Documentation: A few minor fixes to Git User's Manual
[git] / diff-lib.c
1 /*
2  * Copyright (C) 2005 Junio C Hamano
3  */
4 #include "cache.h"
5 #include "quote.h"
6 #include "commit.h"
7 #include "diff.h"
8 #include "diffcore.h"
9 #include "revision.h"
10 #include "cache-tree.h"
11 #include "path-list.h"
12
13 /*
14  * diff-files
15  */
16
17 static int read_directory(const char *path, struct path_list *list)
18 {
19         DIR *dir;
20         struct dirent *e;
21
22         if (!(dir = opendir(path)))
23                 return error("Could not open directory %s", path);
24
25         while ((e = readdir(dir)))
26                 if (strcmp(".", e->d_name) && strcmp("..", e->d_name))
27                         path_list_insert(xstrdup(e->d_name), list);
28
29         closedir(dir);
30         return 0;
31 }
32
33 static int get_mode(const char *path, int *mode)
34 {
35         struct stat st;
36
37         if (!path || !strcmp(path, "/dev/null"))
38                 *mode = 0;
39         else if (!strcmp(path, "-"))
40                 *mode = ntohl(create_ce_mode(0666));
41         else if (stat(path, &st))
42                 return error("Could not access '%s'", path);
43         else
44                 *mode = st.st_mode;
45         return 0;
46 }
47
48 static int queue_diff(struct diff_options *o,
49                 const char *name1, const char *name2)
50 {
51         int mode1 = 0, mode2 = 0;
52
53         if (get_mode(name1, &mode1) || get_mode(name2, &mode2))
54                 return -1;
55
56         if (mode1 && mode2 && S_ISDIR(mode1) != S_ISDIR(mode2))
57                 return error("file/directory conflict: %s, %s", name1, name2);
58
59         if (S_ISDIR(mode1) || S_ISDIR(mode2)) {
60                 char buffer1[PATH_MAX], buffer2[PATH_MAX];
61                 struct path_list p1 = {NULL, 0, 0, 1}, p2 = {NULL, 0, 0, 1};
62                 int len1 = 0, len2 = 0, i1, i2, ret = 0;
63
64                 if (name1 && read_directory(name1, &p1))
65                         return -1;
66                 if (name2 && read_directory(name2, &p2)) {
67                         path_list_clear(&p1, 0);
68                         return -1;
69                 }
70
71                 if (name1) {
72                         len1 = strlen(name1);
73                         if (len1 > 0 && name1[len1 - 1] == '/')
74                                 len1--;
75                         memcpy(buffer1, name1, len1);
76                         buffer1[len1++] = '/';
77                 }
78
79                 if (name2) {
80                         len2 = strlen(name2);
81                         if (len2 > 0 && name2[len2 - 1] == '/')
82                                 len2--;
83                         memcpy(buffer2, name2, len2);
84                         buffer2[len2++] = '/';
85                 }
86
87                 for (i1 = i2 = 0; !ret && (i1 < p1.nr || i2 < p2.nr); ) {
88                         const char *n1, *n2;
89                         int comp;
90
91                         if (i1 == p1.nr)
92                                 comp = 1;
93                         else if (i2 == p2.nr)
94                                 comp = -1;
95                         else
96                                 comp = strcmp(p1.items[i1].path,
97                                         p2.items[i2].path);
98
99                         if (comp > 0)
100                                 n1 = NULL;
101                         else {
102                                 n1 = buffer1;
103                                 strncpy(buffer1 + len1, p1.items[i1++].path,
104                                                 PATH_MAX - len1);
105                         }
106
107                         if (comp < 0)
108                                 n2 = NULL;
109                         else {
110                                 n2 = buffer2;
111                                 strncpy(buffer2 + len2, p2.items[i2++].path,
112                                                 PATH_MAX - len2);
113                         }
114
115                         ret = queue_diff(o, n1, n2);
116                 }
117                 path_list_clear(&p1, 0);
118                 path_list_clear(&p2, 0);
119
120                 return ret;
121         } else {
122                 struct diff_filespec *d1, *d2;
123
124                 if (o->reverse_diff) {
125                         unsigned tmp;
126                         const char *tmp_c;
127                         tmp = mode1; mode1 = mode2; mode2 = tmp;
128                         tmp_c = name1; name1 = name2; name2 = tmp_c;
129                 }
130
131                 if (!name1)
132                         name1 = "/dev/null";
133                 if (!name2)
134                         name2 = "/dev/null";
135                 d1 = alloc_filespec(name1);
136                 d2 = alloc_filespec(name2);
137                 fill_filespec(d1, null_sha1, mode1);
138                 fill_filespec(d2, null_sha1, mode2);
139
140                 diff_queue(&diff_queued_diff, d1, d2);
