ewah: compressed bitmap implementation
[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                                      const struct cache_entry *ce,
68                                      struct stat *st, unsigned ce_option,
69                                      unsigned *dirty_submodule)
70 {
71         int changed = ce_match_stat(ce, st, ce_option);
72         if (S_ISGITLINK(ce->ce_mode)) {
73                 unsigned orig_flags = diffopt->flags;
74                 if (!DIFF_OPT_TST(diffopt, OVERRIDE_SUBMODULE_CONFIG))
75                         set_diffopt_flags_from_submodule_config(diffopt, ce->name);
76                 if (DIFF_OPT_TST(diffopt, IGNORE_SUBMODULES))
77                         changed = 0;
78                 else if (!DIFF_OPT_TST(diffopt, IGNORE_DIRTY_SUBMODULES)
79                     && (!changed || DIFF_OPT_TST(diffopt, DIRTY_SUBMODULES)))
80                         *dirty_submodule = is_submodule_modified(ce->name, DIFF_OPT_TST(diffopt, IGNORE_UNTRACKED_IN_SUBMODULES));
81                 diffopt->flags = orig_flags;
82         }
83         return changed;
84 }
85
86 int run_diff_files(struct rev_info *revs, unsigned int option)
87 {
88         int entries, i;
89         int diff_unmerged_stage = revs->max_count;
90         unsigned ce_option = ((option & DIFF_RACY_IS_MODIFIED)
91                               ? CE_MATCH_RACY_IS_DIRTY : 0);
92
93         if (option & DIFF_SILENT_ON_REMOVED)
94                 handle_deprecated_show_diff_q(&revs->diffopt);
95
96         diff_set_mnemonic_prefix(&revs->diffopt, "i/", "w/");
97
98         if (diff_unmerged_stage < 0)
99                 diff_unmerged_stage = 2;
100         entries = active_nr;
101         for (i = 0; i < entries; i++) {
102                 struct stat st;
103                 unsigned int oldmode, newmode;
104                 struct cache_entry *ce = active_cache[i];
105                 int changed;
106                 unsigned dirty_submodule = 0;
107
108                 if (diff_can_quit_early(&revs->diffopt))
109                         break;
110
111                 if (!ce_path_match(ce, &revs->prune_data))
112                         continue;
113
114                 if (ce_stage(ce)) {
115                         struct combine_diff_path *dpath;
116                         struct diff_filepair *pair;
117                         unsigned int wt_mode = 0;
118                         int num_compare_stages = 0;
119                         size_t path_len;
120
121                         path_len = ce_namelen(ce);
122
123                         dpath = xmalloc(combine_diff_path_size(5, path_len));
124                         dpath->path = (char *) &(dpath->parent[5]);
125
126                         dpath->next = NULL;
127                         dpath->len = path_len;
128                         memcpy(dpath->path, ce->name, path_len);
129                         dpath->path[path_len] = '\0';
130                         hashclr(dpath->sha1);
131                         memset(&(dpath->parent[0]), 0,
132                                sizeof(struct combine_diff_parent)*5);
133
134                         changed = check_removed(ce, &st);
135                         if (!changed)
136                                 wt_mode = ce_mode_from_stat(ce, st.st_mode);
137                         else {
138                                 if (changed < 0) {
139                                         perror(ce->name);
140                                         continue;
141                                 }
142                                 wt_mode = 0;
143                         }
144                         dpath->mode = wt_mode;
145
146                         while (i < entries) {
147                                 struct cache_entry *nce = active_cache[i];
148                                 int stage;
149
150                                 if (strcmp(ce->name, nce->name))
151                                         break;
152
153                                 /* Stage #2 (ours) is the first parent,
154                                  * stage #3 (theirs) is the second.
