Merge part of 'lt/rev-list' into 'fk/blame'
[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 show_valid_bit = 0;
24 static int line_terminator = '\n';
25
26 static int prefix_len = 0, prefix_offset = 0;
27 static const char *prefix = NULL;
28 static const char **pathspec = NULL;
29 static int error_unmatch = 0;
30 static char *ps_matched = NULL;
31
32 static const char *tag_cached = "";
33 static const char *tag_unmerged = "";
34 static const char *tag_removed = "";
35 static const char *tag_other = "";
36 static const char *tag_killed = "";
37 static const char *tag_modified = "";
38
39 static const char *exclude_per_dir = NULL;
40
41 /* We maintain three exclude pattern lists:
42  * EXC_CMDL lists patterns explicitly given on the command line.
43  * EXC_DIRS lists patterns obtained from per-directory ignore files.
44  * EXC_FILE lists patterns from fallback ignore files.
45  */
46 #define EXC_CMDL 0
47 #define EXC_DIRS 1
48 #define EXC_FILE 2
49 static struct exclude_list {
50         int nr;
51         int alloc;
52         struct exclude {
53                 const char *pattern;
54                 const char *base;
55                 int baselen;
56         } **excludes;
57 } exclude_list[3];
58
59 static void add_exclude(const char *string, const char *base,
60                         int baselen, struct exclude_list *which)
61 {
62         struct exclude *x = xmalloc(sizeof (*x));
63
64         x->pattern = string;
65         x->base = base;
66         x->baselen = baselen;
67         if (which->nr == which->alloc) {
68                 which->alloc = alloc_nr(which->alloc);
69                 which->excludes = realloc(which->excludes,
70                                           which->alloc * sizeof(x));
71         }
72         which->excludes[which->nr++] = x;
73 }
74
75 static int add_excludes_from_file_1(const char *fname,
76                                     const char *base,
77                                     int baselen,
78                                     struct exclude_list *which)
79 {
80         int fd, i;
81         long size;
82         char *buf, *entry;
83
84         fd = open(fname, O_RDONLY);
85         if (fd < 0)
86                 goto err;
87         size = lseek(fd, 0, SEEK_END);
88         if (size < 0)
89                 goto err;
90         lseek(fd, 0, SEEK_SET);
91         if (size == 0) {
92                 close(fd);
93                 return 0;
94         }
95         buf = xmalloc(size);
96         if (read(fd, buf, size) != size)
97                 goto err;
98         close(fd);
99
100         entry = buf;
101         for (i = 0; i < size; i++) {
102                 if (buf[i] == '\n') {
103                         if (entry != buf + i && entry[0] != '#') {
104                                 buf[i - (i && buf[i-1] == '\r')] = 0;
105                                 add_exclude(entry, base, baselen, which);
106                         }
107                         entry = buf + i + 1;
108                 }
109         }
110         return 0;
111
112  err:
113         if (0 <= fd)
114                 close(fd);
115         return -1;
116 }
117
118 static void add_excludes_from_file(const char *fname)
119 {
120         if (add_excludes_from_file_1(fname, "", 0,
121                                      &exclude_list[EXC_FILE]) < 0)
122                 die("cannot use %s as an exclude file", fname);
123 }
124
125 static int push_exclude_per_directory(const char *base, int baselen)
126 {
127         char exclude_file[PATH_MAX];
128         struct exclude_list *el = &exclude_list[EXC_DIRS];
129         int current_nr = el->nr;
130
131         if (exclude_per_dir) {
132                 memcpy(exclude_file, base, baselen);
133                 strcpy(exclude_file + baselen, exclude_per_dir);
134                 add_excludes_from_file_1(exclude_file, base, baselen, el);
135         }
136         return current_nr;
137 }
138
139 static void pop_exclude_per_directory(int stk)
140 {
141         struct exclude_list *el = &exclude_list[EXC_DIRS];
142
143         while (stk < el->nr)
144                 free(el->excludes[--el->nr]);
145 }
146
147 /* Scan the list and let the last match determines the fate.
148  * Return 1 for exclude, 0 for include and -1 for undecided.
