dir: warn about trailing spaces in exclude patterns
[git/git.git] / dir.c
CommitLineData
453ec4bd
LT
1/*
2 * This handles recursive filename detection with exclude
3 * files, index knowledge etc..
4 *
95a68344
AS
5 * See Documentation/technical/api-directory-listing.txt
6 *
453ec4bd
LT
7 * Copyright (C) Linus Torvalds, 2005-2006
8 * Junio Hamano, 2005-2006
9 */
453ec4bd
LT
10#include "cache.h"
11#include "dir.h"
09595258 12#include "refs.h"
237ec6e4 13#include "wildmatch.h"
64acde94 14#include "pathspec.h"
453ec4bd 15
9fc42d60
LT
16struct path_simplify {
17 int len;
18 const char *path;
19};
20
defd7c7b
KB
21/*
22 * Tells read_directory_recursive how a file or directory should be treated.
23 * Values are ordered by significance, e.g. if a directory contains both
24 * excluded and untracked files, it is listed as untracked because
25 * path_untracked > path_excluded.
26 */
27enum path_treatment {
28 path_none = 0,
29 path_recurse,
30 path_excluded,
31 path_untracked
32};
33
34static enum path_treatment read_directory_recursive(struct dir_struct *dir,
35 const char *path, int len,
09595258 36 int check_only, const struct path_simplify *simplify);
caa6b782 37static int get_dtype(struct dirent *de, const char *path, int len);
09595258 38
8cf2a84e
JJ
39/* helper string functions with support for the ignore_case flag */
40int strcmp_icase(const char *a, const char *b)
41{
42 return ignore_case ? strcasecmp(a, b) : strcmp(a, b);
43}
44
45int strncmp_icase(const char *a, const char *b, size_t count)
46{
47 return ignore_case ? strncasecmp(a, b, count) : strncmp(a, b, count);
48}
49
50int fnmatch_icase(const char *pattern, const char *string, int flags)
51{
52 return fnmatch(pattern, string, flags | (ignore_case ? FNM_CASEFOLD : 0));
53}
54
bd30c2e4
NTND
55inline int git_fnmatch(const struct pathspec_item *item,
56 const char *pattern, const char *string,
57 int prefix)
5d74762d 58{
5d74762d 59 if (prefix > 0) {
93d93537 60 if (ps_strncmp(item, pattern, string, prefix))
5d74762d
NTND
61 return FNM_NOMATCH;
62 pattern += prefix;
63 string += prefix;
64 }
bd30c2e4 65 if (item->flags & PATHSPEC_ONESTAR) {
8c6abbcd
NTND
66 int pattern_len = strlen(++pattern);
67 int string_len = strlen(string);
68 return string_len < pattern_len ||
93d93537
NTND
69 ps_strcmp(item, pattern,
70 string + string_len - pattern_len);
8c6abbcd 71 }
bd30c2e4 72 if (item->magic & PATHSPEC_GLOB)
93d93537
NTND
73 return wildmatch(pattern, string,
74 WM_PATHNAME |
75 (item->magic & PATHSPEC_ICASE ? WM_CASEFOLD : 0),
76 NULL);
bd30c2e4
NTND
77 else
78 /* wildmatch has not learned no FNM_PATHNAME mode yet */
93d93537
NTND
79 return fnmatch(pattern, string,
80 item->magic & PATHSPEC_ICASE ? FNM_CASEFOLD : 0);
5d74762d
NTND
81}
82
0b6e56df
JH
83static int fnmatch_icase_mem(const char *pattern, int patternlen,
84 const char *string, int stringlen,
85 int flags)
86{
87 int match_status;
88 struct strbuf pat_buf = STRBUF_INIT;
89 struct strbuf str_buf = STRBUF_INIT;
90 const char *use_pat = pattern;
91 const char *use_str = string;
92
93 if (pattern[patternlen]) {
94 strbuf_add(&pat_buf, pattern, patternlen);
95 use_pat = pat_buf.buf;
96 }
97 if (string[stringlen]) {
98 strbuf_add(&str_buf, string, stringlen);
99 use_str = str_buf.buf;
100 }
101
f30366b2
JH
102 if (ignore_case)
103 flags |= WM_CASEFOLD;
104 match_status = wildmatch(use_pat, use_str, flags, NULL);
0b6e56df
JH
105
106 strbuf_release(&pat_buf);
107 strbuf_release(&str_buf);
108
109 return match_status;
110}
111
827f4d6c 112static size_t common_prefix_len(const struct pathspec *pathspec)
3c6a370b 113{
827f4d6c 114 int n;
4a085b16 115 size_t max = 0;
3c6a370b 116
93d93537
NTND
117 /*
118 * ":(icase)path" is treated as a pathspec full of
119 * wildcard. In other words, only prefix is considered common
120 * prefix. If the pathspec is abc/foo abc/bar, running in
121 * subdir xyz, the common prefix is still xyz, not xuz/abc as
122 * in non-:(icase).
123 */
5c6933d2
NTND
124 GUARD_PATHSPEC(pathspec,
125 PATHSPEC_FROMTOP |
126 PATHSPEC_MAXDEPTH |
bd30c2e4 127 PATHSPEC_LITERAL |
93d93537
NTND
128 PATHSPEC_GLOB |
129 PATHSPEC_ICASE);
4a085b16 130
827f4d6c 131 for (n = 0; n < pathspec->nr; n++) {
93d93537
NTND
132 size_t i = 0, len = 0, item_len;
133 if (pathspec->items[n].magic & PATHSPEC_ICASE)
134 item_len = pathspec->items[n].prefix;
135 else
136 item_len = pathspec->items[n].nowildcard_len;
137 while (i < item_len && (n == 0 || i < max)) {
827f4d6c
NTND
138 char c = pathspec->items[n].match[i];
139 if (c != pathspec->items[0].match[i])
4a085b16
JH
140 break;
141 if (c == '/')
142 len = i + 1;
827f4d6c 143 i++;
4a085b16 144 }
827f4d6c 145 if (n == 0 || len < max) {
4a085b16
JH
146 max = len;
147 if (!max)
148 break;
149 }
3c6a370b 150 }
4a085b16 151 return max;
3c6a370b
LT
152}
153
f950eb95
CB
154/*
155 * Returns a copy of the longest leading path common among all
156 * pathspecs.
157 */
827f4d6c 158char *common_prefix(const struct pathspec *pathspec)
f950eb95
CB
159{
160 unsigned long len = common_prefix_len(pathspec);
161
827f4d6c 162 return len ? xmemdupz(pathspec->items[0].match, len) : NULL;
f950eb95
CB
163}
164
7327d3d1 165int fill_directory(struct dir_struct *dir, const struct pathspec *pathspec)
1d8842d9 166{
4a085b16 167 size_t len;
1d8842d9
LT
168
169 /*
170 * Calculate common prefix for the pathspec, and
171 * use that to optimize the directory walk
172 */
4a085b16 173 len = common_prefix_len(pathspec);
1d8842d9
LT
174
175 /* Read the directory and prune it */
b3920bbd 176 read_directory(dir, pathspec->nr ? pathspec->_raw[0] : "", len, pathspec);
dba2e203 177 return len;
1d8842d9
LT
178}
179
bc96cc87
NTND
180int within_depth(const char *name, int namelen,
181 int depth, int max_depth)
182{
183 const char *cp = name, *cpe = name + namelen;
184
185 while (cp < cpe) {
186 if (*cp++ != '/')
187 continue;
188 depth++;
189 if (depth > max_depth)
190 return 0;
191 }
192 return 1;
193}
194
61cf2820
NTND
195/*
196 * Does 'match' match the given name?
197 * A match is found if
198 *
199 * (1) the 'match' string is leading directory of 'name', or
200 * (2) the 'match' string is a wildcard and matches 'name', or
201 * (3) the 'match' string is exactly the same as 'name'.
