Use off_t when we really mean a file offset.
[git/fastimport.git] / diff-lib.c
blob778cf58244d64b1e1b14ead43c2fdf5735bd717f
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(xstrdup(e->d_name), list);
29 closedir(dir);
30 return 0;
33 static int queue_diff(struct diff_options *o,
34 const char *name1, const char *name2)
36 struct stat st;
37 int mode1 = 0, mode2 = 0;
39 if (name1) {
40 if (stat(name1, &st))
41 return error("Could not access '%s'", name1);
42 mode1 = st.st_mode;
44 if (name2) {
45 if (stat(name2, &st))
46 return error("Could not access '%s'", name2);
47 mode2 = st.st_mode;
50 if (mode1 && mode2 && S_ISDIR(mode1) != S_ISDIR(mode2))
51 return error("file/directory conflict: %s, %s", name1, name2);
53 if (S_ISDIR(mode1) || S_ISDIR(mode2)) {
54 char buffer1[PATH_MAX], buffer2[PATH_MAX];
55 struct path_list p1 = {NULL, 0, 0, 1}, p2 = {NULL, 0, 0, 1};
56 int len1 = 0, len2 = 0, i1, i2, ret = 0;
58 if (name1 && read_directory(name1, &p1))
59 return -1;
60 if (name2 && read_directory(name2, &p2)) {
61 path_list_clear(&p1, 0);
62 return -1;
65 if (name1) {
66 len1 = strlen(name1);
67 if (len1 > 0 && name1[len1 - 1] == '/')
68 len1--;
69 memcpy(buffer1, name1, len1);
70 buffer1[len1++] = '/';
73 if (name2) {
74 len2 = strlen(name2);
75 if (len2 > 0 && name2[len2 - 1] == '/')
76 len2--;
77 memcpy(buffer2, name2, len2);
78 buffer2[len2++] = '/';
81 for (i1 = i2 = 0; !ret && (i1 < p1.nr || i2 < p2.nr); ) {
82 const char *n1, *n2;
83 int comp;
85 if (i1 == p1.nr)
86 comp = 1;
87 else if (i2 == p2.nr)
88 comp = -1;
89 else
90 comp = strcmp(p1.items[i1].path,
91 p2.items[i2].path);
93 if (comp > 0)
94 n1 = NULL;
95 else {
96 n1 = buffer1;
97 strncpy(buffer1 + len1, p1.items[i1++].path,
98 PATH_MAX - len1);
101 if (comp < 0)
102 n2 = NULL;
103 else {
104 n2 = buffer2;
105 strncpy(buffer2 + len2, p2.items[i2++].path,
106 PATH_MAX - len2);
109 ret = queue_diff(o, n1, n2);
111 path_list_clear(&p1, 0);
112 path_list_clear(&p2, 0);
114 return ret;
115 } else {
116 struct diff_filespec *d1, *d2;
118 if (o->reverse_diff) {
119 unsigned tmp;
120 const char *tmp_c;
121 tmp = mode1; mode1 = mode2; mode2 = tmp;
122 tmp_c = name1; name1 = name2; name2 = tmp_c;
125 if (!name1)
126 name1 = "/dev/null";
127 if (!name2)
128 name2 = "/dev/null";
129 d1 = alloc_filespec(name1);
130 d2 = alloc_filespec(name2);
131 fill_filespec(d1, null_sha1, mode1);
132 fill_filespec(d2, null_sha1, mode2);
134 diff_queue(&diff_queued_diff, d1, d2);
135 return 0;
139 static int is_in_index(const char *path)
141 int len = strlen(path);
142 int pos = cache_name_pos(path, len);
143 char c;
145 if (pos < 0)
146 return 0;
147 if (strncmp(active_cache[pos]->name, path, len))
148 return 0;
149 c = active_cache[pos]->name[len];
150 return c == '\0' || c == '/';
153 static int handle_diff_files_args(struct rev_info *revs,
154 int argc, const char **argv, int *silent)
156 *silent = 0;
158 /* revs->max_count == -2 means --no-index */
159 while (1 < argc && argv[1][0] == '-') {
160 if (!strcmp(argv[1], "--base"))
161 revs->max_count = 1;
162 else if (!strcmp(argv[1], "--ours"))
163 revs->max_count = 2;
164 else if (!strcmp(argv[1], "--theirs"))
165 revs->max_count = 3;
166 else if (!strcmp(argv[1], "-n") ||
167 !strcmp(argv[1], "--no-index"))
168 revs->max_count = -2;
169 else if (!strcmp(argv[1], "-q"))
170 *silent = 1;
171 else
172 return error("invalid option: %s", argv[1]);
173 argv++; argc--;
176 if (revs->max_count == -1 && revs->diffopt.nr_paths == 2) {
178 * If two files are specified, and at least one is untracked,
179 * default to no-index.