141                 return 0;
142         }
143 }
144
145 static int is_in_index(const char *path)
146 {
147         int len = strlen(path);
148         int pos = cache_name_pos(path, len);
149         char c;
150
151         if (pos < 0)
152                 return 0;
153         if (strncmp(active_cache[pos]->name, path, len))
154                 return 0;
155         c = active_cache[pos]->name[len];
156         return c == '\0' || c == '/';
157 }
158
159 static int handle_diff_files_args(struct rev_info *revs,
160                 int argc, const char **argv, int *silent)
161 {
162         *silent = 0;
163
164         /* revs->max_count == -2 means --no-index */
165         while (1 < argc && argv[1][0] == '-') {
166                 if (!strcmp(argv[1], "--base"))
167                         revs->max_count = 1;
168                 else if (!strcmp(argv[1], "--ours"))
169                         revs->max_count = 2;
170                 else if (!strcmp(argv[1], "--theirs"))
171                         revs->max_count = 3;
172                 else if (!strcmp(argv[1], "-n") ||
173                                 !strcmp(argv[1], "--no-index")) {
174                         revs->max_count = -2;
175                         revs->diffopt.exit_with_status = 1;
176                 }
177                 else if (!strcmp(argv[1], "-q"))
178                         *silent = 1;
179                 else
180                         return error("invalid option: %s", argv[1]);
181                 argv++; argc--;
182         }
183
184         if (revs->max_count == -1 && revs->diffopt.nr_paths == 2) {
185                 /*
186                  * If two files are specified, and at least one is untracked,
187                  * default to no-index.
188                  */
189                 read_cache();
190                 if (!is_in_index(revs->diffopt.paths[0]) ||
191                                         !is_in_index(revs->diffopt.paths[1]))
192                         revs->max_count = -2;
193         }
194
195         /*
196          * Make sure there are NO revision (i.e. pending object) parameter,
197          * rev.max_count is reasonable (0 <= n <= 3),
198          * there is no other revision filtering parameters.
199          */
200         if (revs->pending.nr || revs->max_count > 3 ||
201             revs->min_age != -1 || revs->max_age != -1)
202                 return error("no revision allowed with diff-files");
203
204         if (revs->max_count == -1 &&
205             (revs->diffopt.output_format & DIFF_FORMAT_PATCH))
206                 revs->combine_merges = revs->dense_combined_merges = 1;
207
208         return 0;
209 }
210
211 static int is_outside_repo(const char *path, int nongit, const char *prefix)
212 {
213         int i;
214         if (nongit || !strcmp(path, "-") || path[0] == '/')
215                 return 1;
216         if (prefixcmp(path, "../"))
217                 return 0;
218         if (!prefix)
219                 return 1;
220         for (i = strlen(prefix); !prefixcmp(path, "../"); ) {
221                 while (i > 0 && prefix[i - 1] != '/')
222                         i--;
223                 if (--i < 0)
224                         return 1;
225                 path += 3;
226         }
227         return 0;
228 }
229
230 int setup_diff_no_index(struct rev_info *revs,
231                 int argc, const char ** argv, int nongit, const char *prefix)
232 {
233         int i;
234         for (i = 1; i < argc; i++)
235                 if (argv[i][0] != '-' || argv[i][1] == '\0')
236                         break;
237                 else if (!strcmp(argv[i], "--")) {
238                         i++;
239                         break;
240                 } else if (i < argc - 3 && !strcmp(argv[i], "--no-index")) {
241                         i = argc - 3;
242                         revs->diffopt.exit_with_status = 1;
243                         break;
244                 }
245         if (argc != i + 2 || (!is_outside_repo(argv[i + 1], nongit, prefix) &&
246                                 !is_outside_repo(argv[i], nongit, prefix)))
247                 return -1;
248
249         diff_setup(&revs->diffopt);
250         for (i = 1; i < argc - 2; )
251                 if (!strcmp(argv[i], "--no-index"))
252                         i++;
253                 else {
254                         int j = diff_opt_parse(&revs->diffopt,
255                                         argv + i, argc - i);
256                         if (!j)
257                                 die("invalid diff option/value: %s", argv[i]);
258                         i += j;
259                 }
260
261         if (prefix) {
262                 int len = strlen(prefix);
263
264                 revs->diffopt.paths = xcalloc(2, sizeof(char*));
265                 for (i = 0; i < 2; i++) {
266                         const char *p = argv[argc - 2 + i];
267                         /*
268                          * stdin should be spelled as '-'; if you have
269                          * path that is '-', spell it as ./-.
