Merge branch 'sg/t5310-empty-input-fix'
[git/git-svn.git] / wildmatch.c
blobd074c1be1046f1aba11d21eb5b66a738590bc712
1 /*
2 ** Do shell-style pattern matching for ?, \, [], and * characters.
3 ** It is 8bit clean.
4 **
5 ** Written by Rich $alz, mirror!rs, Wed Nov 26 19:03:17 EST 1986.
6 ** Rich $alz is now <rsalz@bbn.com>.
7 **
8 ** Modified by Wayne Davison to special-case '/' matching, to make '**'
9 ** work differently than '*', and to fix the character-class code.
12 #include "cache.h"
13 #include "wildmatch.h"
15 typedef unsigned char uchar;
17 /* What character marks an inverted character class? */
18 #define NEGATE_CLASS '!'
19 #define NEGATE_CLASS2 '^'
21 #define CC_EQ(class, len, litmatch) ((len) == sizeof (litmatch)-1 \
22 && *(class) == *(litmatch) \
23 && strncmp((char*)class, litmatch, len) == 0)
25 #if defined STDC_HEADERS || !defined isascii
26 # define ISASCII(c) 1
27 #else
28 # define ISASCII(c) isascii(c)
29 #endif
31 #ifdef isblank
32 # define ISBLANK(c) (ISASCII(c) && isblank(c))
33 #else
34 # define ISBLANK(c) ((c) == ' ' || (c) == '\t')
35 #endif
37 #ifdef isgraph
38 # define ISGRAPH(c) (ISASCII(c) && isgraph(c))
39 #else
40 # define ISGRAPH(c) (ISASCII(c) && isprint(c) && !isspace(c))
41 #endif
43 #define ISPRINT(c) (ISASCII(c) && isprint(c))
44 #define ISDIGIT(c) (ISASCII(c) && isdigit(c))
45 #define ISALNUM(c) (ISASCII(c) && isalnum(c))
46 #define ISALPHA(c) (ISASCII(c) && isalpha(c))
47 #define ISCNTRL(c) (ISASCII(c) && iscntrl(c))
48 #define ISLOWER(c) (ISASCII(c) && islower(c))
49 #define ISPUNCT(c) (ISASCII(c) && ispunct(c))
50 #define ISSPACE(c) (ISASCII(c) && isspace(c))
51 #define ISUPPER(c) (ISASCII(c) && isupper(c))
52 #define ISXDIGIT(c) (ISASCII(c) && isxdigit(c))
54 /* Match pattern "p" against "text" */
55 static int dowild(const uchar *p, const uchar *text, unsigned int flags)
57 uchar p_ch;
58 const uchar *pattern = p;
60 for ( ; (p_ch = *p) != '\0'; text++, p++) {
61 int matched, match_slash, negated;
62 uchar t_ch, prev_ch;
63 if ((t_ch = *text) == '\0' && p_ch != '*')
64 return WM_ABORT_ALL;
65 if ((flags & WM_CASEFOLD) && ISUPPER(t_ch))
66 t_ch = tolower(t_ch);
67 if ((flags & WM_CASEFOLD) && ISUPPER(p_ch))
68 p_ch = tolower(p_ch);
69 switch (p_ch) {
70 case '\\':
71 /* Literal match with following character. Note that the test
72 * in "default" handles the p[1] == '\0' failure case. */
73 p_ch = *++p;
74 /* FALLTHROUGH */
75 default:
76 if (t_ch != p_ch)
77 return WM_NOMATCH;
78 continue;
79 case '?':
80 /* Match anything but '/'. */
81 if ((flags & WM_PATHNAME) && t_ch == '/')
82 return WM_NOMATCH;
83 continue;
84 case '*':
85 if (*++p == '*') {
86 const uchar *prev_p = p - 2;
87 while (*++p == '*') {}
88 if (!(flags & WM_PATHNAME))
89 /* without WM_PATHNAME, '*' == '**' */
90 match_slash = 1;
91 else if ((prev_p < pattern || *prev_p == '/') &&
92 (*p == '\0' || *p == '/' ||
93 (p[0] == '\\' && p[1] == '/'))) {
95 * Assuming we already match 'foo/' and are at
96 * <star star slash>, just assume it matches
97 * nothing and go ahead match the rest of the
98 * pattern with the remaining string. This
99 * helps make foo/<*><*>/bar (<> because
100 * otherwise it breaks C comment syntax) match
101 * both foo/bar and foo/a/bar.
103 if (p[0] == '/' &&
104 dowild(p + 1, text, flags) == WM_MATCH)
105 return WM_MATCH;
106 match_slash = 1;
107 } else
108 return WM_ABORT_MALFORMED;
109 } else
110 /* without WM_PATHNAME, '*' == '**' */
111 match_slash = flags & WM_PATHNAME ? 0 : 1;
112 if (*p == '\0') {
113 /* Trailing "**" matches everything. Trailing "*" matches
114 * only if there are no more slash characters. */
115 if (!match_slash) {
116 if (strchr((char*)text, '/') != NULL)
117 return WM_NOMATCH;
119 return WM_MATCH;
120 } else if (!match_slash && *p == '/') {
122 * _one_ asterisk followed by a slash
123 * with WM_PATHNAME matches the next
124 * directory
126 const char *slash = strchr((char*)text, '/');
127 if (!slash)
128 return WM_NOMATCH;
129 text = (const uchar*)slash;
130 /* the slash is consumed by the top-level for loop */
131 break;
133 while (1) {
134 if (t_ch == '\0')
135 break;
137 * Try to advance faster when an asterisk is
138 * followed by a literal. We know in this case
139 * that the string before the literal
140 * must belong to "*".
