Use proper object allocators for unknown object nodes too
[git/fastimport.git] / diff-lib.c
blob7531e20c784c44c0b5d3ecb2057638874a09ce6c
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 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;
193 else if (!strcmp(argv[1], "-q"))
194 *silent = 1;
195 else
196 return error("invalid option: %s", argv[1]);
197 argv++; argc--;
200 if (revs->max_count == -1 && revs->diffopt.nr_paths == 2) {
202 * If two files are specified, and at least one is untracked,
203 * default to no-index.
205 read_cache();
206 if (!is_in_index(revs->diffopt.paths[0]) ||
207 !is_in_index(revs->diffopt.paths[1]))
208 revs->max_count = -2;
212 * Make sure there are NO revision (i.e. pending object) parameter,
213 * rev.max_count is reasonable (0 <= n <= 3),
214 * there is no other revision filtering parameters.
216 if (revs->pending.nr || revs->max_count > 3 ||
217 revs->min_age != -1 || revs->max_age != -1)
218 return error("no revision allowed with diff-files");
220 if (revs->max_count == -1 &&
221 (revs->diffopt.output_format & DIFF_FORMAT_PATCH))
222 revs->combine_merges = revs->dense_combined_merges = 1;
224 return 0;
227 static int is_outside_repo(const char *path, int nongit, const char *prefix)
229 int i;
230 if (nongit || !strcmp(path, "-") || path[0] == '/')
231 return 1;
232 if (prefixcmp(path, "../"))
233 return 0;
234 if (!prefix)
235 return 1;
236 for (i = strlen(prefix); !prefixcmp(path, "../"); ) {
237 while (i > 0 && prefix[i - 1] != '/')
238 i--;
239 if (--i < 0)
240 return 1;
241 path += 3;
243 return 0;
246 int setup_diff_no_index(struct rev_info *revs,
247 int argc, const char ** argv, int nongit, const char *prefix)
249 int i;
250 for (i = 1; i < argc; i++)
251 if (argv[i][0] != '-' || argv[i][1] == '\0')
252 break;
253 else if (!strcmp(argv[i], "--")) {
254 i++;
255 break;
256 } else if (i < argc - 3 && !strcmp(argv[i], "--no-index")) {
257 i = argc - 3;
258 revs->diffopt.exit_with_status = 1;
259 break;
261 if (argc != i + 2 || (!is_outside_repo(argv[i + 1], nongit, prefix) &&
262 !is_outside_repo(argv[i], nongit, prefix)))
263 return -1;
265 diff_setup(&revs->diffopt);
266 for (i = 1; i < argc - 2; )
267 if (!strcmp(argv[i], "--no-index"))
268 i++;
269 else {
270 int j = diff_opt_parse(&revs->diffopt,
271 argv + i, argc - i);
272 if (!j)
273 die("invalid diff option/value: %s", argv[i]);
274 i += j;
277 if (prefix) {
278 int len = strlen(prefix);
280 revs->diffopt.paths = xcalloc(2, sizeof(char*));
281 for (i = 0; i < 2; i++) {
282 const char *p = argv[argc - 2 + i];
284 * stdin should be spelled as '-'; if you have
285 * path that is '-', spell it as ./-.