202 *
203 * and the return value tells which case it was.
204 *
205 * It returns 0 when there is no match.
206 */
207static int match_pathspec_item(const struct pathspec_item *item, int prefix,
208 const char *name, int namelen)
209{
210 /* name/namelen has prefix cut off by caller */
211 const char *match = item->match + prefix;
212 int matchlen = item->len - prefix;
213
93d93537
NTND
214 /*
215 * The normal call pattern is:
216 * 1. prefix = common_prefix_len(ps);
217 * 2. prune something, or fill_directory
218 * 3. match_pathspec_depth()
219 *
220 * 'prefix' at #1 may be shorter than the command's prefix and
221 * it's ok for #2 to match extra files. Those extras will be
222 * trimmed at #3.
223 *
224 * Suppose the pathspec is 'foo' and '../bar' running from
225 * subdir 'xyz'. The common prefix at #1 will be empty, thanks
226 * to "../". We may have xyz/foo _and_ XYZ/foo after #2. The
227 * user does not want XYZ/foo, only the "foo" part should be
228 * case-insensitive. We need to filter out XYZ/foo here. In
229 * other words, we do not trust the caller on comparing the
230 * prefix part when :(icase) is involved. We do exact
231 * comparison ourselves.
232 *
233 * Normally the caller (common_prefix_len() in fact) does
234 * _exact_ matching on name[-prefix+1..-1] and we do not need
235 * to check that part. Be defensive and check it anyway, in
236 * case common_prefix_len is changed, or a new caller is
237 * introduced that does not use common_prefix_len.
238 *
239 * If the penalty turns out too high when prefix is really
240 * long, maybe change it to
241 * strncmp(match, name, item->prefix - prefix)
242 */
243 if (item->prefix && (item->magic & PATHSPEC_ICASE) &&
244 strncmp(item->match, name - prefix, item->prefix))
245 return 0;
246
61cf2820
NTND
247 /* If the match was just the prefix, we matched */
248 if (!*match)
249 return MATCHED_RECURSIVELY;
250
93d93537 251 if (matchlen <= namelen && !ps_strncmp(item, match, name, matchlen)) {
61cf2820
NTND
252 if (matchlen == namelen)
253 return MATCHED_EXACTLY;
254
255 if (match[matchlen-1] == '/' || name[matchlen] == '/')
256 return MATCHED_RECURSIVELY;
257 }
258
5d74762d 259 if (item->nowildcard_len < item->len &&
bd30c2e4 260 !git_fnmatch(item, match, name,
8c6abbcd 261 item->nowildcard_len - prefix))
61cf2820
NTND
262 return MATCHED_FNMATCH;
263
264 return 0;
265}
266
267/*
52ed1894
AS
268 * Given a name and a list of pathspecs, returns the nature of the
269 * closest (i.e. most specific) match of the name to any of the
270 * pathspecs.
271 *
272 * The caller typically calls this multiple times with the same
273 * pathspec and seen[] array but with different name/namelen
274 * (e.g. entries from the index) and is interested in seeing if and
275 * how each pathspec matches all the names it calls this function
276 * with. A mark is left in the seen[] array for each pathspec element
277 * indicating the closest type of match that element achieved, so if
278 * seen[n] remains zero after multiple invocations, that means the nth
279 * pathspec did not match any names, which could indicate that the
280 * user mistyped the nth pathspec.
61cf2820
NTND
281 */
282int match_pathspec_depth(const struct pathspec *ps,
283 const char *name, int namelen,
284 int prefix, char *seen)
285{
286 int i, retval = 0;
287
5c6933d2
NTND
288 GUARD_PATHSPEC(ps,
289 PATHSPEC_FROMTOP |
290 PATHSPEC_MAXDEPTH |
bd30c2e4 291 PATHSPEC_LITERAL |
93d93537
NTND
292 PATHSPEC_GLOB |
293 PATHSPEC_ICASE);
8f4f8f45 294
61cf2820 295 if (!ps->nr) {
6330a171
NTND
296 if (!ps->recursive ||
297 !(ps->magic & PATHSPEC_MAXDEPTH) ||
298 ps->max_depth == -1)
61cf2820
NTND
299 return MATCHED_RECURSIVELY;
300
301 if (within_depth(name, namelen, 0, ps->max_depth))
302 return MATCHED_EXACTLY;
303 else
304 return 0;
305 }
306
307 name += prefix;
308 namelen -= prefix;
309
310 for (i = ps->nr - 1; i >= 0; i--) {
311 int how;
312 if (seen && seen[i] == MATCHED_EXACTLY)
313 continue;
314 how = match_pathspec_item(ps->items+i, prefix, name, namelen);
6330a171
NTND
315 if (ps->recursive &&
316 (ps->magic & PATHSPEC_MAXDEPTH) &&
317 ps->max_depth != -1 &&
61cf2820
NTND
318 how && how != MATCHED_FNMATCH) {
319 int len = ps->items[i].len;
320 if (name[len] == '/')
321 len++;
322 if (within_depth(name+len, namelen-len, 0, ps->max_depth))
323 how = MATCHED_EXACTLY;
324 else
325 how = 0;
326 }
327 if (how) {
328 if (retval < how)
329 retval = how;
330 if (seen && seen[i] < how)
331 seen[i] = how;
332 }
333 }
334 return retval;
335}
336
fcd631ed
NTND
337/*
338 * Return the length of the "simple" part of a path match limiter.
339 */
87323bda 340int simple_length(const char *match)
fcd631ed
NTND
341{
342 int len = -1;
343
344 for (;;) {
345 unsigned char c = *match++;
346 len++;
347 if (c == '\0' || is_glob_special(c))
348 return len;
349 }
350}
351
87323bda 352int no_wildcard(const char *string)
68492fc7 353{
fcd631ed 354 return string[simple_length(string)] == '\0';
68492fc7
LK
355}
356
82dce998
NTND
357void parse_exclude_pattern(const char **pattern,
358 int *patternlen,
359 int *flags,
360 int *nowildcardlen)
84460eec
NTND
361{
362 const char *p = *pattern;
363 size_t i, len;
364
365 *flags = 0;
366 if (*p == '!') {
367 *flags |= EXC_FLAG_NEGATIVE;
368 p++;
369 }
370 len = strlen(p);
371 if (len && p[len - 1] == '/') {
372 len--;
373 *flags |= EXC_FLAG_MUSTBEDIR;
374 }
375 for (i = 0; i < len; i++) {
376 if (p[i] == '/')
377 break;
378 }
379 if (i == len)
380 *flags |= EXC_FLAG_NODIR;
381 *nowildcardlen = simple_length(p);
382 /*
383 * we should have excluded the trailing slash from 'p' too,
384 * but that's one more allocation. Instead just make sure
385 * nowildcardlen does not exceed real patternlen
386 */
387 if (*nowildcardlen > len)
388 *nowildcardlen = len;
389 if (*p == '*' && no_wildcard(p + 1))
390 *flags |= EXC_FLAG_ENDSWITH;
391 *pattern = p;
392 *patternlen = len;
393}
394
453ec4bd 395void add_exclude(const char *string, const char *base,
c04318e4 396 int baselen, struct exclude_list *el, int srcpos)
453ec4bd 397{
d6b8fc30 398 struct exclude *x;
84460eec
NTND
399 int patternlen;
400 int flags;
401 int nowildcardlen;
453ec4bd 402
84460eec
NTND
403 parse_exclude_pattern(&string, &patternlen, &flags, &nowildcardlen);
404 if (flags & EXC_FLAG_MUSTBEDIR) {
d6b8fc30 405 char *s;
84460eec 406 x = xmalloc(sizeof(*x) + patternlen + 1);
4b25d091 407 s = (char *)(x+1);
84460eec
NTND
408 memcpy(s, string, patternlen);
409 s[patternlen] = '\0';
d6b8fc30 410 x->pattern = s;
d6b8fc30
JH
411 } else {
412 x = xmalloc(sizeof(*x));
413 x->pattern = string;
414 }
84460eec
NTND
415 x->patternlen = patternlen;
416 x->nowildcardlen = nowildcardlen;
453ec4bd
LT
417 x->base = base;
418 x->baselen = baselen;
d6b8fc30 419 x->flags = flags;
c04318e4 420 x->srcpos = srcpos;
840fc334
AS
421 ALLOC_GROW(el->excludes, el->nr + 1, el->alloc);
422 el->excludes[el->nr++] = x;
c04318e4 423 x->el = el;
453ec4bd
LT
424}
425
c28b3d6e
NTND
426static void *read_skip_worktree_file_from_index(const char *path, size_t *size)
427{
428 int pos, len;
429 unsigned long sz;
430 enum object_type type;
431 void *data;
c28b3d6e
NTND
432
433 len = strlen(path);
71261027 434 pos = cache_name_pos(path, len);
c28b3d6e
NTND
435 if (pos < 0)
436 return NULL;
71261027 437 if (!ce_skip_worktree(active_cache[pos]))
c28b3d6e 438 return NULL;
71261027 439 data = read_sha1_file(active_cache[pos]->sha1, &type, &sz);
c28b3d6e
NTND
440 if (!data || type != OBJ_BLOB) {
441 free(data);
442 return NULL;
443 }
444 *size = xsize_t(sz);
445 return data;
446}
447
f6198812
AS
448/*
449 * Frees memory within el which was allocated for exclude patterns and
450 * the file buffer. Does not free el itself.
