Merge branch 'jc/ls-files-o'
[git] / ls-files.c
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.
5  *
6  * Copyright (C) Linus Torvalds, 2005
7  */
8 #include <dirent.h>
9 #include <fnmatch.h>
10
11 #include "cache.h"
12 #include "quote.h"
13
14 static int show_deleted = 0;
15 static int show_cached = 0;
16 static int show_others = 0;
17 static int show_ignored = 0;
18 static int show_stage = 0;
19 static int show_unmerged = 0;
20 static int show_modified = 0;
21 static int show_killed = 0;
22 static int show_other_directories = 0;
23 static int line_terminator = '\n';
24
25 static int prefix_len = 0, prefix_offset = 0;
26 static const char *prefix = NULL;
27 static const char **pathspec = NULL;
28
29 static const char *tag_cached = "";
30 static const char *tag_unmerged = "";
31 static const char *tag_removed = "";
32 static const char *tag_other = "";
33 static const char *tag_killed = "";
34 static const char *tag_modified = "";
35
36 static const char *exclude_per_dir = NULL;
37
38 /* We maintain three exclude pattern lists:
39  * EXC_CMDL lists patterns explicitly given on the command line.
40  * EXC_DIRS lists patterns obtained from per-directory ignore files.
41  * EXC_FILE lists patterns from fallback ignore files.
42  */
43 #define EXC_CMDL 0
44 #define EXC_DIRS 1
45 #define EXC_FILE 2
46 static struct exclude_list {
47         int nr;
48         int alloc;
49         struct exclude {
50                 const char *pattern;
51                 const char *base;
52                 int baselen;
53         } **excludes;
54 } exclude_list[3];
55
56 static void add_exclude(const char *string, const char *base,
57                         int baselen, struct exclude_list *which)
58 {
59         struct exclude *x = xmalloc(sizeof (*x));
60
61         x->pattern = string;
62         x->base = base;
63         x->baselen = baselen;
64         if (which->nr == which->alloc) {
65                 which->alloc = alloc_nr(which->alloc);
66                 which->excludes = realloc(which->excludes,
67                                           which->alloc * sizeof(x));
68         }
69         which->excludes[which->nr++] = x;
70 }
71
72 static int add_excludes_from_file_1(const char *fname,
73                                     const char *base,
74                                     int baselen,
75                                     struct exclude_list *which)
76 {
77         int fd, i;
78         long size;
79         char *buf, *entry;
80
81         fd = open(fname, O_RDONLY);
82         if (fd < 0)
83                 goto err;
84         size = lseek(fd, 0, SEEK_END);
85         if (size < 0)
86                 goto err;
87         lseek(fd, 0, SEEK_SET);
88         if (size == 0) {
89                 close(fd);
90                 return 0;
91         }
92         buf = xmalloc(size);
93         if (read(fd, buf, size) != size)
94                 goto err;
95         close(fd);
96
97         entry = buf;
98         for (i = 0; i < size; i++) {
99                 if (buf[i] == '\n') {
100                         if (entry != buf + i && entry[0] != '#') {
101                                 buf[i - (i && buf[i-1] == '\r')] = 0;
102                                 add_exclude(entry, base, baselen, which);
103                         }
104                         entry = buf + i + 1;
105                 }
106         }
107         return 0;
108
109  err:
110         if (0 <= fd)
111                 close(fd);
112         return -1;
113 }
114
115 static void add_excludes_from_file(const char *fname)
116 {
117         if (add_excludes_from_file_1(fname, "", 0,
118                                      &exclude_list[EXC_FILE]) < 0)
119                 die("cannot use %s as an exclude file", fname);
120 }
121
122 static int push_exclude_per_directory(const char *base, int baselen)
123 {
124         char exclude_file[PATH_MAX];
125         struct exclude_list *el = &exclude_list[EXC_DIRS];
126         int current_nr = el->nr;
127
128         if (exclude_per_dir) {
129                 memcpy(exclude_file, base, baselen);
130                 strcpy(exclude_file + baselen, exclude_per_dir);
131                 add_excludes_from_file_1(exclude_file, base, baselen, el);
132         }
133         return current_nr;
134 }
135
136 static void pop_exclude_per_directory(int stk)
137 {
138         struct exclude_list *el = &exclude_list[EXC_DIRS];
139
140         while (stk < el->nr)
141                 free(el->excludes[--el->nr]);
142 }
143
144 /* Scan the list and let the last match determines the fate.
