add a howto document about corrupted blob recovery
[git/repo.git] / diff-lib.c
blobda5571302df6ed418874fd4d7423853a7de5b52c
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"
14 * diff-files
17 static int read_directory(const char *path, struct path_list *list)
19 DIR *dir;
20 struct dirent *e;
22 if (!(dir = opendir(path)))
23 return error("Could not open directory %s", path);
25 while ((e = readdir(dir)))
26 if (strcmp(".", e->d_name) && strcmp("..", e->d_name))
27 path_list_insert(e->d_name, list);
29 closedir(dir);
30 return 0;
33 static int get_mode(const char *path, int *mode)
35 struct stat st;
37 if (!path || !strcmp(path, "/dev/null"))
38 *mode = 0;
39 else if (!strcmp(path, "-"))
40 *mode = ntohl(create_ce_mode(0666));
41 else if (stat(path, &st))
42 return error("Could not access '%s'", path);
43 else
44 *mode = st.st_mode;
45 return 0;
48 static int queue_diff(struct diff_options *o,
49 const char *name1, const char *name2)
51 int mode1 = 0, mode2 = 0;
53 if (get_mode(name1, &mode1) || get_mode(name2, &mode2))
54 return -1;
56 if (mode1 && mode2 && S_ISDIR(mode1) != S_ISDIR(mode2))
57 return error("file/directory conflict: %s, %s", name1, name2);
59 if (S_ISDIR(mode1) || S_ISDIR(mode2)) {
60 char buffer1[PATH_MAX], buffer2[PATH_MAX];
61 struct path_list p1 = {NULL, 0, 0, 1}, p2 = {NULL, 0, 0, 1};
62 int len1 = 0, len2 = 0, i1, i2, ret = 0;
64 if (name1 && read_directory(name1, &p1))
65 return -1;
66 if (name2 && read_directory(name2, &p2)) {
67 path_list_clear(&p1, 0);
68 return -1;
71 if (name1) {
72 len1 = strlen(name1);
73 if (len1 > 0 && name1[len1 - 1] == '/')
74 len1--;
75 memcpy(buffer1, name1, len1);
76 buffer1[len1++] = '/';
79 if (name2) {
80 len2 = strlen(name2);
81 if (len2 > 0 && name2[len2 - 1] == '/')
82 len2--;
83 memcpy(buffer2, name2, len2);
84 buffer2[len2++] = '/';
87 for (i1 = i2 = 0; !ret && (i1 < p1.nr || i2 < p2.nr); ) {
88 const char *n1, *n2;
89 int comp;
91 if (i1 == p1.nr)
92 comp = 1;
93 else if (i2 == p2.nr)
94 comp = -1;
95 else
96 comp = strcmp(p1.items[i1].path,
97 p2.items[i2].path);
99 if (comp > 0)
100 n1 = NULL;
101 else {
102 n1 = buffer1;
103 strncpy(buffer1 + len1, p1.items[i1++].path,
104 PATH_MAX - len1);
107 if (comp < 0)
108 n2 = NULL;
109 else {
110 n2 = buffer2;
111 strncpy(buffer2 + len2, p2.items[i2++].path,
112 PATH_MAX - len2);
115 ret = queue_diff(o, n1, n2);
117 path_list_clear(&p1, 0);
118 path_list_clear(&p2, 0);
120 return ret;
121 } else {
122 struct diff_filespec *d1, *d2;
124 if (o->reverse_diff) {
125 unsigned tmp;
126 const char *tmp_c;
127 tmp = mode1; mode1 = mode2; mode2 = tmp;
128 tmp_c = name1; name1 = name2; name2 = tmp_c;
131 if (!name1)
132 name1 = "/dev/null";
133 if (!name2)
134 name2 = "/dev/null";
135 d1 = alloc_filespec(name1);
136 d2 = alloc_filespec(name2);
137 fill_filespec(d1, null_sha1, mode1);
138 fill_filespec(d2, null_sha1, mode2);
140 diff_queue(&diff_queued_diff, d1, d2);
141 return 0;
146 * Does the path name a blob in the working tree, or a directory
147 * in the working tree?
