Support for pickaxe matching regular expressions
[git/builtin-gsoc.git] / ls-files.c
blob4a4af1ca3b3cc11e2996c2889cb8ea16bbb0dded
1 /*
2 * This merges the file listing in the directory cache index
3 * with the actual working directory list, and shows different
4 * combinations of the two.
6 * Copyright (C) Linus Torvalds, 2005
7 */
8 #include <dirent.h>
9 #include <fnmatch.h>
11 #include "cache.h"
12 #include "quote.h"
14 static int abbrev = 0;
15 static int show_deleted = 0;
16 static int show_cached = 0;
17 static int show_others = 0;
18 static int show_ignored = 0;
19 static int show_stage = 0;
20 static int show_unmerged = 0;
21 static int show_modified = 0;
22 static int show_killed = 0;
23 static int show_other_directories = 0;
24 static int hide_empty_directories = 0;
25 static int show_valid_bit = 0;
26 static int line_terminator = '\n';
28 static int prefix_len = 0, prefix_offset = 0;
29 static const char *prefix = NULL;
30 static const char **pathspec = NULL;
31 static int error_unmatch = 0;
32 static char *ps_matched = NULL;
34 static const char *tag_cached = "";
35 static const char *tag_unmerged = "";
36 static const char *tag_removed = "";
37 static const char *tag_other = "";
38 static const char *tag_killed = "";
39 static const char *tag_modified = "";
41 static const char *exclude_per_dir = NULL;
43 /* We maintain three exclude pattern lists:
44 * EXC_CMDL lists patterns explicitly given on the command line.
45 * EXC_DIRS lists patterns obtained from per-directory ignore files.
46 * EXC_FILE lists patterns from fallback ignore files.
48 #define EXC_CMDL 0
49 #define EXC_DIRS 1
50 #define EXC_FILE 2
51 static struct exclude_list {
52 int nr;
53 int alloc;
54 struct exclude {
55 const char *pattern;
56 const char *base;
57 int baselen;
58 } **excludes;
59 } exclude_list[3];
61 static void add_exclude(const char *string, const char *base,
62 int baselen, struct exclude_list *which)
64 struct exclude *x = xmalloc(sizeof (*x));
66 x->pattern = string;
67 x->base = base;
68 x->baselen = baselen;
69 if (which->nr == which->alloc) {
70 which->alloc = alloc_nr(which->alloc);
71 which->excludes = realloc(which->excludes,
72 which->alloc * sizeof(x));
74 which->excludes[which->nr++] = x;
77 static int add_excludes_from_file_1(const char *fname,
78 const char *base,
79 int baselen,
80 struct exclude_list *which)
82 int fd, i;
83 long size;
84 char *buf, *entry;
86 fd = open(fname, O_RDONLY);
87 if (fd < 0)
88 goto err;
89 size = lseek(fd, 0, SEEK_END);
90 if (size < 0)
91 goto err;
92 lseek(fd, 0, SEEK_SET);
93 if (size == 0) {
94 close(fd);
95 return 0;
97 buf = xmalloc(size+1);
98 if (read(fd, buf, size) != size)
99 goto err;
100 close(fd);
102 buf[size++] = '\n';
103 entry = buf;
104 for (i = 0; i < size; i++) {
105 if (buf[i] == '\n') {
106 if (entry != buf + i && entry[0] != '#') {
107 buf[i - (i && buf[i-1] == '\r')] = 0;
108 add_exclude(entry, base, baselen, which);
110 entry = buf + i + 1;
113 return 0;
115 err:
116 if (0 <= fd)
117 close(fd);
118 return -1;
121 static void add_excludes_from_file(const char *fname)
123 if (add_excludes_from_file_1(fname, "", 0,
124 &exclude_list[EXC_FILE]) < 0)
125 die("cannot use %s as an exclude file", fname);
128 static int push_exclude_per_directory(const char *base, int baselen)
130 char exclude_file[PATH_MAX];
131 struct exclude_list *el = &exclude_list[EXC_DIRS];
132 int current_nr = el->nr;
134 if (exclude_per_dir) {
135 memcpy(exclude_file, base, baselen);
136 strcpy(exclude_file + baselen, exclude_per_dir);
137 add_excludes_from_file_1(exclude_file, base, baselen, el);
139 return current_nr;
142 static void pop_exclude_per_directory(int stk)
144 struct exclude_list *el = &exclude_list[EXC_DIRS];
146 while (stk < el->nr)
147 free(el->excludes[--el->nr]);
150 /* Scan the list and let the last match determines the fate.
