Implement limited context matching in git-apply.
[debian-git.git] / revision.c
blobfe265623817c7c6b5b4c1af399f4522a75fdaf30
1 #include "cache.h"
2 #include "tag.h"
3 #include "blob.h"
4 #include "tree.h"
5 #include "commit.h"
6 #include "diff.h"
7 #include "refs.h"
8 #include "revision.h"
10 static char *path_name(struct name_path *path, const char *name)
12 struct name_path *p;
13 char *n, *m;
14 int nlen = strlen(name);
15 int len = nlen + 1;
17 for (p = path; p; p = p->up) {
18 if (p->elem_len)
19 len += p->elem_len + 1;
21 n = xmalloc(len);
22 m = n + len - (nlen + 1);
23 strcpy(m, name);
24 for (p = path; p; p = p->up) {
25 if (p->elem_len) {
26 m -= p->elem_len + 1;
27 memcpy(m, p->elem, p->elem_len);
28 m[p->elem_len] = '/';
31 return n;
34 struct object_list **add_object(struct object *obj,
35 struct object_list **p,
36 struct name_path *path,
37 const char *name)
39 struct object_list *entry = xmalloc(sizeof(*entry));
40 entry->item = obj;
41 entry->next = *p;
42 entry->name = path_name(path, name);
43 *p = entry;
44 return &entry->next;
47 static void mark_blob_uninteresting(struct blob *blob)
49 if (blob->object.flags & UNINTERESTING)
50 return;
51 blob->object.flags |= UNINTERESTING;
54 void mark_tree_uninteresting(struct tree *tree)
56 struct object *obj = &tree->object;
57 struct tree_entry_list *entry;
59 if (obj->flags & UNINTERESTING)
60 return;
61 obj->flags |= UNINTERESTING;
62 if (!has_sha1_file(obj->sha1))
63 return;
64 if (parse_tree(tree) < 0)
65 die("bad tree %s", sha1_to_hex(obj->sha1));
66 entry = tree->entries;
67 tree->entries = NULL;
68 while (entry) {
69 struct tree_entry_list *next = entry->next;
70 if (entry->directory)
71 mark_tree_uninteresting(entry->item.tree);
72 else
73 mark_blob_uninteresting(entry->item.blob);
74 free(entry);
75 entry = next;
79 void mark_parents_uninteresting(struct commit *commit)
81 struct commit_list *parents = commit->parents;
83 while (parents) {
84 struct commit *commit = parents->item;
85 if (!(commit->object.flags & UNINTERESTING)) {
86 commit->object.flags |= UNINTERESTING;
89 * Normally we haven't parsed the parent
90 * yet, so we won't have a parent of a parent
91 * here. However, it may turn out that we've
92 * reached this commit some other way (where it
93 * wasn't uninteresting), in which case we need
94 * to mark its parents recursively too..
96 if (commit->parents)
97 mark_parents_uninteresting(commit);
101 * A missing commit is ok iff its parent is marked
102 * uninteresting.
104 * We just mark such a thing parsed, so that when
105 * it is popped next time around, we won't be trying
106 * to parse it and get an error.
108 if (!has_sha1_file(commit->object.sha1))
109 commit->object.parsed = 1;
110 parents = parents->next;
114 static void add_pending_object(struct rev_info *revs, struct object *obj, const char *name)
116 add_object(obj, &revs->pending_objects, NULL, name);
119 static struct commit *get_commit_reference(struct rev_info *revs, const char *name, const unsigned char *sha1, unsigned int flags)
121 struct object *object;
123 object = parse_object(sha1);
124 if (!object)
125 die("bad object %s", name);
128 * Tag object? Look what it points to..
130 while (object->type == tag_type) {
131 struct tag *tag = (struct tag *) object;
132 object->flags |= flags;
133 if (revs->tag_objects && !(object->flags & UNINTERESTING))
134 add_pending_object(revs, object, tag->tag);
135 object = parse_object(tag->tagged->sha1);
136 if (!object)
137 die("bad object %s", sha1_to_hex(tag->tagged->sha1));
141 * Commit object? Just return it, we'll do all the complex
142 * reachability crud.
144 if (object->type == commit_type) {
145 struct commit *commit = (struct commit *)object;
146 object->flags |= flags;
147 if (parse_commit(commit) < 0)
148 die("unable to parse commit %s", name);
149 if (flags & UNINTERESTING) {
150 mark_parents_uninteresting(commit);
151 revs->limited = 1;
153 return commit;
157 * Tree object? Either mark it uniniteresting, or add it
158 * to the list of objects to look at later..
