diff-parseopt: convert --word-diff-regex
[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 #include "dir.h"
15 #include "fsmonitor.h"
16
17 /*
18  * diff-files
19  */
20
21 /*
22  * Has the work tree entity been removed?
23  *
24  * Return 1 if it was removed from the work tree, 0 if an entity to be
25  * compared with the cache entry ce still exists (the latter includes
26  * the case where a directory that is not a submodule repository
27  * exists for ce that is a submodule -- it is a submodule that is not
28  * checked out).  Return negative for an error.
29  */
30 static int check_removed(const struct cache_entry *ce, struct stat *st)
31 {
32         if (lstat(ce->name, st) < 0) {
33                 if (!is_missing_file_error(errno))
34                         return -1;
35                 return 1;
36         }
37         if (has_symlink_leading_path(ce->name, ce_namelen(ce)))
38                 return 1;
39         if (S_ISDIR(st->st_mode)) {
40                 struct object_id sub;
41
42                 /*
43                  * If ce is already a gitlink, we can have a plain
44                  * directory (i.e. the submodule is not checked out),
45                  * or a checked out submodule.  Either case this is not
46                  * a case where something was removed from the work tree,
47                  * so we will return 0.
48                  *
49                  * Otherwise, if the directory is not a submodule
50                  * repository, that means ce which was a blob turned into
51                  * a directory --- the blob was removed!
52                  */
53                 if (!S_ISGITLINK(ce->ce_mode) &&
54                     resolve_gitlink_ref(ce->name, "HEAD", &sub))
55                         return 1;
56         }
57         return 0;
58 }
59
60 /*
61  * Has a file changed or has a submodule new commits or a dirty work tree?
62  *
63  * Return 1 when changes are detected, 0 otherwise. If the DIRTY_SUBMODULES
64  * option is set, the caller does not only want to know if a submodule is
65  * modified at all but wants to know all the conditions that are met (new
66  * commits, untracked content and/or modified content).
67  */
68 static int match_stat_with_submodule(struct diff_options *diffopt,
69                                      const struct cache_entry *ce,
70                                      struct stat *st, unsigned ce_option,
71                                      unsigned *dirty_submodule)
72 {
73         int changed = ie_match_stat(diffopt->repo->index, ce, st, ce_option);
74         if (S_ISGITLINK(ce->ce_mode)) {
75                 struct diff_flags orig_flags = diffopt->flags;
76                 if (!diffopt->flags.override_submodule_config)
77                         set_diffopt_flags_from_submodule_config(diffopt, ce->name);
78                 if (diffopt->flags.ignore_submodules)
79                         changed = 0;
80                 else if (!diffopt->flags.ignore_dirty_submodules &&
81                          (!changed || diffopt->flags.dirty_submodules))
82                         *dirty_submodule = is_submodule_modified(ce->name,
83                                                                  diffopt->flags.ignore_untracked_in_submodules);
84                 diffopt->flags = orig_flags;
85         }
86         return changed;
87 }
88
89 int run_diff_files(struct rev_info *revs, unsigned int option)
90 {
91         int entries, i;
92         int diff_unmerged_stage = revs->max_count;
93         unsigned ce_option = ((option & DIFF_RACY_IS_MODIFIED)
94                               ? CE_MATCH_RACY_IS_DIRTY : 0);
95         uint64_t start = getnanotime();
96         struct index_state *istate = revs->diffopt.repo->index;
97
98         diff_set_mnemonic_prefix(&revs->diffopt, "i/", "w/");
99
100         if (diff_unmerged_stage < 0)
101                 diff_unmerged_stage = 2;
102         entries = istate->cache_nr;
103         for (i = 0; i < entries; i++) {
104                 unsigned int oldmode, newmode;
105                 struct cache_entry *ce = istate->cache[i];
106                 int changed;
107                 unsigned dirty_submodule = 0;
108                 const struct object_id *old_oid, *new_oid;
109
110                 if (diff_can_quit_early(&revs->diffopt))
111                         break;
112
113                 if (!ce_path_match(istate, ce, &revs->prune_data, NULL))
114                         continue;
115
116                 if (ce_stage(ce)) {
117                         struct combine_diff_path *dpath;
118                         struct diff_filepair *pair;
119                         unsigned int wt_mode = 0;
120                         int num_compare_stages = 0;
121                         size_t path_len;
122                         struct stat st;
123
124                         path_len = ce_namelen(ce);
125