155                                  */
156                                 stage = ce_stage(nce);
157                                 if (2 <= stage) {
158                                         int mode = nce->ce_mode;
159                                         num_compare_stages++;
160                                         hashcpy(dpath->parent[stage-2].sha1, nce->sha1);
161                                         dpath->parent[stage-2].mode = ce_mode_from_stat(nce, mode);
162                                         dpath->parent[stage-2].status =
163                                                 DIFF_STATUS_MODIFIED;
164                                 }
165
166                                 /* diff against the proper unmerged stage */
167                                 if (stage == diff_unmerged_stage)
168                                         ce = nce;
169                                 i++;
170                         }
171                         /*
172                          * Compensate for loop update
173                          */
174                         i--;
175
176                         if (revs->combine_merges && num_compare_stages == 2) {
177                                 show_combined_diff(dpath, 2,
178                                                    revs->dense_combined_merges,
179                                                    revs);
180                                 free(dpath);
181                                 continue;
182                         }
183                         free(dpath);
184                         dpath = NULL;
185
186                         /*
187                          * Show the diff for the 'ce' if we found the one
188                          * from the desired stage.
189                          */
190                         pair = diff_unmerge(&revs->diffopt, ce->name);
191                         if (wt_mode)
192                                 pair->two->mode = wt_mode;
193                         if (ce_stage(ce) != diff_unmerged_stage)
194                                 continue;
195                 }
196
197                 if (ce_uptodate(ce) || ce_skip_worktree(ce))
198                         continue;
199
200                 /* If CE_VALID is set, don't look at workdir for file removal */
201                 changed = (ce->ce_flags & CE_VALID) ? 0 : check_removed(ce, &st);
202                 if (changed) {
203                         if (changed < 0) {
204                                 perror(ce->name);
205                                 continue;
206                         }
207                         diff_addremove(&revs->diffopt, '-', ce->ce_mode,
208                                        ce->sha1, !is_null_sha1(ce->sha1),
209                                        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                             !is_null_sha1(ce->sha1), (changed ? 0 : !is_null_sha1(ce->sha1)),
224                             ce->name, 0, dirty_submodule);
225
226         }
227         diffcore_std(&revs->diffopt);
228         diff_flush(&revs->diffopt);
229         return 0;
230 }
231
232 /*
233  * diff-index
234  */
235
236 /* A file entry went away or appeared */
237 static void diff_index_show_file(struct rev_info *revs,
238                                  const char *prefix,
239                                  const struct cache_entry *ce,
240                                  const unsigned char *sha1, int sha1_valid,
241                                  unsigned int mode,
242                                  unsigned dirty_submodule)
243 {
244         diff_addremove(&revs->diffopt, prefix[0], mode,
245                        sha1, sha1_valid, ce->name, dirty_submodule);
246 }
247
248 static int get_stat_data(const struct cache_entry *ce,
249                          const unsigned char **sha1p,
250                          unsigned int *modep,
251                          int cached, int match_missing,
252                          unsigned *dirty_submodule, struct diff_options *diffopt)
253 {
254         const unsigned char *sha1 = ce->sha1;
255         unsigned int mode = ce->ce_mode;
256
257         if (!cached && !ce_uptodate(ce)) {
258                 int changed;
259                 struct stat st;
260                 changed = check_removed(ce, &st);
261                 if (changed < 0)
262                         return -1;
263                 else if (changed) {
264                         if (match_missing) {
265                                 *sha1p = sha1;
266                                 *modep = mode;
267                                 return 0;
268                         }
269                         return -1;
270                 }
271                 changed = match_stat_with_submodule(diffopt, ce, &st,
272                                                     0, dirty_submodule);
273                 if (changed) {
274                         mode = ce_mode_from_stat(ce, st.st_mode);
275                         sha1 = null_sha1;
276                 }
277         }
278
279         *sha1p = sha1;
280         *modep = mode;
281         return 0;
282 }
283
284 static void show_new_file(struct rev_info *revs,
285                           const struct cache_entry *new,
286                           int cached, int match_missing)
287 {
288         const unsigned char *sha1;
289         unsigned int mode;
290         unsigned dirty_submodule = 0;
291
292         /*
293          * New file in the index: it might actually be different in
294          * the working tree.
