Merge branch 'jk/diffstat-binary' into 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                 unsigned orig_flags = diffopt->flags;
73                 if (!DIFF_OPT_TST(diffopt, OVERRIDE_SUBMODULE_CONFIG))
74                         set_diffopt_flags_from_submodule_config(diffopt, ce->name);
75                 if (DIFF_OPT_TST(diffopt, IGNORE_SUBMODULES))
76                         changed = 0;
77                 else if (!DIFF_OPT_TST(diffopt, IGNORE_DIRTY_SUBMODULES)
78                     && (!changed || DIFF_OPT_TST(diffopt, DIRTY_SUBMODULES)))
79                         *dirty_submodule = is_submodule_modified(ce->name, DIFF_OPT_TST(diffopt, IGNORE_UNTRACKED_IN_SUBMODULES));
80                 diffopt->flags = orig_flags;
81         }
82         return changed;
83 }
84
85 int run_diff_files(struct rev_info *revs, unsigned int option)
86 {
87         int entries, i;
88         int diff_unmerged_stage = revs->max_count;
89         int silent_on_removed = option & DIFF_SILENT_ON_REMOVED;
90         unsigned ce_option = ((option & DIFF_RACY_IS_MODIFIED)
91                               ? CE_MATCH_RACY_IS_DIRTY : 0);
92
93         diff_set_mnemonic_prefix(&revs->diffopt, "i/", "w/");
94
95         if (diff_unmerged_stage < 0)
96                 diff_unmerged_stage = 2;
97         entries = active_nr;
98         for (i = 0; i < entries; i++) {
99                 struct stat st;
100                 unsigned int oldmode, newmode;
101                 struct cache_entry *ce = active_cache[i];
102                 int changed;
103                 unsigned dirty_submodule = 0;
104
105                 if (DIFF_OPT_TST(&revs->diffopt, QUICK) &&
106                         DIFF_OPT_TST(&revs->diffopt, HAS_CHANGES))
107                         break;
108
109                 if (!ce_path_match(ce, revs->prune_data))
110                         continue;
111
112                 if (ce_stage(ce)) {
113                         struct combine_diff_path *dpath;
114                         int num_compare_stages = 0;
115                         size_t path_len;
116
117                         path_len = ce_namelen(ce);
118
119                         dpath = xmalloc(combine_diff_path_size(5, path_len));
120                         dpath->path = (char *) &(dpath->parent[5]);
121
122                         dpath->next = NULL;
123                         dpath->len = path_len;
124                         memcpy(dpath->path, ce->name, path_len);
125                         dpath->path[path_len] = '\0';
126                         hashclr(dpath->sha1);
127                         memset(&(dpath->parent[0]), 0,
128                                sizeof(struct combine_diff_parent)*5);
129
130                         changed = check_removed(ce, &st);
131                         if (!changed)
132                                 dpath->mode = ce_mode_from_stat(ce, st.st_mode);
133                         else {
134                                 if (changed < 0) {
135                                         perror(ce->name);
136                                         continue;
137                                 }
138                                 if (silent_on_removed)
139                                         continue;
140                         }
141
142                         while (i < entries) {
143                                 struct cache_entry *nce = active_cache[i];
144                                 int stage;
145
146                                 if (strcmp(ce->name, nce->name))
147                                         break;
148
149                                 /* Stage #2 (ours) is the first parent,
150                                  * stage #3 (theirs) is the second.
151                                  */
152                                 stage = ce_stage(nce);
153                                 if (2 <= stage) {
154                                         int mode = nce->ce_mode;
155                                         num_compare_stages++;
156                                         hashcpy(dpath->parent[stage-2].sha1, nce->sha1);
157                                         dpath->parent[stage-2].mode = ce_mode_from_stat(nce, mode);
158                                         dpath->parent[stage-2].status =
159                                                 DIFF_STATUS_MODIFIED;
160                                 }
161
162                                 /* diff against the proper unmerged stage */
163                                 if (stage == diff_unmerged_stage)
164                                         ce = nce;
165                                 i++;
166                         }
167                         /*
168                          * Compensate for loop update
169                          */
170                         i--;
171
172                         if (revs->combine_merges && num_compare_stages == 2) {
173                                 show_combined_diff(dpath, 2,
174                                                    revs->dense_combined_merges,
175                                                    revs);
176                                 free(dpath);
177                                 continue;
178                         }
179                         free(dpath);
180                         dpath = NULL;
181
182                         /*
183                          * Show the diff for the 'ce' if we found the one
184                          * from the desired stage.