126                         dpath = xmalloc(combine_diff_path_size(5, path_len));
127                         dpath->path = (char *) &(dpath->parent[5]);
128
129                         dpath->next = NULL;
130                         memcpy(dpath->path, ce->name, path_len);
131                         dpath->path[path_len] = '\0';
132                         oidclr(&dpath->oid);
133                         memset(&(dpath->parent[0]), 0,
134                                sizeof(struct combine_diff_parent)*5);
135
136                         changed = check_removed(ce, &st);
137                         if (!changed)
138                                 wt_mode = ce_mode_from_stat(ce, st.st_mode);
139                         else {
140                                 if (changed < 0) {
141                                         perror(ce->name);
142                                         continue;
143                                 }
144                                 wt_mode = 0;
145                         }
146                         dpath->mode = wt_mode;
147
148                         while (i < entries) {
149                                 struct cache_entry *nce = istate->cache[i];
150                                 int stage;
151
152                                 if (strcmp(ce->name, nce->name))
153                                         break;
154
155                                 /* Stage #2 (ours) is the first parent,
156                                  * stage #3 (theirs) is the second.
157                                  */
158                                 stage = ce_stage(nce);
159                                 if (2 <= stage) {
160                                         int mode = nce->ce_mode;
161                                         num_compare_stages++;
162                                         oidcpy(&dpath->parent[stage - 2].oid,
163                                                &nce->oid);
164                                         dpath->parent[stage-2].mode = ce_mode_from_stat(nce, mode);
165                                         dpath->parent[stage-2].status =
166                                                 DIFF_STATUS_MODIFIED;
167                                 }
168
169                                 /* diff against the proper unmerged stage */
170                                 if (stage == diff_unmerged_stage)
171                                         ce = nce;
172                                 i++;
173                         }
174                         /*
175                          * Compensate for loop update
176                          */
177                         i--;
178
179                         if (revs->combine_merges && num_compare_stages == 2) {
180                                 show_combined_diff(dpath, 2,
181                                                    revs->dense_combined_merges,
182                                                    revs);
183                                 free(dpath);
184                                 continue;
185                         }
186                         FREE_AND_NULL(dpath);
187
188                         /*
189                          * Show the diff for the 'ce' if we found the one
190                          * from the desired stage.
191                          */
192                         pair = diff_unmerge(&revs->diffopt, ce->name);
193                         if (wt_mode)
194                                 pair->two->mode = wt_mode;
195                         if (ce_stage(ce) != diff_unmerged_stage)
196                                 continue;
197                 }
198
199                 if (ce_uptodate(ce) || ce_skip_worktree(ce))
200                         continue;
201
202                 /* If CE_VALID is set, don't look at workdir for file removal */
203                 if (ce->ce_flags & CE_VALID) {
204                         changed = 0;
205                         newmode = ce->ce_mode;
206                 } else {
207                         struct stat st;
208
209                         changed = check_removed(ce, &st);
210                         if (changed) {
211                                 if (changed < 0) {
212                                         perror(ce->name);
213                                         continue;
214                                 }
215                                 diff_addremove(&revs->diffopt, '-', ce->ce_mode,
216                                                &ce->oid,
217                                                !is_null_oid(&ce->oid),
218                                                ce->name, 0);
219                                 continue;
220                         } else if (revs->diffopt.ita_invisible_in_index &&
221                                    ce_intent_to_add(ce)) {
222                                 diff_addremove(&revs->diffopt, '+', ce->ce_mode,