141 * If match_slash is false, do not look past
142 * the first slash as it cannot belong to '*'.
144 if (!is_glob_special(*p)) {
145 p_ch = *p;
146 if ((flags & WM_CASEFOLD) && ISUPPER(p_ch))
147 p_ch = tolower(p_ch);
148 while ((t_ch = *text) != '\0' &&
149 (match_slash || t_ch != '/')) {
150 if ((flags & WM_CASEFOLD) && ISUPPER(t_ch))
151 t_ch = tolower(t_ch);
152 if (t_ch == p_ch)
153 break;
154 text++;
156 if (t_ch != p_ch)
157 return WM_NOMATCH;
159 if ((matched = dowild(p, text, flags)) != WM_NOMATCH) {
160 if (!match_slash || matched != WM_ABORT_TO_STARSTAR)
161 return matched;
162 } else if (!match_slash && t_ch == '/')
163 return WM_ABORT_TO_STARSTAR;
164 t_ch = *++text;
166 return WM_ABORT_ALL;
167 case '[':
168 p_ch = *++p;
169 #ifdef NEGATE_CLASS2
170 if (p_ch == NEGATE_CLASS2)
171 p_ch = NEGATE_CLASS;
172 #endif
173 /* Assign literal 1/0 because of "matched" comparison. */
174 negated = p_ch == NEGATE_CLASS ? 1 : 0;
175 if (negated) {
176 /* Inverted character class. */
177 p_ch = *++p;
179 prev_ch = 0;
180 matched = 0;
181 do {
182 if (!p_ch)
183 return WM_ABORT_ALL;
184 if (p_ch == '\\') {
185 p_ch = *++p;
186 if (!p_ch)
187 return WM_ABORT_ALL;
188 if (t_ch == p_ch)
189 matched = 1;
190 } else if (p_ch == '-' && prev_ch && p[1] && p[1] != ']') {
191 p_ch = *++p;
192 if (p_ch == '\\') {
193 p_ch = *++p;
194 if (!p_ch)
195 return WM_ABORT_ALL;
197 if (t_ch <= p_ch && t_ch >= prev_ch)
198 matched = 1;
199 else if ((flags & WM_CASEFOLD) && ISLOWER(t_ch)) {
200 uchar t_ch_upper = toupper(t_ch);
201 if (t_ch_upper <= p_ch && t_ch_upper >= prev_ch)
202 matched = 1;
204 p_ch = 0; /* This makes "prev_ch" get set to 0. */
205 } else if (p_ch == '[' && p[1] == ':') {
206 const uchar *s;
207 int i;
208 for (s = p += 2; (p_ch = *p) && p_ch != ']'; p++) {} /*SHARED ITERATOR*/
209 if (!p_ch)
210 return WM_ABORT_ALL;
211 i = p - s - 1;
212 if (i < 0 || p[-1] != ':') {
213 /* Didn't find ":]", so treat like a normal set. */
214 p = s - 2;
215 p_ch = '[';
216 if (t_ch == p_ch)
217 matched = 1;
218 continue;
220 if (CC_EQ(s,i, "alnum")) {
221 if (ISALNUM(t_ch))
222 matched = 1;
223 } else if (CC_EQ(s,i, "alpha")) {
224 if (ISALPHA(t_ch))
225 matched = 1;
226 } else if (CC_EQ(s,i, "blank")) {
227 if (ISBLANK(t_ch))
228 matched = 1;
229 } else if (CC_EQ(s,i, "cntrl")) {
230 if (ISCNTRL(t_ch))
231 matched = 1;
232 } else if (CC_EQ(s,i, "digit")) {
233 if (ISDIGIT(t_ch))
234 matched = 1;
235 } else if (CC_EQ(s,i, "graph")) {
236 if (ISGRAPH(t_ch))
237 matched = 1;
238 } else if (CC_EQ(s,i, "lower")) {
239 if (ISLOWER(t_ch))
240 matched = 1;
241 } else if (CC_EQ(s,i, "print")) {
242 if (ISPRINT(t_ch))
243 matched = 1;
244 } else if (CC_EQ(s,i, "punct")) {
245 if (ISPUNCT(t_ch))
246 matched = 1;
247 } else if (CC_EQ(s,i, "space")) {
248 if (ISSPACE(t_ch))
249 matched = 1;
250 } else if (CC_EQ(s,i, "upper")) {
251 if (ISUPPER(t_ch))
252 matched = 1;
253 else if ((flags & WM_CASEFOLD) && ISLOWER(t_ch))
254 matched = 1;
255 } else if (CC_EQ(s,i, "xdigit")) {
256 if (ISXDIGIT(t_ch))
257 matched = 1;
258 } else /* malformed [:class:] string */
259 return WM_ABORT_ALL;
260 p_ch = 0; /* This makes "prev_ch" get set to 0. */
261 } else if (t_ch == p_ch)
262 matched = 1;
263 } while (prev_ch = p_ch, (p_ch = *++p) != ']');
264 if (matched == negated ||
265 ((flags & WM_PATHNAME) && t_ch == '/'))
266 return WM_NOMATCH;
267 continue;
271 return *text ? WM_NOMATCH : WM_MATCH;
274 /* Match the "pattern" against the "text" string. */
275 int wildmatch(const char *pattern, const char *text, unsigned int flags)
277 return dowild((const uchar*)pattern, (const uchar*)text, flags);