451 */
452void clear_exclude_list(struct exclude_list *el)
0fd0e241
NTND
453{
454 int i;
455
456 for (i = 0; i < el->nr; i++)
457 free(el->excludes[i]);
458 free(el->excludes);
c082df24 459 free(el->filebuf);
0fd0e241
NTND
460
461 el->nr = 0;
462 el->excludes = NULL;
c082df24 463 el->filebuf = NULL;
0fd0e241
NTND
464}
465
16402b99
NTND
466static void check_trailing_spaces(const char *fname, char *buf)
467{
468 int i, last_space = -1, len = strlen(buf);
469 for (i = 0; i < len; i++)
470 if (buf[i] == '\\')
471 i++;
472 else if (buf[i] == ' ')
473 last_space = i;
474 else
475 last_space = -1;
476
477 if (last_space == len - 1)
478 warning(_("%s: trailing spaces in '%s'. Please quote or remove them."),
479 fname, buf);
480}
481
cb097534
NTND
482int add_excludes_from_file_to_list(const char *fname,
483 const char *base,
484 int baselen,
840fc334 485 struct exclude_list *el,
cb097534 486 int check_index)
453ec4bd 487{
c470701a 488 struct stat st;
c04318e4 489 int fd, i, lineno = 1;
9d14017a 490 size_t size = 0;
453ec4bd
LT
491 char *buf, *entry;
492
493 fd = open(fname, O_RDONLY);
c28b3d6e 494 if (fd < 0 || fstat(fd, &st) < 0) {
69660731 495 if (errno != ENOENT)
55b38a48 496 warn_on_inaccessible(fname);
c28b3d6e
NTND
497 if (0 <= fd)
498 close(fd);
499 if (!check_index ||
500 (buf = read_skip_worktree_file_from_index(fname, &size)) == NULL)
501 return -1;
45d76f17
NTND
502 if (size == 0) {
503 free(buf);
504 return 0;
505 }
506 if (buf[size-1] != '\n') {
507 buf = xrealloc(buf, size+1);
508 buf[size++] = '\n';
509 }
453ec4bd 510 }
c28b3d6e
NTND
511 else {
512 size = xsize_t(st.st_size);
513 if (size == 0) {
514 close(fd);
515 return 0;
516 }
45d76f17 517 buf = xmalloc(size+1);
c28b3d6e 518 if (read_in_full(fd, buf, size) != size) {
45d76f17 519 free(buf);
c28b3d6e
NTND
520 close(fd);
521 return -1;
522 }
45d76f17 523 buf[size++] = '\n';
c28b3d6e 524 close(fd);
6ba78238 525 }
453ec4bd 526
c082df24 527 el->filebuf = buf;
453ec4bd 528 entry = buf;
45d76f17
NTND
529 for (i = 0; i < size; i++) {
530 if (buf[i] == '\n') {
453ec4bd
LT
531 if (entry != buf + i && entry[0] != '#') {
532 buf[i - (i && buf[i-1] == '\r')] = 0;
16402b99 533 check_trailing_spaces(fname, entry);
c04318e4 534 add_exclude(entry, base, baselen, el, lineno);
453ec4bd 535 }
c04318e4 536 lineno++;
453ec4bd
LT
537 entry = buf + i + 1;
538 }
539 }
540 return 0;
453ec4bd
LT
541}
542
c04318e4
AS
543struct exclude_list *add_exclude_list(struct dir_struct *dir,
544 int group_type, const char *src)
c082df24
AS
545{
546 struct exclude_list *el;
547 struct exclude_list_group *group;
548
549 group = &dir->exclude_list_group[group_type];
550 ALLOC_GROW(group->el, group->nr + 1, group->alloc);
551 el = &group->el[group->nr++];
552 memset(el, 0, sizeof(*el));
c04318e4 553 el->src = src;
c082df24
AS
554 return el;
555}
556
557/*
558 * Used to set up core.excludesfile and .git/info/exclude lists.
559 */
453ec4bd
LT
560void add_excludes_from_file(struct dir_struct *dir, const char *fname)
561{
c082df24 562 struct exclude_list *el;
c04318e4 563 el = add_exclude_list(dir, EXC_FILE, fname);
c082df24 564 if (add_excludes_from_file_to_list(fname, "", 0, el, 0) < 0)
453ec4bd
LT
565 die("cannot use %s as an exclude file", fname);
566}
567
82dce998
NTND
568int match_basename(const char *basename, int basenamelen,
569 const char *pattern, int prefix, int patternlen,
570 int flags)
593cb880
NTND
571{
572 if (prefix == patternlen) {
0b6e56df
JH
573 if (patternlen == basenamelen &&
574 !strncmp_icase(pattern, basename, basenamelen))
593cb880
NTND
575 return 1;
576 } else if (flags & EXC_FLAG_ENDSWITH) {
0b6e56df 577 /* "*literal" matching against "fooliteral" */
593cb880 578 if (patternlen - 1 <= basenamelen &&
0b6e56df
JH
579 !strncmp_icase(pattern + 1,
580 basename + basenamelen - (patternlen - 1),
581 patternlen - 1))
593cb880
NTND
582 return 1;
583 } else {
0b6e56df
JH
584 if (fnmatch_icase_mem(pattern, patternlen,
585 basename, basenamelen,
586 0) == 0)
593cb880
NTND
587 return 1;
588 }
589 return 0;
590}
591
82dce998
NTND
592int match_pathname(const char *pathname, int pathlen,
593 const char *base, int baselen,
594 const char *pattern, int prefix, int patternlen,
595 int flags)
b5592632
NTND
596{
597 const char *name;
598 int namelen;
599
600 /*
601 * match with FNM_PATHNAME; the pattern has base implicitly
602 * in front of it.
603 */
604 if (*pattern == '/') {
605 pattern++;
982ac873 606 patternlen--;
b5592632
NTND
607 prefix--;
608 }
609
610 /*
611 * baselen does not count the trailing slash. base[] may or
612 * may not end with a trailing slash though.
613 */
614 if (pathlen < baselen + 1 ||
615 (baselen && pathname[baselen] != '/') ||
616 strncmp_icase(pathname, base, baselen))
617 return 0;
618
619 namelen = baselen ? pathlen - baselen - 1 : pathlen;
620 name = pathname + pathlen - namelen;
621
622 if (prefix) {
623 /*
624 * if the non-wildcard part is longer than the
625 * remaining pathname, surely it cannot match.