149 static int is_in_index(const char *path)
151 int len, pos;
152 struct cache_entry *ce;
154 len = strlen(path);
155 while (path[len-1] == '/')
156 len--;
157 if (!len)
158 return 1; /* "." */
159 pos = cache_name_pos(path, len);
160 if (0 <= pos)
161 return 1;
162 pos = -1 - pos;
163 while (pos < active_nr) {
164 ce = active_cache[pos++];
165 if (ce_namelen(ce) <= len ||
166 strncmp(ce->name, path, len) ||
167 (ce->name[len] > '/'))
168 break; /* path cannot be a prefix */
169 if (ce->name[len] == '/')
170 return 1;
172 return 0;
175 static int handle_diff_files_args(struct rev_info *revs,
176 int argc, const char **argv, int *silent)
178 *silent = 0;
180 /* revs->max_count == -2 means --no-index */
181 while (1 < argc && argv[1][0] == '-') {
182 if (!strcmp(argv[1], "--base"))
183 revs->max_count = 1;
184 else if (!strcmp(argv[1], "--ours"))
185 revs->max_count = 2;
186 else if (!strcmp(argv[1], "--theirs"))
187 revs->max_count = 3;
188 else if (!strcmp(argv[1], "-n") ||
189 !strcmp(argv[1], "--no-index")) {
190 revs->max_count = -2;
191 revs->diffopt.exit_with_status = 1;
192 revs->diffopt.no_index = 1;
194 else if (!strcmp(argv[1], "-q"))
195 *silent = 1;
196 else
197 return error("invalid option: %s", argv[1]);
198 argv++; argc--;
201 if (revs->max_count == -1 && revs->diffopt.nr_paths == 2) {
203 * If two files are specified, and at least one is untracked,
204 * default to no-index.
206 read_cache();
207 if (!is_in_index(revs->diffopt.paths[0]) ||
208 !is_in_index(revs->diffopt.paths[1])) {
209 revs->max_count = -2;
210 revs->diffopt.no_index = 1;
215 * Make sure there are NO revision (i.e. pending object) parameter,
216 * rev.max_count is reasonable (0 <= n <= 3),
217 * there is no other revision filtering parameters.
219 if (revs->pending.nr || revs->max_count > 3 ||
220 revs->min_age != -1 || revs->max_age != -1)
221 return error("no revision allowed with diff-files");
223 if (revs->max_count == -1 &&
224 (revs->diffopt.output_format & DIFF_FORMAT_PATCH))
225 revs->combine_merges = revs->dense_combined_merges = 1;
227 return 0;
230 static int is_outside_repo(const char *path, int nongit, const char *prefix)
232 int i;
233 if (nongit || !strcmp(path, "-") || path[0] == '/')
234 return 1;
235 if (prefixcmp(path, "../"))
236 return 0;
237 if (!prefix)
238 return 1;
239 for (i = strlen(prefix); !prefixcmp(path, "../"); ) {
240 while (i > 0 && prefix[i - 1] != '/')
241 i--;
242 if (--i < 0)
243 return 1;
244 path += 3;
246 return 0;
249 int setup_diff_no_index(struct rev_info *revs,
250 int argc, const char ** argv, int nongit, const char *prefix)
252 int i;
253 for (i = 1; i < argc; i++)
254 if (argv[i][0] != '-' || argv[i][1] == '\0')
255 break;
256 else if (!strcmp(argv[i], "--")) {
257 i++;
258 break;
259 } else if (i < argc - 3 && !strcmp(argv[i], "--no-index")) {
260 i = argc - 3;
261 revs->diffopt.exit_with_status = 1;
262 break;
264 if (argc != i + 2 || (!is_outside_repo(argv[i + 1], nongit, prefix) &&
265 !is_outside_repo(argv[i], nongit, prefix)))
266 return -1;
268 diff_setup(&revs->diffopt);
269 for (i = 1; i < argc - 2; )
270 if (!strcmp(argv[i], "--no-index"))
271 i++;
272 else {
273 int j = diff_opt_parse(&revs->diffopt,
274 argv + i, argc - i);
275 if (!j)
276 die("invalid diff option/value: %s", argv[i]);
277 i += j;
280 if (prefix) {
281 int len = strlen(prefix);
283 revs->diffopt.paths = xcalloc(2, sizeof(char*));
284 for (i = 0; i < 2; i++) {
285 const char *p = argv[argc - 2 + i];
287 * stdin should be spelled as '-'; if you have
288 * path that is '-', spell it as ./-.
