Add zlib
[git/pclouds.git] / dir.c
blob8d8faf5d788b402d5fd4f5d724954fd2ee178fdb
1 /*
2 * This handles recursive filename detection with exclude
3 * files, index knowledge etc..
5 * Copyright (C) Linus Torvalds, 2005-2006
6 * Junio Hamano, 2005-2006
7 */
8 #include "cache.h"
9 #include "dir.h"
10 #include "refs.h"
12 struct path_simplify {
13 int len;
14 const char *path;
17 static int read_directory_recursive(struct dir_struct *dir,
18 const char *path, const char *base, int baselen,
19 int check_only, const struct path_simplify *simplify);
21 int common_prefix(const char **pathspec)
23 const char *path, *slash, *next;
24 int prefix;
26 if (!pathspec)
27 return 0;
29 path = *pathspec;
30 slash = strrchr(path, '/');
31 if (!slash)
32 return 0;
34 prefix = slash - path + 1;
35 while ((next = *++pathspec) != NULL) {
36 int len = strlen(next);
37 if (len >= prefix && !memcmp(path, next, prefix))
38 continue;
39 len = prefix - 1;
40 for (;;) {
41 if (!len)
42 return 0;
43 if (next[--len] != '/')
44 continue;
45 if (memcmp(path, next, len+1))
46 continue;
47 prefix = len + 1;
48 break;
51 return prefix;
55 * Does 'match' matches the given name?
56 * A match is found if
58 * (1) the 'match' string is leading directory of 'name', or
59 * (2) the 'match' string is a wildcard and matches 'name', or
60 * (3) the 'match' string is exactly the same as 'name'.
62 * and the return value tells which case it was.
64 * It returns 0 when there is no match.
66 static int match_one(const char *match, const char *name, int namelen)
68 int matchlen;
70 /* If the match was just the prefix, we matched */
71 matchlen = strlen(match);
72 if (!matchlen)
73 return MATCHED_RECURSIVELY;
76 * If we don't match the matchstring exactly,
77 * we need to match by fnmatch
79 if (strncmp(match, name, matchlen))
80 return !fnmatch(match, name, 0) ? MATCHED_FNMATCH : 0;
82 if (!name[matchlen])
83 return MATCHED_EXACTLY;
84 if (match[matchlen-1] == '/' || name[matchlen] == '/')
85 return MATCHED_RECURSIVELY;
86 return 0;
90 * Given a name and a list of pathspecs, see if the name matches
91 * any of the pathspecs. The caller is also interested in seeing
92 * all pathspec matches some names it calls this function with
93 * (otherwise the user could have mistyped the unmatched pathspec),
94 * and a mark is left in seen[] array for pathspec element that
95 * actually matched anything.
97 int match_pathspec(const char **pathspec, const char *name, int namelen, int prefix, char *seen)
99 int retval;
100 const char *match;
102 name += prefix;
103 namelen -= prefix;
105 for (retval = 0; (match = *pathspec++) != NULL; seen++) {
106 int how;
107 if (retval && *seen == MATCHED_EXACTLY)
108 continue;
109 match += prefix;
110 how = match_one(match, name, namelen);
111 if (how) {
112 if (retval < how)
113 retval = how;
114 if (*seen < how)
115 *seen = how;
118 return retval;
121 void add_exclude(const char *string, const char *base,
122 int baselen, struct exclude_list *which)
124 struct exclude *x = xmalloc(sizeof (*x));
126 x->pattern = string;
127 x->base = base;
128 x->baselen = baselen;
129 if (which->nr == which->alloc) {
130 which->alloc = alloc_nr(which->alloc);
131 which->excludes = xrealloc(which->excludes,
132 which->alloc * sizeof(x));
134 which->excludes[which->nr++] = x;
137 static int add_excludes_from_file_1(const char *fname,
138 const char *base,
139 int baselen,
140 struct exclude_list *which)
142 struct stat st;
143 int fd, i;
144 size_t size;
145 char *buf, *entry;
147 fd = open(fname, O_RDONLY);
148 if (fd < 0 || fstat(fd, &st) < 0)
149 goto err;
150 size = xsize_t(st.st_size);
151 if (size == 0) {
152 close(fd);
153 return 0;
155 buf = xmalloc(size+1);
156 if (read_in_full(fd, buf, size) != size)
157 goto err;
158 close(fd);
160 buf[size++] = '\n';
161 entry = buf;
162 for (i = 0; i < size; i++) {
163 if (buf[i] == '\n') {
164 if (entry != buf + i && entry[0] != '#') {
165 buf[i - (i && buf[i-1] == '\r')] = 0;
166 add_exclude(entry, base, baselen, which);
168 entry = buf + i + 1;
171 return 0;
173 err:
174 if (0 <= fd)
175 close(fd);
176 return -1;
179 void add_excludes_from_file(struct dir_struct *dir, const char *fname)
181 if (add_excludes_from_file_1(fname, "", 0,
182 &dir->exclude_list[EXC_FILE]) < 0)
183 die("cannot use %s as an exclude file", fname);
186 int push_exclude_per_directory(struct dir_struct *dir, const char *base, int baselen)
188 char exclude_file[PATH_MAX];
189 struct exclude_list *el = &dir->exclude_list[EXC_DIRS];
190 int current_nr = el->nr;
192 if (dir->exclude_per_dir) {
193 memcpy(exclude_file, base, baselen);
194 strcpy(exclude_file + baselen, dir->exclude_per_dir);
195 add_excludes_from_file_1(exclude_file, base, baselen, el);
197 return current_nr;
200 void pop_exclude_per_directory(struct dir_struct *dir, int stk)
202 struct exclude_list *el = &dir->exclude_list[EXC_DIRS];
204 while (stk < el->nr)
205 free(el->excludes[--el->nr]);
208 /* Scan the list and let the last match determines the fate.