151 * Return 1 for exclude, 0 for include and -1 for undecided.
153 static int excluded_1(const char *pathname,
154 int pathlen,
155 struct exclude_list *el)
157 int i;
159 if (el->nr) {
160 for (i = el->nr - 1; 0 <= i; i--) {
161 struct exclude *x = el->excludes[i];
162 const char *exclude = x->pattern;
163 int to_exclude = 1;
165 if (*exclude == '!') {
166 to_exclude = 0;
167 exclude++;
170 if (!strchr(exclude, '/')) {
171 /* match basename */
172 const char *basename = strrchr(pathname, '/');
173 basename = (basename) ? basename+1 : pathname;
174 if (fnmatch(exclude, basename, 0) == 0)
175 return to_exclude;
177 else {
178 /* match with FNM_PATHNAME:
179 * exclude has base (baselen long) implicitly
180 * in front of it.
182 int baselen = x->baselen;
183 if (*exclude == '/')
184 exclude++;
186 if (pathlen < baselen ||
187 (baselen && pathname[baselen-1] != '/') ||
188 strncmp(pathname, x->base, baselen))
189 continue;
191 if (fnmatch(exclude, pathname+baselen,
192 FNM_PATHNAME) == 0)
193 return to_exclude;
197 return -1; /* undecided */
200 static int excluded(const char *pathname)
202 int pathlen = strlen(pathname);
203 int st;
205 for (st = EXC_CMDL; st <= EXC_FILE; st++) {
206 switch (excluded_1(pathname, pathlen, &exclude_list[st])) {
207 case 0:
208 return 0;
209 case 1:
210 return 1;
213 return 0;
216 struct nond_on_fs {
217 int len;
218 char name[FLEX_ARRAY]; /* more */
221 static struct nond_on_fs **dir;
222 static int nr_dir;
223 static int dir_alloc;
225 static void add_name(const char *pathname, int len)
227 struct nond_on_fs *ent;
229 if (cache_name_pos(pathname, len) >= 0)
230 return;
232 if (nr_dir == dir_alloc) {
233 dir_alloc = alloc_nr(dir_alloc);
234 dir = xrealloc(dir, dir_alloc*sizeof(ent));
236 ent = xmalloc(sizeof(*ent) + len + 1);
237 ent->len = len;
238 memcpy(ent->name, pathname, len);
239 ent->name[len] = 0;
240 dir[nr_dir++] = ent;
243 static int dir_exists(const char *dirname, int len)
245 int pos = cache_name_pos(dirname, len);
246 if (pos >= 0)
247 return 1;
248 pos = -pos-1;
249 if (pos >= active_nr) /* can't */
250 return 0;
251 return !strncmp(active_cache[pos]->name, dirname, len);
255 * Read a directory tree. We currently ignore anything but
256 * directories, regular files and symlinks. That's because git
257 * doesn't handle them at all yet. Maybe that will change some
258 * day.
260 * Also, we ignore the name ".git" (even if it is not a directory).
261 * That likely will not change.
