Fix git-fast-export for zero-sized blobs
[git/jrn.git] / diff-lib.c
blobd85d8f34ba055ad7e02bf5d61ae4a675ce0be1d3
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 (DIFF_OPT_TST(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,
177 unsigned int *options)
179 *options = 0;
181 /* revs->max_count == -2 means --no-index */
182 while (1 < argc && argv[1][0] == '-') {
183 if (!strcmp(argv[1], "--base"))
184 revs->max_count = 1;
185 else if (!strcmp(argv[1], "--ours"))
186 revs->max_count = 2;
187 else if (!strcmp(argv[1], "--theirs"))
188 revs->max_count = 3;
189 else if (!strcmp(argv[1], "-n") ||
190 !strcmp(argv[1], "--no-index")) {
191 revs->max_count = -2;
192 DIFF_OPT_SET(&revs->diffopt, EXIT_WITH_STATUS);
193 DIFF_OPT_SET(&revs->diffopt, NO_INDEX);
195 else if (!strcmp(argv[1], "-q"))
196 *options |= DIFF_SILENT_ON_REMOVED;
197 else
198 return error("invalid option: %s", argv[1]);
199 argv++; argc--;
202 if (revs->max_count == -1 && revs->diffopt.nr_paths == 2) {
204 * If two files are specified, and at least one is untracked,
205 * default to no-index.
207 read_cache();
208 if (!is_in_index(revs->diffopt.paths[0]) ||
209 !is_in_index(revs->diffopt.paths[1])) {
210 revs->max_count = -2;
211 DIFF_OPT_SET(&revs->diffopt, NO_INDEX);
216 * Make sure there are NO revision (i.e. pending object) parameter,
217 * rev.max_count is reasonable (0 <= n <= 3),
218 * there is no other revision filtering parameters.
220 if (revs->pending.nr || revs->max_count > 3 ||
221 revs->min_age != -1 || revs->max_age != -1)
222 return error("no revision allowed with diff-files");
224 if (revs->max_count == -1 &&
225 (revs->diffopt.output_format & DIFF_FORMAT_PATCH))
226 revs->combine_merges = revs->dense_combined_merges = 1;
228 return 0;
231 static int is_outside_repo(const char *path, int nongit, const char *prefix)
233 int i;
234 if (nongit || !strcmp(path, "-") || is_absolute_path(path))
235 return 1;
236 if (prefixcmp(path, "../"))
237 return 0;
238 if (!prefix)
239 return 1;
240 for (i = strlen(prefix); !prefixcmp(path, "../"); ) {
241 while (i > 0 && prefix[i - 1] != '/')
242 i--;
243 if (--i < 0)
244 return 1;
245 path += 3;
247 return 0;
250 int setup_diff_no_index(struct rev_info *revs,
251 int argc, const char ** argv, int nongit, const char *prefix)
253 int i;
254 for (i = 1; i < argc; i++)
255 if (argv[i][0] != '-' || argv[i][1] == '\0')
256 break;
257 else if (!strcmp(argv[i], "--")) {
258 i++;
259 break;
260 } else if (i < argc - 3 && !strcmp(argv[i], "--no-index")) {
261 i = argc - 3;
262 DIFF_OPT_SET(&revs->diffopt, EXIT_WITH_STATUS);
263 break;
265 if (argc != i + 2 || (!is_outside_repo(argv[i + 1], nongit, prefix) &&
266 !is_outside_repo(argv[i], nongit, prefix)))
267 return -1;
269 diff_setup(&revs->diffopt);
270 for (i = 1; i < argc - 2; )
271 if (!strcmp(argv[i], "--no-index"))
272 i++;
273 else {
274 int j = diff_opt_parse(&revs->diffopt,
275 argv + i, argc - i);
276 if (!j)
277 die("invalid diff option/value: %s", argv[i]);
278 i += j;
281 if (prefix) {
282 int len = strlen(prefix);
284 revs->diffopt.paths = xcalloc(2, sizeof(char*));
285 for (i = 0; i < 2; i++) {
286 const char *p = argv[argc - 2 + i];
288 * stdin should be spelled as '-'; if you have
289 * path that is '-', spell it as ./-.