160 if (object->type == tree_type) {
161 struct tree *tree = (struct tree *)object;
162 if (!revs->tree_objects)
163 return NULL;
164 if (flags & UNINTERESTING) {
165 mark_tree_uninteresting(tree);
166 return NULL;
168 add_pending_object(revs, object, "");
169 return NULL;
173 * Blob object? You know the drill by now..
175 if (object->type == blob_type) {
176 struct blob *blob = (struct blob *)object;
177 if (!revs->blob_objects)
178 return NULL;
179 if (flags & UNINTERESTING) {
180 mark_blob_uninteresting(blob);
181 return NULL;
183 add_pending_object(revs, object, "");
184 return NULL;
186 die("%s is unknown object", name);
189 static int everybody_uninteresting(struct commit_list *orig)
191 struct commit_list *list = orig;
192 while (list) {
193 struct commit *commit = list->item;
194 list = list->next;
195 if (commit->object.flags & UNINTERESTING)
196 continue;
197 return 0;
199 return 1;
202 static int tree_difference = REV_TREE_SAME;
204 static void file_add_remove(struct diff_options *options,
205 int addremove, unsigned mode,
206 const unsigned char *sha1,
207 const char *base, const char *path)
209 int diff = REV_TREE_DIFFERENT;
212 * Is it an add of a new file? It means that the old tree
213 * didn't have it at all, so we will turn "REV_TREE_SAME" ->
214 * "REV_TREE_NEW", but leave any "REV_TREE_DIFFERENT" alone
215 * (and if it already was "REV_TREE_NEW", we'll keep it
216 * "REV_TREE_NEW" of course).
218 if (addremove == '+') {
219 diff = tree_difference;
220 if (diff != REV_TREE_SAME)
221 return;
222 diff = REV_TREE_NEW;
224 tree_difference = diff;
227 static void file_change(struct diff_options *options,
228 unsigned old_mode, unsigned new_mode,
229 const unsigned char *old_sha1,
230 const unsigned char *new_sha1,
231 const char *base, const char *path)
233 tree_difference = REV_TREE_DIFFERENT;
236 static struct diff_options diff_opt = {
237 .recursive = 1,
238 .add_remove = file_add_remove,
239 .change = file_change,
242 int rev_compare_tree(struct tree *t1, struct tree *t2)
244 if (!t1)
245 return REV_TREE_NEW;
246 if (!t2)
247 return REV_TREE_DIFFERENT;
248 tree_difference = REV_TREE_SAME;
249 if (diff_tree_sha1(t1->object.sha1, t2->object.sha1, "", &diff_opt) < 0)
250 return REV_TREE_DIFFERENT;
251 return tree_difference;
254 int rev_same_tree_as_empty(struct tree *t1)
256 int retval;
257 void *tree;
258 struct tree_desc empty, real;
260 if (!t1)
261 return 0;
263 tree = read_object_with_reference(t1->object.sha1, tree_type, &real.size, NULL);
264 if (!tree)
265 return 0;
266 real.buf = tree;
268 empty.buf = "";
269 empty.size = 0;
271 tree_difference = 0;
272 retval = diff_tree(&empty, &real, "", &diff_opt);
273 free(tree);
275 return retval >= 0 && !tree_difference;
278 static void try_to_simplify_commit(struct rev_info *revs, struct commit *commit)
280 struct commit_list **pp, *parent;
281 int tree_changed = 0;
283 if (!commit->tree)
284 return;
286 if (!commit->parents) {
287 if (!rev_same_tree_as_empty(commit->tree))
288 commit->object.flags |= TREECHANGE;
289 return;
292 pp = &commit->parents;
293 while ((parent = *pp) != NULL) {
294 struct commit *p = parent->item;
296 parse_commit(p);
297 switch (rev_compare_tree(p->tree, commit->tree)) {
298 case REV_TREE_SAME:
299 if (p->object.flags & UNINTERESTING) {
300 /* Even if a merge with an uninteresting
301 * side branch brought the entire change
302 * we are interested in, we do not want
303 * to lose the other branches of this
304 * merge, so we just keep going.
306 pp = &parent->next;
307 continue;
309 parent->next = NULL;
310 commit->parents = parent;
311 return;
313 case REV_TREE_NEW:
314 if (revs->remove_empty_trees &&
315 rev_same_tree_as_empty(p->tree)) {
316 /* We are adding all the specified
317 * paths from this parent, so the
318 * history beyond this parent is not
319 * interesting. Remove its parents
320 * (they are grandparents for us).