263 static int read_directory(const char *path, const char *base, int baselen)
265 DIR *fdir = opendir(path);
266 int contents = 0;
268 if (fdir) {
269 int exclude_stk;
270 struct dirent *de;
271 char fullname[MAXPATHLEN + 1];
272 memcpy(fullname, base, baselen);
274 exclude_stk = push_exclude_per_directory(base, baselen);
276 while ((de = readdir(fdir)) != NULL) {
277 int len;
279 if ((de->d_name[0] == '.') &&
280 (de->d_name[1] == 0 ||
281 !strcmp(de->d_name + 1, ".") ||
282 !strcmp(de->d_name + 1, "git")))
283 continue;
284 len = strlen(de->d_name);
285 memcpy(fullname + baselen, de->d_name, len+1);
286 if (excluded(fullname) != show_ignored) {
287 if (!show_ignored || DTYPE(de) != DT_DIR) {
288 continue;
292 switch (DTYPE(de)) {
293 struct stat st;
294 int subdir, rewind_base;
295 default:
296 continue;
297 case DT_UNKNOWN:
298 if (lstat(fullname, &st))
299 continue;
300 if (S_ISREG(st.st_mode) || S_ISLNK(st.st_mode))
301 break;
302 if (!S_ISDIR(st.st_mode))
303 continue;
304 /* fallthrough */
305 case DT_DIR:
306 memcpy(fullname + baselen + len, "/", 2);
307 len++;
308 rewind_base = nr_dir;
309 subdir = read_directory(fullname, fullname,
310 baselen + len);
311 if (show_other_directories &&
312 (subdir || !hide_empty_directories) &&
313 !dir_exists(fullname, baselen + len)) {
314 // Rewind the read subdirectory
315 while (nr_dir > rewind_base)
316 free(dir[--nr_dir]);
317 break;
319 contents += subdir;
320 continue;
321 case DT_REG:
322 case DT_LNK:
323 break;
325 add_name(fullname, baselen + len);
326 contents++;
328 closedir(fdir);
330 pop_exclude_per_directory(exclude_stk);
333 return contents;
336 static int cmp_name(const void *p1, const void *p2)
338 const struct nond_on_fs *e1 = *(const struct nond_on_fs **)p1;
339 const struct nond_on_fs *e2 = *(const struct nond_on_fs **)p2;
341 return cache_name_compare(e1->name, e1->len,
342 e2->name, e2->len);
346 * Match a pathspec against a filename. The first "len" characters
347 * are the common prefix
349 static int match(const char **spec, char *ps_matched,
350 const char *filename, int len)
352 const char *m;
354 while ((m = *spec++) != NULL) {
355 int matchlen = strlen(m + len);
357 if (!matchlen)
358 goto matched;
359 if (!strncmp(m + len, filename + len, matchlen)) {
360 if (m[len + matchlen - 1] == '/')
361 goto matched;
362 switch (filename[len + matchlen]) {
363 case '/': case '\0':
364 goto matched;
367 if (!fnmatch(m + len, filename + len, 0))
368 goto matched;
369 if (ps_matched)
370 ps_matched++;
371 continue;
372 matched:
373 if (ps_matched)
374 *ps_matched = 1;
375 return 1;
377 return 0;
380 static void show_dir_entry(const char *tag, struct nond_on_fs *ent)
382 int len = prefix_len;
383 int offset = prefix_offset;
385 if (len >= ent->len)
386 die("git-ls-files: internal error - directory entry not superset of prefix");
388 if (pathspec && !match(pathspec, ps_matched, ent->name, len))
389 return;
391 fputs(tag, stdout);
392 write_name_quoted("", 0, ent->name + offset, line_terminator, stdout);
393 putchar(line_terminator);
396 static void show_other_files(void)
398 int i;
399 for (i = 0; i < nr_dir; i++) {
400 /* We should not have a matching entry, but we
401 * may have an unmerged entry for this path.
403 struct nond_on_fs *ent = dir[i];
404 int pos = cache_name_pos(ent->name, ent->len);
405 struct cache_entry *ce;
406 if (0 <= pos)
407 die("bug in show-other-files");
408 pos = -pos - 1;
409 if (pos < active_nr) {
410 ce = active_cache[pos];
411 if (ce_namelen(ce) == ent->len &&
412 !memcmp(ce->name, ent->name, ent->len))
413 continue; /* Yup, this one exists unmerged */
415 show_dir_entry(tag_other, ent);
419 static void show_killed_files(void)
421 int i;
422 for (i = 0; i < nr_dir; i++) {
423 struct nond_on_fs *ent = dir[i];
424 char *cp, *sp;
425 int pos, len, killed = 0;
427 for (cp = ent->name; cp - ent->name < ent->len; cp = sp + 1) {
428 sp = strchr(cp, '/');
429 if (!sp) {
430 /* If ent->name is prefix of an entry in the
431 * cache, it will be killed.
