gitweb: Better chopping in commit search results
[git.git] / diff-lib.c
blob03eaa7cef35dfaec0a8f61b2b96892e5604ed731
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 "path-list.h"
12 #include "unpack-trees.h"
15 * diff-files
18 static int read_directory(const char *path, struct path_list *list)
20 DIR *dir;
21 struct dirent *e;
23 if (!(dir = opendir(path)))
24 return error("Could not open directory %s", path);
26 while ((e = readdir(dir)))
27 if (strcmp(".", e->d_name) && strcmp("..", e->d_name))
28 path_list_insert(e->d_name, list);
30 closedir(dir);
31 return 0;
34 static int get_mode(const char *path, int *mode)
36 struct stat st;
38 if (!path || !strcmp(path, "/dev/null"))
39 *mode = 0;
40 else if (!strcmp(path, "-"))
41 *mode = create_ce_mode(0666);
42 else if (stat(path, &st))
43 return error("Could not access '%s'", path);
44 else
45 *mode = st.st_mode;
46 return 0;
49 static int queue_diff(struct diff_options *o,
50 const char *name1, const char *name2)
52 int mode1 = 0, mode2 = 0;
54 if (get_mode(name1, &mode1) || get_mode(name2, &mode2))
55 return -1;
57 if (mode1 && mode2 && S_ISDIR(mode1) != S_ISDIR(mode2))
58 return error("file/directory conflict: %s, %s", name1, name2);
60 if (S_ISDIR(mode1) || S_ISDIR(mode2)) {
61 char buffer1[PATH_MAX], buffer2[PATH_MAX];
62 struct path_list p1 = {NULL, 0, 0, 1}, p2 = {NULL, 0, 0, 1};
63 int len1 = 0, len2 = 0, i1, i2, ret = 0;
65 if (name1 && read_directory(name1, &p1))
66 return -1;
67 if (name2 && read_directory(name2, &p2)) {
68 path_list_clear(&p1, 0);
69 return -1;
72 if (name1) {
73 len1 = strlen(name1);
74 if (len1 > 0 && name1[len1 - 1] == '/')
75 len1--;
76 memcpy(buffer1, name1, len1);
77 buffer1[len1++] = '/';
80 if (name2) {
81 len2 = strlen(name2);
82 if (len2 > 0 && name2[len2 - 1] == '/')
83 len2--;
84 memcpy(buffer2, name2, len2);
85 buffer2[len2++] = '/';
88 for (i1 = i2 = 0; !ret && (i1 < p1.nr || i2 < p2.nr); ) {
89 const char *n1, *n2;
90 int comp;
92 if (i1 == p1.nr)
93 comp = 1;
94 else if (i2 == p2.nr)
95 comp = -1;
96 else
97 comp = strcmp(p1.items[i1].path,
98 p2.items[i2].path);
100 if (comp > 0)
101 n1 = NULL;
102 else {
103 n1 = buffer1;
104 strncpy(buffer1 + len1, p1.items[i1++].path,
105 PATH_MAX - len1);
108 if (comp < 0)
109 n2 = NULL;
110 else {
111 n2 = buffer2;
112 strncpy(buffer2 + len2, p2.items[i2++].path,
113 PATH_MAX - len2);
116 ret = queue_diff(o, n1, n2);
118 path_list_clear(&p1, 0);
119 path_list_clear(&p2, 0);
121 return ret;
122 } else {
123 struct diff_filespec *d1, *d2;
125 if (DIFF_OPT_TST(o, REVERSE_DIFF)) {
126 unsigned tmp;
127 const char *tmp_c;
128 tmp = mode1; mode1 = mode2; mode2 = tmp;
129 tmp_c = name1; name1 = name2; name2 = tmp_c;
132 if (!name1)
133 name1 = "/dev/null";
134 if (!name2)
135 name2 = "/dev/null";
136 d1 = alloc_filespec(name1);
137 d2 = alloc_filespec(name2);
138 fill_filespec(d1, null_sha1, mode1);
139 fill_filespec(d2, null_sha1, mode2);
141 diff_queue(&diff_queued_diff, d1, d2);
142 return 0;
147 * Does the path name a blob in the working tree, or a directory
148 * in the working tree?
