1: /* $OpenBSD: fnmatch.c,v 1.15 2011/02/10 21:31:59 stsp Exp $ */
2:
3: /* Copyright (c) 2011, VMware, Inc.
4: * All rights reserved.
5: *
6: * Redistribution and use in source and binary forms, with or without
7: * modification, are permitted provided that the following conditions are met:
8: * * Redistributions of source code must retain the above copyright
9: * notice, this list of conditions and the following disclaimer.
10: * * Redistributions in binary form must reproduce the above copyright
11: * notice, this list of conditions and the following disclaimer in the
12: * documentation and/or other materials provided with the distribution.
13: * * Neither the name of the VMware, Inc. nor the names of its contributors
14: * may be used to endorse or promote products derived from this software
15: * without specific prior written permission.
16: *
17: * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
18: * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
19: * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
20: * ARE DISCLAIMED. IN NO EVENT SHALL VMWARE, INC. OR CONTRIBUTORS BE LIABLE FOR
21: * ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
22: * (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
23: * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND
24: * ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
25: * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF
26: * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
27: */
28:
29: /* Authored by William A. Rowe Jr. <wrowe; apache.org, vmware.com>, April 2011
30: *
31: * Derived from The Open Group Base Specifications Issue 7, IEEE Std 1003.1-2008
32: * as described in;
33: * http://pubs.opengroup.org/onlinepubs/9699919799/functions/fnmatch.html
34: *
35: * Filename pattern matches defined in section 2.13, "Pattern Matching Notation"
36: * from chapter 2. "Shell Command Language"
37: * http://pubs.opengroup.org/onlinepubs/9699919799/utilities/V3_chap02.html#tag_18_13
38: * where; 1. A bracket expression starting with an unquoted <circumflex> '^'
39: * character CONTINUES to specify a non-matching list; 2. an explicit <period> '.'
40: * in a bracket expression matching list, e.g. "[.abc]" does NOT match a leading
41: * <period> in a filename; 3. a <left-square-bracket> '[' which does not introduce
42: * a valid bracket expression is treated as an ordinary character; 4. a differing
43: * number of consecutive slashes within pattern and string will NOT match;
44: * 5. a trailing '\' in FNM_ESCAPE mode is treated as an ordinary '\' character.
45: *
46: * Bracket expansion defined in section 9.3.5, "RE Bracket Expression",
47: * from chapter 9, "Regular Expressions"
48: * http://pubs.opengroup.org/onlinepubs/9699919799/basedefs/V1_chap09.html#tag_09_03_05
49: * with no support for collating symbols, equivalence class expressions or
50: * character class expressions. A partial range expression with a leading
51: * hyphen following a valid range expression will match only the ordinary
52: * <hyphen> and the ending character (e.g. "[a-m-z]" will match characters
53: * 'a' through 'm', a <hyphen> '-', or a 'z').
54: *
55: * Supports BSD extensions FNM_LEADING_DIR to match pattern to the end of one
56: * path segment of string, and FNM_CASEFOLD to ignore alpha case.
57: *
58: * NOTE: Only POSIX/C single byte locales are correctly supported at this time.
59: * Notably, non-POSIX locales with FNM_CASEFOLD produce undefined results,
60: * particularly in ranges of mixed case (e.g. "[A-z]") or spanning alpha and
61: * nonalpha characters within a range.
62: *
63: * XXX comments below indicate porting required for multi-byte character sets
64: * and non-POSIX locale collation orders; requires mbr* APIs to track shift
65: * state of pattern and string (rewinding pattern and string repeatedly).
66: *
67: * Certain parts of the code assume 0x00-0x3F are unique with any MBCS (e.g.
68: * UTF-8, SHIFT-JIS, etc). Any implementation allowing '\' as an alternate
69: * path delimiter must be aware that 0x5C is NOT unique within SHIFT-JIS.
