Merge branch 'maint'
[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 "unpack-trees.h"
12 #include "refs.h"
13 #include "submodule.h"
14
15 /*
16  * diff-files
17  */
18
19 /*
20  * Has the work tree entity been removed?
21  *
22  * Return 1 if it was removed from the work tree, 0 if an entity to be
23  * compared with the cache entry ce still exists (the latter includes
24  * the case where a directory that is not a submodule repository
25  * exists for ce that is a submodule -- it is a submodule that is not
26  * checked out).  Return negative for an error.
27  */
28 static int check_removed(const struct cache_entry *ce, struct stat *st)
29 {
30         if (lstat(ce->name, st) < 0) {
31                 if (errno != ENOENT && errno != ENOTDIR)
32                         return -1;
33                 return 1;
34         }
35         if (has_symlink_leading_path(ce->name, ce_namelen(ce)))
36                 return 1;
37         if (S_ISDIR(st->st_mode)) {
38                 unsigned char sub[20];
39
40                 /*
41                  * If ce is already a gitlink, we can have a plain
42                  * directory (i.e. the submodule is not checked out),
43                  * or a checked out submodule.  Either case this is not
44                  * a case where something was removed from the work tree,
45                  * so we will return 0.
46                  *
47                  * Otherwise, if the directory is not a submodule
48                  * repository, that means ce which was a blob turned into
49                  * a directory --- the blob was removed!
50                  */
51                 if (!S_ISGITLINK(ce->ce_mode) &&
52                     resolve_gitlink_ref(ce->name, "HEAD", sub))
53                         return 1;
54         }
55         return 0;
56 }
57
58 /*
59  * Has a file changed or has a submodule new commits or a dirty work tree?
60  *
61  * Return 1 when changes are detected, 0 otherwise. If the DIRTY_SUBMODULES
62  * option is set, the caller does not only want to know if a submodule is
63  * modified at all but wants to know all the conditions that are met (new
64  * commits, untracked content and/or modified content).
65  */
66 static int match_stat_with_submodule(struct diff_options *diffopt,
67                                       struct cache_entry *ce, struct stat *st,
68                                       unsigned ce_option, unsigned *dirty_submodule)
69 {
70         int changed = ce_match_stat(ce, st, ce_option);
71         if (S_ISGITLINK(ce->ce_mode)
72             && !DIFF_OPT_TST(diffopt, IGNORE_SUBMODULES)
73             && !DIFF_OPT_TST(diffopt, IGNORE_DIRTY_SUBMODULES)
74             && (!changed || DIFF_OPT_TST(diffopt, DIRTY_SUBMODULES))) {
75                 *dirty_submodule = is_submodule_modified(ce->name, DIFF_OPT_TST(diffopt, IGNORE_UNTRACKED_IN_SUBMODULES));
76         }
77         return changed;
78 }
79
80 int run_diff_files(struct rev_info *revs, unsigned int option)
81 {
82         int entries, i;
83         int diff_unmerged_stage = revs->max_count;
84         int silent_on_removed = option & DIFF_SILENT_ON_REMOVED;
85         unsigned ce_option = ((option & DIFF_RACY_IS_MODIFIED)
86                               ? CE_MATCH_RACY_IS_DIRTY : 0);
87
88         diff_set_mnemonic_prefix(&revs->diffopt, "i/", "w/");
89
90         if (diff_unmerged_stage < 0)
91                 diff_unmerged_stage = 2;
92         entries = active_nr;
93         for (i = 0; i < entries; i++) {
94                 struct stat st;
95                 unsigned int oldmode, newmode;
96                 struct cache_entry *ce = active_cache[i];
97                 int changed;
98                 unsigned dirty_submodule = 0;
99
100                 if (DIFF_OPT_TST(&revs->diffopt, QUICK) &&
101                         DIFF_OPT_TST(&revs->diffopt, HAS_CHANGES))
102                         break;
103
104                 if (!ce_path_match(ce, revs->prune_data))
105                         continue;
106
107                 if (ce_stage(ce)) {
108                         struct combine_diff_path *dpath;
109                         int num_compare_stages = 0;
110                         size_t path_len;
111
112                         path_len = ce_namelen(ce);
113
114                         dpath = xmalloc(combine_diff_path_size(5, path_len));
115                         dpath->path = (char *) &(dpath->parent[5]);
116
117                         dpath->next = NULL;
118                         dpath->len = path_len;
119                         memcpy(dpath->path, ce->name, path_len);
120                         dpath->path[path_len] = '\0';
121                         hashclr(dpath->sha1);
122                         memset(&(dpath->parent[0]), 0,
123                                sizeof(struct combine_diff_parent)*5);
124
125                         changed = check_removed(ce, &st);
126                         if (!changed)