181 read_cache();
182 if (!is_in_index(revs->diffopt.paths[0]) ||
183 !is_in_index(revs->diffopt.paths[1]))
184 revs->max_count = -2;
188 * Make sure there are NO revision (i.e. pending object) parameter,
189 * rev.max_count is reasonable (0 <= n <= 3),
190 * there is no other revision filtering parameters.
192 if (revs->pending.nr || revs->max_count > 3 ||
193 revs->min_age != -1 || revs->max_age != -1)
194 return error("no revision allowed with diff-files");
196 if (revs->max_count == -1 &&
197 (revs->diffopt.output_format & DIFF_FORMAT_PATCH))
198 revs->combine_merges = revs->dense_combined_merges = 1;
200 return 0;
203 static int is_outside_repo(const char *path, int nongit, const char *prefix)
205 int i;
206 if (nongit || !strcmp(path, "-") || path[0] == '/')
207 return 1;
208 if (prefixcmp(path, "../"))
209 return 0;
210 if (!prefix)
211 return 1;
212 for (i = strlen(prefix); !prefixcmp(path, "../"); ) {
213 while (i > 0 && prefix[i - 1] != '/')
214 i--;
215 if (--i < 0)
216 return 1;
217 path += 3;
219 return 0;
222 int setup_diff_no_index(struct rev_info *revs,
223 int argc, const char ** argv, int nongit, const char *prefix)
225 int i;
226 for (i = 1; i < argc; i++)
227 if (argv[i][0] != '-')
228 break;
229 else if (!strcmp(argv[i], "--")) {
230 i++;
231 break;
232 } else if (i < argc - 3 && !strcmp(argv[i], "--no-index")) {
233 i = argc - 3;
234 break;
236 if (argc != i + 2 || (!is_outside_repo(argv[i + 1], nongit, prefix) &&
237 !is_outside_repo(argv[i], nongit, prefix)))
238 return -1;
240 diff_setup(&revs->diffopt);
241 for (i = 1; i < argc - 2; )
242 if (!strcmp(argv[i], "--no-index"))
243 i++;
244 else {
245 int j = diff_opt_parse(&revs->diffopt,
246 argv + i, argc - i);
247 if (!j)
248 die("invalid diff option/value: %s", argv[i]);
249 i += j;
252 if (prefix) {
253 int len = strlen(prefix);
255 revs->diffopt.paths = xcalloc(2, sizeof(char*));
256 for (i = 0; i < 2; i++) {
257 const char *p;
258 p = prefix_filename(prefix, len, argv[argc - 2 + i]);
259 revs->diffopt.paths[i] = xstrdup(p);
262 else
263 revs->diffopt.paths = argv + argc - 2;
264 revs->diffopt.nr_paths = 2;
265 revs->max_count = -2;
266 return 0;
269 int run_diff_files_cmd(struct rev_info *revs, int argc, const char **argv)
271 int silent_on_removed;
273 if (handle_diff_files_args(revs, argc, argv, &silent_on_removed))
274 return -1;
276 if (revs->max_count == -2) {
277 if (revs->diffopt.nr_paths != 2)
278 return error("need two files/directories with --no-index");
279 if (queue_diff(&revs->diffopt, revs->diffopt.paths[0],
280 revs->diffopt.paths[1]))
281 return -1;
282 diffcore_std(&revs->diffopt);
283 diff_flush(&revs->diffopt);
285 * The return code for --no-index imitates diff(1):
286 * 0 = no changes, 1 = changes, else error
288 return revs->diffopt.found_changes;
291 if (read_cache() < 0) {
292 perror("read_cache");
293 return -1;
295 return run_diff_files(revs, silent_on_removed);
298 int run_diff_files(struct rev_info *revs, int silent_on_removed)
300 int entries, i;
301 int diff_unmerged_stage = revs->max_count;
303 if (diff_unmerged_stage < 0)
304 diff_unmerged_stage = 2;
305 entries = active_nr;
306 for (i = 0; i < entries; i++) {