209 * Return 1 for exclude, 0 for include and -1 for undecided.
211 static int excluded_1(const char *pathname,
212 int pathlen,
213 struct exclude_list *el)
215 int i;
217 if (el->nr) {
218 for (i = el->nr - 1; 0 <= i; i--) {
219 struct exclude *x = el->excludes[i];
220 const char *exclude = x->pattern;
221 int to_exclude = 1;
223 if (*exclude == '!') {
224 to_exclude = 0;
225 exclude++;
228 if (!strchr(exclude, '/')) {
229 /* match basename */
230 const char *basename = strrchr(pathname, '/');
231 basename = (basename) ? basename+1 : pathname;
232 if (fnmatch(exclude, basename, 0) == 0)
233 return to_exclude;
235 else {
236 /* match with FNM_PATHNAME:
237 * exclude has base (baselen long) implicitly
238 * in front of it.
240 int baselen = x->baselen;
241 if (*exclude == '/')
242 exclude++;
244 if (pathlen < baselen ||
245 (baselen && pathname[baselen-1] != '/') ||
246 strncmp(pathname, x->base, baselen))
247 continue;
249 if (fnmatch(exclude, pathname+baselen,
250 FNM_PATHNAME) == 0)
251 return to_exclude;
255 return -1; /* undecided */
258 int excluded(struct dir_struct *dir, const char *pathname)
260 int pathlen = strlen(pathname);
261 int st;
263 for (st = EXC_CMDL; st <= EXC_FILE; st++) {
264 switch (excluded_1(pathname, pathlen, &dir->exclude_list[st])) {
265 case 0:
266 return 0;
267 case 1:
268 return 1;
271 return 0;
274 static struct dir_entry *dir_entry_new(const char *pathname, int len) {
275 struct dir_entry *ent;
277 ent = xmalloc(sizeof(*ent) + len + 1);
278 ent->len = len;
279 memcpy(ent->name, pathname, len);
280 ent->name[len] = 0;
281 return ent;
284 struct dir_entry *dir_add_name(struct dir_struct *dir, const char *pathname, int len)
286 if (cache_name_pos(pathname, len) >= 0)
287 return NULL;
289 ALLOC_GROW(dir->entries, dir->nr+1, dir->alloc);
290 return dir->entries[dir->nr++] = dir_entry_new(pathname, len);
293 struct dir_entry *dir_add_ignored(struct dir_struct *dir, const char *pathname, int len)
295 if (cache_name_pos(pathname, len) >= 0)
296 return NULL;
298 ALLOC_GROW(dir->ignored, dir->ignored_nr+1, dir->ignored_alloc);
299 return dir->ignored[dir->ignored_nr++] = dir_entry_new(pathname, len);
302 enum exist_status {
303 index_nonexistent = 0,
304 index_directory,
305 index_gitdir,
309 * The index sorts alphabetically by entry name, which
310 * means that a gitlink sorts as '\0' at the end, while
311 * a directory (which is defined not as an entry, but as
312 * the files it contains) will sort with the '/' at the
313 * end.