185                          */
186                         diff_unmerge(&revs->diffopt, ce->name, 0, null_sha1);
187                         if (ce_stage(ce) != diff_unmerged_stage)
188                                 continue;
189                 }
190
191                 if (ce_uptodate(ce) || ce_skip_worktree(ce))
192                         continue;
193
194                 /* If CE_VALID is set, don't look at workdir for file removal */
195                 changed = (ce->ce_flags & CE_VALID) ? 0 : check_removed(ce, &st);
196                 if (changed) {
197                         if (changed < 0) {
198                                 perror(ce->name);
199                                 continue;
200                         }
201                         if (silent_on_removed)
202                                 continue;
203                         diff_addremove(&revs->diffopt, '-', ce->ce_mode,
204                                        ce->sha1, ce->name, 0);
205                         continue;
206                 }
207                 changed = match_stat_with_submodule(&revs->diffopt, ce, &st,
208                                                     ce_option, &dirty_submodule);
209                 if (!changed && !dirty_submodule) {
210                         ce_mark_uptodate(ce);
211                         if (!DIFF_OPT_TST(&revs->diffopt, FIND_COPIES_HARDER))
212                                 continue;
213                 }
214                 oldmode = ce->ce_mode;
215                 newmode = ce_mode_from_stat(ce, st.st_mode);
216                 diff_change(&revs->diffopt, oldmode, newmode,
217                             ce->sha1, (changed ? null_sha1 : ce->sha1),
218                             ce->name, 0, dirty_submodule);
219
220         }
221         diffcore_std(&revs->diffopt);
222         diff_flush(&revs->diffopt);
223         return 0;
224 }
225
226 /*
227  * diff-index
228  */
229
230 /* A file entry went away or appeared */
231 static void diff_index_show_file(struct rev_info *revs,
232                                  const char *prefix,
233                                  struct cache_entry *ce,
234                                  const unsigned char *sha1, unsigned int mode,
235                                  unsigned dirty_submodule)
236 {
237         diff_addremove(&revs->diffopt, prefix[0], mode,
238                        sha1, ce->name, dirty_submodule);
239 }
240
241 static int get_stat_data(struct cache_entry *ce,
242                          const unsigned char **sha1p,
243                          unsigned int *modep,
244                          int cached, int match_missing,
245                          unsigned *dirty_submodule, struct diff_options *diffopt)
246 {
247         const unsigned char *sha1 = ce->sha1;
248         unsigned int mode = ce->ce_mode;
249
250         if (!cached && !ce_uptodate(ce)) {
251                 int changed;
252                 struct stat st;
253                 changed = check_removed(ce, &st);
254                 if (changed < 0)
255                         return -1;
256                 else if (changed) {
257                         if (match_missing) {
258                                 *sha1p = sha1;
259                                 *modep = mode;
260                                 return 0;
261                         }
262                         return -1;
263                 }
264                 changed = match_stat_with_submodule(diffopt, ce, &st,
265                                                     0, dirty_submodule);
266                 if (changed) {
267                         mode = ce_mode_from_stat(ce, st.st_mode);
268                         sha1 = null_sha1;
269                 }
270         }
271
272         *sha1p = sha1;
273         *modep = mode;
274         return 0;
275 }
276
277 static void show_new_file(struct rev_info *revs,
278                           struct cache_entry *new,
279                           int cached, int match_missing)
280 {
281         const unsigned char *sha1;
282         unsigned int mode;
283         unsigned dirty_submodule = 0;
284
285         /*
286          * New file in the index: it might actually be different in
287          * the working copy.