626 */
627 if (prefix > namelen)
628 return 0;
629
630 if (strncmp_icase(pattern, name, prefix))
631 return 0;
632 pattern += prefix;
ab3aebc1 633 patternlen -= prefix;
b5592632
NTND
634 name += prefix;
635 namelen -= prefix;
ab3aebc1
JK
636
637 /*
638 * If the whole pattern did not have a wildcard,
639 * then our prefix match is all we need; we
640 * do not need to call fnmatch at all.
641 */
642 if (!patternlen && !namelen)
643 return 1;
b5592632
NTND
644 }
645
ab3aebc1
JK
646 return fnmatch_icase_mem(pattern, patternlen,
647 name, namelen,
f30366b2 648 WM_PATHNAME) == 0;
b5592632
NTND
649}
650
578cd7c3
AS
651/*
652 * Scan the given exclude list in reverse to see whether pathname
653 * should be ignored. The first match (i.e. the last on the list), if
654 * any, determines the fate. Returns the exclude_list element which
655 * matched, or NULL for undecided.
453ec4bd 656 */
578cd7c3
AS
657static struct exclude *last_exclude_matching_from_list(const char *pathname,
658 int pathlen,
659 const char *basename,
660 int *dtype,
661 struct exclude_list *el)
453ec4bd
LT
662{
663 int i;
664
35a94d44 665 if (!el->nr)
578cd7c3 666 return NULL; /* undefined */
d6b8fc30 667
35a94d44
NTND
668 for (i = el->nr - 1; 0 <= i; i--) {
669 struct exclude *x = el->excludes[i];
b5592632 670 const char *exclude = x->pattern;
b5592632 671 int prefix = x->nowildcardlen;
35a94d44
NTND
672
673 if (x->flags & EXC_FLAG_MUSTBEDIR) {
674 if (*dtype == DT_UNKNOWN)
675 *dtype = get_dtype(NULL, pathname, pathlen);
676 if (*dtype != DT_DIR)
677 continue;
678 }
d6b8fc30 679
35a94d44 680 if (x->flags & EXC_FLAG_NODIR) {
593cb880
NTND
681 if (match_basename(basename,
682 pathlen - (basename - pathname),
683 exclude, prefix, x->patternlen,
684 x->flags))
578cd7c3 685 return x;
35a94d44
NTND
686 continue;
687 }
688
b5592632
NTND
689 assert(x->baselen == 0 || x->base[x->baselen - 1] == '/');
690 if (match_pathname(pathname, pathlen,
691 x->base, x->baselen ? x->baselen - 1 : 0,
692 exclude, prefix, x->patternlen, x->flags))
578cd7c3 693 return x;
453ec4bd 694 }
578cd7c3
AS
695 return NULL; /* undecided */
696}
697
698/*
699 * Scan the list and let the last match determine the fate.
700 * Return 1 for exclude, 0 for include and -1 for undecided.
701 */
702int is_excluded_from_list(const char *pathname,
703 int pathlen, const char *basename, int *dtype,
704 struct exclude_list *el)
705{
706 struct exclude *exclude;
707 exclude = last_exclude_matching_from_list(pathname, pathlen, basename, dtype, el);
708 if (exclude)
709 return exclude->flags & EXC_FLAG_NEGATIVE ? 0 : 1;
453ec4bd
LT
710 return -1; /* undecided */
711}
712
46aa2f95
KB
713static struct exclude *last_exclude_matching_from_lists(struct dir_struct *dir,
714 const char *pathname, int pathlen, const char *basename,
715 int *dtype_p)
716{
717 int i, j;
718 struct exclude_list_group *group;
719 struct exclude *exclude;
720 for (i = EXC_CMDL; i <= EXC_FILE; i++) {
721 group = &dir->exclude_list_group[i];
722 for (j = group->nr - 1; j >= 0; j--) {
723 exclude = last_exclude_matching_from_list(
724 pathname, pathlen, basename, dtype_p,
725 &group->el[j]);
726 if (exclude)
727 return exclude;
728 }
729 }
730 return NULL;
731}
732
6cd5c582
KB
733/*
734 * Loads the per-directory exclude list for the substring of base
735 * which has a char length of baselen.
736 */
737static void prep_exclude(struct dir_struct *dir, const char *base, int baselen)
738{
739 struct exclude_list_group *group;
740 struct exclude_list *el;
741 struct exclude_stack *stk = NULL;
742 int current;
743
6cd5c582
KB
744 group = &dir->exclude_list_group[EXC_DIRS];
745
746 /* Pop the exclude lists from the EXCL_DIRS exclude_list_group
747 * which originate from directories not in the prefix of the
748 * path being checked. */
749 while ((stk = dir->exclude_stack) != NULL) {
750 if (stk->baselen <= baselen &&
751 !strncmp(dir->basebuf, base, stk->baselen))
752 break;
753 el = &group->el[dir->exclude_stack->exclude_ix];
754 dir->exclude_stack = stk->prev;
95c6f271 755 dir->exclude = NULL;
6cd5c582
KB
756 free((char *)el->src); /* see strdup() below */
757 clear_exclude_list(el);
758 free(stk);
759 group->nr--;
760 }
761
95c6f271
KB
762 /* Skip traversing into sub directories if the parent is excluded */
763 if (dir->exclude)
764 return;
765
6cd5c582
KB
766 /* Read from the parent directories and push them down. */
767 current = stk ? stk->baselen : -1;
768 while (current < baselen) {
769 struct exclude_stack *stk = xcalloc(1, sizeof(*stk));
770 const char *cp;
771
772 if (current < 0) {
773 cp = base;
774 current = 0;
775 }
776 else {
777 cp = strchr(base + current + 1, '/');
778 if (!cp)
779 die("oops in prep_exclude");
780 cp++;
781 }
782 stk->prev = dir->exclude_stack;
783 stk->baselen = cp - base;
95c6f271
KB
784 stk->exclude_ix = group->nr;
785 el = add_exclude_list(dir, EXC_DIRS, NULL);
6cd5c582
KB
786 memcpy(dir->basebuf + current, base + current,
787 stk->baselen - current);
95c6f271
KB
788
789 /* Abort if the directory is excluded */
790 if (stk->baselen) {
791 int dt = DT_DIR;
792 dir->basebuf[stk->baselen - 1] = 0;
793 dir->exclude = last_exclude_matching_from_lists(dir,
794 dir->basebuf, stk->baselen - 1,
795 dir->basebuf + current, &dt);
796 dir->basebuf[stk->baselen - 1] = '/';
c3c327de
KB
797 if (dir->exclude &&
798 dir->exclude->flags & EXC_FLAG_NEGATIVE)
799 dir->exclude = NULL;
95c6f271
KB
800 if (dir->exclude) {
801 dir->basebuf[stk->baselen] = 0;
802 dir->exclude_stack = stk;
803 return;
804 }
805 }
806
807 /* Try to read per-directory file unless path is too long */
808 if (dir->exclude_per_dir &&
809 stk->baselen + strlen(dir->exclude_per_dir) < PATH_MAX) {
810 strcpy(dir->basebuf + stk->baselen,
811 dir->exclude_per_dir);
812 /*
813 * dir->basebuf gets reused by the traversal, but we
814 * need fname to remain unchanged to ensure the src
815 * member of each struct exclude correctly
816 * back-references its source file. Other invocations
817 * of add_exclude_list provide stable strings, so we
818 * strdup() and free() here in the caller.
819 */
820 el->src = strdup(dir->basebuf);
821 add_excludes_from_file_to_list(dir->basebuf,
822 dir->basebuf, stk->baselen, el, 1);
823 }
6cd5c582
KB
824 dir->exclude_stack = stk;
825 current = stk->baselen;
826 }
827 dir->basebuf[baselen] = '\0';
828}
829
f4cd69a6
AS
830/*
831 * Loads the exclude lists for the directory containing pathname, then
832 * scans all exclude lists to determine whether pathname is excluded.
