tests: cleanup binary test vector files
[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_can_quit_early(&revs->diffopt))
106                         break;
107
108                 if (!ce_path_match(ce, &revs->prune_data))
109                         continue;
110
111                 if (ce_stage(ce)) {
112                         struct combine_diff_path *dpath;
113                         struct diff_filepair *pair;
114                         unsigned int wt_mode = 0;
115                         int num_compare_stages = 0;
116                         size_t path_len;
117
118                         path_len = ce_namelen(ce);
119
120                         dpath = xmalloc(combine_diff_path_size(5, path_len));
121                         dpath->path = (char *) &(dpath->parent[5]);
122
123                         dpath->next = NULL;
124                         dpath->len = path_len;
125                         memcpy(dpath->path, ce->name, path_len);
126                         dpath->path[path_len] = '\0';
127                         hashclr(dpath->sha1);
128                         memset(&(dpath->parent[0]), 0,
129                                sizeof(struct combine_diff_parent)*5);
130
131                         changed = check_removed(ce, &st);
132                         if (!changed)
133                                 wt_mode = ce_mode_from_stat(ce, st.st_mode);
134                         else {
135                                 if (changed < 0) {
136                                         perror(ce->name);
137                                         continue;
138                                 }
139                                 if (silent_on_removed)
140                                         continue;
141                                 wt_mode = 0;
142                         }
143                         dpath->mode = wt_mode;
144
145                         while (i < entries) {
146                                 struct cache_entry *nce = active_cache[i];
147                                 int stage;
148
149                                 if (strcmp(ce->name, nce->name))
150                                         break;
151
152                                 /* Stage #2 (ours) is the first parent,
153                                  * stage #3 (theirs) is the second.
154                                  */
155                                 stage = ce_stage(nce);
156                                 if (2 <= stage) {
157                                         int mode = nce->ce_mode;
158                                         num_compare_stages++;
159                                         hashcpy(dpath->parent[stage-2].sha1, nce->sha1);
160                                         dpath->parent[stage-2].mode = ce_mode_from_stat(nce, mode);
161                                         dpath->parent[stage-2].status =
162                                                 DIFF_STATUS_MODIFIED;
163                                 }
164
165                                 /* diff against the proper unmerged stage */
166                                 if (stage == diff_unmerged_stage)
167                                         ce = nce;
168                                 i++;
169                         }
170                         /*
171                          * Compensate for loop update
172                          */
173                         i--;
174
175                         if (revs->combine_merges && num_compare_stages == 2) {
176                                 show_combined_diff(dpath, 2,
177                                                    revs->dense_combined_merges,
178                                                    revs);
179                                 free(dpath);
180                                 continue;
181                         }
182                         free(dpath);
183                         dpath = NULL;
184
185                         /*
186                          * Show the diff for the 'ce' if we found the one
187                          * from the desired stage.
