Get rid of the dependency to GNU diff in the tests
[git/git-bigfiles.git] / diff-lib.c
blob5e6bca5906f0fefbe54e0f91c2869c623019e7a7
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;
145 static int is_in_index(const char *path)
147 int len = strlen(path);
148 int pos = cache_name_pos(path, len);
149 char c;
151 if (pos < 0)
152 return 0;
153 if (strncmp(active_cache[pos]->name, path, len))
154 return 0;
155 c = active_cache[pos]->name[len];
156 return c == '\0' || c == '/';
159 static int handle_diff_files_args(struct rev_info *revs,
160 int argc, const char **argv, int *silent)
162 *silent = 0;
164 /* revs->max_count == -2 means --no-index */
165 while (1 < argc && argv[1][0] == '-') {
166 if (!strcmp(argv[1], "--base"))
167 revs->max_count = 1;
168 else if (!strcmp(argv[1], "--ours"))
169 revs->max_count = 2;
170 else if (!strcmp(argv[1], "--theirs"))
171 revs->max_count = 3;
172 else if (!strcmp(argv[1], "-n") ||
173 !strcmp(argv[1], "--no-index"))
174 revs->max_count = -2;
175 else if (!strcmp(argv[1], "-q"))
176 *silent = 1;
177 else
178 return error("invalid option: %s", argv[1]);
179 argv++; argc--;
182 if (revs->max_count == -1 && revs->diffopt.nr_paths == 2) {
184 * If two files are specified, and at least one is untracked,
185 * default to no-index.
187 read_cache();
188 if (!is_in_index(revs->diffopt.paths[0]) ||
189 !is_in_index(revs->diffopt.paths[1]))
190 revs->max_count = -2;
194 * Make sure there are NO revision (i.e. pending object) parameter,
195 * rev.max_count is reasonable (0 <= n <= 3),
196 * there is no other revision filtering parameters.
198 if (revs->pending.nr || revs->max_count > 3 ||
199 revs->min_age != -1 || revs->max_age != -1)
200 return error("no revision allowed with diff-files");
202 if (revs->max_count == -1 &&
203 (revs->diffopt.output_format & DIFF_FORMAT_PATCH))
204 revs->combine_merges = revs->dense_combined_merges = 1;
206 return 0;
209 static int is_outside_repo(const char *path, int nongit, const char *prefix)
211 int i;
212 if (nongit || !strcmp(path, "-") || path[0] == '/')
213 return 1;
214 if (prefixcmp(path, "../"))
215 return 0;
216 if (!prefix)
217 return 1;
218 for (i = strlen(prefix); !prefixcmp(path, "../"); ) {
219 while (i > 0 && prefix[i - 1] != '/')
220 i--;
221 if (--i < 0)
222 return 1;
223 path += 3;
225 return 0;
228 int setup_diff_no_index(struct rev_info *revs,
229 int argc, const char ** argv, int nongit, const char *prefix)
231 int i;
232 for (i = 1; i < argc; i++)
233 if (argv[i][0] != '-' || argv[i][1] == '\0')
234 break;
235 else if (!strcmp(argv[i], "--")) {
236 i++;
237 break;
238 } else if (i < argc - 3 && !strcmp(argv[i], "--no-index")) {
239 i = argc - 3;
240 break;
242 if (argc != i + 2 || (!is_outside_repo(argv[i + 1], nongit, prefix) &&
243 !is_outside_repo(argv[i], nongit, prefix)))
244 return -1;
246 diff_setup(&revs->diffopt);
247 for (i = 1; i < argc - 2; )
248 if (!strcmp(argv[i], "--no-index"))
249 i++;
250 else {
251 int j = diff_opt_parse(&revs->diffopt,
252 argv + i, argc - i);
253 if (!j)
254 die("invalid diff option/value: %s", argv[i]);
255 i += j;
258 if (prefix) {
259 int len = strlen(prefix);
261 revs->diffopt.paths = xcalloc(2, sizeof(char*));
262 for (i = 0; i < 2; i++) {
263 const char *p = argv[argc - 2 + i];
265 * stdin should be spelled as '-'; if you have
266 * path that is '-', spell it as ./-.