290 p = (strcmp(p, "-")
291 ? xstrdup(prefix_filename(prefix, len, p))
292 : p);
293 revs->diffopt.paths[i] = p;
296 else
297 revs->diffopt.paths = argv + argc - 2;
298 revs->diffopt.nr_paths = 2;
299 revs->diffopt.no_index = 1;
300 revs->max_count = -2;
301 if (diff_setup_done(&revs->diffopt) < 0)
302 die("diff_setup_done failed");
303 return 0;
306 int run_diff_files_cmd(struct rev_info *revs, int argc, const char **argv)
308 int silent_on_removed;
310 if (handle_diff_files_args(revs, argc, argv, &silent_on_removed))
311 return -1;
313 if (revs->diffopt.no_index) {
314 if (revs->diffopt.nr_paths != 2)
315 return error("need two files/directories with --no-index");
316 if (queue_diff(&revs->diffopt, revs->diffopt.paths[0],
317 revs->diffopt.paths[1]))
318 return -1;
319 diffcore_std(&revs->diffopt);
320 diff_flush(&revs->diffopt);
322 * The return code for --no-index imitates diff(1):
323 * 0 = no changes, 1 = changes, else error
325 return revs->diffopt.found_changes;
328 if (read_cache() < 0) {
329 perror("read_cache");
330 return -1;
332 return run_diff_files(revs, silent_on_removed);
335 int run_diff_files(struct rev_info *revs, int silent_on_removed)
337 int entries, i;
338 int diff_unmerged_stage = revs->max_count;
340 if (diff_unmerged_stage < 0)
341 diff_unmerged_stage = 2;
342 entries = active_nr;
343 for (i = 0; i < entries; i++) {
344 struct stat st;
345 unsigned int oldmode, newmode;
346 struct cache_entry *ce = active_cache[i];
347 int changed;
349 if (revs->diffopt.quiet && revs->diffopt.has_changes)
350 break;
352 if (!ce_path_match(ce, revs->prune_data))
353 continue;
355 if (ce_stage(ce)) {
356 struct combine_diff_path *dpath;
357 int num_compare_stages = 0;
358 size_t path_len;
360 path_len = ce_namelen(ce);
362 dpath = xmalloc(combine_diff_path_size(5, path_len));
363 dpath->path = (char *) &(dpath->parent[5]);
365 dpath->next = NULL;
366 dpath->len = path_len;
367 memcpy(dpath->path, ce->name, path_len);
368 dpath->path[path_len] = '\0';
369 hashclr(dpath->sha1);
370 memset(&(dpath->parent[0]), 0,
371 sizeof(struct combine_diff_parent)*5);
373 if (lstat(ce->name, &st) < 0) {
374 if (errno != ENOENT && errno != ENOTDIR) {
375 perror(ce->name);
376 continue;
378 if (silent_on_removed)
379 continue;
381 else
382 dpath->mode = ntohl(ce_mode_from_stat(ce, st.st_mode));
384 while (i < entries) {
385 struct cache_entry *nce = active_cache[i];
386 int stage;
388 if (strcmp(ce->name, nce->name))
389 break;
391 /* Stage #2 (ours) is the first parent,
392 * stage #3 (theirs) is the second.
394 stage = ce_stage(nce);
395 if (2 <= stage) {
396 int mode = ntohl(nce->ce_mode);
397 num_compare_stages++;
398 hashcpy(dpath->parent[stage-2].sha1, nce->sha1);
399 dpath->parent[stage-2].mode = ntohl(ce_mode_from_stat(nce, mode));
400 dpath->parent[stage-2].status =
401 DIFF_STATUS_MODIFIED;
404 /* diff against the proper unmerged stage */
405 if (stage == diff_unmerged_stage)
406 ce = nce;
407 i++;
410 * Compensate for loop update
412 i--;
414 if (revs->combine_merges && num_compare_stages == 2) {
415 show_combined_diff(dpath, 2,
416 revs->dense_combined_merges,
417 revs);
418 free(dpath);
419 continue;
421 free(dpath);
422 dpath = NULL;
425 * Show the diff for the 'ce' if we found the one
426 * from the desired stage.