223                                                the_hash_algo->empty_tree, 0,
224                                                ce->name, 0);
225                                 continue;
226                         }
227
228                         changed = match_stat_with_submodule(&revs->diffopt, ce, &st,
229                                                             ce_option, &dirty_submodule);
230                         newmode = ce_mode_from_stat(ce, st.st_mode);
231                 }
232
233                 if (!changed && !dirty_submodule) {
234                         ce_mark_uptodate(ce);
235                         mark_fsmonitor_valid(ce);
236                         if (!revs->diffopt.flags.find_copies_harder)
237                                 continue;
238                 }
239                 oldmode = ce->ce_mode;
240                 old_oid = &ce->oid;
241                 new_oid = changed ? &null_oid : &ce->oid;
242                 diff_change(&revs->diffopt, oldmode, newmode,
243                             old_oid, new_oid,
244                             !is_null_oid(old_oid),
245                             !is_null_oid(new_oid),
246                             ce->name, 0, dirty_submodule);
247
248         }
249         diffcore_std(&revs->diffopt);
250         diff_flush(&revs->diffopt);
251         trace_performance_since(start, "diff-files");
252         return 0;
253 }
254
255 /*
256  * diff-index
257  */
258
259 /* A file entry went away or appeared */
260 static void diff_index_show_file(struct rev_info *revs,
261                                  const char *prefix,
262                                  const struct cache_entry *ce,
263                                  const struct object_id *oid, int oid_valid,
264                                  unsigned int mode,
265                                  unsigned dirty_submodule)
266 {
267         diff_addremove(&revs->diffopt, prefix[0], mode,
268                        oid, oid_valid, ce->name, dirty_submodule);
269 }
270
271 static int get_stat_data(const struct cache_entry *ce,
272                          const struct object_id **oidp,
273                          unsigned int *modep,
274                          int cached, int match_missing,
275                          unsigned *dirty_submodule, struct diff_options *diffopt)
276 {
277         const struct object_id *oid = &ce->oid;
278         unsigned int mode = ce->ce_mode;
279
280         if (!cached && !ce_uptodate(ce)) {
281                 int changed;
282                 struct stat st;
283                 changed = check_removed(ce, &st);
284                 if (changed < 0)
285                         return -1;
286                 else if (changed) {
287                         if (match_missing) {
288                                 *oidp = oid;
289                                 *modep = mode;
290                                 return 0;
291                         }
292                         return -1;
293                 }
294                 changed = match_stat_with_submodule(diffopt, ce, &st,
295                                                     0, dirty_submodule);
296                 if (changed) {
297                         mode = ce_mode_from_stat(ce, st.st_mode);
298                         oid = &null_oid;
299                 }
300         }
301
302         *oidp = oid;
303         *modep = mode;
304         return 0;
305 }
306
307 static void show_new_file(struct rev_info *revs,
308                           const struct cache_entry *new_file,
309                           int cached, int match_missing)
310 {
311         const struct object_id *oid;
312         unsigned int mode;
313         unsigned dirty_submodule = 0;
314
315         /*
316          * New file in the index: it might actually be different in
317          * the working tree.
318          */
319         if (get_stat_data(new_file, &oid, &mode, cached, match_missing,
320             &dirty_submodule, &revs->diffopt) < 0)
321                 return;
322
323         diff_index_show_file(revs, "+", new_file, oid, !is_null_oid(oid), mode, dirty_submodule);
324 }
325
326 static int show_modified(struct rev_info *revs,
327                          const struct cache_entry *old_entry,
328                          const struct cache_entry *new_entry,
329                          int report_missing,
330                          int cached, int match_missing)
331 {
332         unsigned int mode, oldmode;
333         const struct object_id *oid;
334         unsigned dirty_submodule = 0;
335
336         if (get_stat_data(new_entry, &oid, &mode, cached, match_missing,
337                           &dirty_submodule, &revs->diffopt) < 0) {
338                 if (report_missing)
339                         diff_index_show_file(revs, "-", old_entry,
340                                              &old_entry->oid, 1, old_entry->ce_mode,