833 * Returns the exclude_list element which matched, or NULL for
834 * undecided.
835 */
b07bc8c8 836struct exclude *last_exclude_matching(struct dir_struct *dir,
f4cd69a6
AS
837 const char *pathname,
838 int *dtype_p)
453ec4bd
LT
839{
840 int pathlen = strlen(pathname);
68492fc7
LK
841 const char *basename = strrchr(pathname, '/');
842 basename = (basename) ? basename+1 : pathname;
453ec4bd 843
63d285c8 844 prep_exclude(dir, pathname, basename-pathname);
c082df24 845
95c6f271
KB
846 if (dir->exclude)
847 return dir->exclude;
848
46aa2f95
KB
849 return last_exclude_matching_from_lists(dir, pathname, pathlen,
850 basename, dtype_p);
f4cd69a6
AS
851}
852
853/*
854 * Loads the exclude lists for the directory containing pathname, then
855 * scans all exclude lists to determine whether pathname is excluded.
856 * Returns 1 if true, otherwise 0.
857 */
b07bc8c8 858int is_excluded(struct dir_struct *dir, const char *pathname, int *dtype_p)
f4cd69a6
AS
859{
860 struct exclude *exclude =
861 last_exclude_matching(dir, pathname, dtype_p);
862 if (exclude)
863 return exclude->flags & EXC_FLAG_NEGATIVE ? 0 : 1;
453ec4bd
LT
864 return 0;
865}
866
f3fa1838
JH
867static struct dir_entry *dir_entry_new(const char *pathname, int len)
868{
453ec4bd
LT
869 struct dir_entry *ent;
870
453ec4bd
LT
871 ent = xmalloc(sizeof(*ent) + len + 1);
872 ent->len = len;
873 memcpy(ent->name, pathname, len);
874 ent->name[len] = 0;
4d06f8ac 875 return ent;
453ec4bd
LT
876}
877
159b3212 878static struct dir_entry *dir_add_name(struct dir_struct *dir, const char *pathname, int len)
6815e569 879{
ebbd7439 880 if (cache_file_exists(pathname, len, ignore_case))
6815e569
JK
881 return NULL;
882
25fd2f7a 883 ALLOC_GROW(dir->entries, dir->nr+1, dir->alloc);
6815e569
JK
884 return dir->entries[dir->nr++] = dir_entry_new(pathname, len);
885}
886
108da0db 887struct dir_entry *dir_add_ignored(struct dir_struct *dir, const char *pathname, int len)
2abd31b0 888{
6e4f981f 889 if (!cache_name_is_other(pathname, len))
2abd31b0
JK
890 return NULL;
891
25fd2f7a 892 ALLOC_GROW(dir->ignored, dir->ignored_nr+1, dir->ignored_alloc);
2abd31b0
JK
893 return dir->ignored[dir->ignored_nr++] = dir_entry_new(pathname, len);
894}
895
09595258
LT
896enum exist_status {
897 index_nonexistent = 0,
898 index_directory,
4b05548f 899 index_gitdir
09595258
LT
900};
901
5102c617 902/*
71261027 903 * Do not use the alphabetically sorted index to look up
5102c617 904 * the directory name; instead, use the case insensitive
ebbd7439 905 * directory hash.
5102c617
JJ
906 */
907static enum exist_status directory_exists_in_index_icase(const char *dirname, int len)
908{
d28eec26 909 const struct cache_entry *ce = cache_dir_exists(dirname, len);
5102c617
JJ
910 unsigned char endchar;
911
912 if (!ce)
913 return index_nonexistent;
914 endchar = ce->name[len];
915
916 /*
917 * The cache_entry structure returned will contain this dirname
918 * and possibly additional path components.
919 */
920 if (endchar == '/')
921 return index_directory;
922
923 /*
924 * If there are no additional path components, then this cache_entry
925 * represents a submodule. Submodules, despite being directories,
926 * are stored in the cache without a closing slash.
927 */
928 if (!endchar && S_ISGITLINK(ce->ce_mode))
929 return index_gitdir;
930
931 /* This should never be hit, but it exists just in case. */
932 return index_nonexistent;
933}
934
09595258
LT
935/*
936 * The index sorts alphabetically by entry name, which
937 * means that a gitlink sorts as '\0' at the end, while
938 * a directory (which is defined not as an entry, but as
939 * the files it contains) will sort with the '/' at the
940 * end.
941 */
942static enum exist_status directory_exists_in_index(const char *dirname, int len)
453ec4bd 943{
5102c617
JJ
944 int pos;
945
946 if (ignore_case)
947 return directory_exists_in_index_icase(dirname, len);
948
949 pos = cache_name_pos(dirname, len);
09595258
LT
950 if (pos < 0)
951 pos = -pos-1;
952 while (pos < active_nr) {
9c5e6c80 953 const struct cache_entry *ce = active_cache[pos++];
09595258
LT
954 unsigned char endchar;
955
956 if (strncmp(ce->name, dirname, len))
957 break;
958 endchar = ce->name[len];
959 if (endchar > '/')
960 break;
961 if (endchar == '/')
962 return index_directory;
7a51ed66 963 if (!endchar && S_ISGITLINK(ce->ce_mode))
09595258
LT
964 return index_gitdir;
965 }
966 return index_nonexistent;
967}
968
969/*
970 * When we find a directory when traversing the filesystem, we
971 * have three distinct cases:
972 *
973 * - ignore it
974 * - see it as a directory
975 * - recurse into it
976 *
977 * and which one we choose depends on a combination of existing
978 * git index contents and the flags passed into the directory
979 * traversal routine.
980 *
981 * Case 1: If we *already* have entries in the index under that
5bd8e2d8
KB
982 * directory name, we always recurse into the directory to see
983 * all the files.
09595258
LT
984 *
985 * Case 2: If we *already* have that directory name as a gitlink,
986 * we always continue to see it as a gitlink, regardless of whether
987 * there is an actual git directory there or not (it might not
988 * be checked out as a subproject!)
989 *
990 * Case 3: if we didn't have it in the index previously, we
991 * have a few sub-cases:
992 *
993 * (a) if "show_other_directories" is true, we show it as
994 * just a directory, unless "hide_empty_directories" is
defd7c7b
KB
995 * also true, in which case we need to check if it contains any
996 * untracked and / or ignored files.
09595258 997 * (b) if it looks like a git directory, and we don't have
302b9282 998 * 'no_gitlinks' set we treat it as a gitlink, and show it
09595258
LT
999 * as a directory.
1000 * (c) otherwise, we recurse into it.
1001 */
defd7c7b 1002static enum path_treatment treat_directory(struct dir_struct *dir,
721ac4ed 1003 const char *dirname, int len, int exclude,
09595258
LT
1004 const struct path_simplify *simplify)
1005{
1006 /* The "len-1" is to strip the final '/' */
1007 switch (directory_exists_in_index(dirname, len-1)) {
1008 case index_directory:
defd7c7b 1009 return path_recurse;
09595258
LT
1010
1011 case index_gitdir:
26c986e1 1012 return path_none;
09595258
LT
1013
1014 case index_nonexistent:
7c4c97c0 1015 if (dir->flags & DIR_SHOW_OTHER_DIRECTORIES)
09595258 1016 break;
7c4c97c0 1017 if (!(dir->flags & DIR_NO_GITLINKS)) {
09595258
LT
1018 unsigned char sha1[20];
1019 if (resolve_gitlink_ref(dirname, "HEAD", sha1) == 0)
defd7c7b 1020 return path_untracked;
09595258 1021 }
defd7c7b 1022 return path_recurse;
09595258
LT
1023 }
1024
1025 /* This is the "show_other_directories" case */
721ac4ed 1026
184d2a8e 1027 if (!(dir->flags & DIR_HIDE_EMPTY_DIRECTORIES))
defd7c7b
KB
1028 return exclude ? path_excluded : path_untracked;
1029
1030 return read_directory_recursive(dir, dirname, len, 1, simplify);
453ec4bd
LT
1031}
1032
9fc42d60
LT
1033/*
1034 * This is an inexact early pruning of any recursive directory
1035 * reading - if the path cannot possibly be in the pathspec,
1036 * return true, and we'll skip it early.