307 struct stat st;
308 unsigned int oldmode, newmode;
309 struct cache_entry *ce = active_cache[i];
310 int changed;
312 if (!ce_path_match(ce, revs->prune_data))
313 continue;
315 if (ce_stage(ce)) {
316 struct combine_diff_path *dpath;
317 int num_compare_stages = 0;
318 size_t path_len;
320 path_len = ce_namelen(ce);
322 dpath = xmalloc(combine_diff_path_size(5, path_len));
323 dpath->path = (char *) &(dpath->parent[5]);
325 dpath->next = NULL;
326 dpath->len = path_len;
327 memcpy(dpath->path, ce->name, path_len);
328 dpath->path[path_len] = '\0';
329 hashclr(dpath->sha1);
330 memset(&(dpath->parent[0]), 0,
331 sizeof(struct combine_diff_parent)*5);
333 if (lstat(ce->name, &st) < 0) {
334 if (errno != ENOENT && errno != ENOTDIR) {
335 perror(ce->name);
336 continue;
338 if (silent_on_removed)
339 continue;
341 else
342 dpath->mode = canon_mode(st.st_mode);
344 while (i < entries) {
345 struct cache_entry *nce = active_cache[i];
346 int stage;
348 if (strcmp(ce->name, nce->name))
349 break;
351 /* Stage #2 (ours) is the first parent,
352 * stage #3 (theirs) is the second.
354 stage = ce_stage(nce);
355 if (2 <= stage) {
356 int mode = ntohl(nce->ce_mode);
357 num_compare_stages++;
358 hashcpy(dpath->parent[stage-2].sha1, nce->sha1);
359 dpath->parent[stage-2].mode =
360 canon_mode(mode);
361 dpath->parent[stage-2].status =
362 DIFF_STATUS_MODIFIED;
365 /* diff against the proper unmerged stage */
366 if (stage == diff_unmerged_stage)
367 ce = nce;
368 i++;
371 * Compensate for loop update
373 i--;
375 if (revs->combine_merges && num_compare_stages == 2) {
376 show_combined_diff(dpath, 2,
377 revs->dense_combined_merges,
378 revs);
379 free(dpath);
380 continue;
382 free(dpath);
383 dpath = NULL;
386 * Show the diff for the 'ce' if we found the one
387 * from the desired stage.
389 diff_unmerge(&revs->diffopt, ce->name, 0, null_sha1);
390 if (ce_stage(ce) != diff_unmerged_stage)
391 continue;
394 if (lstat(ce->name, &st) < 0) {
395 if (errno != ENOENT && errno != ENOTDIR) {
396 perror(ce->name);
397 continue;
399 if (silent_on_removed)
400 continue;
401 diff_addremove(&revs->diffopt, '-', ntohl(ce->ce_mode),
402 ce->sha1, ce->name, NULL);
403 continue;
405 changed = ce_match_stat(ce, &st, 0);
406 if (!changed && !revs->diffopt.find_copies_harder)
407 continue;
408 oldmode = ntohl(ce->ce_mode);
410 newmode = canon_mode(st.st_mode);
411 if (!trust_executable_bit &&
412 S_ISREG(newmode) && S_ISREG(oldmode) &&
413 ((newmode ^ oldmode) == 0111))
414 newmode = oldmode;
415 else if (!has_symlinks &&
416 S_ISREG(newmode) && S_ISLNK(oldmode))
417 newmode = oldmode;
418 diff_change(&revs->diffopt, oldmode, newmode,
419 ce->sha1, (changed ? null_sha1 : ce->sha1),
420 ce->name, NULL);
423 diffcore_std(&revs->diffopt);
424 diff_flush(&revs->diffopt);
425 return 0;
429 * diff-index
432 /* A file entry went away or appeared */
433 static void diff_index_show_file(struct rev_info *revs,
434 const char *prefix,
435 struct cache_entry *ce,
436 unsigned char *sha1, unsigned int mode)
438 diff_addremove(&revs->diffopt, prefix[0], ntohl(mode),
439 sha1, ce->name, NULL);
442 static int get_stat_data(struct cache_entry *ce,
