREPOSIX_EXP_DEFN void PCRE_CALL_CONVENTION
regfree(regex_t * preg)243 regfree(regex_t *preg)
244 {
245 (PUBL(free))(preg->re_pcre);
246 }
247 
248 
249 
250 
251 /*************************************************
252 *            Compile a regular expression        *
253 *************************************************/
254 
255 /*
256 Arguments:
257   preg        points to a structure for recording the compiled expression
258   pattern     the pattern to compile
259   cflags      compilation flags
260 
261 Returns:      0 on success
262               various non-zero codes on failure
263 */
264 
265 PCREPOSIX_EXP_DEFN int PCRE_CALL_CONVENTION
regcomp(regex_t * preg,const char * pattern,int cflags)266 regcomp(regex_t *preg, const char *pattern, int cflags)
267 {
268 const char *errorptr;
269 int erroffset;
270 int errorcode;
271 int options = 0;
272 int re_nsub = 0;
273 
274 if ((cflags & REG_ICASE) != 0)    options |= PCRE_CASELESS;
275 if ((cflags & REG_NEWLINE) != 0)  options |= PCRE_MULTILINE;
276 if ((cflags & REG_DOTALL) != 0)   options |= PCRE_DOTALL;
277 if ((cflags & REG_NOSUB) != 0)    options |= PCRE_NO_AUTO_CAPTURE;
278 if ((cflags & REG_UTF8) != 0)     options |= PCRE_UTF8;
279 if ((cflags & REG_UCP) != 0)      options |= PCRE_UCP;
280 if ((cflags & REG_UNGREEDY) != 0) options |= PCRE_UNGREEDY;
281 
282 preg->re_pcre = pcre_compile2(pattern, options, &errorcode, &errorptr,
283   &erroffset, NULL);
284 preg->re_erroffset = erroffset;
285 
286 /* Safety: if the error code is too big for the translation vector (which
287 should not happen, but we all make mistakes), return REG_BADPAT. */
288 
289 if (preg->re_pcre == NULL)
290   {
291   return (errorcode < (int)(sizeof(eint)/sizeof(const int)))?
292     eint[errorcode] : REG_BADPAT;
293   }
294 
295 (void)pcre_fullinfo((const pcre *)preg->re_pcre, NULL, PCRE_INFO_CAPTURECOUNT,
296   &re_nsub);
297 preg->re_nsub = (size_t)re_nsub;
298 return 0;
299 }
300 
301 
302 
303 
304 /*************************************************
305 *              Match a regular expression        *
306 *************************************************/
307 
308 /* Unfortunately, PCRE requires 3 ints of working space for each captured
309 substring, so we have to get and release working store instead of just using
310 the POSIX structures as was done in earlier releases when PCRE needed only 2
311 ints. However, if the number of possible capturing brackets is small, use a
312 block of store on the stack, to reduce the use of malloc/free. The threshold is
313 in a macro that can be changed at configure time.
