1#define _BSD_SOURCE
2#include <glob.h>
3#include <fnmatch.h>
4#include <sys/stat.h>
5#include <dirent.h>
6#include <limits.h>
7#include <string.h>
8#include <stdlib.h>
9#include <errno.h>
10#include <stddef.h>
11#include <unistd.h>
12#ifdef __wasilibc_unmodified_upstream // WASI has no usernames
13#include <pwd.h>
14#endif
15
16struct match
17{
18 struct match *next;
19 char name[];
20};
21
22static int append(struct match **tail, const char *name, size_t len, int mark)
23{
24 struct match *new = malloc(sizeof(struct match) + len + 2);
25 if (!new) return -1;
26 (*tail)->next = new;
27 new->next = NULL;
28 memcpy(new->name, name, len+1);
29 if (mark && len && name[len-1]!='/') {
30 new->name[len] = '/';
31 new->name[len+1] = 0;
32 }
33 *tail = new;
34 return 0;
35}
36
37static int do_glob(char *buf, size_t pos, int type, char *pat, int flags, int (*errfunc)(const char *path, int err), struct match **tail)
38{
39 /* If GLOB_MARK is unused, we don't care about type. */
40 if (!type && !(flags & GLOB_MARK)) type = DT_REG;
41
42 /* Special-case the remaining pattern being all slashes, in
43 * which case we can use caller-passed type if it's a dir. */
44 if (*pat && type!=DT_DIR) type = 0;
45 while (pos+1 < PATH_MAX && *pat=='/') buf[pos++] = *pat++;
46
47 /* Consume maximal [escaped-]literal prefix of pattern, copying
48 * and un-escaping it to the running buffer as we go. */
49 ptrdiff_t i=0, j=0;
50 int in_bracket = 0, overflow = 0;
51 for (; pat[i]!='*' && pat[i]!='?' && (!in_bracket || pat[i]!=']'); i++) {
52 if (!pat[i]) {
53 if (overflow) return 0;
54 pat += i;
55 pos += j;
56 i = j = 0;
57 break;
58 } else if (pat[i] == '[') {
59 in_bracket = 1;
60 } else if (pat[i] == '\\' && !(flags & GLOB_NOESCAPE)) {
61 /* Backslashes inside a bracket are (at least by
62 * our interpretation) non-special, so if next
63 * char is ']' we have a complete expression. */
64 if (in_bracket && pat[i+1]==']') break;
65 /* Unpaired final backslash never matches. */
66 if (!pat[i+1]) return 0;
67 i++;
68 }
69 if (pat[i] == '/') {
70 if (overflow) return 0;
71 in_bracket = 0;
72 pat += i+1;
73 i = -1;
74 pos += j+1;
75 j = -1;
76 }
77 /* Only store a character if it fits in the buffer, but if
78 * a potential bracket expression is open, the overflow
79 * must be remembered and handled later only if the bracket
80 * is unterminated (and thereby a literal), so as not to
81 * disallow long bracket expressions with short matches. */
82 if (pos+(j+1) < PATH_MAX) {
83 buf[pos+j++] = pat[i];
84 } else if (in_bracket) {
85 overflow = 1;
86 } else {
87 return 0;
88 }
89 /* If we consume any new components, the caller-passed type
90 * or dummy type from above is no longer valid. */
91 type = 0;
92 }
93 buf[pos] = 0;
94 if (!*pat) {
95 /* If we consumed any components above, or if GLOB_MARK is
96 * requested and we don't yet know if the match is a dir,
97 * we must confirm the file exists and/or determine its type.
