Code

[PATCH] git-ls-files: generalized pathspecs
[git.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>
11 #include "cache.h"
13 static int show_deleted = 0;
14 static int show_cached = 0;
15 static int show_others = 0;
16 static int show_ignored = 0;
17 static int show_stage = 0;
18 static int show_unmerged = 0;
19 static int show_killed = 0;
20 static int line_terminator = '\n';
22 static int prefix_len = 0, prefix_offset = 0;
23 static const char *prefix = NULL;
24 static const char **pathspec = NULL;
26 static const char *tag_cached = "";
27 static const char *tag_unmerged = "";
28 static const char *tag_removed = "";
29 static const char *tag_other = "";
30 static const char *tag_killed = "";
32 static char *exclude_per_dir = NULL;
34 /* We maintain three exclude pattern lists:
35  * EXC_CMDL lists patterns explicitly given on the command line.
36  * EXC_DIRS lists patterns obtained from per-directory ignore files.
37  * EXC_FILE lists patterns from fallback ignore files.
38  */
39 #define EXC_CMDL 0
40 #define EXC_DIRS 1
41 #define EXC_FILE 2
42 static struct exclude_list {
43         int nr;
44         int alloc;
45         struct exclude {
46                 const char *pattern;
47                 const char *base;
48                 int baselen;
49         } **excludes;
50 } exclude_list[3];
52 static void add_exclude(const char *string, const char *base,
53                         int baselen, struct exclude_list *which)
54 {
55         struct exclude *x = xmalloc(sizeof (*x));
57         x->pattern = string;
58         x->base = base;
59         x->baselen = baselen;
60         if (which->nr == which->alloc) {
61                 which->alloc = alloc_nr(which->alloc);
62                 which->excludes = realloc(which->excludes,
63                                           which->alloc * sizeof(x));
64         }
65         which->excludes[which->nr++] = x;
66 }
68 static int add_excludes_from_file_1(const char *fname,
69                                     const char *base,
70                                     int baselen,
71                                     struct exclude_list *which)
72 {
73         int fd, i;
74         long size;
75         char *buf, *entry;
77         fd = open(fname, O_RDONLY);
78         if (fd < 0)
79                 goto err;
80         size = lseek(fd, 0, SEEK_END);
81         if (size < 0)
82                 goto err;
83         lseek(fd, 0, SEEK_SET);
84         if (size == 0) {
85                 close(fd);
86                 return 0;
87         }
88         buf = xmalloc(size);
89         if (read(fd, buf, size) != size)
90                 goto err;
91         close(fd);
93         entry = buf;
94         for (i = 0; i < size; i++) {
95                 if (buf[i] == '\n') {
96                         if (entry != buf + i && entry[0] != '#') {
97                                 buf[i] = 0;
98                                 add_exclude(entry, base, baselen, which);
99                         }
100                         entry = buf + i + 1;
101                 }
102         }
103         return 0;
105  err:
106         if (0 <= fd)
107                 close(fd);
108         return -1;
111 static void add_excludes_from_file(const char *fname)
113         if (add_excludes_from_file_1(fname, "", 0,
114                                      &exclude_list[EXC_FILE]) < 0)
115                 die("cannot use %s as an exclude file", fname);
118 static int push_exclude_per_directory(const char *base, int baselen)
120         char exclude_file[PATH_MAX];
121         struct exclude_list *el = &exclude_list[EXC_DIRS];
122         int current_nr = el->nr;
124         if (exclude_per_dir) {
125                 memcpy(exclude_file, base, baselen);
126                 strcpy(exclude_file + baselen, exclude_per_dir);
127                 add_excludes_from_file_1(exclude_file, base, baselen, el);
128         }
129         return current_nr;
132 static void pop_exclude_per_directory(int stk)
134         struct exclude_list *el = &exclude_list[EXC_DIRS];
136         while (stk < el->nr)
137                 free(el->excludes[--el->nr]);
140 /* Scan the list and let the last match determines the fate.
