Merge branch 'cb/skip-utf8-check-with-pcre1'
[git] / builtin / diff.c
1 /*
2  * Builtin "git diff"
3  *
4  * Copyright (c) 2006 Junio C Hamano
5  */
6 #define USE_THE_INDEX_COMPATIBILITY_MACROS
7 #include "cache.h"
8 #include "config.h"
9 #include "lockfile.h"
10 #include "color.h"
11 #include "commit.h"
12 #include "blob.h"
13 #include "tag.h"
14 #include "diff.h"
15 #include "diffcore.h"
16 #include "revision.h"
17 #include "log-tree.h"
18 #include "builtin.h"
19 #include "submodule.h"
20 #include "sha1-array.h"
21
22 #define DIFF_NO_INDEX_EXPLICIT 1
23 #define DIFF_NO_INDEX_IMPLICIT 2
24
25 static const char builtin_diff_usage[] =
26 "git diff [<options>] [<commit> [<commit>]] [--] [<path>...]";
27
28 static const char *blob_path(struct object_array_entry *entry)
29 {
30         return entry->path ? entry->path : entry->name;
31 }
32
33 static void stuff_change(struct diff_options *opt,
34                          unsigned old_mode, unsigned new_mode,
35                          const struct object_id *old_oid,
36                          const struct object_id *new_oid,
37                          int old_oid_valid,
38                          int new_oid_valid,
39                          const char *old_path,
40                          const char *new_path)
41 {
42         struct diff_filespec *one, *two;
43
44         if (!is_null_oid(old_oid) && !is_null_oid(new_oid) &&
45             oideq(old_oid, new_oid) && (old_mode == new_mode))
46                 return;
47
48         if (opt->flags.reverse_diff) {
49                 SWAP(old_mode, new_mode);
50                 SWAP(old_oid, new_oid);
51                 SWAP(old_path, new_path);
52         }
53
54         if (opt->prefix &&
55             (strncmp(old_path, opt->prefix, opt->prefix_length) ||
56              strncmp(new_path, opt->prefix, opt->prefix_length)))
57                 return;
58
59         one = alloc_filespec(old_path);
60         two = alloc_filespec(new_path);
61         fill_filespec(one, old_oid, old_oid_valid, old_mode);
62         fill_filespec(two, new_oid, new_oid_valid, new_mode);
63
64         diff_queue(&diff_queued_diff, one, two);
65 }
66
67 static int builtin_diff_b_f(struct rev_info *revs,
68                             int argc, const char **argv,
69                             struct object_array_entry **blob)
70 {
71         /* Blob vs file in the working tree*/
72         struct stat st;
73         const char *path;
74
75         if (argc > 1)
76                 usage(builtin_diff_usage);
77
78         GUARD_PATHSPEC(&revs->prune_data, PATHSPEC_FROMTOP | PATHSPEC_LITERAL);
79         path = revs->prune_data.items[0].match;
80
81         if (lstat(path, &st))
82                 die_errno(_("failed to stat '%s'"), path);
83         if (!(S_ISREG(st.st_mode) || S_ISLNK(st.st_mode)))
84                 die(_("'%s': not a regular file or symlink"), path);
85
86         diff_set_mnemonic_prefix(&revs->diffopt, "o/", "w/");
87
88         if (blob[0]->mode == S_IFINVALID)
89                 blob[0]->mode = canon_mode(st.st_mode);
90
91         stuff_change(&revs->diffopt,
92                      blob[0]->mode, canon_mode(st.st_mode),
93                      &blob[0]->item->oid, &null_oid,
94                      1, 0,
95                      blob[0]->path ? blob[0]->path : path,
96                      path);
97         diffcore_std(&revs->diffopt);
98         diff_flush(&revs->diffopt);
99         return 0;
100 }
101
102 static int builtin_diff_blobs(struct rev_info *revs,
103                               int argc, const char **argv,
104                               struct object_array_entry **blob)
105 {
106         const unsigned mode = canon_mode(S_IFREG | 0644);
107
108         if (argc > 1)
109                 usage(builtin_diff_usage);
110
111         if (blob[0]->mode == S_IFINVALID)
112                 blob[0]->mode = mode;
113
114         if (blob[1]->mode == S_IFINVALID)
115                 blob[1]->mode = mode;
116
117         stuff_change(&revs->diffopt,
118                      blob[0]->mode, blob[1]->mode,
119                      &blob[0]->item->oid, &blob[1]->item->oid,
120                      1, 1,
121                      blob_path(blob[0]), blob_path(blob[1]));
122         diffcore_std(&revs->diffopt);
123         diff_flush(&revs->diffopt);
124         return 0;
125 }
126
127 static int builtin_diff_index(struct rev_info *revs,
128                               int argc, const char **argv)
129 {
130         int cached = 0;
131         while (1 < argc) {
132                 const char *arg = argv[1];
133                 if (!strcmp(arg, "--cached") || !strcmp(arg, "--staged"))
134                         cached = 1;
135                 else
136                         usage(builtin_diff_usage);
137                 argv++; argc--;
138         }
139         /*
140          * Make sure there is one revision (i.e. pending object),
141          * and there is no revision filtering parameters.