287 p = (strcmp(p, "-")
288 ? xstrdup(prefix_filename(prefix, len, p))
289 : p);
290 revs->diffopt.paths[i] = p;
293 else
294 revs->diffopt.paths = argv + argc - 2;
295 revs->diffopt.nr_paths = 2;
296 revs->max_count = -2;
297 return 0;
300 int run_diff_files_cmd(struct rev_info *revs, int argc, const char **argv)
302 int silent_on_removed;
304 if (handle_diff_files_args(revs, argc, argv, &silent_on_removed))
305 return -1;
307 if (revs->max_count == -2) {
308 if (revs->diffopt.nr_paths != 2)
309 return error("need two files/directories with --no-index");
310 if (queue_diff(&revs->diffopt, revs->diffopt.paths[0],
311 revs->diffopt.paths[1]))
312 return -1;
313 diffcore_std(&revs->diffopt);
314 diff_flush(&revs->diffopt);
316 * The return code for --no-index imitates diff(1):
317 * 0 = no changes, 1 = changes, else error
319 return revs->diffopt.found_changes;
322 if (read_cache() < 0) {
323 perror("read_cache");
324 return -1;
326 return run_diff_files(revs, silent_on_removed);
329 int run_diff_files(struct rev_info *revs, int silent_on_removed)
331 int entries, i;
332 int diff_unmerged_stage = revs->max_count;
334 if (diff_unmerged_stage < 0)
335 diff_unmerged_stage = 2;
336 entries = active_nr;
337 for (i = 0; i < entries; i++) {
338 struct stat st;
339 unsigned int oldmode, newmode;
340 struct cache_entry *ce = active_cache[i];
341 int changed;
343 if (revs->diffopt.quiet && revs->diffopt.has_changes)
344 break;
346 if (!ce_path_match(ce, revs->prune_data))
347 continue;
349 if (ce_stage(ce)) {
350 struct combine_diff_path *dpath;
351 int num_compare_stages = 0;
352 size_t path_len;
354 path_len = ce_namelen(ce);
356 dpath = xmalloc(combine_diff_path_size(5, path_len));
357 dpath->path = (char *) &(dpath->parent[5]);
359 dpath->next = NULL;
360 dpath->len = path_len;
361 memcpy(dpath->path, ce->name, path_len);
362 dpath->path[path_len] = '\0';
363 hashclr(dpath->sha1);
364 memset(&(dpath->parent[0]), 0,
365 sizeof(struct combine_diff_parent)*5);
367 if (lstat(ce->name, &st) < 0) {
368 if (errno != ENOENT && errno != ENOTDIR) {
369 perror(ce->name);
370 continue;
372 if (silent_on_removed)
373 continue;
375 else
376 dpath->mode = canon_mode(st.st_mode);
378 while (i < entries) {
379 struct cache_entry *nce = active_cache[i];
380 int stage;
382 if (strcmp(ce->name, nce->name))
383 break;
385 /* Stage #2 (ours) is the first parent,
386 * stage #3 (theirs) is the second.
388 stage = ce_stage(nce);
389 if (2 <= stage) {
390 int mode = ntohl(nce->ce_mode);
391 num_compare_stages++;
392 hashcpy(dpath->parent[stage-2].sha1, nce->sha1);
393 dpath->parent[stage-2].mode =
394 canon_mode(mode);
395 dpath->parent[stage-2].status =
396 DIFF_STATUS_MODIFIED;
399 /* diff against the proper unmerged stage */
400 if (stage == diff_unmerged_stage)
401 ce = nce;
402 i++;
405 * Compensate for loop update
407 i--;
409 if (revs->combine_merges && num_compare_stages == 2) {
410 show_combined_diff(dpath, 2,
411 revs->dense_combined_merges,
412 revs);
413 free(dpath);
414 continue;
416 free(dpath);
417 dpath = NULL;
420 * Show the diff for the 'ce' if we found the one
421 * from the desired stage.
423 diff_unmerge(&revs->diffopt, ce->name, 0, null_sha1);
424 if (ce_stage(ce) != diff_unmerged_stage)
425 continue;
428 if (lstat(ce->name, &st) < 0) {
429 if (errno != ENOENT && errno != ENOTDIR) {
430 perror(ce->name);
431 continue;
433 if (silent_on_removed)
434 continue;
435 diff_addremove(&revs->diffopt, '-', ntohl(ce->ce_mode),
436 ce->sha1, ce->name, NULL);
437 continue;
439 changed = ce_match_stat(ce, &st, 0);
440 if (!changed && !revs->diffopt.find_copies_harder)
441 continue;
442 oldmode = ntohl(ce->ce_mode);
444 newmode = canon_mode(st.st_mode);
445 if (!trust_executable_bit &&
446 S_ISREG(newmode) && S_ISREG(oldmode) &&
447 ((newmode ^ oldmode) == 0111))
448 newmode = oldmode;
449 else if (!has_symlinks &&