141  * Return 1 for exclude, 0 for include and -1 for undecided.
142  */
143 static int excluded_1(const char *pathname,
144                       int pathlen,
145                       struct exclude_list *el)
147         int i;
149         if (el->nr) {
150                 for (i = el->nr - 1; 0 <= i; i--) {
151                         struct exclude *x = el->excludes[i];
152                         const char *exclude = x->pattern;
153                         int to_exclude = 1;
155                         if (*exclude == '!') {
156                                 to_exclude = 0;
157                                 exclude++;
158                         }
160                         if (!strchr(exclude, '/')) {
161                                 /* match basename */
162                                 const char *basename = strrchr(pathname, '/');
163                                 basename = (basename) ? basename+1 : pathname;
164                                 if (fnmatch(exclude, basename, 0) == 0)
165                                         return to_exclude;
166                         }
167                         else {
168                                 /* match with FNM_PATHNAME:
169                                  * exclude has base (baselen long) inplicitly
170                                  * in front of it.
171                                  */
172                                 int baselen = x->baselen;
173                                 if (*exclude == '/')
174                                         exclude++;
176                                 if (pathlen < baselen ||
177                                     (baselen && pathname[baselen-1] != '/') ||
178                                     strncmp(pathname, x->base, baselen))
179                                     continue;
181                                 if (fnmatch(exclude, pathname+baselen,
182                                             FNM_PATHNAME) == 0)
183                                         return to_exclude;
184                         }
185                 }
186         }
187         return -1; /* undecided */
190 static int excluded(const char *pathname)
192         int pathlen = strlen(pathname);
193         int st;
195         for (st = EXC_CMDL; st <= EXC_FILE; st++) {
196                 switch (excluded_1(pathname, pathlen, &exclude_list[st])) {
197                 case 0:
198                         return 0;
199                 case 1:
200                         return 1;
201                 }
202         }
203         return 0;
206 struct nond_on_fs {
207         int len;
208         char name[0];
209 };
211 static struct nond_on_fs **dir;
212 static int nr_dir;
213 static int dir_alloc;
215 static void add_name(const char *pathname, int len)
217         struct nond_on_fs *ent;
219         if (cache_name_pos(pathname, len) >= 0)
220                 return;
222         if (nr_dir == dir_alloc) {
223                 dir_alloc = alloc_nr(dir_alloc);
224                 dir = xrealloc(dir, dir_alloc*sizeof(ent));
225         }
226         ent = xmalloc(sizeof(*ent) + len + 1);
227         ent->len = len;
228         memcpy(ent->name, pathname, len);
229         ent->name[len] = 0;
230         dir[nr_dir++] = ent;
233 /*
234  * Read a directory tree. We currently ignore anything but
235  * directories, regular files and symlinks. That's because git
236  * doesn't handle them at all yet. Maybe that will change some
237  * day.
238  *
239  * Also, we ignore the name ".git" (even if it is not a directory).
240  * That likely will not change.