150 static int is_in_index(const char *path)
152 int len, pos;
153 struct cache_entry *ce;
155 len = strlen(path);
156 while (path[len-1] == '/')
157 len--;
158 if (!len)
159 return 1; /* "." */
160 pos = cache_name_pos(path, len);
161 if (0 <= pos)
162 return 1;
163 pos = -1 - pos;
164 while (pos < active_nr) {
165 ce = active_cache[pos++];
166 if (ce_namelen(ce) <= len ||
167 strncmp(ce->name, path, len) ||
168 (ce->name[len] > '/'))
169 break; /* path cannot be a prefix */
170 if (ce->name[len] == '/')
171 return 1;
173 return 0;
176 static int handle_diff_files_args(struct rev_info *revs,
177 int argc, const char **argv,
178 unsigned int *options)
180 *options = 0;
182 /* revs->max_count == -2 means --no-index */
183 while (1 < argc && argv[1][0] == '-') {
184 if (!strcmp(argv[1], "--base"))
185 revs->max_count = 1;
186 else if (!strcmp(argv[1], "--ours"))
187 revs->max_count = 2;
188 else if (!strcmp(argv[1], "--theirs"))
189 revs->max_count = 3;
190 else if (!strcmp(argv[1], "-n") ||
191 !strcmp(argv[1], "--no-index")) {
192 revs->max_count = -2;
193 DIFF_OPT_SET(&revs->diffopt, EXIT_WITH_STATUS);
194 DIFF_OPT_SET(&revs->diffopt, NO_INDEX);
196 else if (!strcmp(argv[1], "-q"))
197 *options |= DIFF_SILENT_ON_REMOVED;
198 else
199 return error("invalid option: %s", argv[1]);
200 argv++; argc--;
203 if (revs->max_count == -1 && revs->diffopt.nr_paths == 2) {
205 * If two files are specified, and at least one is untracked,
206 * default to no-index.
208 read_cache();
209 if (!is_in_index(revs->diffopt.paths[0]) ||
210 !is_in_index(revs->diffopt.paths[1])) {
211 revs->max_count = -2;
212 DIFF_OPT_SET(&revs->diffopt, NO_INDEX);
217 * Make sure there are NO revision (i.e. pending object) parameter,
218 * rev.max_count is reasonable (0 <= n <= 3),
219 * there is no other revision filtering parameters.
221 if (revs->pending.nr || revs->max_count > 3 ||
222 revs->min_age != -1 || revs->max_age != -1)
223 return error("no revision allowed with diff-files");
225 if (revs->max_count == -1 &&
226 (revs->diffopt.output_format & DIFF_FORMAT_PATCH))
227 revs->combine_merges = revs->dense_combined_merges = 1;
229 return 0;
232 static int is_outside_repo(const char *path, int nongit, const char *prefix)
234 int i;
235 if (nongit || !strcmp(path, "-") || is_absolute_path(path))
236 return 1;
237 if (prefixcmp(path, "../"))
238 return 0;
239 if (!prefix)
240 return 1;
241 for (i = strlen(prefix); !prefixcmp(path, "../"); ) {
242 while (i > 0 && prefix[i - 1] != '/')
243 i--;
244 if (--i < 0)
245 return 1;
246 path += 3;
248 return 0;
251 int setup_diff_no_index(struct rev_info *revs,
252 int argc, const char ** argv, int nongit, const char *prefix)
254 int i;
255 for (i = 1; i < argc; i++)
256 if (argv[i][0] != '-' || argv[i][1] == '\0')
257 break;
258 else if (!strcmp(argv[i], "--")) {
259 i++;
260 break;
261 } else if (i < argc - 3 && !strcmp(argv[i], "--no-index")) {
262 i = argc - 3;
263 DIFF_OPT_SET(&revs->diffopt, EXIT_WITH_STATUS);
264 break;
266 if (argc != i + 2 || (!is_outside_repo(argv[i + 1], nongit, prefix) &&
267 !is_outside_repo(argv[i], nongit, prefix)))
268 return -1;
270 diff_setup(&revs->diffopt);
271 for (i = 1; i < argc - 2; )
272 if (!strcmp(argv[i], "--no-index"))
273 i++;
274 else {
275 int j = diff_opt_parse(&revs->diffopt,
276 argv + i, argc - i);
277 if (!j)
278 die("invalid diff option/value: %s", argv[i]);
279 i += j;
282 if (prefix) {
283 int len = strlen(prefix);
285 revs->diffopt.paths = xcalloc(2, sizeof(char*));
286 for (i = 0; i < 2; i++) {
287 const char *p = argv[argc - 2 + i];
289 * stdin should be spelled as '-'; if you have
290 * path that is '-', spell it as ./-.
