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