433 pos = cache_name_pos(ent->name, ent->len);
434 if (0 <= pos)
435 die("bug in show-killed-files");
436 pos = -pos - 1;
437 while (pos < active_nr &&
438 ce_stage(active_cache[pos]))
439 pos++; /* skip unmerged */
440 if (active_nr <= pos)
441 break;
442 /* pos points at a name immediately after
443 * ent->name in the cache. Does it expect
444 * ent->name to be a directory?
446 len = ce_namelen(active_cache[pos]);
447 if ((ent->len < len) &&
448 !strncmp(active_cache[pos]->name,
449 ent->name, ent->len) &&
450 active_cache[pos]->name[ent->len] == '/')
451 killed = 1;
452 break;
454 if (0 <= cache_name_pos(ent->name, sp - ent->name)) {
455 /* If any of the leading directories in
456 * ent->name is registered in the cache,
457 * ent->name will be killed.
459 killed = 1;
460 break;
463 if (killed)
464 show_dir_entry(tag_killed, dir[i]);
468 static void show_ce_entry(const char *tag, struct cache_entry *ce)
470 int len = prefix_len;
471 int offset = prefix_offset;
473 if (len >= ce_namelen(ce))
474 die("git-ls-files: internal error - cache entry not superset of prefix");
476 if (pathspec && !match(pathspec, ps_matched, ce->name, len))
477 return;
479 if (tag && *tag && show_valid_bit &&
480 (ce->ce_flags & htons(CE_VALID))) {
481 static char alttag[4];
482 memcpy(alttag, tag, 3);
483 if (isalpha(tag[0]))
484 alttag[0] = tolower(tag[0]);
485 else if (tag[0] == '?')
486 alttag[0] = '!';
487 else {
488 alttag[0] = 'v';
489 alttag[1] = tag[0];
490 alttag[2] = ' ';
491 alttag[3] = 0;
493 tag = alttag;
496 if (!show_stage) {
497 fputs(tag, stdout);
498 write_name_quoted("", 0, ce->name + offset,
499 line_terminator, stdout);
500 putchar(line_terminator);
502 else {
503 printf("%s%06o %s %d\t",
504 tag,
505 ntohl(ce->ce_mode),
506 abbrev ? find_unique_abbrev(ce->sha1,abbrev)
507 : sha1_to_hex(ce->sha1),
508 ce_stage(ce));
509 write_name_quoted("", 0, ce->name + offset,
510 line_terminator, stdout);
511 putchar(line_terminator);
515 static void show_files(void)
517 int i;
519 /* For cached/deleted files we don't need to even do the readdir */
520 if (show_others || show_killed) {
521 const char *path = ".", *base = "";
522 int baselen = prefix_len;
524 if (baselen) {
525 path = base = prefix;
526 if (exclude_per_dir) {
527 char *p, *pp = xmalloc(baselen+1);
528 memcpy(pp, prefix, baselen+1);
529 p = pp;
530 while (1) {
531 char save = *p;
532 *p = 0;
533 push_exclude_per_directory(pp, p-pp);
534 *p++ = save;
535 if (!save)
536 break;
537 p = strchr(p, '/');
538 if (p)
539 p++;
540 else
541 p = pp + baselen;
543 free(pp);
546 read_directory(path, base, baselen);
547 qsort(dir, nr_dir, sizeof(struct nond_on_fs *), cmp_name);
548 if (show_others)
549 show_other_files();
550 if (show_killed)
551 show_killed_files();
553 if (show_cached | show_stage) {
554 for (i = 0; i < active_nr; i++) {
555 struct cache_entry *ce = active_cache[i];
556 if (excluded(ce->name) != show_ignored)
557 continue;
558 if (show_unmerged && !ce_stage(ce))
559 continue;
560 show_ce_entry(ce_stage(ce) ? tag_unmerged : tag_cached, ce);
563 if (show_deleted | show_modified) {
564 for (i = 0; i < active_nr; i++) {
565 struct cache_entry *ce = active_cache[i];
566 struct stat st;
567 int err;
568 if (excluded(ce->name) != show_ignored)
569 continue;
570 err = lstat(ce->name, &st);
571 if (show_deleted && err)
572 show_ce_entry(tag_removed, ce);
