userdiff/perl: tighten BEGIN/END block pattern to reject here-doc delimiters
[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                     !revs->diffopt.filter &&
107                     DIFF_OPT_TST(&revs->diffopt, HAS_CHANGES))
108                         break;
109
110                 if (!ce_path_match(ce, &revs->prune_data))
111                         continue;
112
113                 if (ce_stage(ce)) {
114                         struct combine_diff_path *dpath;
115                         struct diff_filepair *pair;
116                         unsigned int wt_mode = 0;
117                         int num_compare_stages = 0;
118                         size_t path_len;
119
120                         path_len = ce_namelen(ce);
121
122                         dpath = xmalloc(combine_diff_path_size(5, path_len));
123                         dpath->path = (char *) &(dpath->parent[5]);
124
125                         dpath->next = NULL;
126                         dpath->len = path_len;
127                         memcpy(dpath->path, ce->name, path_len);
128                         dpath->path[path_len] = '\0';
129                         hashclr(dpath->sha1);
130                         memset(&(dpath->parent[0]), 0,
131                                sizeof(struct combine_diff_parent)*5);
132
133                         changed = check_removed(ce, &st);
134                         if (!changed)
135                                 wt_mode = ce_mode_from_stat(ce, st.st_mode);
136                         else {
137                                 if (changed < 0) {
138                                         perror(ce->name);
139                                         continue;
140                                 }
141                                 if (silent_on_removed)
142                                         continue;
143                                 wt_mode = 0;
144                         }
145                         dpath->mode = wt_mode;
146
147                         while (i < entries) {
148                                 struct cache_entry *nce = active_cache[i];
149                                 int stage;
150
151                                 if (strcmp(ce->name, nce->name))
152                                         break;
153
154                                 /* Stage #2 (ours) is the first parent,
155                                  * stage #3 (theirs) is the second.
156                                  */
157                                 stage = ce_stage(nce);
158                                 if (2 <= stage) {
159                                         int mode = nce->ce_mode;
160                                         num_compare_stages++;
161                                         hashcpy(dpath->parent[stage-2].sha1, nce->sha1);
162                                         dpath->parent[stage-2].mode = ce_mode_from_stat(nce, mode);
163                                         dpath->parent[stage-2].status =
164                                                 DIFF_STATUS_MODIFIED;
165                                 }
166
167                                 /* diff against the proper unmerged stage */
168                                 if (stage == diff_unmerged_stage)
169                                         ce = nce;
170                                 i++;
171                         }
172                         /*
173                          * Compensate for loop update
174                          */
175                         i--;
176
177                         if (revs->combine_merges && num_compare_stages == 2) {
178                                 show_combined_diff(dpath, 2,
179                                                    revs->dense_combined_merges,
180                                                    revs);
181                                 free(dpath);
182                                 continue;
183                         }
184                         free(dpath);
185                         dpath = NULL;
186
187                         /*
188                          * Show the diff for the 'ce' if we found the one
189                          * from the desired stage.
190                          */
191                         pair = diff_unmerge(&revs->diffopt, ce->name);
192                         if (wt_mode)
193                                 pair->two->mode = wt_mode;
194                         if (ce_stage(ce) != diff_unmerged_stage)
195                                 continue;
196                 }
197
198                 if (ce_uptodate(ce) || ce_skip_worktree(ce))
199                         continue;
200
201                 /* If CE_VALID is set, don't look at workdir for file removal */
202                 changed = (ce->ce_flags & CE_VALID) ? 0 : check_removed(ce, &st);
203                 if (changed) {
204                         if (changed < 0) {
205                                 perror(ce->name);
206                                 continue;
207                         }
208                         if (silent_on_removed)
209                                 continue;
210                         diff_addremove(&revs->diffopt, '-', ce->ce_mode,
211                                        ce->sha1, ce->name, 0);
212                         continue;
213                 }
214                 changed = match_stat_with_submodule(&revs->diffopt, ce, &st,
215                                                     ce_option, &dirty_submodule);
216                 if (!changed && !dirty_submodule) {
217                         ce_mark_uptodate(ce);
218                         if (!DIFF_OPT_TST(&revs->diffopt, FIND_COPIES_HARDER))
219                                 continue;
220                 }
221                 oldmode = ce->ce_mode;
222                 newmode = ce_mode_from_stat(ce, st.st_mode);
223                 diff_change(&revs->diffopt, oldmode, newmode,
224                             ce->sha1, (changed ? 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, unsigned int mode,
242                                  unsigned dirty_submodule)
243 {
244         diff_addremove(&revs->diffopt, prefix[0], mode,
245                        sha1, ce->name, dirty_submodule);
246 }
247
248 static int get_stat_data(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                           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 copy.