127                                 dpath->mode = ce_mode_from_stat(ce, st.st_mode);
128                         else {
129                                 if (changed < 0) {
130                                         perror(ce->name);
131                                         continue;
132                                 }
133                                 if (silent_on_removed)
134                                         continue;
135                         }
136
137                         while (i < entries) {
138                                 struct cache_entry *nce = active_cache[i];
139                                 int stage;
140
141                                 if (strcmp(ce->name, nce->name))
142                                         break;
143
144                                 /* Stage #2 (ours) is the first parent,
145                                  * stage #3 (theirs) is the second.
146                                  */
147                                 stage = ce_stage(nce);
148                                 if (2 <= stage) {
149                                         int mode = nce->ce_mode;
150                                         num_compare_stages++;
151                                         hashcpy(dpath->parent[stage-2].sha1, nce->sha1);
152                                         dpath->parent[stage-2].mode = ce_mode_from_stat(nce, mode);
153                                         dpath->parent[stage-2].status =
154                                                 DIFF_STATUS_MODIFIED;
155                                 }
156
157                                 /* diff against the proper unmerged stage */
158                                 if (stage == diff_unmerged_stage)
159                                         ce = nce;
160                                 i++;
161                         }
162                         /*
163                          * Compensate for loop update
164                          */
165                         i--;
166
167                         if (revs->combine_merges && num_compare_stages == 2) {
168                                 show_combined_diff(dpath, 2,
169                                                    revs->dense_combined_merges,
170                                                    revs);
171                                 free(dpath);
172                                 continue;
173                         }
174                         free(dpath);
175                         dpath = NULL;
176
177                         /*
178                          * Show the diff for the 'ce' if we found the one
179                          * from the desired stage.
180                          */
181                         diff_unmerge(&revs->diffopt, ce->name, 0, null_sha1);
182                         if (ce_stage(ce) != diff_unmerged_stage)
183                                 continue;
184                 }
185
186                 if (ce_uptodate(ce) || ce_skip_worktree(ce))
187                         continue;
188
189                 /* If CE_VALID is set, don't look at workdir for file removal */
190                 changed = (ce->ce_flags & CE_VALID) ? 0 : check_removed(ce, &st);
191                 if (changed) {
192                         if (changed < 0) {
193                                 perror(ce->name);
194                                 continue;
195                         }
196                         if (silent_on_removed)
197                                 continue;
198                         diff_addremove(&revs->diffopt, '-', ce->ce_mode,
199                                        ce->sha1, ce->name, 0);
200                         continue;
201                 }
202                 changed = match_stat_with_submodule(&revs->diffopt, ce, &st,
203                                                     ce_option, &dirty_submodule);
204                 if (!changed && !dirty_submodule) {
205                         ce_mark_uptodate(ce);
206                         if (!DIFF_OPT_TST(&revs->diffopt, FIND_COPIES_HARDER))
207                                 continue;
208                 }
209                 oldmode = ce->ce_mode;
210                 newmode = ce_mode_from_stat(ce, st.st_mode);
211                 diff_change(&revs->diffopt, oldmode, newmode,
212                             ce->sha1, (changed ? null_sha1 : ce->sha1),
213                             ce->name, 0, dirty_submodule);
214
215         }
216         diffcore_std(&revs->diffopt);
217         diff_flush(&revs->diffopt);
218         return 0;
219 }
220
221 /*
222  * diff-index
223  */
224
225 /* A file entry went away or appeared */
226 static void diff_index_show_file(struct rev_info *revs,
227                                  const char *prefix,
228                                  struct cache_entry *ce,
229                                  const unsigned char *sha1, unsigned int mode,