291 p = (strcmp(p, "-")
292 ? xstrdup(prefix_filename(prefix, len, p))
293 : p);
294 revs->diffopt.paths[i] = p;
297 else
298 revs->diffopt.paths = argv + argc - 2;
299 revs->diffopt.nr_paths = 2;
300 DIFF_OPT_SET(&revs->diffopt, NO_INDEX);
301 revs->max_count = -2;
302 if (diff_setup_done(&revs->diffopt) < 0)
303 die("diff_setup_done failed");
304 return 0;
307 int run_diff_files_cmd(struct rev_info *revs, int argc, const char **argv)
309 unsigned int options;
311 if (handle_diff_files_args(revs, argc, argv, &options))
312 return -1;
314 if (DIFF_OPT_TST(&revs->diffopt, NO_INDEX)) {
315 if (revs->diffopt.nr_paths != 2)
316 return error("need two files/directories with --no-index");
317 if (queue_diff(&revs->diffopt, revs->diffopt.paths[0],
318 revs->diffopt.paths[1]))
319 return -1;
320 diffcore_std(&revs->diffopt);
321 diff_flush(&revs->diffopt);
323 * The return code for --no-index imitates diff(1):
324 * 0 = no changes, 1 = changes, else error
326 return revs->diffopt.found_changes;
329 if (read_cache() < 0) {
330 perror("read_cache");
331 return -1;
333 return run_diff_files(revs, options);
336 int run_diff_files(struct rev_info *revs, unsigned int option)
338 int entries, i;
339 int diff_unmerged_stage = revs->max_count;
340 int silent_on_removed = option & DIFF_SILENT_ON_REMOVED;
341 unsigned ce_option = ((option & DIFF_RACY_IS_MODIFIED)
342 ? CE_MATCH_RACY_IS_DIRTY : 0);
344 if (diff_unmerged_stage < 0)
345 diff_unmerged_stage = 2;
346 entries = active_nr;
347 for (i = 0; i < entries; i++) {
348 struct stat st;
349 unsigned int oldmode, newmode;
350 struct cache_entry *ce = active_cache[i];
351 int changed;
353 if (DIFF_OPT_TST(&revs->diffopt, QUIET) &&
354 DIFF_OPT_TST(&revs->diffopt, HAS_CHANGES))
355 break;
357 if (!ce_path_match(ce, revs->prune_data))
358 continue;
360 if (ce_stage(ce)) {
361 struct combine_diff_path *dpath;
362 int num_compare_stages = 0;
363 size_t path_len;
365 path_len = ce_namelen(ce);
367 dpath = xmalloc(combine_diff_path_size(5, path_len));
368 dpath->path = (char *) &(dpath->parent[5]);
370 dpath->next = NULL;
371 dpath->len = path_len;
372 memcpy(dpath->path, ce->name, path_len);
373 dpath->path[path_len] = '\0';
374 hashclr(dpath->sha1);
375 memset(&(dpath->parent[0]), 0,
376 sizeof(struct combine_diff_parent)*5);
378 if (lstat(ce->name, &st) < 0) {
379 if (errno != ENOENT && errno != ENOTDIR) {
380 perror(ce->name);
381 continue;
383 if (silent_on_removed)
384 continue;
386 else
387 dpath->mode = ntohl(ce_mode_from_stat(ce, st.st_mode));
389 while (i < entries) {
390 struct cache_entry *nce = active_cache[i];
391 int stage;
393 if (strcmp(ce->name, nce->name))
394 break;
396 /* Stage #2 (ours) is the first parent,
397 * stage #3 (theirs) is the second.
399 stage = ce_stage(nce);
400 if (2 <= stage) {
401 int mode = ntohl(nce->ce_mode);
402 num_compare_stages++;
403 hashcpy(dpath->parent[stage-2].sha1, nce->sha1);
404 dpath->parent[stage-2].mode = ntohl(ce_mode_from_stat(nce, mode));
405 dpath->parent[stage-2].status =
406 DIFF_STATUS_MODIFIED;
409 /* diff against the proper unmerged stage */
410 if (stage == diff_unmerged_stage)
411 ce = nce;
412 i++;
415 * Compensate for loop update
417 i--;
419 if (revs->combine_merges && num_compare_stages == 2) {
420 show_combined_diff(dpath, 2,
421 revs->dense_combined_merges,
422 revs);
423 free(dpath);
424 continue;
426 free(dpath);
427 dpath = NULL;
430 * Show the diff for the 'ce' if we found the one
431 * from the desired stage.
