log -L: store the path instead of a diff_filespec
[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, !is_null_sha1(ce->sha1),
210                                        ce->name, 0);
211                         continue;
212                 }
213                 changed = match_stat_with_submodule(&revs->diffopt, ce, &st,
214                                                     ce_option, &dirty_submodule);
215                 if (!changed && !dirty_submodule) {
216                         ce_mark_uptodate(ce);
217                         if (!DIFF_OPT_TST(&revs->diffopt, FIND_COPIES_HARDER))
218                                 continue;
219                 }
220                 oldmode = ce->ce_mode;
221                 newmode = ce_mode_from_stat(ce, st.st_mode);
222                 diff_change(&revs->diffopt, oldmode, newmode,
223                             ce->sha1, (changed ? null_sha1 : ce->sha1),
224                             !is_null_sha1(ce->sha1), (changed ? 0 : !is_null_sha1(ce->sha1)),
225                             ce->name, 0, dirty_submodule);
226
227         }
228         diffcore_std(&revs->diffopt);
229         diff_flush(&revs->diffopt);
230         return 0;
231 }
232
233 /*
234  * diff-index
235  */
236
237 /* A file entry went away or appeared */
238 static void diff_index_show_file(struct rev_info *revs,
239                                  const char *prefix,
240                                  struct cache_entry *ce,
241                                  const unsigned char *sha1, int sha1_valid,
242                                  unsigned int mode,
243                                  unsigned dirty_submodule)
244 {
245         diff_addremove(&revs->diffopt, prefix[0], mode,
246                        sha1, sha1_valid, ce->name, dirty_submodule);
247 }
248
249 static int get_stat_data(struct cache_entry *ce,
250                          const unsigned char **sha1p,
251                          unsigned int *modep,
252                          int cached, int match_missing,
253                          unsigned *dirty_submodule, struct diff_options *diffopt)
254 {
255         const unsigned char *sha1 = ce->sha1;
256         unsigned int mode = ce->ce_mode;
257
258         if (!cached && !ce_uptodate(ce)) {
259                 int changed;
260                 struct stat st;
261                 changed = check_removed(ce, &st);
262                 if (changed < 0)
263                         return -1;
264                 else if (changed) {
265                         if (match_missing) {
266                                 *sha1p = sha1;
267                                 *modep = mode;
268                                 return 0;
269                         }
270                         return -1;
271                 }
272                 changed = match_stat_with_submodule(diffopt, ce, &st,
273                                                     0, dirty_submodule);
274                 if (changed) {
275                         mode = ce_mode_from_stat(ce, st.st_mode);
276                         sha1 = null_sha1;
277                 }
278         }
279
280         *sha1p = sha1;
281         *modep = mode;
282         return 0;
283 }
284
285 static void show_new_file(struct rev_info *revs,
286                           struct cache_entry *new,
287                           int cached, int match_missing)
288 {
289         const unsigned char *sha1;
290         unsigned int mode;
291         unsigned dirty_submodule = 0;
292
293         /*
294          * New file in the index: it might actually be different in
295          * the working tree.
296          */
297         if (get_stat_data(new, &sha1, &mode, cached, match_missing,
298             &dirty_submodule, &revs->diffopt) < 0)
299                 return;
300
301         diff_index_show_file(revs, "+", new, sha1, !is_null_sha1(sha1), mode, dirty_submodule);
302 }
303
304 static int show_modified(struct rev_info *revs,
305                          struct cache_entry *old,
306                          struct cache_entry *new,
307                          int report_missing,
308                          int cached, int match_missing)
309 {
310         unsigned int mode, oldmode;
311         const unsigned char *sha1;
312         unsigned dirty_submodule = 0;
313
314         if (get_stat_data(new, &sha1, &mode, cached, match_missing,
315                           &dirty_submodule, &revs->diffopt) < 0) {
316                 if (report_missing)
317                         diff_index_show_file(revs, "-", old,
318                                              old->sha1, 1, old->ce_mode, 0);
319                 return -1;
320         }
321
322         if (revs->combine_merges && !cached &&
323             (hashcmp(sha1, old->sha1) || hashcmp(old->sha1, new->sha1))) {
324                 struct combine_diff_path *p;
325                 int pathlen = ce_namelen(new);
326
327                 p = xmalloc(combine_diff_path_size(2, pathlen));
328                 p->path = (char *) &p->parent[2];
329                 p->next = NULL;
330                 p->len = pathlen;
331                 memcpy(p->path, new->name, pathlen);
332                 p->path[pathlen] = 0;
333                 p->mode = mode;
334                 hashclr(p->sha1);
335                 memset(p->parent, 0, 2 * sizeof(struct combine_diff_parent));
336                 p->parent[0].status = DIFF_STATUS_MODIFIED;
337                 p->parent[0].mode = new->ce_mode;
338                 hashcpy(p->parent[0].sha1, new->sha1);
339                 p->parent[1].status = DIFF_STATUS_MODIFIED;
340                 p->parent[1].mode = old->ce_mode;
341                 hashcpy(p->parent[1].sha1, old->sha1);
342                 show_combined_diff(p, 2, revs->dense_combined_merges, revs);
343                 free(p);
344                 return 0;
345         }
346
347         oldmode = old->ce_mode;
348         if (mode == oldmode && !hashcmp(sha1, old->sha1) && !dirty_submodule &&
349             !DIFF_OPT_TST(&revs->diffopt, FIND_COPIES_HARDER))
350                 return 0;
351
352         diff_change(&revs->diffopt, oldmode, mode,
353                     old->sha1, sha1, 1, !is_null_sha1(sha1),
354                     old->name, 0, dirty_submodule);
355         return 0;
356 }
357
358 /*
359  * This gets a mix of an existing index and a tree, one pathname entry
360  * at a time. The index entry may be a single stage-0 one, but it could
361  * also be multiple unmerged entries (in which case idx_pos/idx_nr will
362  * give you the position and number of entries in the index).