315 static enum exist_status directory_exists_in_index(const char *dirname, int len)
317 int pos = cache_name_pos(dirname, len);
318 if (pos < 0)
319 pos = -pos-1;
320 while (pos < active_nr) {
321 struct cache_entry *ce = active_cache[pos++];
322 unsigned char endchar;
324 if (strncmp(ce->name, dirname, len))
325 break;
326 endchar = ce->name[len];
327 if (endchar > '/')
328 break;
329 if (endchar == '/')
330 return index_directory;
331 if (!endchar && S_ISGITLINK(ntohl(ce->ce_mode)))
332 return index_gitdir;
334 return index_nonexistent;
338 * When we find a directory when traversing the filesystem, we
339 * have three distinct cases:
341 * - ignore it
342 * - see it as a directory
343 * - recurse into it
345 * and which one we choose depends on a combination of existing
346 * git index contents and the flags passed into the directory
347 * traversal routine.
349 * Case 1: If we *already* have entries in the index under that
350 * directory name, we always recurse into the directory to see
351 * all the files.
353 * Case 2: If we *already* have that directory name as a gitlink,
354 * we always continue to see it as a gitlink, regardless of whether
355 * there is an actual git directory there or not (it might not
356 * be checked out as a subproject!)
358 * Case 3: if we didn't have it in the index previously, we
359 * have a few sub-cases:
361 * (a) if "show_other_directories" is true, we show it as
362 * just a directory, unless "hide_empty_directories" is
363 * also true and the directory is empty, in which case
364 * we just ignore it entirely.
365 * (b) if it looks like a git directory, and we don't have
366 * 'no_gitlinks' set we treat it as a gitlink, and show it
367 * as a directory.
368 * (c) otherwise, we recurse into it.
370 enum directory_treatment {
371 show_directory,
372 ignore_directory,
373 recurse_into_directory,
376 static enum directory_treatment treat_directory(struct dir_struct *dir,
377 const char *dirname, int len,
378 const struct path_simplify *simplify)
380 /* The "len-1" is to strip the final '/' */
381 switch (directory_exists_in_index(dirname, len-1)) {
382 case index_directory:
383 return recurse_into_directory;
385 case index_gitdir:
386 if (dir->show_other_directories)
387 return ignore_directory;
388 return show_directory;
390 case index_nonexistent:
391 if (dir->show_other_directories)
392 break;
393 if (!dir->no_gitlinks) {
394 unsigned char sha1[20];
395 if (resolve_gitlink_ref(dirname, "HEAD", sha1) == 0)
396 return show_directory;
398 return recurse_into_directory;
401 /* This is the "show_other_directories" case */
402 if (!dir->hide_empty_directories)
403 return show_directory;
404 if (!read_directory_recursive(dir, dirname, dirname, len, 1, simplify))
405 return ignore_directory;
406 return show_directory;
410 * This is an inexact early pruning of any recursive directory
411 * reading - if the path cannot possibly be in the pathspec,
412 * return true, and we'll skip it early.
414 static int simplify_away(const char *path, int pathlen, const struct path_simplify *simplify)
416 if (simplify) {
417 for (;;) {
418 const char *match = simplify->path;
419 int len = simplify->len;
421 if (!match)
422 break;
423 if (len > pathlen)
424 len = pathlen;
425 if (!memcmp(path, match, len))
426 return 0;
427 simplify++;
429 return 1;
431 return 0;
434 static int in_pathspec(const char *path, int len, const struct path_simplify *simplify)
436 if (simplify) {
437 for (; simplify->path; simplify++) {
438 if (len == simplify->len
439 && !memcmp(path, simplify->path, len))
440 return 1;
443 return 0;
447 * Read a directory tree. We currently ignore anything but
448 * directories, regular files and symlinks. That's because git
449 * doesn't handle them at all yet. Maybe that will change some
450 * day.
452 * Also, we ignore the name ".git" (even if it is not a directory).
453 * That likely will not change.