268 p = (strcmp(p, "-")
269 ? xstrdup(prefix_filename(prefix, len, p))
270 : p);
271 revs->diffopt.paths[i] = p;
274 else
275 revs->diffopt.paths = argv + argc - 2;
276 revs->diffopt.nr_paths = 2;
277 revs->max_count = -2;
278 return 0;
281 int run_diff_files_cmd(struct rev_info *revs, int argc, const char **argv)
283 int silent_on_removed;
285 if (handle_diff_files_args(revs, argc, argv, &silent_on_removed))
286 return -1;
288 if (revs->max_count == -2) {
289 if (revs->diffopt.nr_paths != 2)
290 return error("need two files/directories with --no-index");
291 if (queue_diff(&revs->diffopt, revs->diffopt.paths[0],
292 revs->diffopt.paths[1]))
293 return -1;
294 diffcore_std(&revs->diffopt);
295 diff_flush(&revs->diffopt);
297 * The return code for --no-index imitates diff(1):
298 * 0 = no changes, 1 = changes, else error
300 return revs->diffopt.found_changes;
303 if (read_cache() < 0) {
304 perror("read_cache");
305 return -1;
307 return run_diff_files(revs, silent_on_removed);
310 int run_diff_files(struct rev_info *revs, int silent_on_removed)
312 int entries, i;
313 int diff_unmerged_stage = revs->max_count;
315 if (diff_unmerged_stage < 0)
316 diff_unmerged_stage = 2;
317 entries = active_nr;
318 for (i = 0; i < entries; i++) {
319 struct stat st;
320 unsigned int oldmode, newmode;
321 struct cache_entry *ce = active_cache[i];
322 int changed;
324 if (!ce_path_match(ce, revs->prune_data))
325 continue;
327 if (ce_stage(ce)) {
328 struct combine_diff_path *dpath;
329 int num_compare_stages = 0;
330 size_t path_len;
332 path_len = ce_namelen(ce);
334 dpath = xmalloc(combine_diff_path_size(5, path_len));
335 dpath->path = (char *) &(dpath->parent[5]);
337 dpath->next = NULL;
338 dpath->len = path_len;
339 memcpy(dpath->path, ce->name, path_len);
340 dpath->path[path_len] = '\0';
341 hashclr(dpath->sha1);
342 memset(&(dpath->parent[0]), 0,
343 sizeof(struct combine_diff_parent)*5);
345 if (lstat(ce->name, &st) < 0) {
346 if (errno != ENOENT && errno != ENOTDIR) {
347 perror(ce->name);
348 continue;
350 if (silent_on_removed)
351 continue;
353 else
354 dpath->mode = canon_mode(st.st_mode);
356 while (i < entries) {
357 struct cache_entry *nce = active_cache[i];
358 int stage;
360 if (strcmp(ce->name, nce->name))
361 break;
363 /* Stage #2 (ours) is the first parent,
364 * stage #3 (theirs) is the second.
366 stage = ce_stage(nce);
367 if (2 <= stage) {
368 int mode = ntohl(nce->ce_mode);
369 num_compare_stages++;
370 hashcpy(dpath->parent[stage-2].sha1, nce->sha1);
371 dpath->parent[stage-2].mode =
372 canon_mode(mode);
373 dpath->parent[stage-2].status =
374 DIFF_STATUS_MODIFIED;
377 /* diff against the proper unmerged stage */
378 if (stage == diff_unmerged_stage)
379 ce = nce;
380 i++;
383 * Compensate for loop update
385 i--;
387 if (revs->combine_merges && num_compare_stages == 2) {
388 show_combined_diff(dpath, 2,
389 revs->dense_combined_merges,
390 revs);
391 free(dpath);
392 continue;
394 free(dpath);
395 dpath = NULL;
398 * Show the diff for the 'ce' if we found the one
399 * from the desired stage.