188                          */
189                         pair = diff_unmerge(&revs->diffopt, ce->name);
190                         if (wt_mode)
191                                 pair->two->mode = wt_mode;
192                         if (ce_stage(ce) != diff_unmerged_stage)
193                                 continue;
194                 }
195
196                 if (ce_uptodate(ce) || ce_skip_worktree(ce))
197                         continue;
198
199                 /* If CE_VALID is set, don't look at workdir for file removal */
200                 changed = (ce->ce_flags & CE_VALID) ? 0 : check_removed(ce, &st);
201                 if (changed) {
202                         if (changed < 0) {
203                                 perror(ce->name);
204                                 continue;
205                         }
206                         if (silent_on_removed)
207                                 continue;
208                         diff_addremove(&revs->diffopt, '-', ce->ce_mode,
209                                        ce->sha1, ce->name, 0);
210                         continue;
211                 }
212                 changed = match_stat_with_submodule(&revs->diffopt, ce, &st,
213                                                     ce_option, &dirty_submodule);
214                 if (!changed && !dirty_submodule) {
215                         ce_mark_uptodate(ce);
216                         if (!DIFF_OPT_TST(&revs->diffopt, FIND_COPIES_HARDER))
217                                 continue;
218                 }
219                 oldmode = ce->ce_mode;
220                 newmode = ce_mode_from_stat(ce, st.st_mode);
221                 diff_change(&revs->diffopt, oldmode, newmode,
222                             ce->sha1, (changed ? null_sha1 : ce->sha1),
223                             ce->name, 0, dirty_submodule);
224
225         }
226         diffcore_std(&revs->diffopt);
227         diff_flush(&revs->diffopt);
228         return 0;
229 }
230
231 /*
232  * diff-index
233  */
234
235 /* A file entry went away or appeared */
236 static void diff_index_show_file(struct rev_info *revs,
237                                  const char *prefix,
238                                  struct cache_entry *ce,
239                                  const unsigned char *sha1, unsigned int mode,
240                                  unsigned dirty_submodule)
241 {
242         diff_addremove(&revs->diffopt, prefix[0], mode,
243                        sha1, ce->name, dirty_submodule);
244 }
245
246 static int get_stat_data(struct cache_entry *ce,
247                          const unsigned char **sha1p,
248                          unsigned int *modep,
249                          int cached, int match_missing,
250                          unsigned *dirty_submodule, struct diff_options *diffopt)
251 {
252         const unsigned char *sha1 = ce->sha1;
253         unsigned int mode = ce->ce_mode;
254
255         if (!cached && !ce_uptodate(ce)) {
256                 int changed;
257                 struct stat st;
258                 changed = check_removed(ce, &st);
259                 if (changed < 0)
260                         return -1;
261                 else if (changed) {
262                         if (match_missing) {
263                                 *sha1p = sha1;
264                                 *modep = mode;
265                                 return 0;
266                         }
267                         return -1;
268                 }
269                 changed = match_stat_with_submodule(diffopt, ce, &st,
270                                                     0, dirty_submodule);
271                 if (changed) {
272                         mode = ce_mode_from_stat(ce, st.st_mode);
273                         sha1 = null_sha1;
274                 }
275         }
276
277         *sha1p = sha1;
278         *modep = mode;
279         return 0;
280 }
281
282 static void show_new_file(struct rev_info *revs,
283                           struct cache_entry *new,
284                           int cached, int match_missing)
285 {
286         const unsigned char *sha1;
287         unsigned int mode;
288         unsigned dirty_submodule = 0;
289
290         /*
291          * New file in the index: it might actually be different in
292          * the working copy.
293          */
294         if (get_stat_data(new, &sha1, &mode, cached, match_missing,
295             &dirty_submodule, &revs->diffopt) < 0)
296                 return;
297
298         diff_index_show_file(revs, "+", new, sha1, mode, dirty_submodule);
299 }
300
301 static int show_modified(struct rev_info *revs,
302                          struct cache_entry *old,
303                          struct cache_entry *new,
304                          int report_missing,
305                          int cached, int match_missing)
306 {
307         unsigned int mode, oldmode;
308         const unsigned char *sha1;
309         unsigned dirty_submodule = 0;
310