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, mode, dirty_submodule);
301 }
302
303 static int show_modified(struct rev_info *revs,
304                          struct cache_entry *old,
305                          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, 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, old->name, 0, dirty_submodule);
353         return 0;
354 }
355
356 /*
357  * This gets a mix of an existing index and a tree, one pathname entry
358  * at a time. The index entry may be a single stage-0 one, but it could
359  * also be multiple unmerged entries (in which case idx_pos/idx_nr will
360  * give you the position and number of entries in the index).
361  */
362 static void do_oneway_diff(struct unpack_trees_options *o,
363         struct cache_entry *idx,
364         struct cache_entry *tree)
365 {
366         struct rev_info *revs = o->unpack_data;
367         int match_missing, cached;
368
369         /* if the entry is not checked out, don't examine work tree */
370         cached = o->index_only ||
371                 (idx && ((idx->ce_flags & CE_VALID) || ce_skip_worktree(idx)));
372         /*
373          * Backward compatibility wart - "diff-index -m" does
374          * not mean "do not ignore merges", but "match_missing".
375          *
376          * But with the revision flag parsing, that's found in
377          * "!revs->ignore_merges".
378          */
379         match_missing = !revs->ignore_merges;
380
381         if (cached && idx && ce_stage(idx)) {
382                 struct diff_filepair *pair;
383                 pair = diff_unmerge(&revs->diffopt, idx->name);
384                 fill_filespec(pair->one, idx->sha1, idx->ce_mode);
385                 return;
386         }
387
388         /*
389          * Something added to the tree?
390          */
391         if (!tree) {
392                 show_new_file(revs, idx, cached, match_missing);
393                 return;
394         }
395
396         /*
397          * Something removed from the tree?
398          */
399         if (!idx) {
400                 diff_index_show_file(revs, "-", tree, tree->sha1, tree->ce_mode, 0);
401                 return;
402         }
403
404         /* Show difference between old and new */
405         show_modified(revs, tree, idx, 1, cached, match_missing);
406 }
407
408 /*
409  * The unpack_trees() interface is designed for merging, so
410  * the different source entries are designed primarily for
411  * the source trees, with the old index being really mainly
412  * used for being replaced by the result.
413  *
414  * For diffing, the index is more important, and we only have a
415  * single tree.
416  *
417  * We're supposed to advance o->pos to skip what we have already processed.
418  *
419  * This wrapper makes it all more readable, and takes care of all
420  * the fairly complex unpack_trees() semantic requirements, including
421  * the skipping, the path matching, the type conflict cases etc.
422  */
423 static int oneway_diff(struct cache_entry **src, struct unpack_trees_options *o)
424 {
425         struct cache_entry *idx = src[0];
426         struct cache_entry *tree = src[1];
427         struct rev_info *revs = o->unpack_data;
428
429         /*
430          * Unpack-trees generates a DF/conflict entry if
431          * there was a directory in the index and a tree
432          * in the tree. From a diff standpoint, that's a
433          * delete of the tree and a create of the file.