288          */
289         if (get_stat_data(new, &sha1, &mode, cached, match_missing,
290             &dirty_submodule, &revs->diffopt) < 0)
291                 return;
292
293         diff_index_show_file(revs, "+", new, sha1, mode, dirty_submodule);
294 }
295
296 static int show_modified(struct rev_info *revs,
297                          struct cache_entry *old,
298                          struct cache_entry *new,
299                          int report_missing,
300                          int cached, int match_missing)
301 {
302         unsigned int mode, oldmode;
303         const unsigned char *sha1;
304         unsigned dirty_submodule = 0;
305
306         if (get_stat_data(new, &sha1, &mode, cached, match_missing,
307                           &dirty_submodule, &revs->diffopt) < 0) {
308                 if (report_missing)
309                         diff_index_show_file(revs, "-", old,
310                                              old->sha1, old->ce_mode, 0);
311                 return -1;
312         }
313
314         if (revs->combine_merges && !cached &&
315             (hashcmp(sha1, old->sha1) || hashcmp(old->sha1, new->sha1))) {
316                 struct combine_diff_path *p;
317                 int pathlen = ce_namelen(new);
318
319                 p = xmalloc(combine_diff_path_size(2, pathlen));
320                 p->path = (char *) &p->parent[2];
321                 p->next = NULL;
322                 p->len = pathlen;
323                 memcpy(p->path, new->name, pathlen);
324                 p->path[pathlen] = 0;
325                 p->mode = mode;
326                 hashclr(p->sha1);
327                 memset(p->parent, 0, 2 * sizeof(struct combine_diff_parent));
328                 p->parent[0].status = DIFF_STATUS_MODIFIED;
329                 p->parent[0].mode = new->ce_mode;
330                 hashcpy(p->parent[0].sha1, new->sha1);
331                 p->parent[1].status = DIFF_STATUS_MODIFIED;
332                 p->parent[1].mode = old->ce_mode;
333                 hashcpy(p->parent[1].sha1, old->sha1);
334                 show_combined_diff(p, 2, revs->dense_combined_merges, revs);
335                 free(p);
336                 return 0;
337         }
338
339         oldmode = old->ce_mode;
340         if (mode == oldmode && !hashcmp(sha1, old->sha1) && !dirty_submodule &&
341             !DIFF_OPT_TST(&revs->diffopt, FIND_COPIES_HARDER))
342                 return 0;
343
344         diff_change(&revs->diffopt, oldmode, mode,
345                     old->sha1, sha1, old->name, 0, dirty_submodule);
346         return 0;
347 }
348
349 /*
350  * This gets a mix of an existing index and a tree, one pathname entry
351  * at a time. The index entry may be a single stage-0 one, but it could
352  * also be multiple unmerged entries (in which case idx_pos/idx_nr will
353  * give you the position and number of entries in the index).
354  */
355 static void do_oneway_diff(struct unpack_trees_options *o,
356         struct cache_entry *idx,
357         struct cache_entry *tree)
358 {
359         struct rev_info *revs = o->unpack_data;
360         int match_missing, cached;
361
362         /* if the entry is not checked out, don't examine work tree */
363         cached = o->index_only ||
364                 (idx && ((idx->ce_flags & CE_VALID) || ce_skip_worktree(idx)));
365         /*
366          * Backward compatibility wart - "diff-index -m" does
367          * not mean "do not ignore merges", but "match_missing".
368          *
369          * But with the revision flag parsing, that's found in
370          * "!revs->ignore_merges".
371          */
372         match_missing = !revs->ignore_merges;
373
374         if (cached && idx && ce_stage(idx)) {
375                 diff_unmerge(&revs->diffopt, idx->name, idx->ce_mode,
376                              idx->sha1);
377                 return;
378         }
379
380         /*
381          * Something added to the tree?
382          */
383         if (!tree) {
384                 show_new_file(revs, idx, cached, match_missing);
385                 return;
386         }
387
388         /*
389          * Something removed from the tree?
390          */
391         if (!idx) {
392                 diff_index_show_file(revs, "-", tree, tree->sha1, tree->ce_mode, 0);
393                 return;
394         }
395
396         /* Show difference between old and new */
397         show_modified(revs, tree, idx, 1, cached, match_missing);
398 }
399
400 /*
401  * The unpack_trees() interface is designed for merging, so
402  * the different source entries are designed primarily for
403  * the source trees, with the old index being really mainly
404  * used for being replaced by the result.
405  *
406  * For diffing, the index is more important, and we only have a
407  * single tree.
408  *
409  * We're supposed to advance o->pos to skip what we have already processed.
410  *
411  * This wrapper makes it all more readable, and takes care of all
412  * the fairly complex unpack_trees() semantic requirements, including
413  * the skipping, the path matching, the type conflict cases etc.
414  */
415 static int oneway_diff(struct cache_entry **src, struct unpack_trees_options *o)
416 {
417         struct cache_entry *idx = src[0];
418         struct cache_entry *tree = src[1];
419         struct rev_info *revs = o->unpack_data;
420
421         /*
422          * Unpack-trees generates a DF/conflict entry if
423          * there was a directory in the index and a tree
424          * in the tree. From a diff standpoint, that's a
425          * delete of the tree and a create of the file.