70: */
71:
72: #include <config.h>
73:
74: #include <sys/types.h>
75:
76: #include <stdio.h>
77: #include <ctype.h>
78: #ifdef HAVE_STRING_H
79: # include <string.h>
80: #endif /* HAVE_STRING_H */
81: #ifdef HAVE_STRINGS_H
82: # include <strings.h>
83: #endif /* HAVE_STRINGS_H */
84: #include <limits.h>
85:
86: #include "missing.h"
87: #include "compat/charclass.h"
88: #include "compat/fnmatch.h"
89:
90: #define RANGE_MATCH 1
91: #define RANGE_NOMATCH 0
92: #define RANGE_ERROR (-1)
93:
94: static int
95: classmatch(const char *pattern, char test, int foldcase, const char **ep)
96: {
97: const char * const mismatch = pattern;
98: const char *colon;
99: struct cclass *cc;
100: int rval = RANGE_NOMATCH;
101: size_t len;
102:
103: if (pattern[0] != '[' || pattern[1] != ':') {
104: *ep = mismatch;
105: return RANGE_ERROR;
106: }
107: pattern += 2;
108:
109: if ((colon = strchr(pattern, ':')) == NULL || colon[1] != ']') {
110: *ep = mismatch;
111: return RANGE_ERROR;
112: }
113: *ep = colon + 2;
114: len = (size_t)(colon - pattern);
115:
116: if (foldcase && strncmp(pattern, "upper:]", 7) == 0)
117: pattern = "lower:]";
118: for (cc = cclasses; cc->name != NULL; cc++) {
119: if (!strncmp(pattern, cc->name, len) && cc->name[len] == '\0') {
120: if (cc->isctype((unsigned char)test))
121: rval = RANGE_MATCH;
122: break;
123: }
124: }
125: if (cc->name == NULL) {
126: /* invalid character class, treat as normal text */
127: *ep = mismatch;
128: rval = RANGE_ERROR;
129: }
130: return rval;
131: }
132:
133: /* Most MBCS/collation/case issues handled here. Wildcard '*' is not handled.
134: * EOS '\0' and the FNM_PATHNAME '/' delimiters are not advanced over,
135: * however the "\/" sequence is advanced to '/'.
136: *
137: * Both pattern and string are **char to support pointer increment of arbitrary
138: * multibyte characters for the given locale, in a later iteration of this code
139: */
140: static int fnmatch_ch(const char **pattern, const char **string, int flags)
141: {
142: const char * const mismatch = *pattern;
143: const int nocase = !!(flags & FNM_CASEFOLD);
144: const int escape = !(flags & FNM_NOESCAPE);
145: const int slash = !!(flags & FNM_PATHNAME);
146: int result = FNM_NOMATCH;
147: const char *startch;
148: int negate;
149:
150: if (**pattern == '[')
151: {
152: ++*pattern;
153:
154: /* Handle negation, either leading ! or ^ operators (never both) */
155: negate = ((**pattern == '!') || (**pattern == '^'));
156: if (negate)
157: ++*pattern;
158:
159: /* ']' is an ordinary character at the start of the range pattern */
160: if (**pattern == ']')
161: goto leadingclosebrace;
162:
163: while (**pattern)
164: {
165: if (**pattern == ']') {
166: ++*pattern;
167: /* XXX: Fix for MBCS character width */
168: ++*string;
169: return (result ^ negate);
170: }
171:
172: if (escape && (**pattern == '\\')) {
173: ++*pattern;
174:
175: /* Patterns must be terminated with ']', not EOS */
176: if (!**pattern)
177: break;
178: }
179:
180: /* Patterns must be terminated with ']' not '/' */
181: if (slash && (**pattern == '/'))
182: break;
183:
184: /* Match character classes. */
185: if (classmatch(*pattern, **string, nocase, pattern)
186: == RANGE_MATCH) {
187: result = 0;
188: continue;
189: }
190:
191: leadingclosebrace:
192: /* Look at only well-formed range patterns;
193: * "x-]" is not allowed unless escaped ("x-\]")
194: * XXX: Fix for locale/MBCS character width
195: */
196: if (((*pattern)[1] == '-') && ((*pattern)[2] != ']'))
197: {
198: startch = *pattern;
199: *pattern += (escape && ((*pattern)[2] == '\\')) ? 3 : 2;
200:
201: /* NOT a properly balanced [expr] pattern, EOS terminated
202: * or ranges containing a slash in FNM_PATHNAME mode pattern
203: * fall out to to the rewind and test '[' literal code path
204: */
205: if (!**pattern || (slash && (**pattern == '/')))
206: break;
207:
208: /* XXX: handle locale/MBCS comparison, advance by MBCS char width */
209: if ((**string >= *startch) && (**string <= **pattern))
210: result = 0;
211: else if (nocase && (isupper((unsigned char)**string) ||
212: isupper((unsigned char)*startch) ||
213: isupper((unsigned char)**pattern))
214: && (tolower((unsigned char)**string) >= tolower((unsigned char)*startch))
215: && (tolower((unsigned char)**string) <= tolower((unsigned char)**pattern)))
216: result = 0;
217:
218: ++*pattern;
219: continue;
220: }
221:
222: /* XXX: handle locale/MBCS comparison, advance by MBCS char width */
223: if ((**string == **pattern))
224: result = 0;
225: else if (nocase && (isupper((unsigned char)**string) ||
226: isupper((unsigned char)**pattern))
227: && (tolower((unsigned char)**string) == tolower((unsigned char)**pattern)))
228: result = 0;
229:
230: ++*pattern;
231: }
232:
233: /* NOT a properly balanced [expr] pattern; Rewind
234: * and reset result to test '[' literal
235: */
236: *pattern = mismatch;
237: result = FNM_NOMATCH;
238: }
239: else if (**pattern == '?') {
240: /* Optimize '?' match before unescaping **pattern */
241: if (!**string || (slash && (**string == '/')))
242: return FNM_NOMATCH;
243: result = 0;
244: goto fnmatch_ch_success;
245: }
246: else if (escape && (**pattern == '\\') && (*pattern)[1]) {
247: ++*pattern;
248: }
249:
250: /* XXX: handle locale/MBCS comparison, advance by the MBCS char width */
251: if (**string == **pattern)
252: result = 0;
253: else if (nocase && (isupper((unsigned char)**string) || isupper((unsigned char)**pattern))
254: && (tolower((unsigned char)**string) == tolower((unsigned char)**pattern)))
255: result = 0;
256:
257: /* Refuse to advance over trailing slash or nulls
258: */
259: if (!**string || !**pattern || (slash && ((**string == '/') || (**pattern == '/'))))
260: return result;
261:
262: fnmatch_ch_success:
263: ++*pattern;
264: ++*string;
265: return result;
266: }
267:
268: int rpl_fnmatch(const char *pattern, const char *string, int flags)
269: {
270: static const char dummystring[2] = {' ', 0};
271: const int escape = !(flags & FNM_NOESCAPE);
272: const int slash = !!(flags & FNM_PATHNAME);
273: const int leading_dir = !!(flags & FNM_LEADING_DIR);
274: const char *strendseg;
275: const char *dummyptr;
276: const char *matchptr;
277: int wild;
278: /* For '*' wild processing only; surpress 'used before initialization'
279: * warnings with dummy initialization values;
280: */
281: const char *strstartseg = NULL;
282: const char *mismatch = NULL;
283: int matchlen = 0;
284:
285: if (strlen(pattern) > PATH_MAX || strlen(string) > PATH_MAX)
286: return FNM_NOMATCH;
287:
288: if (*pattern == '*')
289: goto firstsegment;
290:
291: while (*pattern && *string)
292: {
293: /* Pre-decode "\/" which has no special significance, and
294: * match balanced slashes, starting a new segment pattern
295: */
296: if (slash && escape && (*pattern == '\\') && (pattern[1] == '/'))
297: ++pattern;
298: if (slash && (*pattern == '/') && (*string == '/')) {
299: ++pattern;
300: ++string;
301: }
302:
303: firstsegment:
304: /* At the beginning of each segment, validate leading period behavior.
305: */
306: if ((flags & FNM_PERIOD) && (*string == '.'))
307: {
308: if (*pattern == '.')
309: ++pattern;
310: else if (escape && (*pattern == '\\') && (pattern[1] == '.'))