149  */
150 static int excluded_1(const char *pathname,
151                       int pathlen,
152                       struct exclude_list *el)
153 {
154         int i;
155
156         if (el->nr) {
157                 for (i = el->nr - 1; 0 <= i; i--) {
158                         struct exclude *x = el->excludes[i];
159                         const char *exclude = x->pattern;
160                         int to_exclude = 1;
161
162                         if (*exclude == '!') {
163                                 to_exclude = 0;
164                                 exclude++;
165                         }
166
167                         if (!strchr(exclude, '/')) {
168                                 /* match basename */
169                                 const char *basename = strrchr(pathname, '/');
170                                 basename = (basename) ? basename+1 : pathname;
171                                 if (fnmatch(exclude, basename, 0) == 0)
172                                         return to_exclude;
173                         }
174                         else {
175                                 /* match with FNM_PATHNAME:
176                                  * exclude has base (baselen long) implicitly
177                                  * in front of it.
178                                  */
179                                 int baselen = x->baselen;
180                                 if (*exclude == '/')
181                                         exclude++;
182
183                                 if (pathlen < baselen ||
184                                     (baselen && pathname[baselen-1] != '/') ||
185                                     strncmp(pathname, x->base, baselen))
186                                     continue;
187
188                                 if (fnmatch(exclude, pathname+baselen,
189                                             FNM_PATHNAME) == 0)
190                                         return to_exclude;
191                         }
192                 }
193         }
194         return -1; /* undecided */
195 }
196
197 static int excluded(const char *pathname)
198 {
199         int pathlen = strlen(pathname);
200         int st;
201
202         for (st = EXC_CMDL; st <= EXC_FILE; st++) {
203                 switch (excluded_1(pathname, pathlen, &exclude_list[st])) {
204                 case 0:
205                         return 0;
206                 case 1:
207                         return 1;
208                 }
209         }
210         return 0;
211 }
212
213 struct nond_on_fs {
214         int len;
215         char name[FLEX_ARRAY]; /* more */
216 };
217
218 static struct nond_on_fs **dir;
219 static int nr_dir;
220 static int dir_alloc;
221
222 static void add_name(const char *pathname, int len)
223 {
224         struct nond_on_fs *ent;
225
226         if (cache_name_pos(pathname, len) >= 0)
227                 return;
228
229         if (nr_dir == dir_alloc) {
230                 dir_alloc = alloc_nr(dir_alloc);
231                 dir = xrealloc(dir, dir_alloc*sizeof(ent));
232         }
233         ent = xmalloc(sizeof(*ent) + len + 1);
234         ent->len = len;
235         memcpy(ent->name, pathname, len);
236         ent->name[len] = 0;
237         dir[nr_dir++] = ent;
238 }
239
240 static int dir_exists(const char *dirname, int len)
241 {
242         int pos = cache_name_pos(dirname, len);
243         if (pos >= 0)
244                 return 1;
245         pos = -pos-1;
246         if (pos >= active_nr) /* can't */
247                 return 0;
248         return !strncmp(active_cache[pos]->name, dirname, len);
249 }
250
251 /*
252  * Read a directory tree. We currently ignore anything but
253  * directories, regular files and symlinks. That's because git
254  * doesn't handle them at all yet. Maybe that will change some
255  * day.
256  *
257  * Also, we ignore the name ".git" (even if it is not a directory).
258  * That likely will not change.