428 diff_unmerge(&revs->diffopt, ce->name, 0, null_sha1);
429 if (ce_stage(ce) != diff_unmerged_stage)
430 continue;
433 if (lstat(ce->name, &st) < 0) {
434 if (errno != ENOENT && errno != ENOTDIR) {
435 perror(ce->name);
436 continue;
438 if (silent_on_removed)
439 continue;
440 diff_addremove(&revs->diffopt, '-', ntohl(ce->ce_mode),
441 ce->sha1, ce->name, NULL);
442 continue;
444 changed = ce_match_stat(ce, &st, 0);
445 if (!changed && !revs->diffopt.find_copies_harder)
446 continue;
447 oldmode = ntohl(ce->ce_mode);
448 newmode = ntohl(ce_mode_from_stat(ce, st.st_mode));
449 diff_change(&revs->diffopt, oldmode, newmode,
450 ce->sha1, (changed ? null_sha1 : ce->sha1),
451 ce->name, NULL);
454 diffcore_std(&revs->diffopt);
455 diff_flush(&revs->diffopt);
456 return 0;
460 * diff-index
463 /* A file entry went away or appeared */
464 static void diff_index_show_file(struct rev_info *revs,
465 const char *prefix,
466 struct cache_entry *ce,
467 unsigned char *sha1, unsigned int mode)
469 diff_addremove(&revs->diffopt, prefix[0], ntohl(mode),
470 sha1, ce->name, NULL);
473 static int get_stat_data(struct cache_entry *ce,
474 unsigned char **sha1p,
475 unsigned int *modep,
476 int cached, int match_missing)
478 unsigned char *sha1 = ce->sha1;
479 unsigned int mode = ce->ce_mode;
481 if (!cached) {
482 static unsigned char no_sha1[20];
483 int changed;
484 struct stat st;
485 if (lstat(ce->name, &st) < 0) {
486 if (errno == ENOENT && match_missing) {
487 *sha1p = sha1;
488 *modep = mode;
489 return 0;
491 return -1;
493 changed = ce_match_stat(ce, &st, 0);
494 if (changed) {
495 mode = ce_mode_from_stat(ce, st.st_mode);
496 sha1 = no_sha1;
500 *sha1p = sha1;
501 *modep = mode;
502 return 0;
505 static void show_new_file(struct rev_info *revs,
506 struct cache_entry *new,
507 int cached, int match_missing)
509 unsigned char *sha1;
510 unsigned int mode;
512 /* New file in the index: it might actually be different in
513 * the working copy.
515 if (get_stat_data(new, &sha1, &mode, cached, match_missing) < 0)
516 return;
518 diff_index_show_file(revs, "+", new, sha1, mode);
521 static int show_modified(struct rev_info *revs,
522 struct cache_entry *old,
523 struct cache_entry *new,
524 int report_missing,
525 int cached, int match_missing)
527 unsigned int mode, oldmode;
528 unsigned char *sha1;
530 if (get_stat_data(new, &sha1, &mode, cached, match_missing) < 0) {
531 if (report_missing)
532 diff_index_show_file(revs, "-", old,
533 old->sha1, old->ce_mode);
534 return -1;
537 if (revs->combine_merges && !cached &&
538 (hashcmp(sha1, old->sha1) || hashcmp(old->sha1, new->sha1))) {
539 struct combine_diff_path *p;
540 int pathlen = ce_namelen(new);
542 p = xmalloc(combine_diff_path_size(2, pathlen));
543 p->path = (char *) &p->parent[2];
544 p->next = NULL;
545 p->len = pathlen;
546 memcpy(p->path, new->name, pathlen);
547 p->path[pathlen] = 0;
548 p->mode = ntohl(mode);
549 hashclr(p->sha1);
550 memset(p->parent, 0, 2 * sizeof(struct combine_diff_parent));
551 p->parent[0].status = DIFF_STATUS_MODIFIED;
552 p->parent[0].mode = ntohl(new->ce_mode);
553 hashcpy(p->parent[0].sha1, new->sha1);
554 p->parent[1].status = DIFF_STATUS_MODIFIED;
555 p->parent[1].mode = ntohl(old->ce_mode);
556 hashcpy(p->parent[1].sha1, old->sha1);
557 show_combined_diff(p, 2, revs->dense_combined_merges, revs);
558 free(p);
559 return 0;
562 oldmode = old->ce_mode;
563 if (mode == oldmode && !hashcmp(sha1, old->sha1) &&
564 !revs->diffopt.find_copies_harder)
565 return 0;
567 mode = ntohl(mode);
568 oldmode = ntohl(oldmode);
570 diff_change(&revs->diffopt, oldmode, mode,
571 old->sha1, sha1, old->name, NULL);
572 return 0;
575 static int diff_cache(struct rev_info *revs,
576 struct cache_entry **ac, int entries,
577 const char **pathspec,
578 int cached, int match_missing)
580 while (entries) {
581 struct cache_entry *ce = *ac;
582 int same = (entries > 1) && ce_same_name(ce, ac[1]);
584 if (revs->diffopt.quiet && revs->diffopt.has_changes)
585 break;
587 if (!ce_path_match(ce, pathspec))
588 goto skip_entry;
590 switch (ce_stage(ce)) {
591 case 0:
592 /* No stage 1 entry? That means it's a new file */
593 if (!same) {
594 show_new_file(revs, ce, cached, match_missing);
595 break;
597 /* Show difference between old and new */
598 show_modified(revs, ac[1], ce, 1,
599 cached, match_missing);
600 break;
601 case 1:
602 /* No stage 3 (merge) entry?