573 if (show_modified && ce_modified(ce, &st, 0))
574 show_ce_entry(tag_modified, ce);
580 * Prune the index to only contain stuff starting with "prefix"
582 static void prune_cache(void)
584 int pos = cache_name_pos(prefix, prefix_len);
585 unsigned int first, last;
587 if (pos < 0)
588 pos = -pos-1;
589 active_cache += pos;
590 active_nr -= pos;
591 first = 0;
592 last = active_nr;
593 while (last > first) {
594 int next = (last + first) >> 1;
595 struct cache_entry *ce = active_cache[next];
596 if (!strncmp(ce->name, prefix, prefix_len)) {
597 first = next+1;
598 continue;
600 last = next;
602 active_nr = last;
605 static void verify_pathspec(void)
607 const char **p, *n, *prev;
608 char *real_prefix;
609 unsigned long max;
611 prev = NULL;
612 max = PATH_MAX;
613 for (p = pathspec; (n = *p) != NULL; p++) {
614 int i, len = 0;
615 for (i = 0; i < max; i++) {
616 char c = n[i];
617 if (prev && prev[i] != c)
618 break;
619 if (!c || c == '*' || c == '?')
620 break;
621 if (c == '/')
622 len = i+1;
624 prev = n;
625 if (len < max) {
626 max = len;
627 if (!max)
628 break;
632 if (prefix_offset > max || memcmp(prev, prefix, prefix_offset))
633 die("git-ls-files: cannot generate relative filenames containing '..'");
635 real_prefix = NULL;
636 prefix_len = max;
637 if (max) {
638 real_prefix = xmalloc(max + 1);
639 memcpy(real_prefix, prev, max);
640 real_prefix[max] = 0;
642 prefix = real_prefix;
645 static const char ls_files_usage[] =
646 "git-ls-files [-z] [-t] [-v] (--[cached|deleted|others|stage|unmerged|killed|modified])* "
647 "[ --ignored ] [--exclude=<pattern>] [--exclude-from=<file>] "
648 "[ --exclude-per-directory=<filename> ] [--full-name] [--abbrev] "
649 "[--] [<file>]*";
651 int main(int argc, const char **argv)
653 int i;
654 int exc_given = 0;
656 prefix = setup_git_directory();
657 if (prefix)
658 prefix_offset = strlen(prefix);
659 git_config(git_default_config);
661 for (i = 1; i < argc; i++) {
662 const char *arg = argv[i];
664 if (!strcmp(arg, "--")) {
665 i++;
666 break;
668 if (!strcmp(arg, "-z")) {
669 line_terminator = 0;
670 continue;
672 if (!strcmp(arg, "-t") || !strcmp(arg, "-v")) {
673 tag_cached = "H ";
674 tag_unmerged = "M ";
675 tag_removed = "R ";
676 tag_modified = "C ";
677 tag_other = "? ";
678 tag_killed = "K ";
679 if (arg[1] == 'v')
680 show_valid_bit = 1;
681 continue;
683 if (!strcmp(arg, "-c") || !strcmp(arg, "--cached")) {
684 show_cached = 1;
685 continue;
687 if (!strcmp(arg, "-d") || !strcmp(arg, "--deleted")) {
688 show_deleted = 1;
689 continue;
691 if (!strcmp(arg, "-m") || !strcmp(arg, "--modified")) {
692 show_modified = 1;
693 continue;
695 if (!strcmp(arg, "-o") || !strcmp(arg, "--others")) {
696 show_others = 1;
697 continue;
699 if (!strcmp(arg, "-i") || !strcmp(arg, "--ignored")) {
700 show_ignored = 1;
701 continue;
703 if (!strcmp(arg, "-s") || !strcmp(arg, "--stage")) {
704 show_stage = 1;
705 continue;
707 if (!strcmp(arg, "-k") || !strcmp(arg, "--killed")) {
708 show_killed = 1;
709 continue;
711 if (!strcmp(arg, "--directory")) {
712 show_other_directories = 1;
713 continue;
715 if (!strcmp(arg, "--no-empty-directory")) {
716 hide_empty_directories = 1;
717 continue;
719 if (!strcmp(arg, "-u") || !strcmp(arg, "--unmerged")) {
720 /* There's no point in showing unmerged unless
721 * you also show the stage information.