259  */
260 static void read_directory(const char *path, const char *base, int baselen)
261 {
262         DIR *dir = opendir(path);
263
264         if (dir) {
265                 int exclude_stk;
266                 struct dirent *de;
267                 char fullname[MAXPATHLEN + 1];
268                 memcpy(fullname, base, baselen);
269
270                 exclude_stk = push_exclude_per_directory(base, baselen);
271
272                 while ((de = readdir(dir)) != NULL) {
273                         int len;
274
275                         if ((de->d_name[0] == '.') &&
276                             (de->d_name[1] == 0 ||
277                              !strcmp(de->d_name + 1, ".") ||
278                              !strcmp(de->d_name + 1, "git")))
279                                 continue;
280                         len = strlen(de->d_name);
281                         memcpy(fullname + baselen, de->d_name, len+1);
282                         if (excluded(fullname) != show_ignored) {
283                                 if (!show_ignored || DTYPE(de) != DT_DIR) {
284                                         continue;
285                                 }
286                         }
287
288                         switch (DTYPE(de)) {
289                         struct stat st;
290                         default:
291                                 continue;
292                         case DT_UNKNOWN:
293                                 if (lstat(fullname, &st))
294                                         continue;
295                                 if (S_ISREG(st.st_mode) || S_ISLNK(st.st_mode))
296                                         break;
297                                 if (!S_ISDIR(st.st_mode))
298                                         continue;
299                                 /* fallthrough */
300                         case DT_DIR:
301                                 memcpy(fullname + baselen + len, "/", 2);
302                                 len++;
303                                 if (show_other_directories &&
304                                     !dir_exists(fullname, baselen + len))
305                                         break;
306                                 read_directory(fullname, fullname,
307                                                baselen + len);
308                                 continue;
309                         case DT_REG:
310                         case DT_LNK:
311                                 break;
312                         }
313                         add_name(fullname, baselen + len);
314                 }
315                 closedir(dir);
316
317                 pop_exclude_per_directory(exclude_stk);
318         }
319 }
320
321 static int cmp_name(const void *p1, const void *p2)
322 {
323         const struct nond_on_fs *e1 = *(const struct nond_on_fs **)p1;
324         const struct nond_on_fs *e2 = *(const struct nond_on_fs **)p2;
325
326         return cache_name_compare(e1->name, e1->len,
327                                   e2->name, e2->len);
328 }
329
330 /*
331  * Match a pathspec against a filename. The first "len" characters
332  * are the common prefix
333  */
334 static int match(const char **spec, char *ps_matched,
335                  const char *filename, int len)
336 {
337         const char *m;
338
339         while ((m = *spec++) != NULL) {
340                 int matchlen = strlen(m + len);
341
342                 if (!matchlen)
343                         goto matched;
344                 if (!strncmp(m + len, filename + len, matchlen)) {
345                         if (m[len + matchlen - 1] == '/')
346                                 goto matched;
347                         switch (filename[len + matchlen]) {
348                         case '/': case '\0':
349                                 goto matched;
350                         }
351                 }
352                 if (!fnmatch(m + len, filename + len, 0))
353                         goto matched;
354                 if (ps_matched)
355                         ps_matched++;
356                 continue;
357         matched:
358                 if (ps_matched)
359                         *ps_matched = 1;
360                 return 1;
361         }
362         return 0;
363 }
364
365 static void show_dir_entry(const char *tag, struct nond_on_fs *ent)
366 {
367         int len = prefix_len;
368         int offset = prefix_offset;
369
370         if (len >= ent->len)
371                 die("git-ls-files: internal error - directory entry not superset of prefix");
372
373         if (pathspec && !match(pathspec, ps_matched, ent->name, len))
374                 return;
375
376         fputs(tag, stdout);
377         write_name_quoted("", 0, ent->name + offset, line_terminator, stdout);
378         putchar(line_terminator);
379 }
380
381 static void show_other_files(void)
382 {
383         int i;
384         for (i = 0; i < nr_dir; i++) {
385                 /* We should not have a matching entry, but we
386                  * may have an unmerged entry for this path.
387                  */
388                 struct nond_on_fs *ent = dir[i];
389                 int pos = cache_name_pos(ent->name, ent->len);
390                 struct cache_entry *ce;
391                 if (0 <= pos)
392                         die("bug in show-other-files");
393                 pos = -pos - 1;
394                 if (pos < active_nr) { 
395                         ce = active_cache[pos];
396                         if (ce_namelen(ce) == ent->len &&
397                             !memcmp(ce->name, ent->name, ent->len))
398                                 continue; /* Yup, this one exists unmerged */
399                 }
400                 show_dir_entry(tag_other, ent);
401         }
402 }
403
404 static void show_killed_files(void)
405 {
406         int i;
407         for (i = 0; i < nr_dir; i++) {
408                 struct nond_on_fs *ent = dir[i];
409                 char *cp, *sp;
410                 int pos, len, killed = 0;
411
412                 for (cp = ent->name; cp - ent->name < ent->len; cp = sp + 1) {
413                         sp = strchr(cp, '/');
414                         if (!sp) {
415                                 /* If ent->name is prefix of an entry in the
416                                  * cache, it will be killed.