603 * That means it's been deleted.
605 if (!same) {
606 diff_index_show_file(revs, "-", ce,
607 ce->sha1, ce->ce_mode);
608 break;
610 /* We come here with ce pointing at stage 1
611 * (original tree) and ac[1] pointing at stage
612 * 3 (unmerged). show-modified with
613 * report-missing set to false does not say the
614 * file is deleted but reports true if work
615 * tree does not have it, in which case we
616 * fall through to report the unmerged state.
617 * Otherwise, we show the differences between
618 * the original tree and the work tree.
620 if (!cached &&
621 !show_modified(revs, ce, ac[1], 0,
622 cached, match_missing))
623 break;
624 diff_unmerge(&revs->diffopt, ce->name,
625 ntohl(ce->ce_mode), ce->sha1);
626 break;
627 case 3:
628 diff_unmerge(&revs->diffopt, ce->name,
629 0, null_sha1);
630 break;
632 default:
633 die("impossible cache entry stage");
636 skip_entry:
638 * Ignore all the different stages for this file,
639 * we've handled the relevant cases now.
641 do {
642 ac++;
643 entries--;
644 } while (entries && ce_same_name(ce, ac[0]));
646 return 0;
650 * This turns all merge entries into "stage 3". That guarantees that
651 * when we read in the new tree (into "stage 1"), we won't lose sight
652 * of the fact that we had unmerged entries.
654 static void mark_merge_entries(void)
656 int i;
657 for (i = 0; i < active_nr; i++) {
658 struct cache_entry *ce = active_cache[i];
659 if (!ce_stage(ce))
660 continue;
661 ce->ce_flags |= htons(CE_STAGEMASK);
665 int run_diff_index(struct rev_info *revs, int cached)
667 int ret;
668 struct object *ent;
669 struct tree *tree;
670 const char *tree_name;
671 int match_missing = 0;
674 * Backward compatibility wart - "diff-index -m" does
675 * not mean "do not ignore merges", but totally different.
677 if (!revs->ignore_merges)
678 match_missing = 1;
680 mark_merge_entries();
682 ent = revs->pending.objects[0].item;
683 tree_name = revs->pending.objects[0].name;
684 tree = parse_tree_indirect(ent->sha1);
685 if (!tree)
686 return error("bad tree object %s", tree_name);
687 if (read_tree(tree, 1, revs->prune_data))
688 return error("unable to read tree object %s", tree_name);
689 ret = diff_cache(revs, active_cache, active_nr, revs->prune_data,
690 cached, match_missing);
691 diffcore_std(&revs->diffopt);
692 diff_flush(&revs->diffopt);
693 return ret;
696 int do_diff_cache(const unsigned char *tree_sha1, struct diff_options *opt)
698 struct tree *tree;
699 struct rev_info revs;
700 int i;
701 struct cache_entry **dst;
702 struct cache_entry *last = NULL;
705 * This is used by git-blame to run diff-cache internally;
706 * it potentially needs to repeatedly run this, so we will
707 * start by removing the higher order entries the last round
708 * left behind.
710 dst = active_cache;
711 for (i = 0; i < active_nr; i++) {
712 struct cache_entry *ce = active_cache[i];
713 if (ce_stage(ce)) {
714 if (last && !strcmp(ce->name, last->name))
715 continue;
716 cache_tree_invalidate_path(active_cache_tree,
717 ce->name);
718 last = ce;
719 ce->ce_mode = 0;
720 ce->ce_flags &= ~htons(CE_STAGEMASK);
722 *dst++ = ce;
724 active_nr = dst - active_cache;
726 init_revisions(&revs, NULL);
727 revs.prune_data = opt->paths;
728 tree = parse_tree_indirect(tree_sha1);
729 if (!tree)
730 die("bad tree object %s", sha1_to_hex(tree_sha1));
731 if (read_tree(tree, 1, opt->paths))
732 return error("unable to read tree %s", sha1_to_hex(tree_sha1));
733 return diff_cache(&revs, active_cache, active_nr, revs.prune_data,
734 1, 0);