314 
315 If REG_NOSUB was specified at compile time, the PCRE_NO_AUTO_CAPTURE flag will
316 be set. When this is the case, the nmatch and pmatch arguments are ignored, and
317 the only result is yes/no/error. */
318 
319 PCREPOSIX_EXP_DEFN int PCRE_CALL_CONVENTION
regexec(const regex_t * preg,const char * string,size_t nmatch,regmatch_t pmatch[],int eflags)320 regexec(const regex_t *preg, const char *string, size_t nmatch,
321   regmatch_t pmatch[], int eflags)
322 {
323 int rc, so, eo;
324 int options = 0;
325 int *ovector = NULL;
326 int small_ovector[POSIX_MALLOC_THRESHOLD * 3];
327 BOOL allocated_ovector = FALSE;
328 BOOL nosub =
329   (REAL_PCRE_OPTIONS((const pcre *)preg->re_pcre) & PCRE_NO_AUTO_CAPTURE) != 0;
330 
331 if ((eflags & REG_NOTBOL) != 0) options |= PCRE_NOTBOL;
332 if ((eflags & REG_NOTEOL) != 0) options |= PCRE_NOTEOL;
333 if ((eflags & REG_NOTEMPTY) != 0) options |= PCRE_NOTEMPTY;
334 
335 ((regex_t *)preg)->re_erroffset = (size_t)(-1);  /* Only has meaning after compile */
336 
337 /* When no string data is being returned, or no vector has been passed in which
338 to put it, ensure that nmatch is zero. Otherwise, ensure the vector for holding
339 the return data is large enough. */
340 
341 if (nosub || pmatch == NULL) nmatch = 0;
342 
343 else if (nmatch > 0)
344   {
345   if (nmatch <= POSIX_MALLOC_THRESHOLD)
346     {
347     ovector = &(small_ovector[0]);
348     }
349   else
350     {
351     if (nmatch > INT_MAX/(sizeof(int) * 3)) return REG_ESPACE;
352     ovector = (int *)malloc(sizeof(int) * nmatch * 3);
353     if (ovector == NULL) return REG_ESPACE;
354     allocated_ovector = TRUE;
355     }
356   }
357 
358 /* REG_STARTEND is a BSD extension, to allow for non-NUL-terminated strings.
359 The man page from OS X says "REG_STARTEND affects only the location of the
360 string, not how it is matched". That is why the "so" value is used to bump the
361 start location rather than being passed as a PCRE "starting offset". */
362 
363 if ((eflags & REG_STARTEND) != 0)
364   {
365   so = pmatch[0].rm_so;
366   eo = pmatch[0].rm_eo;
367   }
368 else
369   {
370   so = 0;
371   eo = (int)strlen(string);
372   }
373 
374 rc = pcre_exec((const pcre *)preg->re_pcre, NULL, string + so, (eo - so),
375   0, options, ovector, (int)(nmatch * 3));
376 
377 if (rc == 0) rc = (int)nmatch;    /* All captured slots were filled in */
378 
379 /* Successful match */
380 
381 if (rc >= 0)
382   {
383   size_t i;
384   if (!nosub)
385     {
386     for (i = 0; i < (size_t)rc; i++)
387       {
388       pmatch[i].rm_so = ovector[i*2];
389       pmatch[i].rm_eo = ovector[i*2+1];
390       }
391     if (allocated_ovector) free(ovector);
392     for (; i < nmatch; i++) pmatch[i].rm_so = pmatch[i].rm_eo = -1;
393     }
394   return 0;
395   }
396 
397 /* Unsuccessful match */
398 
399 if (allocated_ovector) free(ovector);
400 switch(rc)
401   {
402 /* ========================================================================== */
403   /* These cases are never obeyed. This is a fudge that causes a compile-time
404   error if the vector eint, which is indexed by compile-time error number, is
405   not the correct length. It seems to be the only way to do such a check at
406   compile time, as the sizeof() operator does not work in the C preprocessor.
407   As all the PCRE_ERROR_xxx values are negative, we can use 0 and 1. */
408 
409   case 0:
410   case (sizeof(eint)/sizeof(int) == ERRCOUNT):
411   return REG_ASSERT;
412 /* ========================================================================== */
413 
414   case PCRE_ERROR_NOMATCH: return REG_NOMATCH;
415   case PCRE_ERROR_NULL: return REG_INVARG;
416   case PCRE_ERROR_BADOPTION: return REG_INVARG;
417   case PCRE_ERROR_BADMAGIC: return REG_INVARG;
418   case PCRE_ERROR_UNKNOWN_NODE: return REG_ASSERT;
419   case PCRE_ERROR_NOMEMORY: return REG_ESPACE;
420   case PCRE_ERROR_MATCHLIMIT: return REG_ESPACE;
421   case PCRE_ERROR_BADUTF8: return REG_INVARG;
422   case PCRE_ERROR_BADUTF8_OFFSET: return REG_INVARG;
423   case PCRE_ERROR_BADMODE: return REG_INVARG;
424   default: return REG_ASSERT;
425   }
426 }
427 
428 /* End of pcreposix.c */
429