311         if (get_stat_data(new, &sha1, &mode, cached, match_missing,
312                           &dirty_submodule, &revs->diffopt) < 0) {
313                 if (report_missing)
314                         diff_index_show_file(revs, "-", old,
315                                              old->sha1, old->ce_mode, 0);
316                 return -1;
317         }
318
319         if (revs->combine_merges && !cached &&
320             (hashcmp(sha1, old->sha1) || hashcmp(old->sha1, new->sha1))) {
321                 struct combine_diff_path *p;
322                 int pathlen = ce_namelen(new);
323
324                 p = xmalloc(combine_diff_path_size(2, pathlen));
325                 p->path = (char *) &p->parent[2];
326                 p->next = NULL;
327                 p->len = pathlen;
328                 memcpy(p->path, new->name, pathlen);
329                 p->path[pathlen] = 0;
330                 p->mode = mode;
331                 hashclr(p->sha1);
332                 memset(p->parent, 0, 2 * sizeof(struct combine_diff_parent));
333                 p->parent[0].status = DIFF_STATUS_MODIFIED;
334                 p->parent[0].mode = new->ce_mode;
335                 hashcpy(p->parent[0].sha1, new->sha1);
336                 p->parent[1].status = DIFF_STATUS_MODIFIED;
337                 p->parent[1].mode = old->ce_mode;
338                 hashcpy(p->parent[1].sha1, old->sha1);
339                 show_combined_diff(p, 2, revs->dense_combined_merges, revs);
340                 free(p);
341                 return 0;
342         }
343
344         oldmode = old->ce_mode;
345         if (mode == oldmode && !hashcmp(sha1, old->sha1) && !dirty_submodule &&
346             !DIFF_OPT_TST(&revs->diffopt, FIND_COPIES_HARDER))
347                 return 0;
348
349         diff_change(&revs->diffopt, oldmode, mode,
350                     old->sha1, sha1, old->name, 0, dirty_submodule);
351         return 0;
352 }
353
354 /*
355  * This gets a mix of an existing index and a tree, one pathname entry
356  * at a time. The index entry may be a single stage-0 one, but it could
357  * also be multiple unmerged entries (in which case idx_pos/idx_nr will
358  * give you the position and number of entries in the index).
359  */
360 static void do_oneway_diff(struct unpack_trees_options *o,
361         struct cache_entry *idx,
362         struct cache_entry *tree)
363 {
364         struct rev_info *revs = o->unpack_data;
365         int match_missing, cached;
366
367         /* if the entry is not checked out, don't examine work tree */
368         cached = o->index_only ||
369                 (idx && ((idx->ce_flags & CE_VALID) || ce_skip_worktree(idx)));
370         /*
371          * Backward compatibility wart - "diff-index -m" does
372          * not mean "do not ignore merges", but "match_missing".
373          *
374          * But with the revision flag parsing, that's found in
375          * "!revs->ignore_merges".
376          */
377         match_missing = !revs->ignore_merges;
378
379         if (cached && idx && ce_stage(idx)) {
380                 struct diff_filepair *pair;
381                 pair = diff_unmerge(&revs->diffopt, idx->name);
382                 fill_filespec(pair->one, idx->sha1, idx->ce_mode);
383                 return;
384         }
385
386         /*
387          * Something added to the tree?
388          */
389         if (!tree) {
390                 show_new_file(revs, idx, cached, match_missing);
391                 return;
392         }
393
394         /*
395          * Something removed from the tree?
396          */
397         if (!idx) {
398                 diff_index_show_file(revs, "-", tree, tree->sha1, tree->ce_mode, 0);
399                 return;
400         }
401
402         /* Show difference between old and new */
403         show_modified(revs, tree, idx, 1, cached, match_missing);
404 }
405
406 /*
407  * The unpack_trees() interface is designed for merging, so
408  * the different source entries are designed primarily for
409  * the source trees, with the old index being really mainly
410  * used for being replaced by the result.
411  *
412  * For diffing, the index is more important, and we only have a
413  * single tree.
414  *
415  * We're supposed to advance o->pos to skip what we have already processed.
416  *
417  * This wrapper makes it all more readable, and takes care of all
418  * the fairly complex unpack_trees() semantic requirements, including
419  * the skipping, the path matching, the type conflict cases etc.
420  */
421 static int oneway_diff(struct cache_entry **src, struct unpack_trees_options *o)
422 {
423         struct cache_entry *idx = src[0];
424         struct cache_entry *tree = src[1];
425         struct rev_info *revs = o->unpack_data;
426
427         /*
428          * Unpack-trees generates a DF/conflict entry if
429          * there was a directory in the index and a tree
430          * in the tree. From a diff standpoint, that's a
431          * delete of the tree and a create of the file.