142          */
143         if (revs->pending.nr != 1 ||
144             revs->max_count != -1 || revs->min_age != -1 ||
145             revs->max_age != -1)
146                 usage(builtin_diff_usage);
147         if (!cached) {
148                 setup_work_tree();
149                 if (read_cache_preload(&revs->diffopt.pathspec) < 0) {
150                         perror("read_cache_preload");
151                         return -1;
152                 }
153         } else if (read_cache() < 0) {
154                 perror("read_cache");
155                 return -1;
156         }
157         return run_diff_index(revs, cached);
158 }
159
160 static int builtin_diff_tree(struct rev_info *revs,
161                              int argc, const char **argv,
162                              struct object_array_entry *ent0,
163                              struct object_array_entry *ent1)
164 {
165         const struct object_id *(oid[2]);
166         int swap = 0;
167
168         if (argc > 1)
169                 usage(builtin_diff_usage);
170
171         /*
172          * We saw two trees, ent0 and ent1.  If ent1 is uninteresting,
173          * swap them.
174          */
175         if (ent1->item->flags & UNINTERESTING)
176                 swap = 1;
177         oid[swap] = &ent0->item->oid;
178         oid[1 - swap] = &ent1->item->oid;
179         diff_tree_oid(oid[0], oid[1], "", &revs->diffopt);
180         log_tree_diff_flush(revs);
181         return 0;
182 }
183
184 static int builtin_diff_combined(struct rev_info *revs,
185                                  int argc, const char **argv,
186                                  struct object_array_entry *ent,
187                                  int ents)
188 {
189         struct oid_array parents = OID_ARRAY_INIT;
190         int i;
191
192         if (argc > 1)
193                 usage(builtin_diff_usage);
194
195         if (!revs->dense_combined_merges && !revs->combine_merges)
196                 revs->dense_combined_merges = revs->combine_merges = 1;
197         for (i = 1; i < ents; i++)
198                 oid_array_append(&parents, &ent[i].item->oid);
199         diff_tree_combined(&ent[0].item->oid, &parents,
200                            revs->dense_combined_merges, revs);
201         oid_array_clear(&parents);
202         return 0;
203 }
204
205 static void refresh_index_quietly(void)
206 {
207         struct lock_file lock_file = LOCK_INIT;
208         int fd;
209
210         fd = hold_locked_index(&lock_file, 0);
211         if (fd < 0)
212                 return;
213         discard_cache();
214         read_cache();
215         refresh_cache(REFRESH_QUIET|REFRESH_UNMERGED);
216         repo_update_index_if_able(the_repository, &lock_file);
217 }
218
219 static int builtin_diff_files(struct rev_info *revs, int argc, const char **argv)
220 {
221         unsigned int options = 0;
222
223         while (1 < argc && argv[1][0] == '-') {
224                 if (!strcmp(argv[1], "--base"))
225                         revs->max_count = 1;
226                 else if (!strcmp(argv[1], "--ours"))
227                         revs->max_count = 2;
228                 else if (!strcmp(argv[1], "--theirs"))
229                         revs->max_count = 3;
230                 else if (!strcmp(argv[1], "-q"))
231                         options |= DIFF_SILENT_ON_REMOVED;
232                 else if (!strcmp(argv[1], "-h"))
233                         usage(builtin_diff_usage);
234                 else
235                         return error(_("invalid option: %s"), argv[1]);
236                 argv++; argc--;
237         }
238
239         /*
240          * "diff --base" should not combine merges because it was not
241          * asked to.  "diff -c" should not densify (if the user wants
242          * dense one, --cc can be explicitly asked for, or just rely
243          * on the default).
