Annotation of embedaddon/php/ext/pcre/pcrelib/pcreposix.c, revision 1.1.1.1
1.1 misho 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) 1997-2010 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{...} sequence 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: };
154:
155: /* Table of texts corresponding to POSIX error codes */
156:
157: static const char *const pstring[] = {
158: "", /* Dummy for value 0 */
159: "internal error", /* REG_ASSERT */
160: "invalid repeat counts in {}", /* BADBR */
161: "pattern error", /* BADPAT */
162: "? * + invalid", /* BADRPT */
163: "unbalanced {}", /* EBRACE */
164: "unbalanced []", /* EBRACK */
165: "collation error - not relevant", /* ECOLLATE */
166: "bad class", /* ECTYPE */
167: "bad escape sequence", /* EESCAPE */
168: "empty expression", /* EMPTY */
169: "unbalanced ()", /* EPAREN */
170: "bad range inside []", /* ERANGE */
171: "expression too big", /* ESIZE */
172: "failed to get memory", /* ESPACE */
173: "bad back reference", /* ESUBREG */
174: "bad argument", /* INVARG */
175: "match failed" /* NOMATCH */
176: };
177:
178:
179:
180:
181: /*************************************************
182: * Translate error code to string *
183: *************************************************/
184:
185: PCREPOSIX_EXP_DEFN size_t PCRE_CALL_CONVENTION
186: regerror(int errcode, const regex_t *preg, char *errbuf, size_t errbuf_size)
187: {
188: const char *message, *addmessage;
189: size_t length, addlength;
190:
191: message = (errcode >= (int)(sizeof(pstring)/sizeof(char *)))?
192: "unknown error code" : pstring[errcode];
193: length = strlen(message) + 1;
194:
195: addmessage = " at offset ";
196: addlength = (preg != NULL && (int)preg->re_erroffset != -1)?
197: strlen(addmessage) + 6 : 0;
198:
199: if (errbuf_size > 0)
200: {
201: if (addlength > 0 && errbuf_size >= length + addlength)
202: sprintf(errbuf, "%s%s%-6d", message, addmessage, (int)preg->re_erroffset);
203: else
204: {
205: strncpy(errbuf, message, errbuf_size - 1);
206: errbuf[errbuf_size-1] = 0;
207: }
208: }
209:
210: return length + addlength;
211: }
212:
213:
214:
215:
216: /*************************************************
217: * Free store held by a regex *
218: *************************************************/
219:
220: PCREPOSIX_EXP_DEFN void PCRE_CALL_CONVENTION
221: regfree(regex_t *preg)
222: {
223: (pcre_free)(preg->re_pcre);
224: }
225:
226:
227:
228:
229: /*************************************************
230: * Compile a regular expression *
231: *************************************************/
232:
233: /*
234: Arguments:
235: preg points to a structure for recording the compiled expression
236: pattern the pattern to compile
237: cflags compilation flags
238:
239: Returns: 0 on success
240: various non-zero codes on failure
241: */
242:
243: PCREPOSIX_EXP_DEFN int PCRE_CALL_CONVENTION
244: regcomp(regex_t *preg, const char *pattern, int cflags)
245: {
246: const char *errorptr;
247: int erroffset;
248: int errorcode;
249: int options = 0;
250:
251: if ((cflags & REG_ICASE) != 0) options |= PCRE_CASELESS;
252: if ((cflags & REG_NEWLINE) != 0) options |= PCRE_MULTILINE;
253: if ((cflags & REG_DOTALL) != 0) options |= PCRE_DOTALL;
254: if ((cflags & REG_NOSUB) != 0) options |= PCRE_NO_AUTO_CAPTURE;
255: if ((cflags & REG_UTF8) != 0) options |= PCRE_UTF8;
256: if ((cflags & REG_UCP) != 0) options |= PCRE_UCP;
257: if ((cflags & REG_UNGREEDY) != 0) options |= PCRE_UNGREEDY;
258:
259: preg->re_pcre = pcre_compile2(pattern, options, &errorcode, &errorptr,
260: &erroffset, NULL);
261: preg->re_erroffset = erroffset;
262:
263: /* Safety: if the error code is too big for the translation vector (which
264: should not happen, but we all make mistakes), return REG_BADPAT. */
265:
266: if (preg->re_pcre == NULL)
267: {
268: return (errorcode < sizeof(eint)/sizeof(const int))?
269: eint[errorcode] : REG_BADPAT;
270: }
271:
272: preg->re_nsub = pcre_info((const pcre *)preg->re_pcre, NULL, NULL);
273: return 0;
274: }
275:
276:
277:
278:
279: /*************************************************
280: * Match a regular expression *
281: *************************************************/
282:
283: /* Unfortunately, PCRE requires 3 ints of working space for each captured
284: substring, so we have to get and release working store instead of just using
285: the POSIX structures as was done in earlier releases when PCRE needed only 2
286: ints. However, if the number of possible capturing brackets is small, use a
287: block of store on the stack, to reduce the use of malloc/free. The threshold is
288: in a macro that can be changed at configure time.