434          */
435         if (tree == o->df_conflict_entry)
436                 tree = NULL;
437
438         if (ce_path_match(idx ? idx : tree, &revs->prune_data))
439                 do_oneway_diff(o, idx, tree);
440
441         return 0;
442 }
443
444 int run_diff_index(struct rev_info *revs, int cached)
445 {
446         struct object *ent;
447         struct tree *tree;
448         const char *tree_name;
449         struct unpack_trees_options opts;
450         struct tree_desc t;
451
452         ent = revs->pending.objects[0].item;
453         tree_name = revs->pending.objects[0].name;
454         tree = parse_tree_indirect(ent->sha1);
455         if (!tree)
456                 return error("bad tree object %s", tree_name);
457
458         memset(&opts, 0, sizeof(opts));
459         opts.head_idx = 1;
460         opts.index_only = cached;
461         opts.diff_index_cached = (cached &&
462                                   !DIFF_OPT_TST(&revs->diffopt, FIND_COPIES_HARDER));
463         opts.merge = 1;
464         opts.fn = oneway_diff;
465         opts.unpack_data = revs;
466         opts.src_index = &the_index;
467         opts.dst_index = NULL;
468
469         init_tree_desc(&t, tree->buffer, tree->size);
470         if (unpack_trees(1, &t, &opts))
471                 exit(128);
472
473         diff_set_mnemonic_prefix(&revs->diffopt, "c/", cached ? "i/" : "w/");
474         diffcore_fix_diff_index(&revs->diffopt);
475         diffcore_std(&revs->diffopt);
476         diff_flush(&revs->diffopt);
477         return 0;
478 }
479
480 int do_diff_cache(const unsigned char *tree_sha1, struct diff_options *opt)
481 {
482         struct tree *tree;
483         struct rev_info revs;
484         int i;
485         struct cache_entry **dst;
486         struct cache_entry *last = NULL;
487         struct unpack_trees_options opts;
488         struct tree_desc t;
489
490         /*
491          * This is used by git-blame to run diff-cache internally;
492          * it potentially needs to repeatedly run this, so we will
493          * start by removing the higher order entries the last round
494          * left behind.
495          */
496         dst = active_cache;
497         for (i = 0; i < active_nr; i++) {
498                 struct cache_entry *ce = active_cache[i];
499                 if (ce_stage(ce)) {
500                         if (last && !strcmp(ce->name, last->name))
501                                 continue;
502                         cache_tree_invalidate_path(active_cache_tree,
503                                                    ce->name);
504                         last = ce;
505                         ce->ce_flags |= CE_REMOVE;
506                 }
507                 *dst++ = ce;
508         }
509         active_nr = dst - active_cache;
510
511         init_revisions(&revs, NULL);
512         init_pathspec(&revs.prune_data, opt->pathspec.raw);
513         tree = parse_tree_indirect(tree_sha1);
514         if (!tree)
515                 die("bad tree object %s", sha1_to_hex(tree_sha1));
516
517         memset(&opts, 0, sizeof(opts));
518         opts.head_idx = 1;
519         opts.index_only = 1;
520         opts.diff_index_cached = !DIFF_OPT_TST(opt, FIND_COPIES_HARDER);
521         opts.merge = 1;
522         opts.fn = oneway_diff;
523         opts.unpack_data = &revs;
524         opts.src_index = &the_index;
525         opts.dst_index = &the_index;
526
527         init_tree_desc(&t, tree->buffer, tree->size);
528         if (unpack_trees(1, &t, &opts))
529                 exit(128);
530         return 0;
531 }
532
533 int index_differs_from(const char *def, int diff_flags)
534 {
535         struct rev_info rev;
536         struct setup_revision_opt opt;
537
538         init_revisions(&rev, NULL);
539         memset(&opt, 0, sizeof(opt));
540         opt.def = def;
541         setup_revisions(0, NULL, &rev, &opt);
542         DIFF_OPT_SET(&rev.diffopt, QUICK);
543         DIFF_OPT_SET(&rev.diffopt, EXIT_WITH_STATUS);
544         rev.diffopt.flags |= diff_flags;
545         run_diff_index(&rev, 1);
546         if (rev.pending.alloc)
547                 free(rev.pending.objects);
548         return (DIFF_OPT_TST(&rev.diffopt, HAS_CHANGES) != 0);
549 }