321 * IOW, we pretend this parent is a
322 * "root" commit.
324 parse_commit(p);
325 p->parents = NULL;
327 /* fallthrough */
328 case REV_TREE_DIFFERENT:
329 tree_changed = 1;
330 pp = &parent->next;
331 continue;
333 die("bad tree compare for commit %s", sha1_to_hex(commit->object.sha1));
335 if (tree_changed)
336 commit->object.flags |= TREECHANGE;
339 static void add_parents_to_list(struct rev_info *revs, struct commit *commit, struct commit_list **list)
341 struct commit_list *parent = commit->parents;
343 if (commit->object.flags & ADDED)
344 return;
345 commit->object.flags |= ADDED;
348 * If the commit is uninteresting, don't try to
349 * prune parents - we want the maximal uninteresting
350 * set.
352 * Normally we haven't parsed the parent
353 * yet, so we won't have a parent of a parent
354 * here. However, it may turn out that we've
355 * reached this commit some other way (where it
356 * wasn't uninteresting), in which case we need
357 * to mark its parents recursively too..
359 if (commit->object.flags & UNINTERESTING) {
360 while (parent) {
361 struct commit *p = parent->item;
362 parent = parent->next;
363 parse_commit(p);
364 p->object.flags |= UNINTERESTING;
365 if (p->parents)
366 mark_parents_uninteresting(p);
367 if (p->object.flags & SEEN)
368 continue;
369 p->object.flags |= SEEN;
370 insert_by_date(p, list);
372 return;
376 * Ok, the commit wasn't uninteresting. Try to
377 * simplify the commit history and find the parent
378 * that has no differences in the path set if one exists.
380 if (revs->prune_fn)
381 revs->prune_fn(revs, commit);
383 parent = commit->parents;
384 while (parent) {
385 struct commit *p = parent->item;
387 parent = parent->next;
389 parse_commit(p);
390 if (p->object.flags & SEEN)
391 continue;
392 p->object.flags |= SEEN;
393 insert_by_date(p, list);
397 static void limit_list(struct rev_info *revs)
399 struct commit_list *list = revs->commits;
400 struct commit_list *newlist = NULL;
401 struct commit_list **p = &newlist;
403 while (list) {
404 struct commit_list *entry = list;
405 struct commit *commit = list->item;
406 struct object *obj = &commit->object;
408 list = list->next;
409 free(entry);
411 if (revs->max_age != -1 && (commit->date < revs->max_age))
412 obj->flags |= UNINTERESTING;
413 if (revs->unpacked && has_sha1_pack(obj->sha1))
414 obj->flags |= UNINTERESTING;
415 add_parents_to_list(revs, commit, &list);
416 if (obj->flags & UNINTERESTING) {
417 mark_parents_uninteresting(commit);
418 if (everybody_uninteresting(list))
419 break;
420 continue;
422 if (revs->min_age != -1 && (commit->date > revs->min_age))
423 continue;
424 p = &commit_list_insert(commit, p)->next;
426 if (revs->boundary) {
427 /* mark the ones that are on the result list first */
428 for (list = newlist; list; list = list->next) {
429 struct commit *commit = list->item;
430 commit->object.flags |= TMP_MARK;
432 for (list = newlist; list; list = list->next) {
433 struct commit *commit = list->item;
434 struct object *obj = &commit->object;
435 struct commit_list *parent;
436 if (obj->flags & UNINTERESTING)
437 continue;
438 for (parent = commit->parents;
439 parent;
440 parent = parent->next) {
441 struct commit *pcommit = parent->item;
442 if (!(pcommit->object.flags & UNINTERESTING))
443 continue;
444 pcommit->object.flags |= BOUNDARY;
445 if (pcommit->object.flags & TMP_MARK)
446 continue;
447 pcommit->object.flags |= TMP_MARK;
448 p = &commit_list_insert(pcommit, p)->next;
451 for (list = newlist; list; list = list->next) {
452 struct commit *commit = list->item;
453 commit->object.flags &= ~TMP_MARK;
456 revs->commits = newlist;
459 static void add_one_commit(struct commit *commit, struct rev_info *revs)
461 if (!commit || (commit->object.flags & SEEN))
462 return;
463 commit->object.flags |= SEEN;
464 commit_list_insert(commit, &revs->commits);
467 static int all_flags;
468 static struct rev_info *all_revs;
470 static int handle_one_ref(const char *path, const unsigned char *sha1)
472 struct commit *commit = get_commit_reference(all_revs, path, sha1, all_flags);
473 add_one_commit(commit, all_revs);
474 return 0;
477 static void handle_all(struct rev_info *revs, unsigned flags)
479 all_revs = revs;
480 all_flags = flags;
481 for_each_ref(handle_one_ref);
484 void init_revisions(struct rev_info *revs)
486 memset(revs, 0, sizeof(*revs));
487 revs->lifo = 1;
488 revs->dense = 1;
489 revs->prefix = setup_git_directory();
490 revs->max_age = -1;
491 revs->min_age = -1;
492 revs->max_count = -1;
494 revs->prune_fn = NULL;
495 revs->prune_data = NULL;
497 revs->topo_setter = topo_sort_default_setter;
498 revs->topo_getter = topo_sort_default_getter;
502 * Parse revision information, filling in the "rev_info" structure,
503 * and removing the used arguments from the argument list.