244          */
245         if (revs->max_count == -1 && !revs->combine_merges &&
246             (revs->diffopt.output_format & DIFF_FORMAT_PATCH))
247                 revs->combine_merges = revs->dense_combined_merges = 1;
248
249         setup_work_tree();
250         if (read_cache_preload(&revs->diffopt.pathspec) < 0) {
251                 perror("read_cache_preload");
252                 return -1;
253         }
254         return run_diff_files(revs, options);
255 }
256
257 int cmd_diff(int argc, const char **argv, const char *prefix)
258 {
259         int i;
260         struct rev_info rev;
261         struct object_array ent = OBJECT_ARRAY_INIT;
262         int blobs = 0, paths = 0;
263         struct object_array_entry *blob[2];
264         int nongit = 0, no_index = 0;
265         int result = 0;
266
267         /*
268          * We could get N tree-ish in the rev.pending_objects list.
269          * Also there could be M blobs there, and P pathspecs.
270          *
271          * N=0, M=0:
272          *      cache vs files (diff-files)
273          * N=0, M=2:
274          *      compare two random blobs.  P must be zero.
275          * N=0, M=1, P=1:
276          *      compare a blob with a working tree file.
277          *
278          * N=1, M=0:
279          *      tree vs cache (diff-index --cached)
280          *
281          * N=2, M=0:
282          *      tree vs tree (diff-tree)
283          *
284          * N=0, M=0, P=2:
285          *      compare two filesystem entities (aka --no-index).
286          *
287          * Other cases are errors.
288          */
289
290         /* Were we asked to do --no-index explicitly? */
291         for (i = 1; i < argc; i++) {
292                 if (!strcmp(argv[i], "--")) {
293                         i++;
294                         break;
295                 }
296                 if (!strcmp(argv[i], "--no-index"))
297                         no_index = DIFF_NO_INDEX_EXPLICIT;
298                 if (argv[i][0] != '-')
299                         break;
300         }
301
302         prefix = setup_git_directory_gently(&nongit);
303
304         if (!no_index) {
305                 /*
306                  * Treat git diff with at least one path outside of the
307                  * repo the same as if the command would have been executed
308                  * outside of a git repository.  In this case it behaves
309                  * the same way as "git diff --no-index <a> <b>", which acts
310                  * as a colourful "diff" replacement.
311                  */
312                 if (nongit || ((argc == i + 2) &&
313                                (!path_inside_repo(prefix, argv[i]) ||
314                                 !path_inside_repo(prefix, argv[i + 1]))))
315                         no_index = DIFF_NO_INDEX_IMPLICIT;
316         }
317
318         init_diff_ui_defaults();
319         git_config(git_diff_ui_config, NULL);
320         precompose_argv(argc, argv);
321
322         repo_init_revisions(the_repository, &rev, prefix);
323
324         /* Set up defaults that will apply to both no-index and regular diffs. */
325         rev.diffopt.stat_width = -1;
326         rev.diffopt.stat_graph_width = -1;
327         rev.diffopt.flags.allow_external = 1;
328         rev.diffopt.flags.allow_textconv = 1;
329
330         /* If this is a no-index diff, just run it and exit there. */
331         if (no_index)
332                 exit(diff_no_index(&rev, no_index == DIFF_NO_INDEX_IMPLICIT,
333                                    argc, argv));
334
335
336         /*
337          * Otherwise, we are doing the usual "git" diff; set up any
338          * further defaults that apply to regular diffs.
339          */
340         rev.diffopt.skip_stat_unmatch = !!diff_auto_refresh_index;
341
342         /*
343          * Default to intent-to-add entries invisible in the
344          * index. This makes them show up as new files in diff-files
345          * and not at all in diff-cached.