270                          */
271                         p = (strcmp(p, "-")
272                              ? xstrdup(prefix_filename(prefix, len, p))
273                              : p);
274                         revs->diffopt.paths[i] = p;
275                 }
276         }
277         else
278                 revs->diffopt.paths = argv + argc - 2;
279         revs->diffopt.nr_paths = 2;
280         revs->max_count = -2;
281         return 0;
282 }
283
284 int run_diff_files_cmd(struct rev_info *revs, int argc, const char **argv)
285 {
286         int silent_on_removed;
287
288         if (handle_diff_files_args(revs, argc, argv, &silent_on_removed))
289                 return -1;
290
291         if (revs->max_count == -2) {
292                 if (revs->diffopt.nr_paths != 2)
293                         return error("need two files/directories with --no-index");
294                 if (queue_diff(&revs->diffopt, revs->diffopt.paths[0],
295                                 revs->diffopt.paths[1]))
296                         return -1;
297                 diffcore_std(&revs->diffopt);
298                 diff_flush(&revs->diffopt);
299                 /*
300                  * The return code for --no-index imitates diff(1):
301                  * 0 = no changes, 1 = changes, else error
302                  */
303                 return revs->diffopt.found_changes;
304         }
305
306         if (read_cache() < 0) {
307                 perror("read_cache");
308                 return -1;
309         }
310         return run_diff_files(revs, silent_on_removed);
311 }
312
313 int run_diff_files(struct rev_info *revs, int silent_on_removed)
314 {
315         int entries, i;
316         int diff_unmerged_stage = revs->max_count;
317
318         if (diff_unmerged_stage < 0)
319                 diff_unmerged_stage = 2;
320         entries = active_nr;
321         for (i = 0; i < entries; i++) {
322                 struct stat st;
323                 unsigned int oldmode, newmode;
324                 struct cache_entry *ce = active_cache[i];
325                 int changed;
326
327                 if (revs->diffopt.quiet && revs->diffopt.has_changes)
328                         break;
329
330                 if (!ce_path_match(ce, revs->prune_data))
331                         continue;
332
333                 if (ce_stage(ce)) {
334                         struct combine_diff_path *dpath;
335                         int num_compare_stages = 0;
336                         size_t path_len;
337
338                         path_len = ce_namelen(ce);
339
340                         dpath = xmalloc(combine_diff_path_size(5, path_len));
341                         dpath->path = (char *) &(dpath->parent[5]);
342
343                         dpath->next = NULL;
344                         dpath->len = path_len;
345                         memcpy(dpath->path, ce->name, path_len);
346                         dpath->path[path_len] = '\0';
347                         hashclr(dpath->sha1);
348                         memset(&(dpath->parent[0]), 0,
349                                sizeof(struct combine_diff_parent)*5);
350
351                         if (lstat(ce->name, &st) < 0) {
352                                 if (errno != ENOENT && errno != ENOTDIR) {
353                                         perror(ce->name);
354                                         continue;
355                                 }
356                                 if (silent_on_removed)
357                                         continue;
358                         }
359                         else
360                                 dpath->mode = canon_mode(st.st_mode);
361
362                         while (i < entries) {
363                                 struct cache_entry *nce = active_cache[i];
364                                 int stage;
365
366                                 if (strcmp(ce->name, nce->name))
367                                         break;
368
369                                 /* Stage #2 (ours) is the first parent,
370                                  * stage #3 (theirs) is the second.