417                                  */
418                                 pos = cache_name_pos(ent->name, ent->len);
419                                 if (0 <= pos)
420                                         die("bug in show-killed-files");
421                                 pos = -pos - 1;
422                                 while (pos < active_nr &&
423                                        ce_stage(active_cache[pos]))
424                                         pos++; /* skip unmerged */
425                                 if (active_nr <= pos)
426                                         break;
427                                 /* pos points at a name immediately after
428                                  * ent->name in the cache.  Does it expect
429                                  * ent->name to be a directory?
430                                  */
431                                 len = ce_namelen(active_cache[pos]);
432                                 if ((ent->len < len) &&
433                                     !strncmp(active_cache[pos]->name,
434                                              ent->name, ent->len) &&
435                                     active_cache[pos]->name[ent->len] == '/')
436                                         killed = 1;
437                                 break;
438                         }
439                         if (0 <= cache_name_pos(ent->name, sp - ent->name)) {
440                                 /* If any of the leading directories in
441                                  * ent->name is registered in the cache,
442                                  * ent->name will be killed.
443                                  */
444                                 killed = 1;
445                                 break;
446                         }
447                 }
448                 if (killed)
449                         show_dir_entry(tag_killed, dir[i]);
450         }
451 }
452
453 static void show_ce_entry(const char *tag, struct cache_entry *ce)
454 {
455         int len = prefix_len;
456         int offset = prefix_offset;
457
458         if (len >= ce_namelen(ce))
459                 die("git-ls-files: internal error - cache entry not superset of prefix");
460
461         if (pathspec && !match(pathspec, ps_matched, ce->name, len))
462                 return;
463
464         if (tag && *tag && show_valid_bit &&
465             (ce->ce_flags & htons(CE_VALID))) {
466                 static char alttag[4];
467                 memcpy(alttag, tag, 3);
468                 if (isalpha(tag[0]))
469                         alttag[0] = tolower(tag[0]);
470                 else if (tag[0] == '?')
471                         alttag[0] = '!';
472                 else {
473                         alttag[0] = 'v';
474                         alttag[1] = tag[0];
475                         alttag[2] = ' ';
476                         alttag[3] = 0;
477                 }
478                 tag = alttag;
479         }
480
481         if (!show_stage) {
482                 fputs(tag, stdout);
483                 write_name_quoted("", 0, ce->name + offset,
484                                   line_terminator, stdout);
485                 putchar(line_terminator);
486         }
487         else {
488                 printf("%s%06o %s %d\t",
489                        tag,
490                        ntohl(ce->ce_mode),
491                        sha1_to_hex(ce->sha1),
492                        ce_stage(ce));
493                 write_name_quoted("", 0, ce->name + offset,
494                                   line_terminator, stdout);
495                 putchar(line_terminator);
496         }
497 }
498
499 static void show_files(void)
500 {
501         int i;
502
503         /* For cached/deleted files we don't need to even do the readdir */
504         if (show_others || show_killed) {
505                 const char *path = ".", *base = "";
506                 int baselen = prefix_len;
507
508                 if (baselen) {
509                         path = base = prefix;
510                         if (exclude_per_dir) {
511                                 char *p, *pp = xmalloc(baselen+1);
512                                 memcpy(pp, prefix, baselen+1);
513                                 p = pp;
514                                 while (1) {
515                                         char save = *p;
516                                         *p = 0;
517                                         push_exclude_per_directory(pp, p-pp);
518                                         *p++ = save;
519                                         if (!save)
520                                                 break;
521                                         p = strchr(p, '/');
522                                         if (p)
523                                                 p++;
524                                         else
525                                                 p = pp + baselen;
526                                 }
527                                 free(pp);
528                         }
529                 }