433 diff_unmerge(&revs->diffopt, ce->name, 0, null_sha1);
434 if (ce_stage(ce) != diff_unmerged_stage)
435 continue;
438 if (lstat(ce->name, &st) < 0) {
439 if (errno != ENOENT && errno != ENOTDIR) {
440 perror(ce->name);
441 continue;
443 if (silent_on_removed)
444 continue;
445 diff_addremove(&revs->diffopt, '-', ntohl(ce->ce_mode),
446 ce->sha1, ce->name, NULL);
447 continue;
449 changed = ce_match_stat(ce, &st, ce_option);
450 if (!changed && !DIFF_OPT_TST(&revs->diffopt, FIND_COPIES_HARDER))
451 continue;
452 oldmode = ntohl(ce->ce_mode);
453 newmode = ntohl(ce_mode_from_stat(ce, st.st_mode));
454 diff_change(&revs->diffopt, oldmode, newmode,
455 ce->sha1, (changed ? null_sha1 : ce->sha1),
456 ce->name, NULL);
459 diffcore_std(&revs->diffopt);
460 diff_flush(&revs->diffopt);
461 return 0;
465 * diff-index
468 /* A file entry went away or appeared */
469 static void diff_index_show_file(struct rev_info *revs,
470 const char *prefix,
471 struct cache_entry *ce,
472 unsigned char *sha1, unsigned int mode)
474 diff_addremove(&revs->diffopt, prefix[0], ntohl(mode),
475 sha1, ce->name, NULL);
478 static int get_stat_data(struct cache_entry *ce,
479 unsigned char **sha1p,
480 unsigned int *modep,
481 int cached, int match_missing)
483 unsigned char *sha1 = ce->sha1;
484 unsigned int mode = ce->ce_mode;
486 if (!cached) {
487 static unsigned char no_sha1[20];
488 int changed;
489 struct stat st;
490 if (lstat(ce->name, &st) < 0) {
491 if (errno == ENOENT && match_missing) {
492 *sha1p = sha1;
493 *modep = mode;
494 return 0;
496 return -1;
498 changed = ce_match_stat(ce, &st, 0);
499 if (changed) {
500 mode = ce_mode_from_stat(ce, st.st_mode);
501 sha1 = no_sha1;
505 *sha1p = sha1;
506 *modep = mode;
507 return 0;
510 static void show_new_file(struct rev_info *revs,
511 struct cache_entry *new,
512 int cached, int match_missing)
514 unsigned char *sha1;
515 unsigned int mode;
517 /* New file in the index: it might actually be different in
518 * the working copy.
520 if (get_stat_data(new, &sha1, &mode, cached, match_missing) < 0)
521 return;
523 diff_index_show_file(revs, "+", new, sha1, mode);
526 static int show_modified(struct rev_info *revs,
527 struct cache_entry *old,
528 struct cache_entry *new,
529 int report_missing,
530 int cached, int match_missing)
532 unsigned int mode, oldmode;
533 unsigned char *sha1;
535 if (get_stat_data(new, &sha1, &mode, cached, match_missing) < 0) {
536 if (report_missing)
537 diff_index_show_file(revs, "-", old,
538 old->sha1, old->ce_mode);
539 return -1;
542 if (revs->combine_merges && !cached &&
543 (hashcmp(sha1, old->sha1) || hashcmp(old->sha1, new->sha1))) {
544 struct combine_diff_path *p;
545 int pathlen = ce_namelen(new);
547 p = xmalloc(combine_diff_path_size(2, pathlen));
548 p->path = (char *) &p->parent[2];
549 p->next = NULL;
550 p->len = pathlen;
551 memcpy(p->path, new->name, pathlen);
552 p->path[pathlen] = 0;
553 p->mode = ntohl(mode);
554 hashclr(p->sha1);
555 memset(p->parent, 0, 2 * sizeof(struct combine_diff_parent));
556 p->parent[0].status = DIFF_STATUS_MODIFIED;
557 p->parent[0].mode = ntohl(new->ce_mode);
558 hashcpy(p->parent[0].sha1, new->sha1);
559 p->parent[1].status = DIFF_STATUS_MODIFIED;
560 p->parent[1].mode = ntohl(old->ce_mode);
561 hashcpy(p->parent[1].sha1, old->sha1);
562 show_combined_diff(p, 2, revs->dense_combined_merges, revs);
563 free(p);
564 return 0;
567 oldmode = old->ce_mode;
568 if (mode == oldmode && !hashcmp(sha1, old->sha1) &&
569 !DIFF_OPT_TST(&revs->diffopt, FIND_COPIES_HARDER))
570 return 0;
572 mode = ntohl(mode);
573 oldmode = ntohl(oldmode);
575 diff_change(&revs->diffopt, oldmode, mode,
576 old->sha1, sha1, old->name, NULL);
577 return 0;
580 static int diff_cache(struct rev_info *revs,
581 struct cache_entry **ac, int entries,
582 const char **pathspec,
583 int cached, int match_missing)
585 while (entries) {
586 struct cache_entry *ce = *ac;
587 int same = (entries > 1) && ce_same_name(ce, ac[1]);
589 if (DIFF_OPT_TST(&revs->diffopt, QUIET) &&
590 DIFF_OPT_TST(&revs->diffopt, HAS_CHANGES))
591 break;
593 if (!ce_path_match(ce, pathspec))
594 goto skip_entry;
596 switch (ce_stage(ce)) {
597 case 0:
598 /* No stage 1 entry? That means it's a new file */
599 if (!same) {
600 show_new_file(revs, ce, cached, match_missing);
601 break;
603 /* Show difference between old and new */
604 show_modified(revs, ac[1], ce, 1,
605 cached, match_missing);
606 break;
607 case 1:
608 /* No stage 3 (merge) entry?