432          */
433         if (tree == o->df_conflict_entry)
434                 tree = NULL;
435
436         if (ce_path_match(idx ? idx : tree, &revs->prune_data))
437                 do_oneway_diff(o, idx, tree);
438
439         return 0;
440 }
441
442 int run_diff_index(struct rev_info *revs, int cached)
443 {
444         struct object *ent;
445         struct tree *tree;
446         const char *tree_name;
447         struct unpack_trees_options opts;
448         struct tree_desc t;
449
450         ent = revs->pending.objects[0].item;
451         tree_name = revs->pending.objects[0].name;
452         tree = parse_tree_indirect(ent->sha1);
453         if (!tree)
454                 return error("bad tree object %s", tree_name);
455
456         memset(&opts, 0, sizeof(opts));
457         opts.head_idx = 1;
458         opts.index_only = cached;
459         opts.diff_index_cached = (cached &&
460                                   !DIFF_OPT_TST(&revs->diffopt, FIND_COPIES_HARDER));
461         opts.merge = 1;
462         opts.fn = oneway_diff;
463         opts.unpack_data = revs;
464         opts.src_index = &the_index;
465         opts.dst_index = NULL;
466
467         init_tree_desc(&t, tree->buffer, tree->size);
468         if (unpack_trees(1, &t, &opts))
469                 exit(128);
470
471         diff_set_mnemonic_prefix(&revs->diffopt, "c/", cached ? "i/" : "w/");
472         diffcore_fix_diff_index(&revs->diffopt);
473         diffcore_std(&revs->diffopt);
474         diff_flush(&revs->diffopt);
475         return 0;
476 }
477
478 int do_diff_cache(const unsigned char *tree_sha1, struct diff_options *opt)
479 {
480         struct tree *tree;
481         struct rev_info revs;
482         int i;
483         struct cache_entry **dst;
484         struct cache_entry *last = NULL;
485         struct unpack_trees_options opts;
486         struct tree_desc t;
487
488         /*
489          * This is used by git-blame to run diff-cache internally;
490          * it potentially needs to repeatedly run this, so we will
491          * start by removing the higher order entries the last round
492          * left behind.
493          */
494         dst = active_cache;
495         for (i = 0; i < active_nr; i++) {
496                 struct cache_entry *ce = active_cache[i];
497                 if (ce_stage(ce)) {
498                         if (last && !strcmp(ce->name, last->name))
499                                 continue;
500                         cache_tree_invalidate_path(active_cache_tree,
501                                                    ce->name);
502                         last = ce;
503                         ce->ce_flags |= CE_REMOVE;
504                 }
505                 *dst++ = ce;
506         }
507         active_nr = dst - active_cache;
508
509         init_revisions(&revs, NULL);
510         init_pathspec(&revs.prune_data, opt->pathspec.raw);
511         tree = parse_tree_indirect(tree_sha1);
512         if (!tree)
513                 die("bad tree object %s", sha1_to_hex(tree_sha1));
514
515         memset(&opts, 0, sizeof(opts));
516         opts.head_idx = 1;
517         opts.index_only = 1;
518         opts.diff_index_cached = !DIFF_OPT_TST(opt, FIND_COPIES_HARDER);
519         opts.merge = 1;
520         opts.fn = oneway_diff;
521         opts.unpack_data = &revs;
522         opts.src_index = &the_index;
523         opts.dst_index = &the_index;
524
525         init_tree_desc(&t, tree->buffer, tree->size);
526         if (unpack_trees(1, &t, &opts))
527                 exit(128);
528         return 0;
529 }
530
531 int index_differs_from(const char *def, int diff_flags)
532 {
533         struct rev_info rev;
534         struct setup_revision_opt opt;
535
536         init_revisions(&rev, NULL);
537         memset(&opt, 0, sizeof(opt));
538         opt.def = def;
539         setup_revisions(0, NULL, &rev, &opt);
540         DIFF_OPT_SET(&rev.diffopt, QUICK);
541         DIFF_OPT_SET(&rev.diffopt, EXIT_WITH_STATUS);
542         rev.diffopt.flags |= diff_flags;
543         run_diff_index(&rev, 1);
544         if (rev.pending.alloc)
545                 free(rev.pending.objects);
546         return (DIFF_OPT_TST(&rev.diffopt, HAS_CHANGES) != 0);
547 }