530                 read_directory(path, base, baselen);
531                 qsort(dir, nr_dir, sizeof(struct nond_on_fs *), cmp_name);
532                 if (show_others)
533                         show_other_files();
534                 if (show_killed)
535                         show_killed_files();
536         }
537         if (show_cached | show_stage) {
538                 for (i = 0; i < active_nr; i++) {
539                         struct cache_entry *ce = active_cache[i];
540                         if (excluded(ce->name) != show_ignored)
541                                 continue;
542                         if (show_unmerged && !ce_stage(ce))
543                                 continue;
544                         show_ce_entry(ce_stage(ce) ? tag_unmerged : tag_cached, ce);
545                 }
546         }
547         if (show_deleted | show_modified) {
548                 for (i = 0; i < active_nr; i++) {
549                         struct cache_entry *ce = active_cache[i];
550                         struct stat st;
551                         int err;
552                         if (excluded(ce->name) != show_ignored)
553                                 continue;
554                         err = lstat(ce->name, &st);
555                         if (show_deleted && err)
556                                 show_ce_entry(tag_removed, ce);
557                         if (show_modified && ce_modified(ce, &st, 0))
558                                 show_ce_entry(tag_modified, ce);
559                 }
560         }
561 }
562
563 /*
564  * Prune the index to only contain stuff starting with "prefix"
565  */
566 static void prune_cache(void)
567 {
568         int pos = cache_name_pos(prefix, prefix_len);
569         unsigned int first, last;
570
571         if (pos < 0)
572                 pos = -pos-1;
573         active_cache += pos;
574         active_nr -= pos;
575         first = 0;
576         last = active_nr;
577         while (last > first) {
578                 int next = (last + first) >> 1;
579                 struct cache_entry *ce = active_cache[next];
580                 if (!strncmp(ce->name, prefix, prefix_len)) {
581                         first = next+1;
582                         continue;
583                 }
584                 last = next;
585         }
586         active_nr = last;
587 }
588
589 static void verify_pathspec(void)
590 {
591         const char **p, *n, *prev;
592         char *real_prefix;
593         unsigned long max;
594
595         prev = NULL;
596         max = PATH_MAX;
597         for (p = pathspec; (n = *p) != NULL; p++) {
598                 int i, len = 0;
599                 for (i = 0; i < max; i++) {
600                         char c = n[i];
601                         if (prev && prev[i] != c)
602                                 break;
603                         if (!c || c == '*' || c == '?')
604                                 break;
605                         if (c == '/')
606                                 len = i+1;
607                 }
608                 prev = n;
609                 if (len < max) {
610                         max = len;
611                         if (!max)
612                                 break;
613                 }
614         }
615
616         if (prefix_offset > max || memcmp(prev, prefix, prefix_offset))
617                 die("git-ls-files: cannot generate relative filenames containing '..'");
618
619         real_prefix = NULL;
620         prefix_len = max;
621         if (max) {
622                 real_prefix = xmalloc(max + 1);
623                 memcpy(real_prefix, prev, max);
624                 real_prefix[max] = 0;
625         }
626         prefix = real_prefix;
627 }
628
629 static const char ls_files_usage[] =
630         "git-ls-files [-z] [-t] [-v] (--[cached|deleted|others|stage|unmerged|killed|modified])* "
631         "[ --ignored ] [--exclude=<pattern>] [--exclude-from=<file>] "
632         "[ --exclude-per-directory=<filename> ] [--full-name] [--] [<file>]*";
633
634 int main(int argc, const char **argv)
635 {
636         int i;
637         int exc_given = 0;
638
639         prefix = setup_git_directory();
640         if (prefix)
641                 prefix_offset = strlen(prefix);
642         git_config(git_default_config);
643
644         for (i = 1; i < argc; i++) {
645                 const char *arg = argv[i];
646
647                 if (!strcmp(arg, "--")) {
648                         i++;
649                         break;
650                 }
651                 if (!strcmp(arg, "-z")) {
652                         line_terminator = 0;
653                         continue;
654                 }
655                 if (!strcmp(arg, "-t") || !strcmp(arg, "-v")) {
656                         tag_cached = "H ";
657                         tag_unmerged = "M ";
658                         tag_removed = "R ";
659                         tag_modified = "C ";
660                         tag_other = "? ";
661                         tag_killed = "K ";
662                         if (arg[1] == 'v')
663                                 show_valid_bit = 1;
664                         continue;