346          */
347         rev.diffopt.ita_invisible_in_index = 1;
348
349         if (nongit)
350                 die(_("Not a git repository"));
351         argc = setup_revisions(argc, argv, &rev, NULL);
352         if (!rev.diffopt.output_format) {
353                 rev.diffopt.output_format = DIFF_FORMAT_PATCH;
354                 diff_setup_done(&rev.diffopt);
355         }
356
357         rev.diffopt.flags.recursive = 1;
358
359         setup_diff_pager(&rev.diffopt);
360
361         /*
362          * Do we have --cached and not have a pending object, then
363          * default to HEAD by hand.  Eek.
364          */
365         if (!rev.pending.nr) {
366                 int i;
367                 for (i = 1; i < argc; i++) {
368                         const char *arg = argv[i];
369                         if (!strcmp(arg, "--"))
370                                 break;
371                         else if (!strcmp(arg, "--cached") ||
372                                  !strcmp(arg, "--staged")) {
373                                 add_head_to_pending(&rev);
374                                 if (!rev.pending.nr) {
375                                         struct tree *tree;
376                                         tree = lookup_tree(the_repository,
377                                                            the_repository->hash_algo->empty_tree);
378                                         add_pending_object(&rev, &tree->object, "HEAD");
379                                 }
380                                 break;
381                         }
382                 }
383         }
384
385         for (i = 0; i < rev.pending.nr; i++) {
386                 struct object_array_entry *entry = &rev.pending.objects[i];
387                 struct object *obj = entry->item;
388                 const char *name = entry->name;
389                 int flags = (obj->flags & UNINTERESTING);
390                 if (!obj->parsed)
391                         obj = parse_object(the_repository, &obj->oid);
392                 obj = deref_tag(the_repository, obj, NULL, 0);
393                 if (!obj)
394                         die(_("invalid object '%s' given."), name);
395                 if (obj->type == OBJ_COMMIT)
396                         obj = &get_commit_tree(((struct commit *)obj))->object;
397
398                 if (obj->type == OBJ_TREE) {
399                         obj->flags |= flags;
400                         add_object_array(obj, name, &ent);
401                 } else if (obj->type == OBJ_BLOB) {
402                         if (2 <= blobs)
403                                 die(_("more than two blobs given: '%s'"), name);
404                         blob[blobs] = entry;
405                         blobs++;
406
407                 } else {
408                         die(_("unhandled object '%s' given."), name);
409                 }
410         }
411         if (rev.prune_data.nr)
412                 paths += rev.prune_data.nr;
413
414         /*
415          * Now, do the arguments look reasonable?
416          */
417         if (!ent.nr) {
418                 switch (blobs) {
419                 case 0:
420                         result = builtin_diff_files(&rev, argc, argv);
421                         break;
422                 case 1:
423                         if (paths != 1)
424                                 usage(builtin_diff_usage);
425                         result = builtin_diff_b_f(&rev, argc, argv, blob);
426                         break;
427                 case 2:
428                         if (paths)
429                                 usage(builtin_diff_usage);
430                         result = builtin_diff_blobs(&rev, argc, argv, blob);
431                         break;
432                 default:
433                         usage(builtin_diff_usage);
434                 }
435         }
436         else if (blobs)
437                 usage(builtin_diff_usage);
438         else if (ent.nr == 1)
439                 result = builtin_diff_index(&rev, argc, argv);
440         else if (ent.nr == 2)
441                 result = builtin_diff_tree(&rev, argc, argv,
442                                            &ent.objects[0], &ent.objects[1]);
443         else if (ent.objects[0].item->flags & UNINTERESTING) {
444                 /*
445                  * diff A...B where there is at least one merge base
446                  * between A and B.  We have ent.objects[0] ==
447                  * merge-base, ent.objects[ents-2] == A, and
448                  * ent.objects[ents-1] == B.  Show diff between the
449                  * base and B.  Note that we pick one merge base at
450                  * random if there are more than one.
451                  */
452                 result = builtin_diff_tree(&rev, argc, argv,
453                                            &ent.objects[0],
454                                            &ent.objects[ent.nr-1]);
455         } else
456                 result = builtin_diff_combined(&rev, argc, argv,
457                                                ent.objects, ent.nr);
458         result = diff_result_code(&rev.diffopt, result);
459         if (1 < rev.diffopt.skip_stat_unmatch)
460                 refresh_index_quietly();
461         UNLEAK(rev);
462         UNLEAK(ent);
463         UNLEAK(blob);
464         return result;
465 }