426          */
427         if (tree == o->df_conflict_entry)
428                 tree = NULL;
429
430         if (ce_path_match(idx ? idx : tree, revs->prune_data))
431                 do_oneway_diff(o, idx, tree);
432
433         return 0;
434 }
435
436 int run_diff_index(struct rev_info *revs, int cached)
437 {
438         struct object *ent;
439         struct tree *tree;
440         const char *tree_name;
441         struct unpack_trees_options opts;
442         struct tree_desc t;
443
444         ent = revs->pending.objects[0].item;
445         tree_name = revs->pending.objects[0].name;
446         tree = parse_tree_indirect(ent->sha1);
447         if (!tree)
448                 return error("bad tree object %s", tree_name);
449
450         memset(&opts, 0, sizeof(opts));
451         opts.head_idx = 1;
452         opts.index_only = cached;
453         opts.diff_index_cached = (cached &&
454                                   !DIFF_OPT_TST(&revs->diffopt, FIND_COPIES_HARDER));
455         opts.merge = 1;
456         opts.fn = oneway_diff;
457         opts.unpack_data = revs;
458         opts.src_index = &the_index;
459         opts.dst_index = NULL;
460
461         init_tree_desc(&t, tree->buffer, tree->size);
462         if (unpack_trees(1, &t, &opts))
463                 exit(128);
464
465         diff_set_mnemonic_prefix(&revs->diffopt, "c/", cached ? "i/" : "w/");
466         diffcore_fix_diff_index(&revs->diffopt);
467         diffcore_std(&revs->diffopt);
468         diff_flush(&revs->diffopt);
469         return 0;
470 }
471
472 int do_diff_cache(const unsigned char *tree_sha1, struct diff_options *opt)
473 {
474         struct tree *tree;
475         struct rev_info revs;
476         int i;
477         struct cache_entry **dst;
478         struct cache_entry *last = NULL;
479         struct unpack_trees_options opts;
480         struct tree_desc t;
481
482         /*
483          * This is used by git-blame to run diff-cache internally;
484          * it potentially needs to repeatedly run this, so we will
485          * start by removing the higher order entries the last round
486          * left behind.
487          */
488         dst = active_cache;
489         for (i = 0; i < active_nr; i++) {
490                 struct cache_entry *ce = active_cache[i];
491                 if (ce_stage(ce)) {
492                         if (last && !strcmp(ce->name, last->name))
493                                 continue;
494                         cache_tree_invalidate_path(active_cache_tree,
495                                                    ce->name);
496                         last = ce;
497                         ce->ce_flags |= CE_REMOVE;
498                 }
499                 *dst++ = ce;
500         }
501         active_nr = dst - active_cache;
502
503         init_revisions(&revs, NULL);
504         revs.prune_data = opt->paths;
505         tree = parse_tree_indirect(tree_sha1);
506         if (!tree)
507                 die("bad tree object %s", sha1_to_hex(tree_sha1));
508
509         memset(&opts, 0, sizeof(opts));
510         opts.head_idx = 1;
511         opts.index_only = 1;
512         opts.diff_index_cached = !DIFF_OPT_TST(opt, FIND_COPIES_HARDER);
513         opts.merge = 1;
514         opts.fn = oneway_diff;
515         opts.unpack_data = &revs;
516         opts.src_index = &the_index;
517         opts.dst_index = &the_index;
518
519         init_tree_desc(&t, tree->buffer, tree->size);
520         if (unpack_trees(1, &t, &opts))
521                 exit(128);
522         return 0;
523 }
524
525 int index_differs_from(const char *def, int diff_flags)
526 {
527         struct rev_info rev;
528         struct setup_revision_opt opt;
529
530         init_revisions(&rev, NULL);
531         memset(&opt, 0, sizeof(opt));
532         opt.def = def;
533         setup_revisions(0, NULL, &rev, &opt);
534         DIFF_OPT_SET(&rev.diffopt, QUICK);
535         DIFF_OPT_SET(&rev.diffopt, EXIT_WITH_STATUS);
536         rev.diffopt.flags |= diff_flags;
537         run_diff_index(&rev, 1);
538         if (rev.pending.alloc)
539                 free(rev.pending.objects);
540         return (DIFF_OPT_TST(&rev.diffopt, HAS_CHANGES) != 0);
541 }