145  * Return 1 for exclude, 0 for include and -1 for undecided.
146  */
147 static int excluded_1(const char *pathname,
148                       int pathlen,
149                       struct exclude_list *el)
150 {
151         int i;
152
153         if (el->nr) {
154                 for (i = el->nr - 1; 0 <= i; i--) {
155                         struct exclude *x = el->excludes[i];
156                         const char *exclude = x->pattern;
157                         int to_exclude = 1;
158
159                         if (*exclude == '!') {
160                                 to_exclude = 0;
161                                 exclude++;
162                         }
163
164                         if (!strchr(exclude, '/')) {
165                                 /* match basename */
166                                 const char *basename = strrchr(pathname, '/');
167                                 basename = (basename) ? basename+1 : pathname;
168                                 if (fnmatch(exclude, basename, 0) == 0)
169                                         return to_exclude;
170                         }
171                         else {
172                                 /* match with FNM_PATHNAME:
173                                  * exclude has base (baselen long) implicitly
174                                  * in front of it.
175                                  */
176                                 int baselen = x->baselen;
177                                 if (*exclude == '/')
178                                         exclude++;
179
180                                 if (pathlen < baselen ||
181                                     (baselen && pathname[baselen-1] != '/') ||
182                                     strncmp(pathname, x->base, baselen))
183                                     continue;
184
185                                 if (fnmatch(exclude, pathname+baselen,
186                                             FNM_PATHNAME) == 0)
187                                         return to_exclude;
188                         }
189                 }
190         }
191         return -1; /* undecided */
192 }
193
194 static int excluded(const char *pathname)
195 {
196         int pathlen = strlen(pathname);
197         int st;
198
199         for (st = EXC_CMDL; st <= EXC_FILE; st++) {
200                 switch (excluded_1(pathname, pathlen, &exclude_list[st])) {
201                 case 0:
202                         return 0;
203                 case 1:
204                         return 1;
205                 }
206         }
207         return 0;
208 }
209
210 struct nond_on_fs {
211         int len;
212         char name[FLEX_ARRAY]; /* more */
213 };
214
215 static struct nond_on_fs **dir;
216 static int nr_dir;
217 static int dir_alloc;
218
219 static void add_name(const char *pathname, int len)
220 {
221         struct nond_on_fs *ent;
222
223         if (cache_name_pos(pathname, len) >= 0)
224                 return;
225
226         if (nr_dir == dir_alloc) {
227                 dir_alloc = alloc_nr(dir_alloc);
228                 dir = xrealloc(dir, dir_alloc*sizeof(ent));
229         }
230         ent = xmalloc(sizeof(*ent) + len + 1);
231         ent->len = len;
232         memcpy(ent->name, pathname, len);
233         ent->name[len] = 0;
234         dir[nr_dir++] = ent;
235 }
236
237 static int dir_exists(const char *dirname, int len)
238 {
239         int pos = cache_name_pos(dirname, len);
240         if (pos >= 0)
241                 return 1;
242         pos = -pos-1;
243         if (pos >= active_nr) /* can't */
244                 return 0;
245         return !strncmp(active_cache[pos]->name, dirname, len);
246 }
247
248 /*
249  * Read a directory tree. We currently ignore anything but
250  * directories, regular files and symlinks. That's because git
251  * doesn't handle them at all yet. Maybe that will change some
252  * day.
253  *
254  * Also, we ignore the name ".git" (even if it is not a directory).
255  * That likely will not change.