665                 }
666                 if (!strcmp(arg, "-c") || !strcmp(arg, "--cached")) {
667                         show_cached = 1;
668                         continue;
669                 }
670                 if (!strcmp(arg, "-d") || !strcmp(arg, "--deleted")) {
671                         show_deleted = 1;
672                         continue;
673                 }
674                 if (!strcmp(arg, "-m") || !strcmp(arg, "--modified")) {
675                         show_modified = 1;
676                         continue;
677                 }
678                 if (!strcmp(arg, "-o") || !strcmp(arg, "--others")) {
679                         show_others = 1;
680                         continue;
681                 }
682                 if (!strcmp(arg, "-i") || !strcmp(arg, "--ignored")) {
683                         show_ignored = 1;
684                         continue;
685                 }
686                 if (!strcmp(arg, "-s") || !strcmp(arg, "--stage")) {
687                         show_stage = 1;
688                         continue;
689                 }
690                 if (!strcmp(arg, "-k") || !strcmp(arg, "--killed")) {
691                         show_killed = 1;
692                         continue;
693                 }
694                 if (!strcmp(arg, "--directory")) {
695                         show_other_directories = 1;
696                         continue;
697                 }
698                 if (!strcmp(arg, "-u") || !strcmp(arg, "--unmerged")) {
699                         /* There's no point in showing unmerged unless
700                          * you also show the stage information.
701                          */
702                         show_stage = 1;
703                         show_unmerged = 1;
704                         continue;
705                 }
706                 if (!strcmp(arg, "-x") && i+1 < argc) {
707                         exc_given = 1;
708                         add_exclude(argv[++i], "", 0, &exclude_list[EXC_CMDL]);
709                         continue;
710                 }
711                 if (!strncmp(arg, "--exclude=", 10)) {
712                         exc_given = 1;
713                         add_exclude(arg+10, "", 0, &exclude_list[EXC_CMDL]);
714                         continue;
715                 }
716                 if (!strcmp(arg, "-X") && i+1 < argc) {
717                         exc_given = 1;
718                         add_excludes_from_file(argv[++i]);
719                         continue;
720                 }
721                 if (!strncmp(arg, "--exclude-from=", 15)) {
722                         exc_given = 1;
723                         add_excludes_from_file(arg+15);
724                         continue;
725                 }
726                 if (!strncmp(arg, "--exclude-per-directory=", 24)) {
727                         exc_given = 1;
728                         exclude_per_dir = arg + 24;
729                         continue;
730                 }
731                 if (!strcmp(arg, "--full-name")) {
732                         prefix_offset = 0;
733                         continue;
734                 }
735                 if (!strcmp(arg, "--error-unmatch")) {
736                         error_unmatch = 1;
737                         continue;
738                 }
739                 if (*arg == '-')
740                         usage(ls_files_usage);
741                 break;
742         }
743
744         pathspec = get_pathspec(prefix, argv + i);
745
746         /* Verify that the pathspec matches the prefix */
747         if (pathspec)
748                 verify_pathspec();
749
750         /* Treat unmatching pathspec elements as errors */
751         if (pathspec && error_unmatch) {
752                 int num;
753                 for (num = 0; pathspec[num]; num++)
754                         ;
755                 ps_matched = xcalloc(1, num);
756         }
757
758         if (show_ignored && !exc_given) {
759                 fprintf(stderr, "%s: --ignored needs some exclude pattern\n",
760                         argv[0]);
761                 exit(1);
762         }
763
764         /* With no flags, we default to showing the cached files */
765         if (!(show_stage | show_deleted | show_others | show_unmerged |
766               show_killed | show_modified))
767                 show_cached = 1;
768
769         read_cache();
770         if (prefix)
771                 prune_cache();
772         show_files();
773
774         if (ps_matched) {
775                 /* We need to make sure all pathspec matched otherwise
776                  * it is an error.
777                  */
778                 int num, errors = 0;
779                 for (num = 0; pathspec[num]; num++) {
780                         if (ps_matched[num])
781                                 continue;
782                         error("pathspec '%s' did not match any.",
783                               pathspec[num] + prefix_offset);
784                         errors++;
785                 }
786                 return errors ? 1 : 0;
787         }
788
789         return 0;
790 }