289:
290: If REG_NOSUB was specified at compile time, the PCRE_NO_AUTO_CAPTURE flag will
291: be set. When this is the case, the nmatch and pmatch arguments are ignored, and
292: the only result is yes/no/error. */
293:
294: PCREPOSIX_EXP_DEFN int PCRE_CALL_CONVENTION
295: regexec(const regex_t *preg, const char *string, size_t nmatch,
296: regmatch_t pmatch[], int eflags)
297: {
298: int rc, so, eo;
299: int options = 0;
300: int *ovector = NULL;
301: int small_ovector[POSIX_MALLOC_THRESHOLD * 3];
302: BOOL allocated_ovector = FALSE;
303: BOOL nosub =
304: (((const pcre *)preg->re_pcre)->options & PCRE_NO_AUTO_CAPTURE) != 0;
305:
306: if ((eflags & REG_NOTBOL) != 0) options |= PCRE_NOTBOL;
307: if ((eflags & REG_NOTEOL) != 0) options |= PCRE_NOTEOL;
308: if ((eflags & REG_NOTEMPTY) != 0) options |= PCRE_NOTEMPTY;
309:
310: ((regex_t *)preg)->re_erroffset = (size_t)(-1); /* Only has meaning after compile */
311:
312: /* When no string data is being returned, or no vector has been passed in which
313: to put it, ensure that nmatch is zero. Otherwise, ensure the vector for holding
314: the return data is large enough. */
315:
316: if (nosub || pmatch == NULL) nmatch = 0;
317:
318: else if (nmatch > 0)
319: {
320: if (nmatch <= POSIX_MALLOC_THRESHOLD)
321: {
322: ovector = &(small_ovector[0]);
323: }
324: else
325: {
326: if (nmatch > INT_MAX/(sizeof(int) * 3)) return REG_ESPACE;
327: ovector = (int *)malloc(sizeof(int) * nmatch * 3);
328: if (ovector == NULL) return REG_ESPACE;
329: allocated_ovector = TRUE;
330: }
331: }
332:
333: /* REG_STARTEND is a BSD extension, to allow for non-NUL-terminated strings.
334: The man page from OS X says "REG_STARTEND affects only the location of the
335: string, not how it is matched". That is why the "so" value is used to bump the
336: start location rather than being passed as a PCRE "starting offset". */
337:
338: if ((eflags & REG_STARTEND) != 0)
339: {
340: so = pmatch[0].rm_so;
341: eo = pmatch[0].rm_eo;
342: }
343: else
344: {
345: so = 0;
346: eo = (int)strlen(string);
347: }
348:
349: rc = pcre_exec((const pcre *)preg->re_pcre, NULL, string + so, (eo - so),
350: 0, options, ovector, (int)(nmatch * 3));
351:
352: if (rc == 0) rc = (int)nmatch; /* All captured slots were filled in */
353:
354: /* Successful match */
355:
356: if (rc >= 0)
357: {
358: size_t i;
359: if (!nosub)
360: {
361: for (i = 0; i < (size_t)rc; i++)
362: {
363: pmatch[i].rm_so = ovector[i*2];
364: pmatch[i].rm_eo = ovector[i*2+1];
365: }
366: if (allocated_ovector) free(ovector);
367: for (; i < nmatch; i++) pmatch[i].rm_so = pmatch[i].rm_eo = -1;
368: }
369: return 0;
370: }
371:
372: /* Unsuccessful match */
373:
374: if (allocated_ovector) free(ovector);
375: switch(rc)
376: {
377: /* ========================================================================== */
378: /* These cases are never obeyed. This is a fudge that causes a compile-time
379: error if the vector eint, which is indexed by compile-time error number, is
380: not the correct length. It seems to be the only way to do such a check at
381: compile time, as the sizeof() operator does not work in the C preprocessor.
382: As all the PCRE_ERROR_xxx values are negative, we can use 0 and 1. */
383:
384: case 0:
385: case (sizeof(eint)/sizeof(int) == ERRCOUNT):
386: return REG_ASSERT;
387: /* ========================================================================== */
388:
389: case PCRE_ERROR_NOMATCH: return REG_NOMATCH;
390: case PCRE_ERROR_NULL: return REG_INVARG;
391: case PCRE_ERROR_BADOPTION: return REG_INVARG;
392: case PCRE_ERROR_BADMAGIC: return REG_INVARG;
393: case PCRE_ERROR_UNKNOWN_NODE: return REG_ASSERT;
394: case PCRE_ERROR_NOMEMORY: return REG_ESPACE;
395: case PCRE_ERROR_MATCHLIMIT: return REG_ESPACE;
396: case PCRE_ERROR_BADUTF8: return REG_INVARG;
397: case PCRE_ERROR_BADUTF8_OFFSET: return REG_INVARG;
398: default: return REG_ASSERT;
399: }
400: }
401:
402: /* End of pcreposix.c */
FreeBSD-CVSweb <freebsd-cvsweb@FreeBSD.org>