1037 */
1038static int simplify_away(const char *path, int pathlen, const struct path_simplify *simplify)
1039{
1040 if (simplify) {
1041 for (;;) {
1042 const char *match = simplify->path;
1043 int len = simplify->len;
1044
1045 if (!match)
1046 break;
1047 if (len > pathlen)
1048 len = pathlen;
1049 if (!memcmp(path, match, len))
1050 return 0;
1051 simplify++;
1052 }
1053 return 1;
1054 }
1055 return 0;
1056}
1057
29209cbe
JK
1058/*
1059 * This function tells us whether an excluded path matches a
1060 * list of "interesting" pathspecs. That is, whether a path matched
1061 * by any of the pathspecs could possibly be ignored by excluding
1062 * the specified path. This can happen if:
1063 *
1064 * 1. the path is mentioned explicitly in the pathspec
1065 *
1066 * 2. the path is a directory prefix of some element in the
1067 * pathspec
1068 */
1069static int exclude_matches_pathspec(const char *path, int len,
1070 const struct path_simplify *simplify)
e96980ef
JK
1071{
1072 if (simplify) {
1073 for (; simplify->path; simplify++) {
1074 if (len == simplify->len
1075 && !memcmp(path, simplify->path, len))
1076 return 1;
29209cbe
JK
1077 if (len < simplify->len
1078 && simplify->path[len] == '/'
1079 && !memcmp(path, simplify->path, len))
1080 return 1;
e96980ef
JK
1081 }
1082 }
1083 return 0;
1084}
1085
443e061a
LT
1086static int get_index_dtype(const char *path, int len)
1087{
1088 int pos;
9c5e6c80 1089 const struct cache_entry *ce;
443e061a 1090
ebbd7439 1091 ce = cache_file_exists(path, len, 0);
443e061a
LT
1092 if (ce) {
1093 if (!ce_uptodate(ce))
1094 return DT_UNKNOWN;
1095 if (S_ISGITLINK(ce->ce_mode))
1096 return DT_DIR;
1097 /*
1098 * Nobody actually cares about the
1099 * difference between DT_LNK and DT_REG
1100 */
1101 return DT_REG;
1102 }
1103
1104 /* Try to look it up as a directory */
1105 pos = cache_name_pos(path, len);
1106 if (pos >= 0)
1107 return DT_UNKNOWN;
1108 pos = -pos-1;
1109 while (pos < active_nr) {
1110 ce = active_cache[pos++];
1111 if (strncmp(ce->name, path, len))
1112 break;
1113 if (ce->name[len] > '/')
1114 break;
1115 if (ce->name[len] < '/')
1116 continue;
1117 if (!ce_uptodate(ce))
1118 break; /* continue? */
1119 return DT_DIR;
1120 }
1121 return DT_UNKNOWN;
1122}
1123
caa6b782 1124static int get_dtype(struct dirent *de, const char *path, int len)
07134421 1125{
6831a88a 1126 int dtype = de ? DTYPE(de) : DT_UNKNOWN;
07134421
LT
1127 struct stat st;
1128
1129 if (dtype != DT_UNKNOWN)
1130 return dtype;
443e061a
LT
1131 dtype = get_index_dtype(path, len);
1132 if (dtype != DT_UNKNOWN)
1133 return dtype;
1134 if (lstat(path, &st))
07134421
LT
1135 return dtype;
1136 if (S_ISREG(st.st_mode))
1137 return DT_REG;
1138 if (S_ISDIR(st.st_mode))
1139 return DT_DIR;
1140 if (S_ISLNK(st.st_mode))
1141 return DT_LNK;
1142 return dtype;
1143}
1144
16e2cfa9 1145static enum path_treatment treat_one_path(struct dir_struct *dir,
49dc2cc2 1146 struct strbuf *path,
16e2cfa9
JH
1147 const struct path_simplify *simplify,
1148 int dtype, struct dirent *de)
53cc5356 1149{
8aaf8d77 1150 int exclude;
ebbd7439 1151 int has_path_in_index = !!cache_file_exists(path->buf, path->len, ignore_case);
2eac2a4c 1152
8aaf8d77
KB
1153 if (dtype == DT_UNKNOWN)
1154 dtype = get_dtype(de, path->buf, path->len);
1155
1156 /* Always exclude indexed files */
2eac2a4c 1157 if (dtype != DT_DIR && has_path_in_index)
defd7c7b 1158 return path_none;
8aaf8d77 1159
2eac2a4c
JH
1160 /*
1161 * When we are looking at a directory P in the working tree,
1162 * there are three cases:
1163 *
1164 * (1) P exists in the index. Everything inside the directory P in
1165 * the working tree needs to go when P is checked out from the
1166 * index.
1167 *
1168 * (2) P does not exist in the index, but there is P/Q in the index.
1169 * We know P will stay a directory when we check out the contents
1170 * of the index, but we do not know yet if there is a directory
1171 * P/Q in the working tree to be killed, so we need to recurse.
1172 *
1173 * (3) P does not exist in the index, and there is no P/Q in the index
1174 * to require P to be a directory, either. Only in this case, we
1175 * know that everything inside P will not be killed without
1176 * recursing.
1177 */
1178 if ((dir->flags & DIR_COLLECT_KILLED_ONLY) &&
1179 (dtype == DT_DIR) &&
de372b1b
ES
1180 !has_path_in_index &&
1181 (directory_exists_in_index(path->buf, path->len) == index_nonexistent))
1182 return path_none;
8aaf8d77
KB
1183
1184 exclude = is_excluded(dir, path->buf, &dtype);
53cc5356
JH
1185
1186 /*
1187 * Excluded? If we don't explicitly want to show
1188 * ignored files, ignore it
1189 */
0aaf62b6 1190 if (exclude && !(dir->flags & (DIR_SHOW_IGNORED|DIR_SHOW_IGNORED_TOO)))
defd7c7b 1191 return path_excluded;
53cc5356 1192
53cc5356
JH
1193 switch (dtype) {
1194 default:
defd7c7b 1195 return path_none;
53cc5356 1196 case DT_DIR:
49dc2cc2 1197 strbuf_addch(path, '/');
defd7c7b
KB
1198 return treat_directory(dir, path->buf, path->len, exclude,
1199 simplify);
53cc5356
JH
1200 case DT_REG:
1201 case DT_LNK:
defd7c7b 1202 return exclude ? path_excluded : path_untracked;
53cc5356 1203 }
53cc5356
JH
1204}
1205
16e2cfa9
JH
1206static enum path_treatment treat_path(struct dir_struct *dir,
1207 struct dirent *de,
49dc2cc2 1208 struct strbuf *path,
16e2cfa9 1209 int baselen,
49dc2cc2 1210 const struct path_simplify *simplify)
16e2cfa9
JH
1211{
1212 int dtype;
1213
1214 if (is_dot_or_dotdot(de->d_name) || !strcmp(de->d_name, ".git"))
defd7c7b 1215 return path_none;
49dc2cc2
RS
1216 strbuf_setlen(path, baselen);
1217 strbuf_addstr(path, de->d_name);
1218 if (simplify_away(path->buf, path->len, simplify))
defd7c7b 1219 return path_none;
16e2cfa9
JH
1220
1221 dtype = DTYPE(de);
49dc2cc2 1222 return treat_one_path(dir, path, simplify, dtype, de);
16e2cfa9
JH
1223}
1224
453ec4bd
LT
1225/*
1226 * Read a directory tree. We currently ignore anything but
1227 * directories, regular files and symlinks. That's because git
1228 * doesn't handle them at all yet. Maybe that will change some
1229 * day.