341                                              0);
342                 return -1;
343         }
344
345         if (revs->combine_merges && !cached &&
346             (!oideq(oid, &old_entry->oid) || !oideq(&old_entry->oid, &new_entry->oid))) {
347                 struct combine_diff_path *p;
348                 int pathlen = ce_namelen(new_entry);
349
350                 p = xmalloc(combine_diff_path_size(2, pathlen));
351                 p->path = (char *) &p->parent[2];
352                 p->next = NULL;
353                 memcpy(p->path, new_entry->name, pathlen);
354                 p->path[pathlen] = 0;
355                 p->mode = mode;
356                 oidclr(&p->oid);
357                 memset(p->parent, 0, 2 * sizeof(struct combine_diff_parent));
358                 p->parent[0].status = DIFF_STATUS_MODIFIED;
359                 p->parent[0].mode = new_entry->ce_mode;
360                 oidcpy(&p->parent[0].oid, &new_entry->oid);
361                 p->parent[1].status = DIFF_STATUS_MODIFIED;
362                 p->parent[1].mode = old_entry->ce_mode;
363                 oidcpy(&p->parent[1].oid, &old_entry->oid);
364                 show_combined_diff(p, 2, revs->dense_combined_merges, revs);
365                 free(p);
366                 return 0;
367         }
368
369         oldmode = old_entry->ce_mode;
370         if (mode == oldmode && oideq(oid, &old_entry->oid) && !dirty_submodule &&
371             !revs->diffopt.flags.find_copies_harder)
372                 return 0;
373
374         diff_change(&revs->diffopt, oldmode, mode,
375                     &old_entry->oid, oid, 1, !is_null_oid(oid),
376                     old_entry->name, 0, dirty_submodule);
377         return 0;
378 }
379
380 /*
381  * This gets a mix of an existing index and a tree, one pathname entry
382  * at a time. The index entry may be a single stage-0 one, but it could
383  * also be multiple unmerged entries (in which case idx_pos/idx_nr will
384  * give you the position and number of entries in the index).
385  */
386 static void do_oneway_diff(struct unpack_trees_options *o,
387                            const struct cache_entry *idx,
388                            const struct cache_entry *tree)
389 {
390         struct rev_info *revs = o->unpack_data;
391         int match_missing, cached;
392
393         /*
394          * i-t-a entries do not actually exist in the index (if we're
395          * looking at its content)
396          */
397         if (o->index_only &&
398             revs->diffopt.ita_invisible_in_index &&
399             idx && ce_intent_to_add(idx)) {
400                 idx = NULL;
401                 if (!tree)
402                         return; /* nothing to diff.. */
403         }
404
405         /* if the entry is not checked out, don't examine work tree */
406         cached = o->index_only ||
407                 (idx && ((idx->ce_flags & CE_VALID) || ce_skip_worktree(idx)));
408         /*
409          * Backward compatibility wart - "diff-index -m" does
410          * not mean "do not ignore merges", but "match_missing".
411          *
412          * But with the revision flag parsing, that's found in
413          * "!revs->ignore_merges".
414          */
415         match_missing = !revs->ignore_merges;
416
417         if (cached && idx && ce_stage(idx)) {
418                 struct diff_filepair *pair;
419                 pair = diff_unmerge(&revs->diffopt, idx->name);
420                 if (tree)
421                         fill_filespec(pair->one, &tree->oid, 1,
422                                       tree->ce_mode);
423                 return;
424         }
425
426         /*
427          * Something added to the tree?
428          */
429         if (!tree) {
430                 show_new_file(revs, idx, cached, match_missing);
431                 return;
432         }
433
434         /*
435          * Something removed from the tree?
436          */
437         if (!idx) {
438                 diff_index_show_file(revs, "-", tree, &tree->oid, 1,
439                                      tree->ce_mode, 0);
440                 return;
441         }
442
443         /* Show difference between old and new */
444         show_modified(revs, tree, idx, 1, cached, match_missing);
445 }
446
447 /*
448  * The unpack_trees() interface is designed for merging, so
449  * the different source entries are designed primarily for
450  * the source trees, with the old index being really mainly
451  * used for being replaced by the result.
452  *
453  * For diffing, the index is more important, and we only have a
454  * single tree.
455  *
456  * We're supposed to advance o->pos to skip what we have already processed.
457  *
458  * This wrapper makes it all more readable, and takes care of all
459  * the fairly complex unpack_trees() semantic requirements, including
460  * the skipping, the path matching, the type conflict cases etc.