371                                  */
372                                 stage = ce_stage(nce);
373                                 if (2 <= stage) {
374                                         int mode = ntohl(nce->ce_mode);
375                                         num_compare_stages++;
376                                         hashcpy(dpath->parent[stage-2].sha1, nce->sha1);
377                                         dpath->parent[stage-2].mode =
378                                                 canon_mode(mode);
379                                         dpath->parent[stage-2].status =
380                                                 DIFF_STATUS_MODIFIED;
381                                 }
382
383                                 /* diff against the proper unmerged stage */
384                                 if (stage == diff_unmerged_stage)
385                                         ce = nce;
386                                 i++;
387                         }
388                         /*
389                          * Compensate for loop update
390                          */
391                         i--;
392
393                         if (revs->combine_merges && num_compare_stages == 2) {
394                                 show_combined_diff(dpath, 2,
395                                                    revs->dense_combined_merges,
396                                                    revs);
397                                 free(dpath);
398                                 continue;
399                         }
400                         free(dpath);
401                         dpath = NULL;
402
403                         /*
404                          * Show the diff for the 'ce' if we found the one
405                          * from the desired stage.
406                          */
407                         diff_unmerge(&revs->diffopt, ce->name, 0, null_sha1);
408                         if (ce_stage(ce) != diff_unmerged_stage)
409                                 continue;
410                 }
411
412                 if (lstat(ce->name, &st) < 0) {
413                         if (errno != ENOENT && errno != ENOTDIR) {
414                                 perror(ce->name);
415                                 continue;
416                         }
417                         if (silent_on_removed)
418                                 continue;
419                         diff_addremove(&revs->diffopt, '-', ntohl(ce->ce_mode),
420                                        ce->sha1, ce->name, NULL);
421                         continue;
422                 }
423                 changed = ce_match_stat(ce, &st, 0);
424                 if (!changed && !revs->diffopt.find_copies_harder)
425                         continue;
426                 oldmode = ntohl(ce->ce_mode);
427
428                 newmode = canon_mode(st.st_mode);
429                 if (!trust_executable_bit &&
430                     S_ISREG(newmode) && S_ISREG(oldmode) &&
431                     ((newmode ^ oldmode) == 0111))
432                         newmode = oldmode;
433                 else if (!has_symlinks &&
434                     S_ISREG(newmode) && S_ISLNK(oldmode))
435                         newmode = oldmode;
436                 diff_change(&revs->diffopt, oldmode, newmode,
437                             ce->sha1, (changed ? null_sha1 : ce->sha1),
438                             ce->name, NULL);
439
440         }
441         diffcore_std(&revs->diffopt);
442         diff_flush(&revs->diffopt);
443         return 0;
444 }
445
446 /*
447  * diff-index
448  */
449
450 /* A file entry went away or appeared */
451 static void diff_index_show_file(struct rev_info *revs,
452                                  const char *prefix,
453                                  struct cache_entry *ce,
454                                  unsigned char *sha1, unsigned int mode)
455 {
456         diff_addremove(&revs->diffopt, prefix[0], ntohl(mode),
457                        sha1, ce->name, NULL);
458 }
459
460 static int get_stat_data(struct cache_entry *ce,
461                          unsigned char **sha1p,
462                          unsigned int *modep,
463                          int cached, int match_missing)
464 {
465         unsigned char *sha1 = ce->sha1;
466         unsigned int mode = ce->ce_mode;
467
468         if (!cached) {
469                 static unsigned char no_sha1[20];
470                 int changed;
471                 struct stat st;
472                 if (lstat(ce->name, &st) < 0) {
473                         if (errno == ENOENT && match_missing) {
474                                 *sha1p = sha1;
475                                 *modep = mode;
476                                 return 0;
477                         }
478                         return -1;
479                 }
480                 changed = ce_match_stat(ce, &st, 0);
481                 if (changed) {
482                         mode = ce_mode_from_stat(ce, st.st_mode);
483                         sha1 = no_sha1;
484                 }
485         }
486
487         *sha1p = sha1;
488         *modep = mode;
489         return 0;
490 }
491
492 static void show_new_file(struct rev_info *revs,
493                           struct cache_entry *new,
494                           int cached, int match_missing)
495 {
496         unsigned char *sha1;
497         unsigned int mode;
498
499         /* New file in the index: it might actually be different in
500          * the working copy.
