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
12 struct path_simplify {
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;
30 slash = strrchr(path, '/');
34 prefix = slash - path + 1;
35 while ((next = *++pathspec) != NULL) {
36 int len = strlen(next);
37 if (len >= prefix && !memcmp(path, next, prefix))
43 if (next[--len] != '/')
45 if (memcmp(path, next, len+1))
55 * Does 'match' matches the given name?
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)
70 /* If the match was just the prefix, we matched */
71 matchlen = strlen(match);
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;
83 return MATCHED_EXACTLY;
84 if (match[matchlen-1] == '/' || name[matchlen] == '/')
85 return MATCHED_RECURSIVELY;
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)
105 for (retval = 0; (match = *pathspec++) != NULL; seen++) {
107 if (retval && *seen == MATCHED_EXACTLY)
110 how = match_one(match, name, namelen);
121 void add_exclude(const char *string, const char *base,
122 int baselen, struct exclude_list *which)
124 struct exclude *x = xmalloc(sizeof (*x));
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,
140 struct exclude_list *which)
147 fd = open(fname, O_RDONLY);
148 if (fd < 0 || fstat(fd, &st) < 0)
150 size = xsize_t(st.st_size);
155 buf = xmalloc(size+1);
156 if (read_in_full(fd, buf, size) != size)
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);
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);
200 void pop_exclude_per_directory(struct dir_struct *dir, int stk)
202 struct exclude_list *el = &dir->exclude_list[EXC_DIRS];
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,
213 struct exclude_list *el)
218 for (i = el->nr - 1; 0 <= i; i--) {
219 struct exclude *x = el->excludes[i];
220 const char *exclude = x->pattern;
223 if (*exclude == '!') {
228 if (!strchr(exclude, '/')) {
230 const char *basename = strrchr(pathname, '/');
231 basename = (basename) ? basename+1 : pathname;
232 if (fnmatch(exclude, basename, 0) == 0)
236 /* match with FNM_PATHNAME:
237 * exclude has base (baselen long) implicitly
240 int baselen = x->baselen;
244 if (pathlen < baselen ||
245 (baselen && pathname[baselen-1] != '/') ||
246 strncmp(pathname, x->base, baselen))
249 if (fnmatch(exclude, pathname+baselen,
255 return -1; /* undecided */
258 int excluded(struct dir_struct *dir, const char *pathname)
260 int pathlen = strlen(pathname);
263 for (st = EXC_CMDL; st <= EXC_FILE; st++) {
264 switch (excluded_1(pathname, pathlen, &dir->exclude_list[st])) {
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->ignored = ent->ignored_dir = 0;
280 memcpy(ent->name, pathname, len);
285 struct dir_entry *dir_add_name(struct dir_struct *dir, const char *pathname, int len)
287 if (cache_name_pos(pathname, len) >= 0)
290 ALLOC_GROW(dir->entries, dir->nr, dir->alloc);
291 return dir->entries[dir->nr++] = dir_entry_new(pathname, len);
295 index_nonexistent = 0,
301 * The index sorts alphabetically by entry name, which
302 * means that a gitlink sorts as '\0' at the end, while
303 * a directory (which is defined not as an entry, but as
304 * the files it contains) will sort with the '/' at the
307 static enum exist_status directory_exists_in_index(const char *dirname, int len)
309 int pos = cache_name_pos(dirname, len);
312 while (pos < active_nr) {
313 struct cache_entry *ce = active_cache[pos++];
314 unsigned char endchar;
316 if (strncmp(ce->name, dirname, len))
318 endchar = ce->name[len];
322 return index_directory;
323 if (!endchar && S_ISGITLINK(ntohl(ce->ce_mode)))
326 return index_nonexistent;
330 * When we find a directory when traversing the filesystem, we
331 * have three distinct cases:
334 * - see it as a directory
337 * and which one we choose depends on a combination of existing
338 * git index contents and the flags passed into the directory
341 * Case 1: If we *already* have entries in the index under that
342 * directory name, we always recurse into the directory to see
345 * Case 2: If we *already* have that directory name as a gitlink,
346 * we always continue to see it as a gitlink, regardless of whether
347 * there is an actual git directory there or not (it might not
348 * be checked out as a subproject!)
350 * Case 3: if we didn't have it in the index previously, we
351 * have a few sub-cases:
353 * (a) if "show_other_directories" is true, we show it as
354 * just a directory, unless "hide_empty_directories" is
355 * also true and the directory is empty, in which case
356 * we just ignore it entirely.
357 * (b) if it looks like a git directory, and we don't have
358 * 'no_gitlinks' set we treat it as a gitlink, and show it
360 * (c) otherwise, we recurse into it.
