Add back md5sum, which is required by git tests
[msysgit/historical-msysgit.git] / git / diff-lib.c
blob92c0e39ad6f21e0511ee84052e2b762d6e447f61
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;
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 = ntohl(ce_mode_from_stat(ce, 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 = ntohl(ce_mode_from_stat(nce, mode));
394 dpath->parent[stage-2].status =
395 DIFF_STATUS_MODIFIED;
398 /* diff against the proper unmerged stage */
399 if (stage == diff_unmerged_stage)
400 ce = nce;
401 i++;
404 * Compensate for loop update
406 i--;
408 if (revs->combine_merges && num_compare_stages == 2) {
409 show_combined_diff(dpath, 2,
410 revs->dense_combined_merges,
411 revs);
412 free(dpath);
413 continue;
415 free(dpath);
416 dpath = NULL;
419 * Show the diff for the 'ce' if we found the one
420 * from the desired stage.
422 diff_unmerge(&revs->diffopt, ce->name, 0, null_sha1);
423 if (ce_stage(ce) != diff_unmerged_stage)
424 continue;
427 if (lstat(ce->name, &st) < 0) {
428 if (errno != ENOENT && errno != ENOTDIR) {
429 perror(ce->name);
430 continue;
432 if (silent_on_removed)
433 continue;
434 diff_addremove(&revs->diffopt, '-', ntohl(ce->ce_mode),
435 ce->sha1, ce->name, NULL);
436 continue;
438 changed = ce_match_stat(ce, &st, 0);
439 if (!changed && !revs->diffopt.find_copies_harder)
440 continue;
441 oldmode = ntohl(ce->ce_mode);
442 newmode = ntohl(ce_mode_from_stat(ce, st.st_mode));
443 diff_change(&revs->diffopt, oldmode, newmode,
444 ce->sha1, (changed ? null_sha1 : ce->sha1),
445 ce->name, NULL);
448 diffcore_std(&revs->diffopt);
449 diff_flush(&revs->diffopt);
450 return 0;
454 * diff-index
457 /* A file entry went away or appeared */
458 static void diff_index_show_file(struct rev_info *revs,
459 const char *prefix,
460 struct cache_entry *ce,
461 unsigned char *sha1, unsigned int mode)
463 diff_addremove(&revs->diffopt, prefix[0], ntohl(mode),
464 sha1, ce->name, NULL);
467 static int get_stat_data(struct cache_entry *ce,
468 unsigned char **sha1p,
469 unsigned int *modep,
470 int cached, int match_missing)
472 unsigned char *sha1 = ce->sha1;
473 unsigned int mode = ce->ce_mode;
475 if (!cached) {
476 static unsigned char no_sha1[20];
477 int changed;
478 struct stat st;
479 if (lstat(ce->name, &st) < 0) {
480 if (errno == ENOENT && match_missing) {
481 *sha1p = sha1;
482 *modep = mode;
483 return 0;
485 return -1;
487 changed = ce_match_stat(ce, &st, 0);
488 if (changed) {
489 mode = ce_mode_from_stat(ce, st.st_mode);
490 sha1 = no_sha1;
494 *sha1p = sha1;
495 *modep = mode;
496 return 0;
499 static void show_new_file(struct rev_info *revs,
500 struct cache_entry *new,
501 int cached, int match_missing)
503 unsigned char *sha1;
504 unsigned int mode;
506 /* New file in the index: it might actually be different in
507 * the working copy.
509 if (get_stat_data(new, &sha1, &mode, cached, match_missing) < 0)
510 return;
512 diff_index_show_file(revs, "+", new, sha1, mode);
515 static int show_modified(struct rev_info *revs,
516 struct cache_entry *old,
517 struct cache_entry *new,
518 int report_missing,
519 int cached, int match_missing)
521 unsigned int mode, oldmode;
522 unsigned char *sha1;
524 if (get_stat_data(new, &sha1, &mode, cached, match_missing) < 0) {
525 if (report_missing)
526 diff_index_show_file(revs, "-", old,
527 old->sha1, old->ce_mode);
528 return -1;
531 if (revs->combine_merges && !cached &&
532 (hashcmp(sha1, old->sha1) || hashcmp(old->sha1, new->sha1))) {
533 struct combine_diff_path *p;
534 int pathlen = ce_namelen(new);
536 p = xmalloc(combine_diff_path_size(2, pathlen));
537 p->path = (char *) &p->parent[2];
538 p->next = NULL;
539 p->len = pathlen;
540 memcpy(p->path, new->name, pathlen);
541 p->path[pathlen] = 0;
542 p->mode = ntohl(mode);
543 hashclr(p->sha1);
544 memset(p->parent, 0, 2 * sizeof(struct combine_diff_parent));
545 p->parent[0].status = DIFF_STATUS_MODIFIED;
546 p->parent[0].mode = ntohl(new->ce_mode);
547 hashcpy(p->parent[0].sha1, new->sha1);
548 p->parent[1].status = DIFF_STATUS_MODIFIED;
549 p->parent[1].mode = ntohl(old->ce_mode);
550 hashcpy(p->parent[1].sha1, old->sha1);
551 show_combined_diff(p, 2, revs->dense_combined_merges, revs);
552 free(p);
553 return 0;
556 oldmode = old->ce_mode;
557 if (mode == oldmode && !hashcmp(sha1, old->sha1) &&
558 !revs->diffopt.find_copies_harder)
559 return 0;
561 mode = ntohl(mode);
562 oldmode = ntohl(oldmode);
564 diff_change(&revs->diffopt, oldmode, mode,
565 old->sha1, sha1, old->name, NULL);
566 return 0;
569 static int diff_cache(struct rev_info *revs,
570 struct cache_entry **ac, int entries,
571 const char **pathspec,
572 int cached, int match_missing)
574 while (entries) {
575 struct cache_entry *ce = *ac;
576 int same = (entries > 1) && ce_same_name(ce, ac[1]);
578 if (revs->diffopt.quiet && revs->diffopt.has_changes)
579 break;
581 if (!ce_path_match(ce, pathspec))
582 goto skip_entry;
584 switch (ce_stage(ce)) {
585 case 0:
586 /* No stage 1 entry? That means it's a new file */
587 if (!same) {
588 show_new_file(revs, ce, cached, match_missing);
589 break;
591 /* Show difference between old and new */
592 show_modified(revs, ac[1], ce, 1,
593 cached, match_missing);
594 break;
595 case 1:
596 /* No stage 3 (merge) entry?