501          */
502         if (get_stat_data(new, &sha1, &mode, cached, match_missing) < 0)
503                 return;
504
505         diff_index_show_file(revs, "+", new, sha1, mode);
506 }
507
508 static int show_modified(struct rev_info *revs,
509                          struct cache_entry *old,
510                          struct cache_entry *new,
511                          int report_missing,
512                          int cached, int match_missing)
513 {
514         unsigned int mode, oldmode;
515         unsigned char *sha1;
516
517         if (get_stat_data(new, &sha1, &mode, cached, match_missing) < 0) {
518                 if (report_missing)
519                         diff_index_show_file(revs, "-", old,
520                                              old->sha1, old->ce_mode);
521                 return -1;
522         }
523
524         if (revs->combine_merges && !cached &&
525             (hashcmp(sha1, old->sha1) || hashcmp(old->sha1, new->sha1))) {
526                 struct combine_diff_path *p;
527                 int pathlen = ce_namelen(new);
528
529                 p = xmalloc(combine_diff_path_size(2, pathlen));
530                 p->path = (char *) &p->parent[2];
531                 p->next = NULL;
532                 p->len = pathlen;
533                 memcpy(p->path, new->name, pathlen);
534                 p->path[pathlen] = 0;
535                 p->mode = ntohl(mode);
536                 hashclr(p->sha1);
537                 memset(p->parent, 0, 2 * sizeof(struct combine_diff_parent));
538                 p->parent[0].status = DIFF_STATUS_MODIFIED;
539                 p->parent[0].mode = ntohl(new->ce_mode);
540                 hashcpy(p->parent[0].sha1, new->sha1);
541                 p->parent[1].status = DIFF_STATUS_MODIFIED;
542                 p->parent[1].mode = ntohl(old->ce_mode);
543                 hashcpy(p->parent[1].sha1, old->sha1);
544                 show_combined_diff(p, 2, revs->dense_combined_merges, revs);
545                 free(p);
546                 return 0;
547         }
548
549         oldmode = old->ce_mode;
550         if (mode == oldmode && !hashcmp(sha1, old->sha1) &&
551             !revs->diffopt.find_copies_harder)
552                 return 0;
553
554         mode = ntohl(mode);
555         oldmode = ntohl(oldmode);
556
557         diff_change(&revs->diffopt, oldmode, mode,
558                     old->sha1, sha1, old->name, NULL);
559         return 0;
560 }
561
562 static int diff_cache(struct rev_info *revs,
563                       struct cache_entry **ac, int entries,
564                       const char **pathspec,
565                       int cached, int match_missing)
566 {
567         while (entries) {
568                 struct cache_entry *ce = *ac;
569                 int same = (entries > 1) && ce_same_name(ce, ac[1]);
570
571                 if (revs->diffopt.quiet && revs->diffopt.has_changes)
572                         break;
573
574                 if (!ce_path_match(ce, pathspec))
575                         goto skip_entry;
576
577                 switch (ce_stage(ce)) {
578                 case 0:
579                         /* No stage 1 entry? That means it's a new file */
580                         if (!same) {
581                                 show_new_file(revs, ce, cached, match_missing);
582                                 break;
583                         }
584                         /* Show difference between old and new */
585                         show_modified(revs, ac[1], ce, 1,
586                                       cached, match_missing);
587                         break;
588                 case 1:
589                         /* No stage 3 (merge) entry?
590                          * That means it's been deleted.
591                          */
592                         if (!same) {
593                                 diff_index_show_file(revs, "-", ce,
594                                                      ce->sha1, ce->ce_mode);
595                                 break;
596                         }
597                         /* We come here with ce pointing at stage 1
598                          * (original tree) and ac[1] pointing at stage
599                          * 3 (unmerged).  show-modified with
600                          * report-missing set to false does not say the
601                          * file is deleted but reports true if work
602                          * tree does not have it, in which case we
603                          * fall through to report the unmerged state.