362 enum directory_treatment {
365 recurse_into_directory,
368 static enum directory_treatment treat_directory(struct dir_struct *dir,
369 const char *dirname, int len,
370 const struct path_simplify *simplify)
372 /* The "len-1" is to strip the final '/' */
373 switch (directory_exists_in_index(dirname, len-1)) {
374 case index_directory:
375 return recurse_into_directory;
378 if (dir->show_other_directories)
379 return ignore_directory;
380 return show_directory;
382 case index_nonexistent:
383 if (dir->show_other_directories)
385 if (!dir->no_gitlinks) {
386 unsigned char sha1[20];
387 if (resolve_gitlink_ref(dirname, "HEAD", sha1) == 0)
388 return show_directory;
390 return recurse_into_directory;
393 /* This is the "show_other_directories" case */
394 if (!dir->hide_empty_directories)
395 return show_directory;
396 if (!read_directory_recursive(dir, dirname, dirname, len, 1, simplify))
397 return ignore_directory;
398 return show_directory;
402 * This is an inexact early pruning of any recursive directory
403 * reading - if the path cannot possibly be in the pathspec,
404 * return true, and we'll skip it early.
406 static int simplify_away(const char *path, int pathlen, const struct path_simplify *simplify)
410 const char *match = simplify->path;
411 int len = simplify->len;
417 if (!memcmp(path, match, len))
427 * Read a directory tree. We currently ignore anything but
428 * directories, regular files and symlinks. That's because git
429 * doesn't handle them at all yet. Maybe that will change some
432 * Also, we ignore the name ".git" (even if it is not a directory).
433 * That likely will not change.
435 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)
437 DIR *fdir = opendir(path);
443 char fullname[PATH_MAX + 1];
444 memcpy(fullname, base, baselen);
446 exclude_stk = push_exclude_per_directory(dir, base, baselen);
448 while ((de = readdir(fdir)) != NULL) {
452 if ((de->d_name[0] == '.') &&
453 (de->d_name[1] == 0 ||
454 !strcmp(de->d_name + 1, ".") ||
455 !strcmp(de->d_name + 1, "git")))
457 len = strlen(de->d_name);
458 /* Ignore overly long pathnames! */
459 if (len + baselen + 8 > sizeof(fullname))
461 memcpy(fullname + baselen, de->d_name, len+1);
462 if (simplify_away(fullname, baselen + len, simplify))
465 exclude = excluded(dir, fullname);
466 if (exclude != dir->show_ignored) {
467 if (!dir->show_ignored || DTYPE(de) != DT_DIR) {
477 if (lstat(fullname, &st))
479 if (S_ISREG(st.st_mode) || S_ISLNK(st.st_mode))
481 if (!S_ISDIR(st.st_mode))
485 memcpy(fullname + baselen + len, "/", 2);
487 switch (treat_directory(dir, fullname, baselen + len, simplify)) {
489 if (exclude != dir->show_ignored)
492 case recurse_into_directory:
493 contents += read_directory_recursive(dir,
494 fullname, fullname, baselen + len, 0, simplify);
496 case ignore_directory:
508 dir_add_name(dir, fullname, baselen + len);
513 pop_exclude_per_directory(dir, exclude_stk);
519 static int cmp_name(const void *p1, const void *p2)
521 const struct dir_entry *e1 = *(const struct dir_entry **)p1;
522 const struct dir_entry *e2 = *(const struct dir_entry **)p2;
524 return cache_name_compare(e1->name, e1->len,
529 * Return the length of the "simple" part of a path match limiter.
531 static int simple_length(const char *match)
533 const char special[256] = {
535 ['\\'] = 1, ['*'] = 1,
541 unsigned char c = *match++;
548 static struct path_simplify *create_simplify(const char **pathspec)
551 struct path_simplify *simplify = NULL;
556 for (nr = 0 ; ; nr++) {
559 alloc = alloc_nr(alloc);
560 simplify = xrealloc(simplify, alloc * sizeof(*simplify));
565 simplify[nr].path = match;
566 simplify[nr].len = simple_length(match);
568 simplify[nr].path = NULL;
569 simplify[nr].len = 0;
573 static void free_simplify(struct path_simplify *simplify)
579 int read_directory(struct dir_struct *dir, const char *path, const char *base, int baselen, const char **pathspec)
581 struct path_simplify *simplify = create_simplify(pathspec);
584 * Make sure to do the per-directory exclude for all the
585 * directories leading up to our base.
588 if (dir->exclude_per_dir) {
589 char *p, *pp = xmalloc(baselen+1);
590 memcpy(pp, base, baselen+1);
595 push_exclude_per_directory(dir, pp, p-pp);
609 read_directory_recursive(dir, path, base, baselen, 0, simplify);
610 free_simplify(simplify);
611 qsort(dir->entries, dir->nr, sizeof(struct dir_entry *), cmp_name);
616 file_exists(const char *f)
619 return stat(f, &sb) == 0;