292 p = (strcmp(p, "-")
293 ? xstrdup(prefix_filename(prefix, len, p))
294 : p);
295 revs->diffopt.paths[i] = p;
298 else
299 revs->diffopt.paths = argv + argc - 2;
300 revs->diffopt.nr_paths = 2;
301 DIFF_OPT_SET(&revs->diffopt, NO_INDEX);
302 revs->max_count = -2;
303 if (diff_setup_done(&revs->diffopt) < 0)
304 die("diff_setup_done failed");
305 return 0;
308 int run_diff_files_cmd(struct rev_info *revs, int argc, const char **argv)
310 unsigned int options;
312 if (handle_diff_files_args(revs, argc, argv, &options))
313 return -1;
315 if (DIFF_OPT_TST(&revs->diffopt, NO_INDEX)) {
316 if (revs->diffopt.nr_paths != 2)
317 return error("need two files/directories with --no-index");
318 if (queue_diff(&revs->diffopt, revs->diffopt.paths[0],
319 revs->diffopt.paths[1]))
320 return -1;
321 diffcore_std(&revs->diffopt);
322 diff_flush(&revs->diffopt);
324 * The return code for --no-index imitates diff(1):
325 * 0 = no changes, 1 = changes, else error
327 return revs->diffopt.found_changes;
330 if (read_cache() < 0) {
331 perror("read_cache");
332 return -1;
334 return run_diff_files(revs, options);
337 int run_diff_files(struct rev_info *revs, unsigned int option)
339 int entries, i;
340 int diff_unmerged_stage = revs->max_count;
341 int silent_on_removed = option & DIFF_SILENT_ON_REMOVED;
342 unsigned ce_option = ((option & DIFF_RACY_IS_MODIFIED)
343 ? CE_MATCH_RACY_IS_DIRTY : 0);
345 if (diff_unmerged_stage < 0)
346 diff_unmerged_stage = 2;
347 entries = active_nr;
348 for (i = 0; i < entries; i++) {
349 struct stat st;
350 unsigned int oldmode, newmode;
351 struct cache_entry *ce = active_cache[i];
352 int changed;
354 if (DIFF_OPT_TST(&revs->diffopt, QUIET) &&
355 DIFF_OPT_TST(&revs->diffopt, HAS_CHANGES))
356 break;
358 if (!ce_path_match(ce, revs->prune_data))
359 continue;
361 if (ce_stage(ce)) {
362 struct combine_diff_path *dpath;
363 int num_compare_stages = 0;
364 size_t path_len;
366 path_len = ce_namelen(ce);
368 dpath = xmalloc(combine_diff_path_size(5, path_len));
369 dpath->path = (char *) &(dpath->parent[5]);
371 dpath->next = NULL;
372 dpath->len = path_len;
373 memcpy(dpath->path, ce->name, path_len);
374 dpath->path[path_len] = '\0';
375 hashclr(dpath->sha1);
376 memset(&(dpath->parent[0]), 0,
377 sizeof(struct combine_diff_parent)*5);
379 if (lstat(ce->name, &st) < 0) {
380 if (errno != ENOENT && errno != ENOTDIR) {
381 perror(ce->name);
382 continue;
384 if (silent_on_removed)
385 continue;
387 else
388 dpath->mode = ce_mode_from_stat(ce, st.st_mode);
390 while (i < entries) {
391 struct cache_entry *nce = active_cache[i];
392 int stage;
394 if (strcmp(ce->name, nce->name))
395 break;
397 /* Stage #2 (ours) is the first parent,
398 * stage #3 (theirs) is the second.