256  */
257 static void read_directory(const char *path, const char *base, int baselen)
258 {
259         DIR *dir = opendir(path);
260
261         if (dir) {
262                 int exclude_stk;
263                 struct dirent *de;
264                 char fullname[MAXPATHLEN + 1];
265                 memcpy(fullname, base, baselen);
266
267                 exclude_stk = push_exclude_per_directory(base, baselen);
268
269                 while ((de = readdir(dir)) != NULL) {
270                         int len;
271
272                         if ((de->d_name[0] == '.') &&
273                             (de->d_name[1] == 0 ||
274                              !strcmp(de->d_name + 1, ".") ||
275                              !strcmp(de->d_name + 1, "git")))
276                                 continue;
277                         len = strlen(de->d_name);
278                         memcpy(fullname + baselen, de->d_name, len+1);
279                         if (excluded(fullname) != show_ignored)
280                                 continue;
281
282                         switch (DTYPE(de)) {
283                         struct stat st;
284                         default:
285                                 continue;
286                         case DT_UNKNOWN:
287                                 if (lstat(fullname, &st))
288                                         continue;
289                                 if (S_ISREG(st.st_mode) || S_ISLNK(st.st_mode))
290                                         break;
291                                 if (!S_ISDIR(st.st_mode))
292                                         continue;
293                                 /* fallthrough */
294                         case DT_DIR:
295                                 memcpy(fullname + baselen + len, "/", 2);
296                                 len++;
297                                 if (show_other_directories &&
298                                     !dir_exists(fullname, baselen + len))
299                                         break;
300                                 read_directory(fullname, fullname,
301                                                baselen + len);
302                                 continue;
303                         case DT_REG:
304                         case DT_LNK:
305                                 break;
306                         }
307                         add_name(fullname, baselen + len);
308                 }
309                 closedir(dir);
310
311                 pop_exclude_per_directory(exclude_stk);
312         }
313 }
314
315 static int cmp_name(const void *p1, const void *p2)
316 {
317         const struct nond_on_fs *e1 = *(const struct nond_on_fs **)p1;
318         const struct nond_on_fs *e2 = *(const struct nond_on_fs **)p2;
319
320         return cache_name_compare(e1->name, e1->len,
321                                   e2->name, e2->len);
322 }
323
324 /*
325  * Match a pathspec against a filename. The first "len" characters
326  * are the common prefix
327  */
328 static int match(const char **spec, const char *filename, int len)
329 {
330         const char *m;
331
332         while ((m = *spec++) != NULL) {
333                 int matchlen = strlen(m + len);
334
335                 if (!matchlen)
336                         return 1;
337                 if (!strncmp(m + len, filename + len, matchlen)) {
338                         if (m[len + matchlen - 1] == '/')
339                                 return 1;
340                         switch (filename[len + matchlen]) {
341                         case '/': case '\0':
342                                 return 1;
343                         }
344                 }
345                 if (!fnmatch(m + len, filename + len, 0))
346                         return 1;
347         }
348         return 0;
349 }
350
351 static void show_dir_entry(const char *tag, struct nond_on_fs *ent)
352 {
353         int len = prefix_len;
354         int offset = prefix_offset;
355
356         if (len >= ent->len)
357                 die("git-ls-files: internal error - directory entry not superset of prefix");
358
359         if (pathspec && !match(pathspec, ent->name, len))
360                 return;
361
362         fputs(tag, stdout);
363         write_name_quoted("", 0, ent->name + offset, line_terminator, stdout);
364         putchar(line_terminator);
365 }
366
367 static void show_other_files(void)
368 {
369         int i;
370         for (i = 0; i < nr_dir; i++) {
371                 /* We should not have a matching entry, but we
372                  * may have an unmerged entry for this path.