597 * That means it's been deleted.
599 if (!same) {
600 diff_index_show_file(revs, "-", ce,
601 ce->sha1, ce->ce_mode);
602 break;
604 /* We come here with ce pointing at stage 1
605 * (original tree) and ac[1] pointing at stage
606 * 3 (unmerged). show-modified with
607 * report-missing set to false does not say the
608 * file is deleted but reports true if work
609 * tree does not have it, in which case we
610 * fall through to report the unmerged state.
611 * Otherwise, we show the differences between
612 * the original tree and the work tree.
614 if (!cached &&
615 !show_modified(revs, ce, ac[1], 0,
616 cached, match_missing))
617 break;
618 diff_unmerge(&revs->diffopt, ce->name,
619 ntohl(ce->ce_mode), ce->sha1);
620 break;
621 case 3:
622 diff_unmerge(&revs->diffopt, ce->name,
623 0, null_sha1);
624 break;
626 default:
627 die("impossible cache entry stage");
630 skip_entry:
632 * Ignore all the different stages for this file,
633 * we've handled the relevant cases now.
635 do {
636 ac++;
637 entries--;
638 } while (entries && ce_same_name(ce, ac[0]));
640 return 0;
644 * This turns all merge entries into "stage 3". That guarantees that
645 * when we read in the new tree (into "stage 1"), we won't lose sight
646 * of the fact that we had unmerged entries.
648 static void mark_merge_entries(void)
650 int i;
651 for (i = 0; i < active_nr; i++) {
652 struct cache_entry *ce = active_cache[i];
653 if (!ce_stage(ce))
654 continue;
655 ce->ce_flags |= htons(CE_STAGEMASK);
659 int run_diff_index(struct rev_info *revs, int cached)
661 int ret;
662 struct object *ent;
663 struct tree *tree;
664 const char *tree_name;
665 int match_missing = 0;
668 * Backward compatibility wart - "diff-index -m" does
669 * not mean "do not ignore merges", but totally different.
671 if (!revs->ignore_merges)
672 match_missing = 1;
674 mark_merge_entries();
676 ent = revs->pending.objects[0].item;
677 tree_name = revs->pending.objects[0].name;
678 tree = parse_tree_indirect(ent->sha1);
679 if (!tree)
680 return error("bad tree object %s", tree_name);
681 if (read_tree(tree, 1, revs->prune_data))
682 return error("unable to read tree object %s", tree_name);
683 ret = diff_cache(revs, active_cache, active_nr, revs->prune_data,
684 cached, match_missing);
685 diffcore_std(&revs->diffopt);
686 diff_flush(&revs->diffopt);
687 return ret;
690 int do_diff_cache(const unsigned char *tree_sha1, struct diff_options *opt)
692 struct tree *tree;
693 struct rev_info revs;
694 int i;
695 struct cache_entry **dst;
696 struct cache_entry *last = NULL;
699 * This is used by git-blame to run diff-cache internally;
700 * it potentially needs to repeatedly run this, so we will
701 * start by removing the higher order entries the last round
702 * left behind.
704 dst = active_cache;
705 for (i = 0; i < active_nr; i++) {
706 struct cache_entry *ce = active_cache[i];
707 if (ce_stage(ce)) {
708 if (last && !strcmp(ce->name, last->name))
709 continue;
710 cache_tree_invalidate_path(active_cache_tree,
711 ce->name);
712 last = ce;
713 ce->ce_mode = 0;
714 ce->ce_flags &= ~htons(CE_STAGEMASK);
716 *dst++ = ce;
718 active_nr = dst - active_cache;
720 init_revisions(&revs, NULL);
721 revs.prune_data = opt->paths;
722 tree = parse_tree_indirect(tree_sha1);
723 if (!tree)
724 die("bad tree object %s", sha1_to_hex(tree_sha1));
725 if (read_tree(tree, 1, opt->paths))
726 return error("unable to read tree %s", sha1_to_hex(tree_sha1));
727 return diff_cache(&revs, active_cache, active_nr, revs.prune_data,
728 1, 0);