400 stage = ce_stage(nce);
401 if (2 <= stage) {
402 int mode = nce->ce_mode;
403 num_compare_stages++;
404 hashcpy(dpath->parent[stage-2].sha1, nce->sha1);
405 dpath->parent[stage-2].mode = ce_mode_from_stat(nce, mode);
406 dpath->parent[stage-2].status =
407 DIFF_STATUS_MODIFIED;
410 /* diff against the proper unmerged stage */
411 if (stage == diff_unmerged_stage)
412 ce = nce;
413 i++;
416 * Compensate for loop update
418 i--;
420 if (revs->combine_merges && num_compare_stages == 2) {
421 show_combined_diff(dpath, 2,
422 revs->dense_combined_merges,
423 revs);
424 free(dpath);
425 continue;
427 free(dpath);
428 dpath = NULL;
431 * Show the diff for the 'ce' if we found the one
432 * from the desired stage.
434 diff_unmerge(&revs->diffopt, ce->name, 0, null_sha1);
435 if (ce_stage(ce) != diff_unmerged_stage)
436 continue;
439 if (ce_uptodate(ce))
440 continue;
441 if (lstat(ce->name, &st) < 0) {
442 if (errno != ENOENT && errno != ENOTDIR) {
443 perror(ce->name);
444 continue;
446 if (silent_on_removed)
447 continue;
448 diff_addremove(&revs->diffopt, '-', ce->ce_mode,
449 ce->sha1, ce->name, NULL);
450 continue;
452 changed = ce_match_stat(ce, &st, ce_option);
453 if (!changed && !DIFF_OPT_TST(&revs->diffopt, FIND_COPIES_HARDER))
454 continue;
455 oldmode = ce->ce_mode;
456 newmode = ce_mode_from_stat(ce, st.st_mode);
457 diff_change(&revs->diffopt, oldmode, newmode,
458 ce->sha1, (changed ? null_sha1 : ce->sha1),
459 ce->name, NULL);
462 diffcore_std(&revs->diffopt);
463 diff_flush(&revs->diffopt);
464 return 0;
468 * diff-index
471 /* A file entry went away or appeared */
472 static void diff_index_show_file(struct rev_info *revs,
473 const char *prefix,
474 struct cache_entry *ce,
475 unsigned char *sha1, unsigned int mode)
477 diff_addremove(&revs->diffopt, prefix[0], mode,
478 sha1, ce->name, NULL);
481 static int get_stat_data(struct cache_entry *ce,
482 unsigned char **sha1p,
483 unsigned int *modep,
484 int cached, int match_missing)
486 unsigned char *sha1 = ce->sha1;
487 unsigned int mode = ce->ce_mode;
489 if (!cached) {
490 static unsigned char no_sha1[20];
491 int changed;
492 struct stat st;
493 if (lstat(ce->name, &st) < 0) {
494 if (errno == ENOENT && match_missing) {
495 *sha1p = sha1;
496 *modep = mode;
497 return 0;
499 return -1;
501 changed = ce_match_stat(ce, &st, 0);
502 if (changed) {
503 mode = ce_mode_from_stat(ce, st.st_mode);
504 sha1 = no_sha1;
508 *sha1p = sha1;
509 *modep = mode;
510 return 0;
513 static void show_new_file(struct rev_info *revs,
514 struct cache_entry *new,
515 int cached, int match_missing)
517 unsigned char *sha1;
518 unsigned int mode;
520 /* New file in the index: it might actually be different in
521 * the working copy.