609 * That means it's been deleted.
611 if (!same) {
612 diff_index_show_file(revs, "-", ce,
613 ce->sha1, ce->ce_mode);
614 break;
616 /* We come here with ce pointing at stage 1
617 * (original tree) and ac[1] pointing at stage
618 * 3 (unmerged). show-modified with
619 * report-missing set to false does not say the
620 * file is deleted but reports true if work
621 * tree does not have it, in which case we
622 * fall through to report the unmerged state.
623 * Otherwise, we show the differences between
624 * the original tree and the work tree.
626 if (!cached &&
627 !show_modified(revs, ce, ac[1], 0,
628 cached, match_missing))
629 break;
630 diff_unmerge(&revs->diffopt, ce->name,
631 ntohl(ce->ce_mode), ce->sha1);
632 break;
633 case 3:
634 diff_unmerge(&revs->diffopt, ce->name,
635 0, null_sha1);
636 break;
638 default:
639 die("impossible cache entry stage");
642 skip_entry:
644 * Ignore all the different stages for this file,
645 * we've handled the relevant cases now.
647 do {
648 ac++;
649 entries--;
650 } while (entries && ce_same_name(ce, ac[0]));
652 return 0;
656 * This turns all merge entries into "stage 3". That guarantees that
657 * when we read in the new tree (into "stage 1"), we won't lose sight
658 * of the fact that we had unmerged entries.
660 static void mark_merge_entries(void)
662 int i;
663 for (i = 0; i < active_nr; i++) {
664 struct cache_entry *ce = active_cache[i];
665 if (!ce_stage(ce))
666 continue;
667 ce->ce_flags |= htons(CE_STAGEMASK);
671 int run_diff_index(struct rev_info *revs, int cached)
673 int ret;
674 struct object *ent;
675 struct tree *tree;
676 const char *tree_name;
677 int match_missing = 0;
680 * Backward compatibility wart - "diff-index -m" does
681 * not mean "do not ignore merges", but totally different.
683 if (!revs->ignore_merges)
684 match_missing = 1;
686 mark_merge_entries();
688 ent = revs->pending.objects[0].item;
689 tree_name = revs->pending.objects[0].name;
690 tree = parse_tree_indirect(ent->sha1);
691 if (!tree)
692 return error("bad tree object %s", tree_name);
693 if (read_tree(tree, 1, revs->prune_data))
694 return error("unable to read tree object %s", tree_name);
695 ret = diff_cache(revs, active_cache, active_nr, revs->prune_data,
696 cached, match_missing);
697 diffcore_std(&revs->diffopt);
698 diff_flush(&revs->diffopt);
699 return ret;
702 int do_diff_cache(const unsigned char *tree_sha1, struct diff_options *opt)
704 struct tree *tree;
705 struct rev_info revs;
706 int i;
707 struct cache_entry **dst;
708 struct cache_entry *last = NULL;
711 * This is used by git-blame to run diff-cache internally;
712 * it potentially needs to repeatedly run this, so we will
713 * start by removing the higher order entries the last round
714 * left behind.
716 dst = active_cache;
717 for (i = 0; i < active_nr; i++) {
718 struct cache_entry *ce = active_cache[i];
719 if (ce_stage(ce)) {
720 if (last && !strcmp(ce->name, last->name))
721 continue;
722 cache_tree_invalidate_path(active_cache_tree,
723 ce->name);
724 last = ce;
725 ce->ce_mode = 0;
726 ce->ce_flags &= ~htons(CE_STAGEMASK);
728 *dst++ = ce;
730 active_nr = dst - active_cache;
732 init_revisions(&revs, NULL);
733 revs.prune_data = opt->paths;
734 tree = parse_tree_indirect(tree_sha1);
735 if (!tree)
736 die("bad tree object %s", sha1_to_hex(tree_sha1));
737 if (read_tree(tree, 1, opt->paths))
738 return error("unable to read tree %s", sha1_to_hex(tree_sha1));
739 return diff_cache(&revs, active_cache, active_nr, revs.prune_data,
740 1, 0);