241  */
242 static void read_directory(const char *path, const char *base, int baselen)
244         DIR *dir = opendir(path);
246         if (dir) {
247                 int exclude_stk;
248                 struct dirent *de;
249                 char fullname[MAXPATHLEN + 1];
250                 memcpy(fullname, base, baselen);
252                 exclude_stk = push_exclude_per_directory(base, baselen);
254                 while ((de = readdir(dir)) != NULL) {
255                         int len;
257                         if ((de->d_name[0] == '.') &&
258                             (de->d_name[1] == 0 ||
259                              !strcmp(de->d_name + 1, ".") ||
260                              !strcmp(de->d_name + 1, "git")))
261                                 continue;
262                         len = strlen(de->d_name);
263                         memcpy(fullname + baselen, de->d_name, len+1);
264                         if (excluded(fullname) != show_ignored)
265                                 continue;
267                         switch (DTYPE(de)) {
268                         struct stat st;
269                         default:
270                                 continue;
271                         case DT_UNKNOWN:
272                                 if (lstat(fullname, &st))
273                                         continue;
274                                 if (S_ISREG(st.st_mode) || S_ISLNK(st.st_mode))
275                                         break;
276                                 if (!S_ISDIR(st.st_mode))
277                                         continue;
278                                 /* fallthrough */
279                         case DT_DIR:
280                                 memcpy(fullname + baselen + len, "/", 2);
281                                 read_directory(fullname, fullname,
282                                                baselen + len + 1);
283                                 continue;
284                         case DT_REG:
285                         case DT_LNK:
286                                 break;
287                         }
288                         add_name(fullname, baselen + len);
289                 }
290                 closedir(dir);
292                 pop_exclude_per_directory(exclude_stk);
293         }
296 static int cmp_name(const void *p1, const void *p2)
298         const struct nond_on_fs *e1 = *(const struct nond_on_fs **)p1;
299         const struct nond_on_fs *e2 = *(const struct nond_on_fs **)p2;
301         return cache_name_compare(e1->name, e1->len,
302                                   e2->name, e2->len);
305 /*
306  * Match a pathspec against a filename. The first "len" characters
307  * are the common prefix
308  */
309 static int match(const char **spec, const char *filename, int len)
311         const char *m;
313         while ((m = *spec++) != NULL) {
314                 int matchlen = strlen(m + len);
316                 if (!matchlen)
317                         return 1;
318                 if (!strncmp(m + len, filename + len, matchlen)) {
319                         if (m[len + matchlen - 1] == '/')
320                                 return 1;
321                         switch (filename[len + matchlen]) {
322                         case '/': case '\0':
323                                 return 1;
324                         }
325                 }
326                 if (!fnmatch(m + len, filename + len, 0))
327                         return 1;
328         }
329         return 0;
332 static void show_dir_entry(const char *tag, struct nond_on_fs *ent)
334         int len = prefix_len;
335         int offset = prefix_offset;
337         if (len >= ent->len)
338                 die("git-ls-files: internal error - directory entry not superset of prefix");
340         if (pathspec && !match(pathspec, ent->name, len))
341                 return;
343         printf("%s%s%c", tag, ent->name + offset, line_terminator);
346 static void show_killed_files(void)
348         int i;
349         for (i = 0; i < nr_dir; i++) {
350                 struct nond_on_fs *ent = dir[i];
351                 char *cp, *sp;
352                 int pos, len, killed = 0;
354                 for (cp = ent->name; cp - ent->name < ent->len; cp = sp + 1) {
355                         sp = strchr(cp, '/');
356                         if (!sp) {
357                                 /* If ent->name is prefix of an entry in the
358                                  * cache, it will be killed.
359                                  */
360                                 pos = cache_name_pos(ent->name, ent->len);
361                                 if (0 <= pos)
362                                         die("bug in show-killed-files");
363                                 pos = -pos - 1;
364                                 while (pos < active_nr &&
365                                        ce_stage(active_cache[pos]))
366                                         pos++; /* skip unmerged */
367                                 if (active_nr <= pos)
368                                         break;
369                                 /* pos points at a name immediately after
370                                  * ent->name in the cache.  Does it expect
371                                  * ent->name to be a directory?
372                                  */
373                                 len = ce_namelen(active_cache[pos]);
374                                 if ((ent->len < len) &&
375                                     !strncmp(active_cache[pos]->name,
376                                              ent->name, ent->len) &&
377                                     active_cache[pos]->name[ent->len] == '/')
378                                         killed = 1;
379                                 break;
380                         }
381                         if (0 <= cache_name_pos(ent->name, sp - ent->name)) {
382                                 /* If any of the leading directories in
383                                  * ent->name is registered in the cache,
384                                  * ent->name will be killed.