1230 *
1231 * Also, we ignore the name ".git" (even if it is not a directory).
1232 * That likely will not change.
defd7c7b
KB
1233 *
1234 * Returns the most significant path_treatment value encountered in the scan.
453ec4bd 1235 */
defd7c7b 1236static enum path_treatment read_directory_recursive(struct dir_struct *dir,
53cc5356
JH
1237 const char *base, int baselen,
1238 int check_only,
1239 const struct path_simplify *simplify)
453ec4bd 1240{
1528d247 1241 DIR *fdir;
defd7c7b 1242 enum path_treatment state, subdir_state, dir_state = path_none;
02cb6753 1243 struct dirent *de;
bef36921 1244 struct strbuf path = STRBUF_INIT;
453ec4bd 1245
bef36921 1246 strbuf_add(&path, base, baselen);
02cb6753 1247
1528d247
RS
1248 fdir = opendir(path.len ? path.buf : ".");
1249 if (!fdir)
1250 goto out;
1251
02cb6753 1252 while ((de = readdir(fdir)) != NULL) {
defd7c7b
KB
1253 /* check how the file or directory should be treated */
1254 state = treat_path(dir, de, &path, baselen, simplify);
1255 if (state > dir_state)
1256 dir_state = state;
1257
1258 /* recurse into subdir if instructed by treat_path */
1259 if (state == path_recurse) {
1260 subdir_state = read_directory_recursive(dir, path.buf,
289ff559 1261 path.len, check_only, simplify);
defd7c7b
KB
1262 if (subdir_state > dir_state)
1263 dir_state = subdir_state;
1264 }
1265
1266 if (check_only) {
1267 /* abort early if maximum state has been reached */
1268 if (dir_state == path_untracked)
1269 break;
1270 /* skip the dir_add_* part */
02cb6753 1271 continue;
453ec4bd 1272 }
defd7c7b
KB
1273
1274 /* add the path to the appropriate result list */
1275 switch (state) {
1276 case path_excluded:
1277 if (dir->flags & DIR_SHOW_IGNORED)
1278 dir_add_name(dir, path.buf, path.len);
0aaf62b6
KB
1279 else if ((dir->flags & DIR_SHOW_IGNORED_TOO) ||
1280 ((dir->flags & DIR_COLLECT_IGNORED) &&
1281 exclude_matches_pathspec(path.buf, path.len,
1282 simplify)))
1283 dir_add_ignored(dir, path.buf, path.len);
defd7c7b
KB
1284 break;
1285
1286 case path_untracked:
1287 if (!(dir->flags & DIR_SHOW_IGNORED))
1288 dir_add_name(dir, path.buf, path.len);
1528d247 1289 break;
defd7c7b
KB
1290
1291 default:
1292 break;
1293 }
453ec4bd 1294 }
02cb6753 1295 closedir(fdir);
1528d247 1296 out:
bef36921 1297 strbuf_release(&path);
453ec4bd 1298
defd7c7b 1299 return dir_state;
453ec4bd
LT
1300}
1301
1302static int cmp_name(const void *p1, const void *p2)
1303{
1304 const struct dir_entry *e1 = *(const struct dir_entry **)p1;
1305 const struct dir_entry *e2 = *(const struct dir_entry **)p2;
1306
1307 return cache_name_compare(e1->name, e1->len,
1308 e2->name, e2->len);
1309}
1310
9fc42d60
LT
1311static struct path_simplify *create_simplify(const char **pathspec)
1312{
1313 int nr, alloc = 0;
1314 struct path_simplify *simplify = NULL;
1315
1316 if (!pathspec)
1317 return NULL;
1318
1319 for (nr = 0 ; ; nr++) {
1320 const char *match;
1321 if (nr >= alloc) {
1322 alloc = alloc_nr(alloc);
1323 simplify = xrealloc(simplify, alloc * sizeof(*simplify));
1324 }
1325 match = *pathspec++;
1326 if (!match)
1327 break;
1328 simplify[nr].path = match;
1329 simplify[nr].len = simple_length(match);
1330 }
1331 simplify[nr].path = NULL;
1332 simplify[nr].len = 0;
1333 return simplify;
1334}
1335
1336static void free_simplify(struct path_simplify *simplify)
1337{
8e0f7003 1338 free(simplify);
9fc42d60
LT
1339}
1340
48ffef96
JH
1341static int treat_leading_path(struct dir_struct *dir,
1342 const char *path, int len,
1343 const struct path_simplify *simplify)
1344{
49dc2cc2
RS
1345 struct strbuf sb = STRBUF_INIT;
1346 int baselen, rc = 0;
48ffef96 1347 const char *cp;
be8a84c5 1348 int old_flags = dir->flags;
48ffef96
JH
1349
1350 while (len && path[len - 1] == '/')
1351 len--;
1352 if (!len)
1353 return 1;
1354 baselen = 0;
be8a84c5 1355 dir->flags &= ~DIR_SHOW_OTHER_DIRECTORIES;
48ffef96
JH
1356 while (1) {
1357 cp = path + baselen + !!baselen;
1358 cp = memchr(cp, '/', path + len - cp);
1359 if (!cp)
1360 baselen = len;
1361 else
1362 baselen = cp - path;
49dc2cc2
RS
1363 strbuf_setlen(&sb, 0);
1364 strbuf_add(&sb, path, baselen);
1365 if (!is_directory(sb.buf))
1366 break;
1367 if (simplify_away(sb.buf, sb.len, simplify))
1368 break;
1369 if (treat_one_path(dir, &sb, simplify,
defd7c7b 1370 DT_DIR, NULL) == path_none)
49dc2cc2
RS
1371 break; /* do not recurse into it */
1372 if (len <= baselen) {
1373 rc = 1;
1374 break; /* finished checking */
1375 }
48ffef96 1376 }
49dc2cc2 1377 strbuf_release(&sb);
be8a84c5 1378 dir->flags = old_flags;
49dc2cc2 1379 return rc;
48ffef96
JH
1380}
1381
7327d3d1 1382int read_directory(struct dir_struct *dir, const char *path, int len, const struct pathspec *pathspec)
9fc42d60 1383{
725b0605 1384 struct path_simplify *simplify;
b4189aa8 1385
7327d3d1
NTND
1386 /*
1387 * Check out create_simplify()
1388 */
1389 if (pathspec)
5c6933d2
NTND
1390 GUARD_PATHSPEC(pathspec,
1391 PATHSPEC_FROMTOP |
1392 PATHSPEC_MAXDEPTH |
bd30c2e4 1393 PATHSPEC_LITERAL |
93d93537
NTND
1394 PATHSPEC_GLOB |
1395 PATHSPEC_ICASE);
7327d3d1 1396
dba2e203 1397 if (has_symlink_leading_path(path, len))
725b0605
JH
1398 return dir->nr;
1399
b3920bbd 1400 simplify = create_simplify(pathspec ? pathspec->_raw : NULL);
48ffef96
JH
1401 if (!len || treat_leading_path(dir, path, len, simplify))
1402 read_directory_recursive(dir, path, len, 0, simplify);
9fc42d60 1403 free_simplify(simplify);
453ec4bd 1404 qsort(dir->entries, dir->nr, sizeof(struct dir_entry *), cmp_name);
2abd31b0 1405 qsort(dir->ignored, dir->ignored_nr, sizeof(struct dir_entry *), cmp_name);
453ec4bd
LT
1406 return dir->nr;
1407}
c91f0d92 1408
686a4a06 1409int file_exists(const char *f)
c91f0d92 1410{
686a4a06 1411 struct stat sb;
a50f9fc5 1412 return lstat(f, &sb) == 0;
c91f0d92 1413}
e6636747
JS
1414
1415/*
9b125da4
NTND
1416 * Given two normalized paths (a trailing slash is ok), if subdir is
1417 * outside dir, return -1. Otherwise return the offset in subdir that
1418 * can be used as relative path to dir.