723 show_stage = 1;
724 show_unmerged = 1;
725 continue;
727 if (!strcmp(arg, "-x") && i+1 < argc) {
728 exc_given = 1;
729 add_exclude(argv[++i], "", 0, &exclude_list[EXC_CMDL]);
730 continue;
732 if (!strncmp(arg, "--exclude=", 10)) {
733 exc_given = 1;
734 add_exclude(arg+10, "", 0, &exclude_list[EXC_CMDL]);
735 continue;
737 if (!strcmp(arg, "-X") && i+1 < argc) {
738 exc_given = 1;
739 add_excludes_from_file(argv[++i]);
740 continue;
742 if (!strncmp(arg, "--exclude-from=", 15)) {
743 exc_given = 1;
744 add_excludes_from_file(arg+15);
745 continue;
747 if (!strncmp(arg, "--exclude-per-directory=", 24)) {
748 exc_given = 1;
749 exclude_per_dir = arg + 24;
750 continue;
752 if (!strcmp(arg, "--full-name")) {
753 prefix_offset = 0;
754 continue;
756 if (!strcmp(arg, "--error-unmatch")) {
757 error_unmatch = 1;
758 continue;
760 if (!strncmp(arg, "--abbrev=", 9)) {
761 abbrev = strtoul(arg+9, NULL, 10);
762 if (abbrev && abbrev < MINIMUM_ABBREV)
763 abbrev = MINIMUM_ABBREV;
764 else if (abbrev > 40)
765 abbrev = 40;
766 continue;
768 if (!strcmp(arg, "--abbrev")) {
769 abbrev = DEFAULT_ABBREV;
770 continue;
772 if (*arg == '-')
773 usage(ls_files_usage);
774 break;
777 pathspec = get_pathspec(prefix, argv + i);
779 /* Verify that the pathspec matches the prefix */
780 if (pathspec)
781 verify_pathspec();
783 /* Treat unmatching pathspec elements as errors */
784 if (pathspec && error_unmatch) {
785 int num;
786 for (num = 0; pathspec[num]; num++)
788 ps_matched = xcalloc(1, num);
791 if (show_ignored && !exc_given) {
792 fprintf(stderr, "%s: --ignored needs some exclude pattern\n",
793 argv[0]);
794 exit(1);
797 /* With no flags, we default to showing the cached files */
798 if (!(show_stage | show_deleted | show_others | show_unmerged |
799 show_killed | show_modified))
800 show_cached = 1;
802 read_cache();
803 if (prefix)
804 prune_cache();
805 show_files();
807 if (ps_matched) {
808 /* We need to make sure all pathspec matched otherwise
809 * it is an error.
811 int num, errors = 0;
812 for (num = 0; pathspec[num]; num++) {
813 if (ps_matched[num])
814 continue;
815 error("pathspec '%s' did not match any.",
816 pathspec[num] + prefix_offset);
817 errors++;
819 return errors ? 1 : 0;
822 return 0;