385                                  */
386                                 killed = 1;
387                                 break;
388                         }
389                 }
390                 if (killed)
391                         show_dir_entry(tag_killed, dir[i]);
392         }
395 static void show_ce_entry(const char *tag, struct cache_entry *ce)
397         int len = prefix_len;
398         int offset = prefix_offset;
400         if (len >= ce_namelen(ce))
401                 die("git-ls-files: internal error - cache entry not superset of prefix");
403         if (pathspec && !match(pathspec, ce->name, len))
404                 return;
406         if (!show_stage)
407                 printf("%s%s%c", tag, ce->name + offset, line_terminator);
408         else
409                 printf("%s%06o %s %d\t%s%c",
410                        tag,
411                        ntohl(ce->ce_mode),
412                        sha1_to_hex(ce->sha1),
413                        ce_stage(ce),
414                        ce->name + offset, line_terminator); 
417 static void show_files(void)
419         int i;
421         /* For cached/deleted files we don't need to even do the readdir */
422         if (show_others || show_killed) {
423                 const char *path = ".", *base = "";
424                 int baselen = prefix_len;
426                 if (baselen)
427                         path = base = prefix;
428                 read_directory(path, base, baselen);
429                 qsort(dir, nr_dir, sizeof(struct nond_on_fs *), cmp_name);
430                 if (show_others)
431                         for (i = 0; i < nr_dir; i++)
432                                 show_dir_entry(tag_other, dir[i]);
433                 if (show_killed)
434                         show_killed_files();
435         }
436         if (show_cached | show_stage) {
437                 for (i = 0; i < active_nr; i++) {
438                         struct cache_entry *ce = active_cache[i];
439                         if (excluded(ce->name) != show_ignored)
440                                 continue;
441                         if (show_unmerged && !ce_stage(ce))
442                                 continue;
443                         show_ce_entry(ce_stage(ce) ? tag_unmerged : tag_cached, ce);
444                 }
445         }
446         if (show_deleted) {
447                 for (i = 0; i < active_nr; i++) {
448                         struct cache_entry *ce = active_cache[i];
449                         struct stat st;
450                         if (excluded(ce->name) != show_ignored)
451                                 continue;
452                         if (!lstat(ce->name, &st))
453                                 continue;
454                         show_ce_entry(tag_removed, ce);
455                 }
456         }
459 /*
460  * Prune the index to only contain stuff starting with "prefix"
461  */
462 static void prune_cache(void)
464         int pos = cache_name_pos(prefix, prefix_len);
465         unsigned int first, last;
467         if (pos < 0)
468                 pos = -pos-1;
469         active_cache += pos;
470         active_nr -= pos;
471         first = 0;
472         last = active_nr;
473         while (last > first) {
474                 int next = (last + first) >> 1;
475                 struct cache_entry *ce = active_cache[next];
476                 if (!strncmp(ce->name, prefix, prefix_len)) {
477                         first = next+1;
478                         continue;
479                 }
480                 last = next;
481         }
482         active_nr = last;
485 static void verify_pathspec(void)
487         const char **p, *n, *prev;
488         char *real_prefix;
489         unsigned long max;
491         prev = NULL;
492         max = PATH_MAX;
493         for (p = pathspec; (n = *p) != NULL; p++) {
494                 int i, len = 0;
495                 for (i = 0; i < max; i++) {
496                         char c = n[i];
497                         if (prev && prev[i] != c)
498                                 break;
499                         if (c == '*' || c == '?')