e6636747 1419 */
9b125da4 1420int dir_inside_of(const char *subdir, const char *dir)
e6636747 1421{
9b125da4 1422 int offset = 0;
e6636747 1423
9b125da4 1424 assert(dir && subdir && *dir && *subdir);
e6636747 1425
9b125da4 1426 while (*dir && *subdir && *dir == *subdir) {
e6636747 1427 dir++;
9b125da4
NTND
1428 subdir++;
1429 offset++;
490544b1 1430 }
9b125da4
NTND
1431
1432 /* hel[p]/me vs hel[l]/yeah */
1433 if (*dir && *subdir)
1434 return -1;
1435
1436 if (!*subdir)
1437 return !*dir ? offset : -1; /* same dir */
1438
1439 /* foo/[b]ar vs foo/[] */
1440 if (is_dir_sep(dir[-1]))
1441 return is_dir_sep(subdir[-1]) ? offset : -1;
1442
1443 /* foo[/]bar vs foo[] */
1444 return is_dir_sep(*subdir) ? offset + 1 : -1;
e6636747
JS
1445}
1446
1447int is_inside_dir(const char *dir)
1448{
b892913d
NTND
1449 char cwd[PATH_MAX];
1450 if (!dir)
1451 return 0;
1452 if (!getcwd(cwd, sizeof(cwd)))
1453 die_errno("can't find the current directory");
1454 return dir_inside_of(cwd, dir) >= 0;
e6636747 1455}
7155b727 1456
55892d23
AP
1457int is_empty_dir(const char *path)
1458{
1459 DIR *dir = opendir(path);
1460 struct dirent *e;
1461 int ret = 1;
1462
1463 if (!dir)
1464 return 0;
1465
1466 while ((e = readdir(dir)) != NULL)
1467 if (!is_dot_or_dotdot(e->d_name)) {
1468 ret = 0;
1469 break;
1470 }
1471
1472 closedir(dir);
1473 return ret;
1474}
1475
ae2f203e 1476static int remove_dir_recurse(struct strbuf *path, int flag, int *kept_up)
7155b727 1477{
a0f4afbe 1478 DIR *dir;
7155b727 1479 struct dirent *e;
ae2f203e 1480 int ret = 0, original_len = path->len, len, kept_down = 0;
a0f4afbe 1481 int only_empty = (flag & REMOVE_DIR_EMPTY_ONLY);
c844a803 1482 int keep_toplevel = (flag & REMOVE_DIR_KEEP_TOPLEVEL);
a0f4afbe 1483 unsigned char submodule_head[20];
7155b727 1484
a0f4afbe 1485 if ((flag & REMOVE_DIR_KEEP_NESTED_GIT) &&
ae2f203e 1486 !resolve_gitlink_ref(path->buf, "HEAD", submodule_head)) {
a0f4afbe 1487 /* Do not descend and nuke a nested git work tree. */
ae2f203e
JH
1488 if (kept_up)
1489 *kept_up = 1;
a0f4afbe 1490 return 0;
ae2f203e 1491 }
a0f4afbe 1492
ae2f203e 1493 flag &= ~REMOVE_DIR_KEEP_TOPLEVEL;
a0f4afbe 1494 dir = opendir(path->buf);
c844a803 1495 if (!dir) {
ae2f203e 1496 /* an empty dir could be removed even if it is unreadble */
c844a803
JH
1497 if (!keep_toplevel)
1498 return rmdir(path->buf);
1499 else
1500 return -1;
1501 }
7155b727
JS
1502 if (path->buf[original_len - 1] != '/')
1503 strbuf_addch(path, '/');
1504
1505 len = path->len;
1506 while ((e = readdir(dir)) != NULL) {
1507 struct stat st;
8ca12c0d
AP
1508 if (is_dot_or_dotdot(e->d_name))
1509 continue;
7155b727
JS
1510
1511 strbuf_setlen(path, len);
1512 strbuf_addstr(path, e->d_name);
1513 if (lstat(path->buf, &st))
1514 ; /* fall thru */
1515 else if (S_ISDIR(st.st_mode)) {
ae2f203e 1516 if (!remove_dir_recurse(path, flag, &kept_down))
7155b727
JS
1517 continue; /* happy */
1518 } else if (!only_empty && !unlink(path->buf))
1519 continue; /* happy, too */
1520
1521 /* path too long, stat fails, or non-directory still exists */
1522 ret = -1;
1523 break;
1524 }
1525 closedir(dir);
1526
1527 strbuf_setlen(path, original_len);
ae2f203e 1528 if (!ret && !keep_toplevel && !kept_down)
7155b727 1529 ret = rmdir(path->buf);
ae2f203e
JH
1530 else if (kept_up)
1531 /*
1532 * report the uplevel that it is not an error that we
1533 * did not rmdir() our directory.
1534 */
1535 *kept_up = !ret;
7155b727
JS
1536 return ret;
1537}
039bc64e 1538
ae2f203e
JH
1539int remove_dir_recursively(struct strbuf *path, int flag)
1540{
1541 return remove_dir_recurse(path, flag, NULL);
1542}
1543
039bc64e
JH
1544void setup_standard_excludes(struct dir_struct *dir)
1545{
1546 const char *path;
dc79687e 1547 char *xdg_path;
039bc64e
JH
1548
1549 dir->exclude_per_dir = ".gitignore";
1550 path = git_path("info/exclude");
dc79687e
HKNN
1551 if (!excludes_file) {
1552 home_config_paths(NULL, &xdg_path, "ignore");
1553 excludes_file = xdg_path;
1554 }
4698c8fe 1555 if (!access_or_warn(path, R_OK, 0))
039bc64e 1556 add_excludes_from_file(dir, path);
4698c8fe 1557 if (excludes_file && !access_or_warn(excludes_file, R_OK, 0))
039bc64e
JH
1558 add_excludes_from_file(dir, excludes_file);
1559}
4a92d1bf
AR
1560
1561int remove_path(const char *name)
1562{
1563 char *slash;
1564
9a6728d4 1565 if (unlink(name) && errno != ENOENT && errno != ENOTDIR)
4a92d1bf
AR
1566 return -1;
1567
1568 slash = strrchr(name, '/');
1569 if (slash) {
1570 char *dirs = xstrdup(name);
1571 slash = dirs + (slash - name);
1572 do {
1573 *slash = '\0';
3fc0d131 1574 } while (rmdir(dirs) == 0 && (slash = strrchr(dirs, '/')));
4a92d1bf
AR
1575 free(dirs);
1576 }
1577 return 0;
1578}
1579
270be816
AS
1580/*
1581 * Frees memory within dir which was allocated for exclude lists and
1582 * the exclude_stack. Does not free dir itself.
1583 */
1584void clear_directory(struct dir_struct *dir)
1585{
1586 int i, j;
1587 struct exclude_list_group *group;
1588 struct exclude_list *el;
1589 struct exclude_stack *stk;
1590
1591 for (i = EXC_CMDL; i <= EXC_FILE; i++) {
1592 group = &dir->exclude_list_group[i];
1593 for (j = 0; j < group->nr; j++) {
1594 el = &group->el[j];
1595 if (i == EXC_DIRS)
1596 free((char *)el->src);
1597 clear_exclude_list(el);
1598 }
1599 free(group->el);
1600 }
1601
1602 stk = dir->exclude_stack;
1603 while (stk) {
1604 struct exclude_stack *prev = stk->prev;
1605 free(stk);
1606 stk = prev;
1607 }
1608}