523 if (get_stat_data(new, &sha1, &mode, cached, match_missing) < 0)
524 return;
526 diff_index_show_file(revs, "+", new, sha1, mode);
529 static int show_modified(struct rev_info *revs,
530 struct cache_entry *old,
531 struct cache_entry *new,
532 int report_missing,
533 int cached, int match_missing)
535 unsigned int mode, oldmode;
536 unsigned char *sha1;
538 if (get_stat_data(new, &sha1, &mode, cached, match_missing) < 0) {
539 if (report_missing)
540 diff_index_show_file(revs, "-", old,
541 old->sha1, old->ce_mode);
542 return -1;
545 if (revs->combine_merges && !cached &&
546 (hashcmp(sha1, old->sha1) || hashcmp(old->sha1, new->sha1))) {
547 struct combine_diff_path *p;
548 int pathlen = ce_namelen(new);
550 p = xmalloc(combine_diff_path_size(2, pathlen));
551 p->path = (char *) &p->parent[2];
552 p->next = NULL;
553 p->len = pathlen;
554 memcpy(p->path, new->name, pathlen);
555 p->path[pathlen] = 0;
556 p->mode = mode;
557 hashclr(p->sha1);
558 memset(p->parent, 0, 2 * sizeof(struct combine_diff_parent));
559 p->parent[0].status = DIFF_STATUS_MODIFIED;
560 p->parent[0].mode = new->ce_mode;
561 hashcpy(p->parent[0].sha1, new->sha1);
562 p->parent[1].status = DIFF_STATUS_MODIFIED;
563 p->parent[1].mode = old->ce_mode;
564 hashcpy(p->parent[1].sha1, old->sha1);
565 show_combined_diff(p, 2, revs->dense_combined_merges, revs);
566 free(p);
567 return 0;
570 oldmode = old->ce_mode;
571 if (mode == oldmode && !hashcmp(sha1, old->sha1) &&
572 !DIFF_OPT_TST(&revs->diffopt, FIND_COPIES_HARDER))
573 return 0;
575 diff_change(&revs->diffopt, oldmode, mode,
576 old->sha1, sha1, old->name, NULL);
577 return 0;
581 * This turns all merge entries into "stage 3". That guarantees that
582 * when we read in the new tree (into "stage 1"), we won't lose sight
583 * of the fact that we had unmerged entries.
585 static void mark_merge_entries(void)
587 int i;
588 for (i = 0; i < active_nr; i++) {
589 struct cache_entry *ce = active_cache[i];
590 if (!ce_stage(ce))
591 continue;
592 ce->ce_flags |= CE_STAGEMASK;
597 * This gets a mix of an existing index and a tree, one pathname entry
598 * at a time. The index entry may be a single stage-0 one, but it could
599 * also be multiple unmerged entries (in which case idx_pos/idx_nr will
600 * give you the position and number of entries in the index).
602 static void do_oneway_diff(struct unpack_trees_options *o,
603 struct cache_entry *idx,
604 struct cache_entry *tree,
605 int idx_pos, int idx_nr)
607 struct rev_info *revs = o->unpack_data;
608 int match_missing, cached;
611 * Backward compatibility wart - "diff-index -m" does
612 * not mean "do not ignore merges", but "match_missing".
614 * But with the revision flag parsing, that's found in
615 * "!revs->ignore_merges".
617 cached = o->index_only;
618 match_missing = !revs->ignore_merges;
620 if (cached && idx && ce_stage(idx)) {
621 if (tree)
622 diff_unmerge(&revs->diffopt, idx->name, idx->ce_mode, idx->sha1);
623 return;
627 * Something added to the tree?
629 if (!tree) {
630 show_new_file(revs, idx, cached, match_missing);
631 return;
635 * Something removed from the tree?
637 if (!idx) {
638 diff_index_show_file(revs, "-", tree, tree->sha1, tree->ce_mode);
639 return;
642 /* Show difference between old and new */
643 show_modified(revs, tree, idx, 1, cached, match_missing);
647 * Count how many index entries go with the first one
649 static inline int count_skip(const struct cache_entry *src, int pos)
651 int skip = 1;
653 /* We can only have multiple entries if the first one is not stage-0 */
654 if (ce_stage(src)) {
655 struct cache_entry **p = active_cache + pos;
656 int namelen = ce_namelen(src);
658 for (;;) {
659 const struct cache_entry *ce;
660 pos++;
661 if (pos >= active_nr)
662 break;
663 ce = *++p;
664 if (ce_namelen(ce) != namelen)
665 break;
666 if (memcmp(ce->name, src->name, namelen))
667 break;
668 skip++;
671 return skip;
675 * The unpack_trees() interface is designed for merging, so
676 * the different source entries are designed primarily for
677 * the source trees, with the old index being really mainly
678 * used for being replaced by the result.
