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