295          */
296         if (get_stat_data(new, &sha1, &mode, cached, match_missing,
297             &dirty_submodule, &revs->diffopt) < 0)
298                 return;
299
300         diff_index_show_file(revs, "+", new, sha1, !is_null_sha1(sha1), mode, dirty_submodule);
301 }
302
303 static int show_modified(struct rev_info *revs,
304                          const struct cache_entry *old,
305                          const struct cache_entry *new,
306                          int report_missing,
307                          int cached, int match_missing)
308 {
309         unsigned int mode, oldmode;
310         const unsigned char *sha1;
311         unsigned dirty_submodule = 0;
312
313         if (get_stat_data(new, &sha1, &mode, cached, match_missing,
314                           &dirty_submodule, &revs->diffopt) < 0) {
315                 if (report_missing)
316                         diff_index_show_file(revs, "-", old,
317                                              old->sha1, 1, old->ce_mode, 0);
318                 return -1;
319         }
320
321         if (revs->combine_merges && !cached &&
322             (hashcmp(sha1, old->sha1) || hashcmp(old->sha1, new->sha1))) {
323                 struct combine_diff_path *p;
324                 int pathlen = ce_namelen(new);
325
326                 p = xmalloc(combine_diff_path_size(2, pathlen));
327                 p->path = (char *) &p->parent[2];
328                 p->next = NULL;
329                 p->len = pathlen;
330                 memcpy(p->path, new->name, pathlen);
331                 p->path[pathlen] = 0;
332                 p->mode = mode;
333                 hashclr(p->sha1);
334                 memset(p->parent, 0, 2 * sizeof(struct combine_diff_parent));
335                 p->parent[0].status = DIFF_STATUS_MODIFIED;
336                 p->parent[0].mode = new->ce_mode;
337                 hashcpy(p->parent[0].sha1, new->sha1);
338                 p->parent[1].status = DIFF_STATUS_MODIFIED;
339                 p->parent[1].mode = old->ce_mode;
340                 hashcpy(p->parent[1].sha1, old->sha1);
341                 show_combined_diff(p, 2, revs->dense_combined_merges, revs);
342                 free(p);
343                 return 0;
344         }
345
346         oldmode = old->ce_mode;
347         if (mode == oldmode && !hashcmp(sha1, old->sha1) && !dirty_submodule &&
348             !DIFF_OPT_TST(&revs->diffopt, FIND_COPIES_HARDER))
349                 return 0;
350
351         diff_change(&revs->diffopt, oldmode, mode,
352                     old->sha1, sha1, 1, !is_null_sha1(sha1),
353                     old->name, 0, dirty_submodule);
354         return 0;
355 }
356
357 /*
358  * This gets a mix of an existing index and a tree, one pathname entry
359  * at a time. The index entry may be a single stage-0 one, but it could
360  * also be multiple unmerged entries (in which case idx_pos/idx_nr will
361  * give you the position and number of entries in the index).
362  */
363 static void do_oneway_diff(struct unpack_trees_options *o,
364                            const struct cache_entry *idx,
365                            const struct cache_entry *tree)
366 {
367         struct rev_info *revs = o->unpack_data;
368         int match_missing, cached;
369
370         /* if the entry is not checked out, don't examine work tree */
371         cached = o->index_only ||
372                 (idx && ((idx->ce_flags & CE_VALID) || ce_skip_worktree(idx)));
373         /*
374          * Backward compatibility wart - "diff-index -m" does
375          * not mean "do not ignore merges", but "match_missing".
376          *
377          * But with the revision flag parsing, that's found in
378          * "!revs->ignore_merges".
379          */
380         match_missing = !revs->ignore_merges;
381
382         if (cached && idx && ce_stage(idx)) {
383                 struct diff_filepair *pair;
384                 pair = diff_unmerge(&revs->diffopt, idx->name);
385                 if (tree)
386                         fill_filespec(pair->one, tree->sha1, 1, tree->ce_mode);
387                 return;
388         }
389
390         /*
391          * Something added to the tree?
392          */
393         if (!tree) {
394                 show_new_file(revs, idx, cached, match_missing);
395                 return;
396         }
397
398         /*
399          * Something removed from the tree?
400          */
401         if (!idx) {
402                 diff_index_show_file(revs, "-", tree, tree->sha1, 1, tree->ce_mode, 0);
403                 return;
404         }
405
406         /* Show difference between old and new */
407         show_modified(revs, tree, idx, 1, cached, match_missing);
408 }
409
410 /*
411  * The unpack_trees() interface is designed for merging, so
412  * the different source entries are designed primarily for
413  * the source trees, with the old index being really mainly
414  * used for being replaced by the result.