373                  */
374                 struct nond_on_fs *ent = dir[i];
375                 int pos = cache_name_pos(ent->name, ent->len);
376                 struct cache_entry *ce;
377                 if (0 <= pos)
378                         die("bug in show-other-files");
379                 pos = -pos - 1;
380                 if (pos < active_nr) { 
381                         ce = active_cache[pos];
382                         if (ce_namelen(ce) == ent->len &&
383                             !memcmp(ce->name, ent->name, ent->len))
384                                 continue; /* Yup, this one exists unmerged */
385                 }
386                 show_dir_entry(tag_other, ent);
387         }
388 }
389
390 static void show_killed_files(void)
391 {
392         int i;
393         for (i = 0; i < nr_dir; i++) {
394                 struct nond_on_fs *ent = dir[i];
395                 char *cp, *sp;
396                 int pos, len, killed = 0;
397
398                 for (cp = ent->name; cp - ent->name < ent->len; cp = sp + 1) {
399                         sp = strchr(cp, '/');
400                         if (!sp) {
401                                 /* If ent->name is prefix of an entry in the
402                                  * cache, it will be killed.
403                                  */
404                                 pos = cache_name_pos(ent->name, ent->len);
405                                 if (0 <= pos)
406                                         die("bug in show-killed-files");
407                                 pos = -pos - 1;
408                                 while (pos < active_nr &&
409                                        ce_stage(active_cache[pos]))
410                                         pos++; /* skip unmerged */
411                                 if (active_nr <= pos)
412                                         break;
413                                 /* pos points at a name immediately after
414                                  * ent->name in the cache.  Does it expect
415                                  * ent->name to be a directory?
416                                  */
417                                 len = ce_namelen(active_cache[pos]);
418                                 if ((ent->len < len) &&
419                                     !strncmp(active_cache[pos]->name,
420                                              ent->name, ent->len) &&
421                                     active_cache[pos]->name[ent->len] == '/')
422                                         killed = 1;
423                                 break;
424                         }
425                         if (0 <= cache_name_pos(ent->name, sp - ent->name)) {
426                                 /* If any of the leading directories in
427                                  * ent->name is registered in the cache,
428                                  * ent->name will be killed.
429                                  */
430                                 killed = 1;
431                                 break;
432                         }
433                 }
434                 if (killed)
435                         show_dir_entry(tag_killed, dir[i]);
436         }
437 }
438
439 static void show_ce_entry(const char *tag, struct cache_entry *ce)
440 {
441         int len = prefix_len;
442         int offset = prefix_offset;
443
444         if (len >= ce_namelen(ce))
445                 die("git-ls-files: internal error - cache entry not superset of prefix");
446
447         if (pathspec && !match(pathspec, ce->name, len))
448                 return;
449
450         if (tag && *tag && (ce->ce_flags & htons(CE_VALID))) {
451                 static char alttag[4];
452                 memcpy(alttag, tag, 3);
453                 if (isalpha(tag[0]))
454                         alttag[0] = tolower(tag[0]);
455                 else if (tag[0] == '?')
456                         alttag[0] = '!';
457                 else {
458                         alttag[0] = 'v';
459                         alttag[1] = tag[0];
460                         alttag[2] = ' ';
461                         alttag[3] = 0;
462                 }
463                 tag = alttag;
464         }
465
466         if (!show_stage) {
467                 fputs(tag, stdout);
468                 write_name_quoted("", 0, ce->name + offset,
469                                   line_terminator, stdout);
470                 putchar(line_terminator);
471         }
472         else {
473                 printf("%s%06o %s %d\t",
474                        tag,
475                        ntohl(ce->ce_mode),
476                        sha1_to_hex(ce->sha1),
477                        ce_stage(ce));
478                 write_name_quoted("", 0, ce->name + offset,
479                                   line_terminator, stdout);
480                 putchar(line_terminator);
481         }
482 }
483
484 static void show_files(void)
485 {
486         int i;
487
488         /* For cached/deleted files we don't need to even do the readdir */
489         if (show_others || show_killed) {
490                 const char *path = ".", *base = "";