230                                  unsigned dirty_submodule)
231 {
232         diff_addremove(&revs->diffopt, prefix[0], mode,
233                        sha1, ce->name, dirty_submodule);
234 }
235
236 static int get_stat_data(struct cache_entry *ce,
237                          const unsigned char **sha1p,
238                          unsigned int *modep,
239                          int cached, int match_missing,
240                          unsigned *dirty_submodule, struct diff_options *diffopt)
241 {
242         const unsigned char *sha1 = ce->sha1;
243         unsigned int mode = ce->ce_mode;
244
245         if (!cached && !ce_uptodate(ce)) {
246                 int changed;
247                 struct stat st;
248                 changed = check_removed(ce, &st);
249                 if (changed < 0)
250                         return -1;
251                 else if (changed) {
252                         if (match_missing) {
253                                 *sha1p = sha1;
254                                 *modep = mode;
255                                 return 0;
256                         }
257                         return -1;
258                 }
259                 changed = match_stat_with_submodule(diffopt, ce, &st,
260                                                     0, dirty_submodule);
261                 if (changed) {
262                         mode = ce_mode_from_stat(ce, st.st_mode);
263                         sha1 = null_sha1;
264                 }
265         }
266
267         *sha1p = sha1;
268         *modep = mode;
269         return 0;
270 }
271
272 static void show_new_file(struct rev_info *revs,
273                           struct cache_entry *new,
274                           int cached, int match_missing)
275 {
276         const unsigned char *sha1;
277         unsigned int mode;
278         unsigned dirty_submodule = 0;
279
280         /*
281          * New file in the index: it might actually be different in
282          * the working copy.
283          */
284         if (get_stat_data(new, &sha1, &mode, cached, match_missing,
285             &dirty_submodule, &revs->diffopt) < 0)
286                 return;
287
288         diff_index_show_file(revs, "+", new, sha1, mode, dirty_submodule);
289 }
290
291 static int show_modified(struct rev_info *revs,
292                          struct cache_entry *old,
293                          struct cache_entry *new,
294                          int report_missing,
295                          int cached, int match_missing)
296 {
297         unsigned int mode, oldmode;
298         const unsigned char *sha1;
299         unsigned dirty_submodule = 0;
300
301         if (get_stat_data(new, &sha1, &mode, cached, match_missing,
302                           &dirty_submodule, &revs->diffopt) < 0) {
303                 if (report_missing)
304                         diff_index_show_file(revs, "-", old,
305                                              old->sha1, old->ce_mode, 0);
306                 return -1;
307         }
308
309         if (revs->combine_merges && !cached &&
310             (hashcmp(sha1, old->sha1) || hashcmp(old->sha1, new->sha1))) {
311                 struct combine_diff_path *p;
312                 int pathlen = ce_namelen(new);
313
314                 p = xmalloc(combine_diff_path_size(2, pathlen));
315                 p->path = (char *) &p->parent[2];
316                 p->next = NULL;
317                 p->len = pathlen;
318                 memcpy(p->path, new->name, pathlen);
319                 p->path[pathlen] = 0;
320                 p->mode = mode;
321                 hashclr(p->sha1);
322                 memset(p->parent, 0, 2 * sizeof(struct combine_diff_parent));
323                 p->parent[0].status = DIFF_STATUS_MODIFIED;
324                 p->parent[0].mode = new->ce_mode;
325                 hashcpy(p->parent[0].sha1, new->sha1);
326                 p->parent[1].status = DIFF_STATUS_MODIFIED;
327                 p->parent[1].mode = old->ce_mode;
328                 hashcpy(p->parent[1].sha1, old->sha1);
329                 show_combined_diff(p, 2, revs->dense_combined_merges, revs);
330                 free(p);
331                 return 0;
332         }
333
334         oldmode = old->ce_mode;
335         if (mode == oldmode && !hashcmp(sha1, old->sha1) && !dirty_submodule &&
336             !DIFF_OPT_TST(&revs->diffopt, FIND_COPIES_HARDER))
337                 return 0;
338
339         diff_change(&revs->diffopt, oldmode, mode,
340                     old->sha1, sha1, old->name, 0, dirty_submodule);
341         return 0;
342 }
343
344 /*
345  * This gets a mix of an existing index and a tree, one pathname entry
346  * at a time. The index entry may be a single stage-0 one, but it could
347  * also be multiple unmerged entries (in which case idx_pos/idx_nr will
348  * give you the position and number of entries in the index).