311: pattern += 2;
312: else
313: return FNM_NOMATCH;
314: ++string;
315: }
316:
317: /* Determine the end of string segment
318: *
319: * Presumes '/' character is unique, not composite in any MBCS encoding
320: */
321: if (slash) {
322: strendseg = strchr(string, '/');
323: if (!strendseg)
324: strendseg = strchr(string, '\0');
325: }
326: else {
327: strendseg = strchr(string, '\0');
328: }
329:
330: /* Allow pattern '*' to be consumed even with no remaining string to match
331: */
332: while (*pattern)
333: {
334: if ((string > strendseg)
335: || ((string == strendseg) && (*pattern != '*')))
336: break;
337:
338: if (slash && ((*pattern == '/')
339: || (escape && (*pattern == '\\')
340: && (pattern[1] == '/'))))
341: break;
342:
343: /* Reduce groups of '*' and '?' to n '?' matches
344: * followed by one '*' test for simplicity
345: */
346: for (wild = 0; ((*pattern == '*') || (*pattern == '?')); ++pattern)
347: {
348: if (*pattern == '*') {
349: wild = 1;
350: }
351: else if (string < strendseg) { /* && (*pattern == '?') */
352: /* XXX: Advance 1 char for MBCS locale */
353: ++string;
354: }
355: else { /* (string >= strendseg) && (*pattern == '?') */
356: return FNM_NOMATCH;
357: }
358: }
359:
360: if (wild)
361: {
362: strstartseg = string;
363: mismatch = pattern;
364:
365: /* Count fixed (non '*') char matches remaining in pattern
366: * excluding '/' (or "\/") and '*'
367: */
368: for (matchptr = pattern, matchlen = 0; 1; ++matchlen)
369: {
370: if ((*matchptr == '\0')
371: || (slash && ((*matchptr == '/')
372: || (escape && (*matchptr == '\\')
373: && (matchptr[1] == '/')))))
374: {
375: /* Compare precisely this many trailing string chars,
376: * the resulting match needs no wildcard loop
377: */
378: /* XXX: Adjust for MBCS */
379: if (string + matchlen > strendseg)
380: return FNM_NOMATCH;
381:
382: string = strendseg - matchlen;
383: wild = 0;
384: break;
385: }
386:
387: if (*matchptr == '*')
388: {
389: /* Ensure at least this many trailing string chars remain
390: * for the first comparison
391: */
392: /* XXX: Adjust for MBCS */
393: if (string + matchlen > strendseg)
394: return FNM_NOMATCH;
395:
396: /* Begin first wild comparison at the current position */
397: break;
398: }
399:
400: /* Skip forward in pattern by a single character match
401: * Use a dummy fnmatch_ch() test to count one "[range]" escape
402: */
403: /* XXX: Adjust for MBCS */
404: if (escape && (*matchptr == '\\') && matchptr[1]) {
405: matchptr += 2;
406: }
407: else if (*matchptr == '[') {
408: dummyptr = dummystring;
409: fnmatch_ch(&matchptr, &dummyptr, flags);
410: }
411: else {
412: ++matchptr;
413: }
414: }
415: }
416:
417: /* Incrementally match string against the pattern
418: */
419: while (*pattern && (string < strendseg))
420: {
421: /* Success; begin a new wild pattern search
422: */
423: if (*pattern == '*')
424: break;
425:
426: if (slash && ((*string == '/')
427: || (*pattern == '/')
428: || (escape && (*pattern == '\\')
429: && (pattern[1] == '/'))))
430: break;
431:
432: /* Compare ch's (the pattern is advanced over "\/" to the '/',
433: * but slashes will mismatch, and are not consumed)
434: */
435: if (!fnmatch_ch(&pattern, &string, flags))
436: continue;
437:
438: /* Failed to match, loop against next char offset of string segment
439: * until not enough string chars remain to match the fixed pattern
440: */
441: if (wild) {
442: /* XXX: Advance 1 char for MBCS locale */
443: string = ++strstartseg;
444: if (string + matchlen > strendseg)
445: return FNM_NOMATCH;
446:
447: pattern = mismatch;
448: continue;
449: }
450: else
451: return FNM_NOMATCH;
452: }
453: }
454:
455: if (*string && !((slash || leading_dir) && (*string == '/')))
456: return FNM_NOMATCH;
457:
458: if (*pattern && !(slash && ((*pattern == '/')
459: || (escape && (*pattern == '\\')
460: && (pattern[1] == '/')))))
461: return FNM_NOMATCH;
462:
463: if (leading_dir && !*pattern && *string == '/')
464: return 0;
465: }
466:
467: /* Where both pattern and string are at EOS, declare success
468: */
469: if (!*string && !*pattern)
470: return 0;
471:
472: /* pattern didn't match to the end of string */
473: return FNM_NOMATCH;
474: }
FreeBSD-CVSweb <freebsd-cvsweb@FreeBSD.org>