491                 int baselen = prefix_len;
492
493                 if (baselen) {
494                         path = base = prefix;
495                         if (exclude_per_dir) {
496                                 char *p, *pp = xmalloc(baselen+1);
497                                 memcpy(pp, prefix, baselen+1);
498                                 p = pp;
499                                 while (1) {
500                                         char save = *p;
501                                         *p = 0;
502                                         push_exclude_per_directory(pp, p-pp);
503                                         *p++ = save;
504                                         if (!save)
505                                                 break;
506                                         p = strchr(p, '/');
507                                         if (p)
508                                                 p++;
509                                         else
510                                                 p = pp + baselen;
511                                 }
512                                 free(pp);
513                         }
514                 }
515                 read_directory(path, base, baselen);
516                 qsort(dir, nr_dir, sizeof(struct nond_on_fs *), cmp_name);
517                 if (show_others)
518                         show_other_files();
519                 if (show_killed)
520                         show_killed_files();
521         }
522         if (show_cached | show_stage) {
523                 for (i = 0; i < active_nr; i++) {
524                         struct cache_entry *ce = active_cache[i];
525                         if (excluded(ce->name) != show_ignored)
526                                 continue;
527                         if (show_unmerged && !ce_stage(ce))
528                                 continue;
529                         show_ce_entry(ce_stage(ce) ? tag_unmerged : tag_cached, ce);
530                 }
531         }
532         if (show_deleted | show_modified) {
533                 for (i = 0; i < active_nr; i++) {
534                         struct cache_entry *ce = active_cache[i];
535                         struct stat st;
536                         int err;
537                         if (excluded(ce->name) != show_ignored)
538                                 continue;
539                         err = lstat(ce->name, &st);
540                         if (show_deleted && err)
541                                 show_ce_entry(tag_removed, ce);
542                         if (show_modified && ce_modified(ce, &st, 0))
543                                 show_ce_entry(tag_modified, ce);
544                 }
545         }
546 }
547
548 /*
549  * Prune the index to only contain stuff starting with "prefix"
550  */
551 static void prune_cache(void)
552 {
553         int pos = cache_name_pos(prefix, prefix_len);
554         unsigned int first, last;
555
556         if (pos < 0)
557                 pos = -pos-1;
558         active_cache += pos;
559         active_nr -= pos;
560         first = 0;
561         last = active_nr;
562         while (last > first) {
563                 int next = (last + first) >> 1;
564                 struct cache_entry *ce = active_cache[next];
565                 if (!strncmp(ce->name, prefix, prefix_len)) {
566                         first = next+1;
567                         continue;
568                 }
569                 last = next;
570         }
571         active_nr = last;
572 }
573
574 static void verify_pathspec(void)
575 {
576         const char **p, *n, *prev;
577         char *real_prefix;
578         unsigned long max;
579
580         prev = NULL;
581         max = PATH_MAX;
582         for (p = pathspec; (n = *p) != NULL; p++) {
583                 int i, len = 0;
584                 for (i = 0; i < max; i++) {
585                         char c = n[i];
586                         if (prev && prev[i] != c)
587                                 break;
588                         if (!c || c == '*' || c == '?')
589                                 break;
590                         if (c == '/')
591                                 len = i+1;
592                 }
593                 prev = n;
594                 if (len < max) {
595                         max = len;
596                         if (!max)
597                                 break;
598                 }
599         }
600
601         if (prefix_offset > max || memcmp(prev, prefix, prefix_offset))
602                 die("git-ls-files: cannot generate relative filenames containing '..'");
603
604         real_prefix = NULL;
605         prefix_len = max;
606         if (max) {
607                 real_prefix = xmalloc(max + 1);
608                 memcpy(real_prefix, prev, max);
609                 real_prefix[max] = 0;
610         }
611         prefix = real_prefix;
612 }
613
614 static const char ls_files_usage[] =
615         "git-ls-files [-z] [-t] (--[cached|deleted|others|stage|unmerged|killed|modified])* "
616         "[ --ignored ] [--exclude=<pattern>] [--exclude-from=<file>] "