604                          * Otherwise, we show the differences between
605                          * the original tree and the work tree.
606                          */
607                         if (!cached &&
608                             !show_modified(revs, ce, ac[1], 0,
609                                            cached, match_missing))
610                                 break;
611                         diff_unmerge(&revs->diffopt, ce->name,
612                                      ntohl(ce->ce_mode), ce->sha1);
613                         break;
614                 case 3:
615                         diff_unmerge(&revs->diffopt, ce->name,
616                                      0, null_sha1);
617                         break;
618
619                 default:
620                         die("impossible cache entry stage");
621                 }
622
623 skip_entry:
624                 /*
625                  * Ignore all the different stages for this file,
626                  * we've handled the relevant cases now.
627                  */
628                 do {
629                         ac++;
630                         entries--;
631                 } while (entries && ce_same_name(ce, ac[0]));
632         }
633         return 0;
634 }
635
636 /*
637  * This turns all merge entries into "stage 3". That guarantees that
638  * when we read in the new tree (into "stage 1"), we won't lose sight
639  * of the fact that we had unmerged entries.
640  */
641 static void mark_merge_entries(void)
642 {
643         int i;
644         for (i = 0; i < active_nr; i++) {
645                 struct cache_entry *ce = active_cache[i];
646                 if (!ce_stage(ce))
647                         continue;
648                 ce->ce_flags |= htons(CE_STAGEMASK);
649         }
650 }
651
652 int run_diff_index(struct rev_info *revs, int cached)
653 {
654         int ret;
655         struct object *ent;
656         struct tree *tree;
657         const char *tree_name;
658         int match_missing = 0;
659
660         /* 
661          * Backward compatibility wart - "diff-index -m" does
662          * not mean "do not ignore merges", but totally different.
663          */
664         if (!revs->ignore_merges)
665                 match_missing = 1;
666
667         mark_merge_entries();
668
669         ent = revs->pending.objects[0].item;
670         tree_name = revs->pending.objects[0].name;
671         tree = parse_tree_indirect(ent->sha1);
672         if (!tree)
673                 return error("bad tree object %s", tree_name);
674         if (read_tree(tree, 1, revs->prune_data))
675                 return error("unable to read tree object %s", tree_name);
676         ret = diff_cache(revs, active_cache, active_nr, revs->prune_data,
677                          cached, match_missing);
678         diffcore_std(&revs->diffopt);
679         diff_flush(&revs->diffopt);
680         return ret;
681 }
682
683 int do_diff_cache(const unsigned char *tree_sha1, struct diff_options *opt)
684 {
685         struct tree *tree;
686         struct rev_info revs;
687         int i;
688         struct cache_entry **dst;
689         struct cache_entry *last = NULL;
690
691         /*
692          * This is used by git-blame to run diff-cache internally;
693          * it potentially needs to repeatedly run this, so we will
694          * start by removing the higher order entries the last round
695          * left behind.
696          */
697         dst = active_cache;
698         for (i = 0; i < active_nr; i++) {
699                 struct cache_entry *ce = active_cache[i];
700                 if (ce_stage(ce)) {
701                         if (last && !strcmp(ce->name, last->name))
702                                 continue;
703                         cache_tree_invalidate_path(active_cache_tree,
704                                                    ce->name);
705                         last = ce;
706                         ce->ce_mode = 0;
707                         ce->ce_flags &= ~htons(CE_STAGEMASK);
708                 }
709                 *dst++ = ce;
710         }
711         active_nr = dst - active_cache;
712
713         init_revisions(&revs, NULL);
714         revs.prune_data = opt->paths;
715         tree = parse_tree_indirect(tree_sha1);
716         if (!tree)
717                 die("bad tree object %s", sha1_to_hex(tree_sha1));
718         if (read_tree(tree, 1, opt->paths))
719                 return error("unable to read tree %s", sha1_to_hex(tree_sha1));
720         return diff_cache(&revs, active_cache, active_nr, revs.prune_data,
721                           1, 0);
722 }