349  */
350 static void do_oneway_diff(struct unpack_trees_options *o,
351         struct cache_entry *idx,
352         struct cache_entry *tree)
353 {
354         struct rev_info *revs = o->unpack_data;
355         int match_missing, cached;
356
357         /* if the entry is not checked out, don't examine work tree */
358         cached = o->index_only ||
359                 (idx && ((idx->ce_flags & CE_VALID) || ce_skip_worktree(idx)));
360         /*
361          * Backward compatibility wart - "diff-index -m" does
362          * not mean "do not ignore merges", but "match_missing".
363          *
364          * But with the revision flag parsing, that's found in
365          * "!revs->ignore_merges".
366          */
367         match_missing = !revs->ignore_merges;
368
369         if (cached && idx && ce_stage(idx)) {
370                 diff_unmerge(&revs->diffopt, idx->name, idx->ce_mode,
371                              idx->sha1);
372                 return;
373         }
374
375         /*
376          * Something added to the tree?
377          */
378         if (!tree) {
379                 show_new_file(revs, idx, cached, match_missing);
380                 return;
381         }
382
383         /*
384          * Something removed from the tree?
385          */
386         if (!idx) {
387                 diff_index_show_file(revs, "-", tree, tree->sha1, tree->ce_mode, 0);
388                 return;
389         }
390
391         /* Show difference between old and new */
392         show_modified(revs, tree, idx, 1, cached, match_missing);
393 }
394
395 /*
396  * The unpack_trees() interface is designed for merging, so
397  * the different source entries are designed primarily for
398  * the source trees, with the old index being really mainly
399  * used for being replaced by the result.
400  *
401  * For diffing, the index is more important, and we only have a
402  * single tree.
403  *
404  * We're supposed to advance o->pos to skip what we have already processed.
405  *
406  * This wrapper makes it all more readable, and takes care of all
407  * the fairly complex unpack_trees() semantic requirements, including
408  * the skipping, the path matching, the type conflict cases etc.
409  */
410 static int oneway_diff(struct cache_entry **src, struct unpack_trees_options *o)
411 {
412         struct cache_entry *idx = src[0];
413         struct cache_entry *tree = src[1];
414         struct rev_info *revs = o->unpack_data;
415
416         /*
417          * Unpack-trees generates a DF/conflict entry if
418          * there was a directory in the index and a tree
419          * in the tree. From a diff standpoint, that's a
420          * delete of the tree and a create of the file.