680 * For diffing, the index is more important, and we only have a
681 * single tree.
683 * We're supposed to return how many index entries we want to skip.
685 * This wrapper makes it all more readable, and takes care of all
686 * the fairly complex unpack_trees() semantic requirements, including
687 * the skipping, the path matching, the type conflict cases etc.
689 static int oneway_diff(struct cache_entry **src,
690 struct unpack_trees_options *o,
691 int index_pos)
693 int skip = 0;
694 struct cache_entry *idx = src[0];
695 struct cache_entry *tree = src[1];
696 struct rev_info *revs = o->unpack_data;
698 if (index_pos >= 0)
699 skip = count_skip(idx, index_pos);
702 * Unpack-trees generates a DF/conflict entry if
703 * there was a directory in the index and a tree
704 * in the tree. From a diff standpoint, that's a
705 * delete of the tree and a create of the file.
707 if (tree == o->df_conflict_entry)
708 tree = NULL;
710 if (ce_path_match(idx ? idx : tree, revs->prune_data))
711 do_oneway_diff(o, idx, tree, index_pos, skip);
713 return skip;
716 int run_diff_index(struct rev_info *revs, int cached)
718 struct object *ent;
719 struct tree *tree;
720 const char *tree_name;
721 struct unpack_trees_options opts;
722 struct tree_desc t;
724 mark_merge_entries();
726 ent = revs->pending.objects[0].item;
727 tree_name = revs->pending.objects[0].name;
728 tree = parse_tree_indirect(ent->sha1);
729 if (!tree)
730 return error("bad tree object %s", tree_name);
732 memset(&opts, 0, sizeof(opts));
733 opts.head_idx = 1;
734 opts.index_only = cached;
735 opts.merge = 1;
736 opts.fn = oneway_diff;
737 opts.unpack_data = revs;
739 init_tree_desc(&t, tree->buffer, tree->size);
740 unpack_trees(1, &t, &opts);
742 diffcore_std(&revs->diffopt);
743 diff_flush(&revs->diffopt);
744 return 0;
747 int do_diff_cache(const unsigned char *tree_sha1, struct diff_options *opt)
749 struct tree *tree;
750 struct rev_info revs;
751 int i;
752 struct cache_entry **dst;
753 struct cache_entry *last = NULL;
754 struct unpack_trees_options opts;
755 struct tree_desc t;
758 * This is used by git-blame to run diff-cache internally;
759 * it potentially needs to repeatedly run this, so we will
760 * start by removing the higher order entries the last round
761 * left behind.
763 dst = active_cache;
764 for (i = 0; i < active_nr; i++) {
765 struct cache_entry *ce = active_cache[i];
766 if (ce_stage(ce)) {
767 if (last && !strcmp(ce->name, last->name))
768 continue;
769 cache_tree_invalidate_path(active_cache_tree,
770 ce->name);
771 last = ce;
772 ce->ce_flags |= CE_REMOVE;
774 *dst++ = ce;
776 active_nr = dst - active_cache;
778 init_revisions(&revs, NULL);
779 revs.prune_data = opt->paths;
780 tree = parse_tree_indirect(tree_sha1);
781 if (!tree)
782 die("bad tree object %s", sha1_to_hex(tree_sha1));
784 memset(&opts, 0, sizeof(opts));
785 opts.head_idx = 1;
786 opts.index_only = 1;
787 opts.merge = 1;
788 opts.fn = oneway_diff;
789 opts.unpack_data = &revs;
791 init_tree_desc(&t, tree->buffer, tree->size);
792 unpack_trees(1, &t, &opts);
793 return 0;