617         "[ --exclude-per-directory=<filename> ] [--full-name] [--] [<file>]*";
618
619 int main(int argc, const char **argv)
620 {
621         int i;
622         int exc_given = 0;
623
624         prefix = setup_git_directory();
625         if (prefix)
626                 prefix_offset = strlen(prefix);
627         git_config(git_default_config);
628
629         for (i = 1; i < argc; i++) {
630                 const char *arg = argv[i];
631
632                 if (!strcmp(arg, "--")) {
633                         i++;
634                         break;
635                 }
636                 if (!strcmp(arg, "-z")) {
637                         line_terminator = 0;
638                         continue;
639                 }
640                 if (!strcmp(arg, "-t")) {
641                         tag_cached = "H ";
642                         tag_unmerged = "M ";
643                         tag_removed = "R ";
644                         tag_modified = "C ";
645                         tag_other = "? ";
646                         tag_killed = "K ";
647                         continue;
648                 }
649                 if (!strcmp(arg, "-c") || !strcmp(arg, "--cached")) {
650                         show_cached = 1;
651                         continue;
652                 }
653                 if (!strcmp(arg, "-d") || !strcmp(arg, "--deleted")) {
654                         show_deleted = 1;
655                         continue;
656                 }
657                 if (!strcmp(arg, "-m") || !strcmp(arg, "--modified")) {
658                         show_modified = 1;
659                         continue;
660                 }
661                 if (!strcmp(arg, "-o") || !strcmp(arg, "--others")) {
662                         show_others = 1;
663                         continue;
664                 }
665                 if (!strcmp(arg, "-i") || !strcmp(arg, "--ignored")) {
666                         show_ignored = 1;
667                         continue;
668                 }
669                 if (!strcmp(arg, "-s") || !strcmp(arg, "--stage")) {
670                         show_stage = 1;
671                         continue;
672                 }
673                 if (!strcmp(arg, "-k") || !strcmp(arg, "--killed")) {
674                         show_killed = 1;
675                         continue;
676                 }
677                 if (!strcmp(arg, "--directory")) {
678                         show_other_directories = 1;
679                         continue;
680                 }
681                 if (!strcmp(arg, "-u") || !strcmp(arg, "--unmerged")) {
682                         /* There's no point in showing unmerged unless
683                          * you also show the stage information.
684                          */
685                         show_stage = 1;
686                         show_unmerged = 1;
687                         continue;
688                 }
689                 if (!strcmp(arg, "-x") && i+1 < argc) {
690                         exc_given = 1;
691                         add_exclude(argv[++i], "", 0, &exclude_list[EXC_CMDL]);
692                         continue;
693                 }
694                 if (!strncmp(arg, "--exclude=", 10)) {
695                         exc_given = 1;
696                         add_exclude(arg+10, "", 0, &exclude_list[EXC_CMDL]);
697                         continue;
698                 }
699                 if (!strcmp(arg, "-X") && i+1 < argc) {
700                         exc_given = 1;
701                         add_excludes_from_file(argv[++i]);
702                         continue;
703                 }
704                 if (!strncmp(arg, "--exclude-from=", 15)) {
705                         exc_given = 1;
706                         add_excludes_from_file(arg+15);
707                         continue;
708                 }
709                 if (!strncmp(arg, "--exclude-per-directory=", 24)) {
710                         exc_given = 1;
711                         exclude_per_dir = arg + 24;
712                         continue;
713                 }
714                 if (!strcmp(arg, "--full-name")) {
715                         prefix_offset = 0;
716                         continue;
717                 }
718                 if (*arg == '-')
719                         usage(ls_files_usage);
720                 break;
721         }
722
723         pathspec = get_pathspec(prefix, argv + i);
724
725         /* Verify that the pathspec matches the prefix */
726         if (pathspec)
727                 verify_pathspec();
728
729         if (show_ignored && !exc_given) {
730                 fprintf(stderr, "%s: --ignored needs some exclude pattern\n",
731                         argv[0]);
732                 exit(1);
733         }
734
735         /* With no flags, we default to showing the cached files */
736         if (!(show_stage | show_deleted | show_others | show_unmerged |
737               show_killed | show_modified))
738                 show_cached = 1;
739
740         read_cache();
741         if (prefix)
742                 prune_cache();
743         show_files();
744         return 0;
745 }