443 unsigned char **sha1p,
444 unsigned int *modep,
445 int cached, int match_missing)
447 unsigned char *sha1 = ce->sha1;
448 unsigned int mode = ce->ce_mode;
450 if (!cached) {
451 static unsigned char no_sha1[20];
452 int changed;
453 struct stat st;
454 if (lstat(ce->name, &st) < 0) {
455 if (errno == ENOENT && match_missing) {
456 *sha1p = sha1;
457 *modep = mode;
458 return 0;
460 return -1;
462 changed = ce_match_stat(ce, &st, 0);
463 if (changed) {
464 mode = ce_mode_from_stat(ce, st.st_mode);
465 sha1 = no_sha1;
469 *sha1p = sha1;
470 *modep = mode;
471 return 0;
474 static void show_new_file(struct rev_info *revs,
475 struct cache_entry *new,
476 int cached, int match_missing)
478 unsigned char *sha1;
479 unsigned int mode;
481 /* New file in the index: it might actually be different in
482 * the working copy.
484 if (get_stat_data(new, &sha1, &mode, cached, match_missing) < 0)
485 return;
487 diff_index_show_file(revs, "+", new, sha1, mode);
490 static int show_modified(struct rev_info *revs,
491 struct cache_entry *old,
492 struct cache_entry *new,
493 int report_missing,
494 int cached, int match_missing)
496 unsigned int mode, oldmode;
497 unsigned char *sha1;
499 if (get_stat_data(new, &sha1, &mode, cached, match_missing) < 0) {
500 if (report_missing)
501 diff_index_show_file(revs, "-", old,
502 old->sha1, old->ce_mode);
503 return -1;
506 if (revs->combine_merges && !cached &&
507 (hashcmp(sha1, old->sha1) || hashcmp(old->sha1, new->sha1))) {
508 struct combine_diff_path *p;
509 int pathlen = ce_namelen(new);
511 p = xmalloc(combine_diff_path_size(2, pathlen));
512 p->path = (char *) &p->parent[2];
513 p->next = NULL;
514 p->len = pathlen;
515 memcpy(p->path, new->name, pathlen);
516 p->path[pathlen] = 0;
517 p->mode = ntohl(mode);
518 hashclr(p->sha1);
519 memset(p->parent, 0, 2 * sizeof(struct combine_diff_parent));
520 p->parent[0].status = DIFF_STATUS_MODIFIED;
521 p->parent[0].mode = ntohl(new->ce_mode);
522 hashcpy(p->parent[0].sha1, new->sha1);
523 p->parent[1].status = DIFF_STATUS_MODIFIED;
524 p->parent[1].mode = ntohl(old->ce_mode);
525 hashcpy(p->parent[1].sha1, old->sha1);
526 show_combined_diff(p, 2, revs->dense_combined_merges, revs);
527 free(p);
528 return 0;
531 oldmode = old->ce_mode;
532 if (mode == oldmode && !hashcmp(sha1, old->sha1) &&
533 !revs->diffopt.find_copies_harder)
534 return 0;
536 mode = ntohl(mode);
537 oldmode = ntohl(oldmode);
539 diff_change(&revs->diffopt, oldmode, mode,
540 old->sha1, sha1, old->name, NULL);
541 return 0;
544 static int diff_cache(struct rev_info *revs,
545 struct cache_entry **ac, int entries,
546 const char **pathspec,
547 int cached, int match_missing)
549 while (entries) {
550 struct cache_entry *ce = *ac;
551 int same = (entries > 1) && ce_same_name(ce, ac[1]);
553 if (!ce_path_match(ce, pathspec))
554 goto skip_entry;
556 switch (ce_stage(ce)) {
557 case 0:
558 /* No stage 1 entry? That means it's a new file */
559 if (!same) {
560 show_new_file(revs, ce, cached, match_missing);
561 break;
563 /* Show difference between old and new */
564 show_modified(revs, ac[1], ce, 1,
565 cached, match_missing);
566 break;
567 case 1:
568 /* No stage 3 (merge) entry?