401 diff_unmerge(&revs->diffopt, ce->name, 0, null_sha1);
402 if (ce_stage(ce) != diff_unmerged_stage)
403 continue;
406 if (lstat(ce->name, &st) < 0) {
407 if (errno != ENOENT && errno != ENOTDIR) {
408 perror(ce->name);
409 continue;
411 if (silent_on_removed)
412 continue;
413 diff_addremove(&revs->diffopt, '-', ntohl(ce->ce_mode),
414 ce->sha1, ce->name, NULL);
415 continue;
417 changed = ce_match_stat(ce, &st, 0);
418 if (!changed && !revs->diffopt.find_copies_harder)
419 continue;
420 oldmode = ntohl(ce->ce_mode);
422 newmode = canon_mode(st.st_mode);
423 if (!trust_executable_bit &&
424 S_ISREG(newmode) && S_ISREG(oldmode) &&
425 ((newmode ^ oldmode) == 0111))
426 newmode = oldmode;
427 diff_change(&revs->diffopt, oldmode, newmode,
428 ce->sha1, (changed ? null_sha1 : ce->sha1),
429 ce->name, NULL);
432 diffcore_std(&revs->diffopt);
433 diff_flush(&revs->diffopt);
434 return 0;
438 * diff-index
441 /* A file entry went away or appeared */
442 static void diff_index_show_file(struct rev_info *revs,
443 const char *prefix,
444 struct cache_entry *ce,
445 unsigned char *sha1, unsigned int mode)
447 diff_addremove(&revs->diffopt, prefix[0], ntohl(mode),
448 sha1, ce->name, NULL);
451 static int get_stat_data(struct cache_entry *ce,
452 unsigned char **sha1p,
453 unsigned int *modep,
454 int cached, int match_missing)
456 unsigned char *sha1 = ce->sha1;
457 unsigned int mode = ce->ce_mode;
459 if (!cached) {
460 static unsigned char no_sha1[20];
461 int changed;
462 struct stat st;
463 if (lstat(ce->name, &st) < 0) {
464 if (errno == ENOENT && match_missing) {
465 *sha1p = sha1;
466 *modep = mode;
467 return 0;
469 return -1;
471 changed = ce_match_stat(ce, &st, 0);
472 if (changed) {
473 mode = ce_mode_from_stat(ce, st.st_mode);
474 sha1 = no_sha1;
478 *sha1p = sha1;
479 *modep = mode;
480 return 0;
483 static void show_new_file(struct rev_info *revs,
484 struct cache_entry *new,
485 int cached, int match_missing)
487 unsigned char *sha1;
488 unsigned int mode;
490 /* New file in the index: it might actually be different in
491 * the working copy.
493 if (get_stat_data(new, &sha1, &mode, cached, match_missing) < 0)
494 return;
496 diff_index_show_file(revs, "+", new, sha1, mode);
499 static int show_modified(struct rev_info *revs,
500 struct cache_entry *old,
501 struct cache_entry *new,
502 int report_missing,
503 int cached, int match_missing)
505 unsigned int mode, oldmode;
506 unsigned char *sha1;
508 if (get_stat_data(new, &sha1, &mode, cached, match_missing) < 0) {
509 if (report_missing)
510 diff_index_show_file(revs, "-", old,
511 old->sha1, old->ce_mode);
512 return -1;
515 if (revs->combine_merges && !cached &&
516 (hashcmp(sha1, old->sha1) || hashcmp(old->sha1, new->sha1))) {
517 struct combine_diff_path *p;
518 int pathlen = ce_namelen(new);
520 p = xmalloc(combine_diff_path_size(2, pathlen));
521 p->path = (char *) &p->parent[2];
522 p->next = NULL;
523 p->len = pathlen;
524 memcpy(p->path, new->name, pathlen);
525 p->path[pathlen] = 0;
526 p->mode = ntohl(mode);
527 hashclr(p->sha1);
528 memset(p->parent, 0, 2 * sizeof(struct combine_diff_parent));
529 p->parent[0].status = DIFF_STATUS_MODIFIED;
530 p->parent[0].mode = ntohl(new->ce_mode);
531 hashcpy(p->parent[0].sha1, new->sha1);
532 p->parent[1].status = DIFF_STATUS_MODIFIED;
533 p->parent[1].mode = ntohl(old->ce_mode);
534 hashcpy(p->parent[1].sha1, old->sha1);
535 show_combined_diff(p, 2, revs->dense_combined_merges, revs);
536 free(p);
537 return 0;
540 oldmode = old->ce_mode;
541 if (mode == oldmode && !hashcmp(sha1, old->sha1) &&
542 !revs->diffopt.find_copies_harder)
543 return 0;
545 mode = ntohl(mode);
546 oldmode = ntohl(oldmode);
548 diff_change(&revs->diffopt, oldmode, mode,
549 old->sha1, sha1, old->name, NULL);
550 return 0;
553 static int diff_cache(struct rev_info *revs,
554 struct cache_entry **ac, int entries,
555 const char **pathspec,
556 int cached, int match_missing)
558 while (entries) {
559 struct cache_entry *ce = *ac;
560 int same = (entries > 1) && ce_same_name(ce, ac[1]);
562 if (!ce_path_match(ce, pathspec))
563 goto skip_entry;
565 switch (ce_stage(ce)) {
566 case 0:
567 /* No stage 1 entry? That means it's a new file */
568 if (!same) {
569 show_new_file(revs, ce, cached, match_missing);
570 break;
572 /* Show difference between old and new */
573 show_modified(revs, ac[1], ce, 1,
574 cached, match_missing);
575 break;
576 case 1:
577 /* No stage 3 (merge) entry?