500                                 break;
501                         if (c == '/')
502                                 len = i+1;
503                 }
504                 prev = n;
505                 if (len < max) {
506                         max = len;
507                         if (!max)
508                                 break;
509                 }
510         }
512         if (prefix_offset > max || memcmp(prev, prefix, prefix_offset))
513                 die("git-ls-files: cannot generate relative filenames containing '..'");
515         real_prefix = NULL;
516         prefix_len = max;
517         if (max) {
518                 real_prefix = xmalloc(max + 1);
519                 memcpy(real_prefix, prev, max);
520                 real_prefix[max] = 0;
521         }
522         prefix = real_prefix;
525 static const char ls_files_usage[] =
526         "git-ls-files [-z] [-t] (--[cached|deleted|others|stage|unmerged|killed])* "
527         "[ --ignored ] [--exclude=<pattern>] [--exclude-from=<file>] "
528         "[ --exclude-per-directory=<filename> ]";
530 int main(int argc, char **argv)
532         int i;
533         int exc_given = 0;
535         prefix = setup_git_directory();
536         if (prefix)
537                 prefix_offset = strlen(prefix);
539         for (i = 1; i < argc; i++) {
540                 char *arg = argv[i];
542                 if (!strcmp(arg, "-z")) {
543                         line_terminator = 0;
544                         continue;
545                 }
546                 if (!strcmp(arg, "-t")) {
547                         tag_cached = "H ";
548                         tag_unmerged = "M ";
549                         tag_removed = "R ";
550                         tag_other = "? ";
551                         tag_killed = "K ";
552                         continue;
553                 }
554                 if (!strcmp(arg, "-c") || !strcmp(arg, "--cached")) {
555                         show_cached = 1;
556                         continue;
557                 }
558                 if (!strcmp(arg, "-d") || !strcmp(arg, "--deleted")) {
559                         show_deleted = 1;
560                         continue;
561                 }
562                 if (!strcmp(arg, "-o") || !strcmp(arg, "--others")) {
563                         show_others = 1;
564                         continue;
565                 }
566                 if (!strcmp(arg, "-i") || !strcmp(arg, "--ignored")) {
567                         show_ignored = 1;
568                         continue;
569                 }
570                 if (!strcmp(arg, "-s") || !strcmp(arg, "--stage")) {
571                         show_stage = 1;
572                         continue;
573                 }
574                 if (!strcmp(arg, "-k") || !strcmp(arg, "--killed")) {
575                         show_killed = 1;
576                         continue;
577                 }
578                 if (!strcmp(arg, "-u") || !strcmp(arg, "--unmerged")) {
579                         /* There's no point in showing unmerged unless
580                          * you also show the stage information.
581                          */
582                         show_stage = 1;
583                         show_unmerged = 1;
584                         continue;
585                 }
586                 if (!strcmp(arg, "-x") && i+1 < argc) {
587                         exc_given = 1;
588                         add_exclude(argv[++i], "", 0, &exclude_list[EXC_CMDL]);
589                         continue;
590                 }
591                 if (!strncmp(arg, "--exclude=", 10)) {
592                         exc_given = 1;
593                         add_exclude(arg+10, "", 0, &exclude_list[EXC_CMDL]);
594                         continue;
595                 }
596                 if (!strcmp(arg, "-X") && i+1 < argc) {
597                         exc_given = 1;
598                         add_excludes_from_file(argv[++i]);
599                         continue;
600                 }
601                 if (!strncmp(arg, "--exclude-from=", 15)) {
602                         exc_given = 1;
603                         add_excludes_from_file(arg+15);
604                         continue;
605                 }
606                 if (!strncmp(arg, "--exclude-per-directory=", 24)) {
607                         exc_given = 1;
608                         exclude_per_dir = arg + 24;
609                         continue;
610                 }
611                 if (!strcmp(arg, "--full-name")) {
612                         prefix_offset = 0;
613                         continue;
614                 }
615                 if (*arg == '-')
616                         usage(ls_files_usage);
617                 break;
618         }
620         pathspec = get_pathspec(prefix, argv + i);
622         /* Verify that the pathspec matches the prefix */
623         if (pathspec)
624                 verify_pathspec();
626         if (show_ignored && !exc_given) {
627                 fprintf(stderr, "%s: --ignored needs some exclude pattern\n",
628                         argv[0]);
629                 exit(1);
630         }
632         /* With no flags, we default to showing the cached files */
633         if (!(show_stage | show_deleted | show_others | show_unmerged | show_killed))
634                 show_cached = 1;
636         read_cache();
637         if (prefix)
638                 prune_cache();
639         show_files();
640         return 0;