569 * That means it's been deleted.
571 if (!same) {
572 diff_index_show_file(revs, "-", ce,
573 ce->sha1, ce->ce_mode);
574 break;
576 /* We come here with ce pointing at stage 1
577 * (original tree) and ac[1] pointing at stage
578 * 3 (unmerged). show-modified with
579 * report-missing set to false does not say the
580 * file is deleted but reports true if work
581 * tree does not have it, in which case we
582 * fall through to report the unmerged state.
583 * Otherwise, we show the differences between
584 * the original tree and the work tree.
586 if (!cached &&
587 !show_modified(revs, ce, ac[1], 0,
588 cached, match_missing))
589 break;
590 diff_unmerge(&revs->diffopt, ce->name,
591 ntohl(ce->ce_mode), ce->sha1);
592 break;
593 case 3:
594 diff_unmerge(&revs->diffopt, ce->name,
595 0, null_sha1);
596 break;
598 default:
599 die("impossible cache entry stage");
602 skip_entry:
604 * Ignore all the different stages for this file,
605 * we've handled the relevant cases now.
607 do {
608 ac++;
609 entries--;
610 } while (entries && ce_same_name(ce, ac[0]));
612 return 0;
616 * This turns all merge entries into "stage 3". That guarantees that
617 * when we read in the new tree (into "stage 1"), we won't lose sight
618 * of the fact that we had unmerged entries.
620 static void mark_merge_entries(void)
622 int i;
623 for (i = 0; i < active_nr; i++) {
624 struct cache_entry *ce = active_cache[i];
625 if (!ce_stage(ce))
626 continue;
627 ce->ce_flags |= htons(CE_STAGEMASK);
631 int run_diff_index(struct rev_info *revs, int cached)
633 int ret;
634 struct object *ent;
635 struct tree *tree;
636 const char *tree_name;
637 int match_missing = 0;
640 * Backward compatibility wart - "diff-index -m" does
641 * not mean "do not ignore merges", but totally different.
643 if (!revs->ignore_merges)
644 match_missing = 1;
646 mark_merge_entries();
648 ent = revs->pending.objects[0].item;
649 tree_name = revs->pending.objects[0].name;
650 tree = parse_tree_indirect(ent->sha1);
651 if (!tree)
652 return error("bad tree object %s", tree_name);
653 if (read_tree(tree, 1, revs->prune_data))
654 return error("unable to read tree object %s", tree_name);
655 ret = diff_cache(revs, active_cache, active_nr, revs->prune_data,
656 cached, match_missing);
657 diffcore_std(&revs->diffopt);
658 diff_flush(&revs->diffopt);
659 return ret;
662 int do_diff_cache(const unsigned char *tree_sha1, struct diff_options *opt)
664 struct tree *tree;
665 struct rev_info revs;
666 int i;
667 struct cache_entry **dst;
668 struct cache_entry *last = NULL;
671 * This is used by git-blame to run diff-cache internally;
672 * it potentially needs to repeatedly run this, so we will
673 * start by removing the higher order entries the last round
674 * left behind.
676 dst = active_cache;
677 for (i = 0; i < active_nr; i++) {
678 struct cache_entry *ce = active_cache[i];
679 if (ce_stage(ce)) {
680 if (last && !strcmp(ce->name, last->name))
681 continue;
682 cache_tree_invalidate_path(active_cache_tree,
683 ce->name);
684 last = ce;
685 ce->ce_mode = 0;
686 ce->ce_flags &= ~htons(CE_STAGEMASK);
688 *dst++ = ce;
690 active_nr = dst - active_cache;
692 init_revisions(&revs, NULL);
693 revs.prune_data = opt->paths;
694 tree = parse_tree_indirect(tree_sha1);
695 if (!tree)
696 die("bad tree object %s", sha1_to_hex(tree_sha1));
697 if (read_tree(tree, 1, opt->paths))
698 return error("unable to read tree %s", sha1_to_hex(tree_sha1));
699 return diff_cache(&revs, active_cache, active_nr, revs.prune_data,
700 1, 0);