455 static int read_directory_recursive(struct dir_struct *dir, const char *path, const char *base, int baselen, int check_only, const struct path_simplify *simplify)
457 DIR *fdir = opendir(path);
458 int contents = 0;
460 if (fdir) {
461 int exclude_stk;
462 struct dirent *de;
463 char fullname[PATH_MAX + 1];
464 memcpy(fullname, base, baselen);
466 exclude_stk = push_exclude_per_directory(dir, base, baselen);
468 while ((de = readdir(fdir)) != NULL) {
469 int len;
470 int exclude;
472 if ((de->d_name[0] == '.') &&
473 (de->d_name[1] == 0 ||
474 !strcmp(de->d_name + 1, ".") ||
475 !strcmp(de->d_name + 1, "git")))
476 continue;
477 len = strlen(de->d_name);
478 /* Ignore overly long pathnames! */
479 if (len + baselen + 8 > sizeof(fullname))
480 continue;
481 memcpy(fullname + baselen, de->d_name, len+1);
482 if (simplify_away(fullname, baselen + len, simplify))
483 continue;
485 exclude = excluded(dir, fullname);
486 if (exclude && dir->collect_ignored
487 && in_pathspec(fullname, baselen + len, simplify))
488 dir_add_ignored(dir, fullname, baselen + len);
489 if (exclude != dir->show_ignored) {
490 if (!dir->show_ignored || DTYPE(de) != DT_DIR) {
491 continue;
495 switch (DTYPE(de)) {
496 struct stat st;
497 default:
498 continue;
499 case DT_UNKNOWN:
500 if (lstat(fullname, &st))
501 continue;
502 if (S_ISREG(st.st_mode) || S_ISLNK(st.st_mode))
503 break;
504 if (!S_ISDIR(st.st_mode))
505 continue;
506 /* fallthrough */
507 case DT_DIR:
508 memcpy(fullname + baselen + len, "/", 2);
509 len++;
510 switch (treat_directory(dir, fullname, baselen + len, simplify)) {
511 case show_directory:
512 if (exclude != dir->show_ignored)
513 continue;
514 break;
515 case recurse_into_directory:
516 contents += read_directory_recursive(dir,
517 fullname, fullname, baselen + len, 0, simplify);
518 continue;
519 case ignore_directory:
520 continue;
522 break;
523 case DT_REG:
524 case DT_LNK:
525 break;
527 contents++;
528 if (check_only)
529 goto exit_early;
530 else
531 dir_add_name(dir, fullname, baselen + len);
533 exit_early:
534 closedir(fdir);
536 pop_exclude_per_directory(dir, exclude_stk);
539 return contents;
542 static int cmp_name(const void *p1, const void *p2)
544 const struct dir_entry *e1 = *(const struct dir_entry **)p1;
545 const struct dir_entry *e2 = *(const struct dir_entry **)p2;
547 return cache_name_compare(e1->name, e1->len,
548 e2->name, e2->len);
552 * Return the length of the "simple" part of a path match limiter.
554 static int simple_length(const char *match)
556 const char special[256] = {
557 [0] = 1, ['?'] = 1,
558 ['\\'] = 1, ['*'] = 1,
559 ['['] = 1
561 int len = -1;
563 for (;;) {
564 unsigned char c = *match++;
565 len++;
566 if (special[c])
567 return len;
571 static struct path_simplify *create_simplify(const char **pathspec)
573 int nr, alloc = 0;
574 struct path_simplify *simplify = NULL;
576 if (!pathspec)
577 return NULL;
579 for (nr = 0 ; ; nr++) {
580 const char *match;
581 if (nr >= alloc) {
582 alloc = alloc_nr(alloc);
583 simplify = xrealloc(simplify, alloc * sizeof(*simplify));
585 match = *pathspec++;
586 if (!match)
587 break;
588 simplify[nr].path = match;
589 simplify[nr].len = simple_length(match);
591 simplify[nr].path = NULL;
592 simplify[nr].len = 0;
593 return simplify;
596 static void free_simplify(struct path_simplify *simplify)
598 if (simplify)
599 free(simplify);
602 int read_directory(struct dir_struct *dir, const char *path, const char *base, int baselen, const char **pathspec)
604 struct path_simplify *simplify = create_simplify(pathspec);
607 * Make sure to do the per-directory exclude for all the
608 * directories leading up to our base.
610 if (baselen) {
611 if (dir->exclude_per_dir) {
612 char *p, *pp = xmalloc(baselen+1);
613 memcpy(pp, base, baselen+1);
614 p = pp;
615 while (1) {
616 char save = *p;
617 *p = 0;
618 push_exclude_per_directory(dir, pp, p-pp);
619 *p++ = save;
620 if (!save)
621 break;
622 p = strchr(p, '/');
623 if (p)
624 p++;
625 else
626 p = pp + baselen;
628 free(pp);
632 read_directory_recursive(dir, path, base, baselen, 0, simplify);
633 free_simplify(simplify);
634 qsort(dir->entries, dir->nr, sizeof(struct dir_entry *), cmp_name);
635 qsort(dir->ignored, dir->ignored_nr, sizeof(struct dir_entry *), cmp_name);
636 return dir->nr;
640 file_exists(const char *f)
642 struct stat sb;
643 return stat(f, &sb) == 0;