421          */
422         if (tree == o->df_conflict_entry)
423                 tree = NULL;
424
425         if (ce_path_match(idx ? idx : tree, revs->prune_data))
426                 do_oneway_diff(o, idx, tree);
427
428         return 0;
429 }
430
431 int run_diff_index(struct rev_info *revs, int cached)
432 {
433         struct object *ent;
434         struct tree *tree;
435         const char *tree_name;
436         struct unpack_trees_options opts;
437         struct tree_desc t;
438
439         ent = revs->pending.objects[0].item;
440         tree_name = revs->pending.objects[0].name;
441         tree = parse_tree_indirect(ent->sha1);
442         if (!tree)
443                 return error("bad tree object %s", tree_name);
444
445         memset(&opts, 0, sizeof(opts));
446         opts.head_idx = 1;
447         opts.index_only = cached;
448         opts.diff_index_cached = (cached &&
449                                   !DIFF_OPT_TST(&revs->diffopt, FIND_COPIES_HARDER));
450         opts.merge = 1;
451         opts.fn = oneway_diff;
452         opts.unpack_data = revs;
453         opts.src_index = &the_index;
454         opts.dst_index = NULL;
455
456         init_tree_desc(&t, tree->buffer, tree->size);
457         if (unpack_trees(1, &t, &opts))
458                 exit(128);
459
460         diff_set_mnemonic_prefix(&revs->diffopt, "c/", cached ? "i/" : "w/");
461         diffcore_fix_diff_index(&revs->diffopt);
462         diffcore_std(&revs->diffopt);
463         diff_flush(&revs->diffopt);
464         return 0;
465 }
466
467 int do_diff_cache(const unsigned char *tree_sha1, struct diff_options *opt)
468 {
469         struct tree *tree;
470         struct rev_info revs;
471         int i;
472         struct cache_entry **dst;
473         struct cache_entry *last = NULL;
474         struct unpack_trees_options opts;
475         struct tree_desc t;
476
477         /*
478          * This is used by git-blame to run diff-cache internally;
479          * it potentially needs to repeatedly run this, so we will
480          * start by removing the higher order entries the last round
481          * left behind.
482          */
483         dst = active_cache;
484         for (i = 0; i < active_nr; i++) {
485                 struct cache_entry *ce = active_cache[i];
486                 if (ce_stage(ce)) {
487                         if (last && !strcmp(ce->name, last->name))
488                                 continue;
489                         cache_tree_invalidate_path(active_cache_tree,
490                                                    ce->name);
491                         last = ce;
492                         ce->ce_flags |= CE_REMOVE;
493                 }
494                 *dst++ = ce;
495         }
496         active_nr = dst - active_cache;
497
498         init_revisions(&revs, NULL);
499         revs.prune_data = opt->paths;
500         tree = parse_tree_indirect(tree_sha1);
501         if (!tree)
502                 die("bad tree object %s", sha1_to_hex(tree_sha1));
503
504         memset(&opts, 0, sizeof(opts));
505         opts.head_idx = 1;
506         opts.index_only = 1;
507         opts.diff_index_cached = !DIFF_OPT_TST(opt, FIND_COPIES_HARDER);
508         opts.merge = 1;
509         opts.fn = oneway_diff;
510         opts.unpack_data = &revs;
511         opts.src_index = &the_index;
512         opts.dst_index = &the_index;
513
514         init_tree_desc(&t, tree->buffer, tree->size);
515         if (unpack_trees(1, &t, &opts))
516                 exit(128);
517         return 0;
518 }
519
520 int index_differs_from(const char *def, int diff_flags)
521 {
522         struct rev_info rev;
523         struct setup_revision_opt opt;
524
525         init_revisions(&rev, NULL);
526         memset(&opt, 0, sizeof(opt));
527         opt.def = def;
528         setup_revisions(0, NULL, &rev, &opt);
529         DIFF_OPT_SET(&rev.diffopt, QUICK);
530         DIFF_OPT_SET(&rev.diffopt, EXIT_WITH_STATUS);
531         rev.diffopt.flags |= diff_flags;
532         run_diff_index(&rev, 1);
533         if (rev.pending.alloc)
534                 free(rev.pending.objects);
535         return (DIFF_OPT_TST(&rev.diffopt, HAS_CHANGES) != 0);
536 }