415  *
416  * For diffing, the index is more important, and we only have a
417  * single tree.
418  *
419  * We're supposed to advance o->pos to skip what we have already processed.
420  *
421  * This wrapper makes it all more readable, and takes care of all
422  * the fairly complex unpack_trees() semantic requirements, including
423  * the skipping, the path matching, the type conflict cases etc.
424  */
425 static int oneway_diff(const struct cache_entry * const *src,
426                        struct unpack_trees_options *o)
427 {
428         const struct cache_entry *idx = src[0];
429         const struct cache_entry *tree = src[1];
430         struct rev_info *revs = o->unpack_data;
431
432         /*
433          * Unpack-trees generates a DF/conflict entry if
434          * there was a directory in the index and a tree
435          * in the tree. From a diff standpoint, that's a
436          * delete of the tree and a create of the file.
437          */
438         if (tree == o->df_conflict_entry)
439                 tree = NULL;
440
441         if (ce_path_match(idx ? idx : tree, &revs->prune_data)) {
442                 do_oneway_diff(o, idx, tree);
443                 if (diff_can_quit_early(&revs->diffopt)) {
444                         o->exiting_early = 1;
445                         return -1;
446                 }
447         }
448
449         return 0;
450 }
451
452 static int diff_cache(struct rev_info *revs,
453                       const unsigned char *tree_sha1,
454                       const char *tree_name,
455                       int cached)
456 {
457         struct tree *tree;
458         struct tree_desc t;
459         struct unpack_trees_options opts;
460
461         tree = parse_tree_indirect(tree_sha1);
462         if (!tree)
463                 return error("bad tree object %s",
464                              tree_name ? tree_name : sha1_to_hex(tree_sha1));
465         memset(&opts, 0, sizeof(opts));
466         opts.head_idx = 1;
467         opts.index_only = cached;
468         opts.diff_index_cached = (cached &&
469                                   !DIFF_OPT_TST(&revs->diffopt, FIND_COPIES_HARDER));
470         opts.merge = 1;
471         opts.fn = oneway_diff;
472         opts.unpack_data = revs;
473         opts.src_index = &the_index;
474         opts.dst_index = NULL;
475         opts.pathspec = &revs->diffopt.pathspec;
476         opts.pathspec->recursive = 1;
477
478         init_tree_desc(&t, tree->buffer, tree->size);
479         return unpack_trees(1, &t, &opts);
480 }
481
482 int run_diff_index(struct rev_info *revs, int cached)
483 {
484         struct object_array_entry *ent;
485
486         ent = revs->pending.objects;
487         if (diff_cache(revs, ent->item->sha1, ent->name, cached))
488                 exit(128);
489
490         diff_set_mnemonic_prefix(&revs->diffopt, "c/", cached ? "i/" : "w/");
491         diffcore_fix_diff_index(&revs->diffopt);
492         diffcore_std(&revs->diffopt);
493         diff_flush(&revs->diffopt);
494         return 0;
495 }
496
497 int do_diff_cache(const unsigned char *tree_sha1, struct diff_options *opt)
498 {
499         struct rev_info revs;
500
501         init_revisions(&revs, NULL);
502         copy_pathspec(&revs.prune_data, &opt->pathspec);
503         revs.diffopt = *opt;
504
505         if (diff_cache(&revs, tree_sha1, NULL, 1))
506                 exit(128);
507         return 0;
508 }
509
510 int index_differs_from(const char *def, int diff_flags)
511 {
512         struct rev_info rev;
513         struct setup_revision_opt opt;
514
515         init_revisions(&rev, NULL);
516         memset(&opt, 0, sizeof(opt));
517         opt.def = def;
518         setup_revisions(0, NULL, &rev, &opt);
519         DIFF_OPT_SET(&rev.diffopt, QUICK);
520         DIFF_OPT_SET(&rev.diffopt, EXIT_WITH_STATUS);
521         rev.diffopt.flags |= diff_flags;
522         run_diff_index(&rev, 1);
523         if (rev.pending.alloc)
524                 free(rev.pending.objects);
525         return (DIFF_OPT_TST(&rev.diffopt, HAS_CHANGES) != 0);
526 }