450 S_ISREG(newmode) && S_ISLNK(oldmode))
451 newmode = oldmode;
452 diff_change(&revs->diffopt, oldmode, newmode,
453 ce->sha1, (changed ? null_sha1 : ce->sha1),
454 ce->name, NULL);
457 diffcore_std(&revs->diffopt);
458 diff_flush(&revs->diffopt);
459 return 0;
463 * diff-index
466 /* A file entry went away or appeared */
467 static void diff_index_show_file(struct rev_info *revs,
468 const char *prefix,
469 struct cache_entry *ce,
470 unsigned char *sha1, unsigned int mode)
472 diff_addremove(&revs->diffopt, prefix[0], ntohl(mode),
473 sha1, ce->name, NULL);
476 static int get_stat_data(struct cache_entry *ce,
477 unsigned char **sha1p,
478 unsigned int *modep,
479 int cached, int match_missing)
481 unsigned char *sha1 = ce->sha1;
482 unsigned int mode = ce->ce_mode;
484 if (!cached) {
485 static unsigned char no_sha1[20];
486 int changed;
487 struct stat st;
488 if (lstat(ce->name, &st) < 0) {
489 if (errno == ENOENT && match_missing) {
490 *sha1p = sha1;
491 *modep = mode;
492 return 0;
494 return -1;
496 changed = ce_match_stat(ce, &st, 0);
497 if (changed) {
498 mode = ce_mode_from_stat(ce, st.st_mode);
499 sha1 = no_sha1;
503 *sha1p = sha1;
504 *modep = mode;
505 return 0;
508 static void show_new_file(struct rev_info *revs,
509 struct cache_entry *new,
510 int cached, int match_missing)
512 unsigned char *sha1;
513 unsigned int mode;
515 /* New file in the index: it might actually be different in
516 * the working copy.
518 if (get_stat_data(new, &sha1, &mode, cached, match_missing) < 0)
519 return;
521 diff_index_show_file(revs, "+", new, sha1, mode);
524 static int show_modified(struct rev_info *revs,
525 struct cache_entry *old,
526 struct cache_entry *new,
527 int report_missing,
528 int cached, int match_missing)
530 unsigned int mode, oldmode;
531 unsigned char *sha1;
533 if (get_stat_data(new, &sha1, &mode, cached, match_missing) < 0) {
534 if (report_missing)
535 diff_index_show_file(revs, "-", old,
536 old->sha1, old->ce_mode);
537 return -1;
540 if (revs->combine_merges && !cached &&
541 (hashcmp(sha1, old->sha1) || hashcmp(old->sha1, new->sha1))) {
542 struct combine_diff_path *p;
543 int pathlen = ce_namelen(new);
545 p = xmalloc(combine_diff_path_size(2, pathlen));
546 p->path = (char *) &p->parent[2];
547 p->next = NULL;
548 p->len = pathlen;
549 memcpy(p->path, new->name, pathlen);
550 p->path[pathlen] = 0;
551 p->mode = ntohl(mode);
552 hashclr(p->sha1);
553 memset(p->parent, 0, 2 * sizeof(struct combine_diff_parent));
554 p->parent[0].status = DIFF_STATUS_MODIFIED;
555 p->parent[0].mode = ntohl(new->ce_mode);
556 hashcpy(p->parent[0].sha1, new->sha1);
557 p->parent[1].status = DIFF_STATUS_MODIFIED;
558 p->parent[1].mode = ntohl(old->ce_mode);
559 hashcpy(p->parent[1].sha1, old->sha1);
560 show_combined_diff(p, 2, revs->dense_combined_merges, revs);
561 free(p);
562 return 0;
565 oldmode = old->ce_mode;
566 if (mode == oldmode && !hashcmp(sha1, old->sha1) &&
567 !revs->diffopt.find_copies_harder)
568 return 0;
570 mode = ntohl(mode);
571 oldmode = ntohl(oldmode);
573 diff_change(&revs->diffopt, oldmode, mode,
574 old->sha1, sha1, old->name, NULL);
575 return 0;
578 static int diff_cache(struct rev_info *revs,
579 struct cache_entry **ac, int entries,
580 const char **pathspec,
581 int cached, int match_missing)
583 while (entries) {
584 struct cache_entry *ce = *ac;
585 int same = (entries > 1) && ce_same_name(ce, ac[1]);
587 if (revs->diffopt.quiet && revs->diffopt.has_changes)
588 break;
590 if (!ce_path_match(ce, pathspec))
591 goto skip_entry;
593 switch (ce_stage(ce)) {
594 case 0:
595 /* No stage 1 entry? That means it's a new file */
596 if (!same) {
597 show_new_file(revs, ce, cached, match_missing);
598 break;
600 /* Show difference between old and new */
601 show_modified(revs, ac[1], ce, 1,
602 cached, match_missing);
603 break;
604 case 1:
605 /* No stage 3 (merge) entry?