98 *
99 * If marking dirs, symlink type is inconclusive; we need the
100 * type for the symlink target, and therefore must try stat
101 * first unless type is known not to be a symlink. Otherwise,
102 * or if that fails, use lstat for determining existence to
103 * avoid false negatives in the case of broken symlinks. */
104 struct stat st;
105 if ((flags & GLOB_MARK) && (!type||type==DT_LNK) && !stat(buf, &st)) {
106 if (S_ISDIR(st.st_mode)) type = DT_DIR;
107 else type = DT_REG;
108 }
109 if (!type && lstat(buf, &st)) {
110 if (errno!=ENOENT && (errfunc(buf, errno) || (flags & GLOB_ERR)))
111 return GLOB_ABORTED;
112 return 0;
113 }
114 if (append(tail, buf, pos, (flags & GLOB_MARK) && type==DT_DIR))
115 return GLOB_NOSPACE;
116 return 0;
117 }
118 char *p2 = strchr(pat, '/'), saved_sep = '/';
119 /* Check if the '/' was escaped and, if so, remove the escape char
120 * so that it will not be unpaired when passed to fnmatch. */
121 if (p2 && !(flags & GLOB_NOESCAPE)) {
122 char *p;
123 for (p=p2; p>pat && p[-1]=='\\'; p--);
124 if ((p2-p)%2) {
125 p2--;
126 saved_sep = '\\';
127 }
128 }
129 DIR *dir = opendir(pos ? buf : ".");
130 if (!dir) {
131 if (errfunc(buf, errno) || (flags & GLOB_ERR))
132 return GLOB_ABORTED;
133 return 0;
134 }
135 int old_errno = errno;
136 struct dirent *de;
137 while (errno=0, de=readdir(dir)) {
138 /* Quickly skip non-directories when there's pattern left. */
139 if (p2 && de->d_type && de->d_type!=DT_DIR && de->d_type!=DT_LNK)
140 continue;
141
142 size_t l = strlen(de->d_name);
143 if (l >= PATH_MAX-pos) continue;
144
145 if (p2) *p2 = 0;
146
147 int fnm_flags= ((flags & GLOB_NOESCAPE) ? FNM_NOESCAPE : 0)
148 | ((!(flags & GLOB_PERIOD)) ? FNM_PERIOD : 0);
149
150 if (fnmatch(pat, de->d_name, fnm_flags))
151 continue;
152
153 /* With GLOB_PERIOD, don't allow matching . or .. unless
154 * fnmatch would match them with FNM_PERIOD rules in effect. */
155 if (p2 && (flags & GLOB_PERIOD) && de->d_name[0]=='.'
156 && (!de->d_name[1] || de->d_name[1]=='.' && !de->d_name[2])
157 && fnmatch(pat, de->d_name, fnm_flags | FNM_PERIOD))
158 continue;
159
160 memcpy(buf+pos, de->d_name, l+1);
161 if (p2) *p2 = saved_sep;
162 int r = do_glob(buf, pos+l, de->d_type, p2 ? p2 : "", flags, errfunc, tail);
163 if (r) {
164 closedir(dir);
165 return r;
166 }
167 }
168 int readerr = errno;
169 if (p2) *p2 = saved_sep;
170 closedir(dir);
171 if (readerr && (errfunc(buf, errno) || (flags & GLOB_ERR)))
172 return GLOB_ABORTED;
173 errno = old_errno;
174 return 0;
175}
176
177static int ignore_err(const char *path, int err)
178{
179 return 0;
180}
181
182static void freelist(struct match *head)
183{
184 struct match *match, *next;
185 for (match=head->next; match; match=next) {
186 next = match->next;
187 free(match);
188 }
189}
190
191static int sort(const void *a, const void *b)
192{
193 return strcmp(*(const char **)a, *(const char **)b);
194}
195
196#ifdef __wasilibc_unmodified_upstream // WASI has no usernames