363  */
364 static void do_oneway_diff(struct unpack_trees_options *o,
365         struct cache_entry *idx,
366         struct cache_entry *tree)
367 {
368         struct rev_info *revs = o->unpack_data;
369         int match_missing, cached;
370
371         /* if the entry is not checked out, don't examine work tree */
372         cached = o->index_only ||
373                 (idx && ((idx->ce_flags & CE_VALID) || ce_skip_worktree(idx)));
374         /*
375          * Backward compatibility wart - "diff-index -m" does
376          * not mean "do not ignore merges", but "match_missing".
377          *
378          * But with the revision flag parsing, that's found in
379          * "!revs->ignore_merges".
380          */
381         match_missing = !revs->ignore_merges;
382
383         if (cached && idx && ce_stage(idx)) {
384                 struct diff_filepair *pair;
385                 pair = diff_unmerge(&revs->diffopt, idx->name);
386                 if (tree)
387                         fill_filespec(pair->one, tree->sha1, 1, tree->ce_mode);
388                 return;
389         }
390
391         /*
392          * Something added to the tree?
393          */
394         if (!tree) {
395                 show_new_file(revs, idx, cached, match_missing);
396                 return;
397         }
398
399         /*
400          * Something removed from the tree?
401          */
402         if (!idx) {
403                 diff_index_show_file(revs, "-", tree, tree->sha1, 1, tree->ce_mode, 0);
404                 return;
405         }
406
407         /* Show difference between old and new */
408         show_modified(revs, tree, idx, 1, cached, match_missing);
409 }
410
411 /*
412  * The unpack_trees() interface is designed for merging, so
413  * the different source entries are designed primarily for
414  * the source trees, with the old index being really mainly
415  * used for being replaced by the result.
416  *
417  * For diffing, the index is more important, and we only have a
418  * single tree.
419  *
420  * We're supposed to advance o->pos to skip what we have already processed.
421  *
422  * This wrapper makes it all more readable, and takes care of all
423  * the fairly complex unpack_trees() semantic requirements, including
424  * the skipping, the path matching, the type conflict cases etc.
425  */
426 static int oneway_diff(struct cache_entry **src, struct unpack_trees_options *o)
427 {
428         struct cache_entry *idx = src[0];
429         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         opts.pathspec->max_depth = -1;
478
479         init_tree_desc(&t, tree->buffer, tree->size);
480         return unpack_trees(1, &t, &opts);
481 }
482
483 int run_diff_index(struct rev_info *revs, int cached)
484 {
485         struct object_array_entry *ent;
486
487         ent = revs->pending.objects;
488         if (diff_cache(revs, ent->item->sha1, ent->name, cached))
489                 exit(128);
490
491         diff_set_mnemonic_prefix(&revs->diffopt, "c/", cached ? "i/" : "w/");
492         diffcore_fix_diff_index(&revs->diffopt);
493         diffcore_std(&revs->diffopt);
494         diff_flush(&revs->diffopt);
495         return 0;
496 }
497
498 int do_diff_cache(const unsigned char *tree_sha1, struct diff_options *opt)
499 {
500         struct rev_info revs;
501
502         init_revisions(&revs, NULL);
503         init_pathspec(&revs.prune_data, opt->pathspec.raw);
504         revs.diffopt = *opt;
505
506         if (diff_cache(&revs, tree_sha1, NULL, 1))
507                 exit(128);
508         return 0;
509 }
510
511 int index_differs_from(const char *def, int diff_flags)
512 {
513         struct rev_info rev;
514         struct setup_revision_opt opt;
515
516         init_revisions(&rev, NULL);
517         memset(&opt, 0, sizeof(opt));
518         opt.def = def;
519         setup_revisions(0, NULL, &rev, &opt);
520         DIFF_OPT_SET(&rev.diffopt, QUICK);
521         DIFF_OPT_SET(&rev.diffopt, EXIT_WITH_STATUS);
522         rev.diffopt.flags |= diff_flags;
523         run_diff_index(&rev, 1);
524         if (rev.pending.alloc)
525                 free(rev.pending.objects);
526         return (DIFF_OPT_TST(&rev.diffopt, HAS_CHANGES) != 0);
527 }