578 * That means it's been deleted.
580 if (!same) {
581 diff_index_show_file(revs, "-", ce,
582 ce->sha1, ce->ce_mode);
583 break;
585 /* We come here with ce pointing at stage 1
586 * (original tree) and ac[1] pointing at stage
587 * 3 (unmerged). show-modified with
588 * report-missing set to false does not say the
589 * file is deleted but reports true if work
590 * tree does not have it, in which case we
591 * fall through to report the unmerged state.
592 * Otherwise, we show the differences between
593 * the original tree and the work tree.
595 if (!cached &&
596 !show_modified(revs, ce, ac[1], 0,
597 cached, match_missing))
598 break;
599 diff_unmerge(&revs->diffopt, ce->name,
600 ntohl(ce->ce_mode), ce->sha1);
601 break;
602 case 3:
603 diff_unmerge(&revs->diffopt, ce->name,
604 0, null_sha1);
605 break;
607 default:
608 die("impossible cache entry stage");
611 skip_entry:
613 * Ignore all the different stages for this file,
614 * we've handled the relevant cases now.
616 do {
617 ac++;
618 entries--;
619 } while (entries && ce_same_name(ce, ac[0]));
621 return 0;
625 * This turns all merge entries into "stage 3". That guarantees that
626 * when we read in the new tree (into "stage 1"), we won't lose sight
627 * of the fact that we had unmerged entries.
629 static void mark_merge_entries(void)
631 int i;
632 for (i = 0; i < active_nr; i++) {
633 struct cache_entry *ce = active_cache[i];
634 if (!ce_stage(ce))
635 continue;
636 ce->ce_flags |= htons(CE_STAGEMASK);
640 int run_diff_index(struct rev_info *revs, int cached)
642 int ret;
643 struct object *ent;
644 struct tree *tree;
645 const char *tree_name;
646 int match_missing = 0;
649 * Backward compatibility wart - "diff-index -m" does
650 * not mean "do not ignore merges", but totally different.
652 if (!revs->ignore_merges)
653 match_missing = 1;
655 mark_merge_entries();
657 ent = revs->pending.objects[0].item;
658 tree_name = revs->pending.objects[0].name;
659 tree = parse_tree_indirect(ent->sha1);
660 if (!tree)
661 return error("bad tree object %s", tree_name);
662 if (read_tree(tree, 1, revs->prune_data))
663 return error("unable to read tree object %s", tree_name);
664 ret = diff_cache(revs, active_cache, active_nr, revs->prune_data,
665 cached, match_missing);
666 diffcore_std(&revs->diffopt);
667 diff_flush(&revs->diffopt);
668 return ret;
671 int do_diff_cache(const unsigned char *tree_sha1, struct diff_options *opt)
673 struct tree *tree;
674 struct rev_info revs;
675 int i;
676 struct cache_entry **dst;
677 struct cache_entry *last = NULL;
680 * This is used by git-blame to run diff-cache internally;
681 * it potentially needs to repeatedly run this, so we will
682 * start by removing the higher order entries the last round
683 * left behind.
685 dst = active_cache;
686 for (i = 0; i < active_nr; i++) {
687 struct cache_entry *ce = active_cache[i];
688 if (ce_stage(ce)) {
689 if (last && !strcmp(ce->name, last->name))
690 continue;
691 cache_tree_invalidate_path(active_cache_tree,
692 ce->name);
693 last = ce;
694 ce->ce_mode = 0;
695 ce->ce_flags &= ~htons(CE_STAGEMASK);
697 *dst++ = ce;
699 active_nr = dst - active_cache;
701 init_revisions(&revs, NULL);
702 revs.prune_data = opt->paths;
703 tree = parse_tree_indirect(tree_sha1);
704 if (!tree)
705 die("bad tree object %s", sha1_to_hex(tree_sha1));
706 if (read_tree(tree, 1, opt->paths))
707 return error("unable to read tree %s", sha1_to_hex(tree_sha1));
708 return diff_cache(&revs, active_cache, active_nr, revs.prune_data,
709 1, 0);