197static int expand_tilde(char **pat, char *buf, size_t *pos)
198{
199 char *p = *pat + 1;
200 size_t i = 0;
201
202 char delim, *name_end = __strchrnul(p, '/');
203 if ((delim = *name_end)) *name_end++ = 0;
204 *pat = name_end;
205
206 char *home = *p ? NULL : getenv("HOME");
207 if (!home) {
208 struct passwd pw, *res;
209 switch (*p ? getpwnam_r(p, &pw, buf, PATH_MAX, &res)
210 : getpwuid_r(getuid(), &pw, buf, PATH_MAX, &res)) {
211 case ENOMEM:
212 return GLOB_NOSPACE;
213 case 0:
214 if (!res)
215 default:
216 return GLOB_NOMATCH;
217 }
218 home = pw.pw_dir;
219 }
220 while (i < PATH_MAX - 2 && *home)
221 buf[i++] = *home++;
222 if (*home)
223 return GLOB_NOMATCH;
224 if ((buf[i] = delim))
225 buf[++i] = 0;
226 *pos = i;
227 return 0;
228}
229#endif
230
231int glob(const char *restrict pat, int flags, int (*errfunc)(const char *path, int err), glob_t *restrict g)
232{
233 struct match head = { .next = NULL }, *tail = &head;
234 size_t cnt, i;
235 size_t offs = (flags & GLOB_DOOFFS) ? g->gl_offs : 0;
236 int error = 0;
237 char buf[PATH_MAX];
238
239 if (!errfunc) errfunc = ignore_err;
240
241 if (!(flags & GLOB_APPEND)) {
242 g->gl_offs = offs;
243 g->gl_pathc = 0;
244 g->gl_pathv = NULL;
245 }
246
247 if (*pat) {
248 char *p = strdup(pat);
249 if (!p) return GLOB_NOSPACE;
250 buf[0] = 0;
251 size_t pos = 0;
252 char *s = p;
253#ifdef __wasilibc_unmodified_upstream // WASI has no usernames
254 if ((flags & (GLOB_TILDE | GLOB_TILDE_CHECK)) && *p == '~')
255 error = expand_tilde(&s, buf, &pos);
256#endif
257 if (!error)
258 error = do_glob(buf, pos, 0, s, flags, errfunc, &tail);
259 free(p);
260 }
261
262 if (error == GLOB_NOSPACE) {
263 freelist(&head);
264 return error;
265 }
266
267 for (cnt=0, tail=head.next; tail; tail=tail->next, cnt++);
268 if (!cnt) {
269 if (flags & GLOB_NOCHECK) {
270 tail = &head;
271 if (append(&tail, pat, strlen(pat), 0))
272 return GLOB_NOSPACE;
273 cnt++;
274 } else if (!error)
275 return GLOB_NOMATCH;
276 }
277
278 if (flags & GLOB_APPEND) {
279 char **pathv = realloc(g->gl_pathv, (offs + g->gl_pathc + cnt + 1) * sizeof(char *));
280 if (!pathv) {
281 freelist(&head);
282 return GLOB_NOSPACE;
283 }
284 g->gl_pathv = pathv;
285 offs += g->gl_pathc;
286 } else {
287 g->gl_pathv = malloc((offs + cnt + 1) * sizeof(char *));
288 if (!g->gl_pathv) {
289 freelist(&head);
290 return GLOB_NOSPACE;
291 }
292 for (i=0; i<offs; i++)
293 g->gl_pathv[i] = NULL;
294 }
295 for (i=0, tail=head.next; i<cnt; tail=tail->next, i++)
296 g->gl_pathv[offs + i] = tail->name;
297 g->gl_pathv[offs + i] = NULL;
298 g->gl_pathc += cnt;
299
300 if (!(flags & GLOB_NOSORT))
301 qsort(g->gl_pathv+offs, cnt, sizeof(char *), sort);
302
303 return error;
304}
305
306void globfree(glob_t *g)
307{
308 size_t i;
309 for (i=0; i<g->gl_pathc; i++)
310 free(g->gl_pathv[g->gl_offs + i] - offsetof(struct match, name));
311 free(g->gl_pathv);
312 g->gl_pathc = 0;
313 g->gl_pathv = NULL;
314}
315
316weak_alias(glob, glob64);
317weak_alias(globfree, globfree64);