461  */
462 static int oneway_diff(const struct cache_entry * const *src,
463                        struct unpack_trees_options *o)
464 {
465         const struct cache_entry *idx = src[0];
466         const struct cache_entry *tree = src[1];
467         struct rev_info *revs = o->unpack_data;
468
469         /*
470          * Unpack-trees generates a DF/conflict entry if
471          * there was a directory in the index and a tree
472          * in the tree. From a diff standpoint, that's a
473          * delete of the tree and a create of the file.
474          */
475         if (tree == o->df_conflict_entry)
476                 tree = NULL;
477
478         if (ce_path_match(revs->diffopt.repo->index,
479                           idx ? idx : tree,
480                           &revs->prune_data, NULL)) {
481                 do_oneway_diff(o, idx, tree);
482                 if (diff_can_quit_early(&revs->diffopt)) {
483                         o->exiting_early = 1;
484                         return -1;
485                 }
486         }
487
488         return 0;
489 }
490
491 static int diff_cache(struct rev_info *revs,
492                       const struct object_id *tree_oid,
493                       const char *tree_name,
494                       int cached)
495 {
496         struct tree *tree;
497         struct tree_desc t;
498         struct unpack_trees_options opts;
499
500         tree = parse_tree_indirect(tree_oid);
501         if (!tree)
502                 return error("bad tree object %s",
503                              tree_name ? tree_name : oid_to_hex(tree_oid));
504         memset(&opts, 0, sizeof(opts));
505         opts.head_idx = 1;
506         opts.index_only = cached;
507         opts.diff_index_cached = (cached &&
508                                   !revs->diffopt.flags.find_copies_harder);
509         opts.merge = 1;
510         opts.fn = oneway_diff;
511         opts.unpack_data = revs;
512         opts.src_index = revs->diffopt.repo->index;
513         opts.dst_index = NULL;
514         opts.pathspec = &revs->diffopt.pathspec;
515         opts.pathspec->recursive = 1;
516
517         init_tree_desc(&t, tree->buffer, tree->size);
518         return unpack_trees(1, &t, &opts);
519 }
520
521 int run_diff_index(struct rev_info *revs, int cached)
522 {
523         struct object_array_entry *ent;
524
525         if (revs->pending.nr != 1)
526                 BUG("run_diff_index must be passed exactly one tree");
527
528         trace_performance_enter();
529         ent = revs->pending.objects;
530         if (diff_cache(revs, &ent->item->oid, ent->name, cached))
531                 exit(128);
532
533         diff_set_mnemonic_prefix(&revs->diffopt, "c/", cached ? "i/" : "w/");
534         diffcore_fix_diff_index(&revs->diffopt);
535         diffcore_std(&revs->diffopt);
536         diff_flush(&revs->diffopt);
537         trace_performance_leave("diff-index");
538         return 0;
539 }
540
541 int do_diff_cache(const struct object_id *tree_oid, struct diff_options *opt)
542 {
543         struct rev_info revs;
544
545         repo_init_revisions(opt->repo, &revs, NULL);
546         copy_pathspec(&revs.prune_data, &opt->pathspec);
547         revs.diffopt = *opt;
548
549         if (diff_cache(&revs, tree_oid, NULL, 1))
550                 exit(128);
551         return 0;
552 }
553
554 int index_differs_from(struct repository *r,
555                        const char *def, const struct diff_flags *flags,
556                        int ita_invisible_in_index)
557 {
558         struct rev_info rev;
559         struct setup_revision_opt opt;
560
561         repo_init_revisions(r, &rev, NULL);
562         memset(&opt, 0, sizeof(opt));
563         opt.def = def;
564         setup_revisions(0, NULL, &rev, &opt);
565         rev.diffopt.flags.quick = 1;
566         rev.diffopt.flags.exit_with_status = 1;
567         if (flags)
568                 diff_flags_or(&rev.diffopt.flags, flags);
569         rev.diffopt.ita_invisible_in_index = ita_invisible_in_index;
570         run_diff_index(&rev, 1);
571         object_array_clear(&rev.pending);
572         return (rev.diffopt.flags.has_changes != 0);
573 }