606 * That means it's been deleted.
608 if (!same) {
609 diff_index_show_file(revs, "-", ce,
610 ce->sha1, ce->ce_mode);
611 break;
613 /* We come here with ce pointing at stage 1
614 * (original tree) and ac[1] pointing at stage
615 * 3 (unmerged). show-modified with
616 * report-missing set to false does not say the
617 * file is deleted but reports true if work
618 * tree does not have it, in which case we
619 * fall through to report the unmerged state.
620 * Otherwise, we show the differences between
621 * the original tree and the work tree.
623 if (!cached &&
624 !show_modified(revs, ce, ac[1], 0,
625 cached, match_missing))
626 break;
627 diff_unmerge(&revs->diffopt, ce->name,
628 ntohl(ce->ce_mode), ce->sha1);
629 break;
630 case 3:
631 diff_unmerge(&revs->diffopt, ce->name,
632 0, null_sha1);
633 break;
635 default:
636 die("impossible cache entry stage");
639 skip_entry:
641 * Ignore all the different stages for this file,
642 * we've handled the relevant cases now.
644 do {
645 ac++;
646 entries--;
647 } while (entries && ce_same_name(ce, ac[0]));
649 return 0;
653 * This turns all merge entries into "stage 3". That guarantees that
654 * when we read in the new tree (into "stage 1"), we won't lose sight
655 * of the fact that we had unmerged entries.
657 static void mark_merge_entries(void)
659 int i;
660 for (i = 0; i < active_nr; i++) {
661 struct cache_entry *ce = active_cache[i];
662 if (!ce_stage(ce))
663 continue;
664 ce->ce_flags |= htons(CE_STAGEMASK);
668 int run_diff_index(struct rev_info *revs, int cached)
670 int ret;
671 struct object *ent;
672 struct tree *tree;
673 const char *tree_name;
674 int match_missing = 0;
677 * Backward compatibility wart - "diff-index -m" does
678 * not mean "do not ignore merges", but totally different.
680 if (!revs->ignore_merges)
681 match_missing = 1;
683 mark_merge_entries();
685 ent = revs->pending.objects[0].item;
686 tree_name = revs->pending.objects[0].name;
687 tree = parse_tree_indirect(ent->sha1);
688 if (!tree)
689 return error("bad tree object %s", tree_name);
690 if (read_tree(tree, 1, revs->prune_data))
691 return error("unable to read tree object %s", tree_name);
692 ret = diff_cache(revs, active_cache, active_nr, revs->prune_data,
693 cached, match_missing);
694 diffcore_std(&revs->diffopt);
695 diff_flush(&revs->diffopt);
696 return ret;
699 int do_diff_cache(const unsigned char *tree_sha1, struct diff_options *opt)
701 struct tree *tree;
702 struct rev_info revs;
703 int i;
704 struct cache_entry **dst;
705 struct cache_entry *last = NULL;
708 * This is used by git-blame to run diff-cache internally;
709 * it potentially needs to repeatedly run this, so we will
710 * start by removing the higher order entries the last round
711 * left behind.
713 dst = active_cache;
714 for (i = 0; i < active_nr; i++) {
715 struct cache_entry *ce = active_cache[i];
716 if (ce_stage(ce)) {
717 if (last && !strcmp(ce->name, last->name))
718 continue;
719 cache_tree_invalidate_path(active_cache_tree,
720 ce->name);
721 last = ce;
722 ce->ce_mode = 0;
723 ce->ce_flags &= ~htons(CE_STAGEMASK);
725 *dst++ = ce;
727 active_nr = dst - active_cache;
729 init_revisions(&revs, NULL);
730 revs.prune_data = opt->paths;
731 tree = parse_tree_indirect(tree_sha1);
732 if (!tree)
733 die("bad tree object %s", sha1_to_hex(tree_sha1));
734 if (read_tree(tree, 1, opt->paths))
735 return error("unable to read tree %s", sha1_to_hex(tree_sha1));
736 return diff_cache(&revs, active_cache, active_nr, revs.prune_data,
737 1, 0);