505 * Returns the number of arguments left that weren't recognized
506 * (which are also moved to the head of the argument list)
508 int setup_revisions(int argc, const char **argv, struct rev_info *revs, const char *def)
510 int i, flags, seen_dashdash;
511 const char **unrecognized = argv + 1;
512 int left = 1;
514 init_revisions(revs);
516 /* First, search for "--" */
517 seen_dashdash = 0;
518 for (i = 1; i < argc; i++) {
519 const char *arg = argv[i];
520 if (strcmp(arg, "--"))
521 continue;
522 argv[i] = NULL;
523 argc = i;
524 revs->prune_data = get_pathspec(revs->prefix, argv + i + 1);
525 seen_dashdash = 1;
526 break;
529 flags = 0;
530 for (i = 1; i < argc; i++) {
531 struct commit *commit;
532 const char *arg = argv[i];
533 unsigned char sha1[20];
534 char *dotdot;
535 int local_flags;
537 if (*arg == '-') {
538 if (!strncmp(arg, "--max-count=", 12)) {
539 revs->max_count = atoi(arg + 12);
540 continue;
542 /* accept -<digit>, like traditilnal "head" */
543 if ((*arg == '-') && isdigit(arg[1])) {
544 revs->max_count = atoi(arg + 1);
545 continue;
547 if (!strcmp(arg, "-n")) {
548 if (argc <= i + 1)
549 die("-n requires an argument");
550 revs->max_count = atoi(argv[++i]);
551 continue;
553 if (!strncmp(arg,"-n",2)) {
554 revs->max_count = atoi(arg + 2);
555 continue;
557 if (!strncmp(arg, "--max-age=", 10)) {
558 revs->max_age = atoi(arg + 10);
559 continue;
561 if (!strncmp(arg, "--since=", 8)) {
562 revs->max_age = approxidate(arg + 8);
563 continue;
565 if (!strncmp(arg, "--after=", 8)) {
566 revs->max_age = approxidate(arg + 8);
567 continue;
569 if (!strncmp(arg, "--min-age=", 10)) {
570 revs->min_age = atoi(arg + 10);
571 continue;
573 if (!strncmp(arg, "--before=", 9)) {
574 revs->min_age = approxidate(arg + 9);
575 continue;
577 if (!strncmp(arg, "--until=", 8)) {
578 revs->min_age = approxidate(arg + 8);
579 continue;
581 if (!strcmp(arg, "--all")) {
582 handle_all(revs, flags);
583 continue;
585 if (!strcmp(arg, "--not")) {
586 flags ^= UNINTERESTING;
587 continue;
589 if (!strcmp(arg, "--default")) {
590 if (++i >= argc)
591 die("bad --default argument");
592 def = argv[i];
593 continue;
595 if (!strcmp(arg, "--topo-order")) {
596 revs->topo_order = 1;
597 continue;
599 if (!strcmp(arg, "--date-order")) {
600 revs->lifo = 0;
601 revs->topo_order = 1;
602 continue;
604 if (!strcmp(arg, "--parents")) {
605 revs->parents = 1;
606 continue;
608 if (!strcmp(arg, "--dense")) {
609 revs->dense = 1;
610 continue;
612 if (!strcmp(arg, "--sparse")) {
613 revs->dense = 0;
614 continue;
616 if (!strcmp(arg, "--remove-empty")) {
617 revs->remove_empty_trees = 1;
618 continue;
620 if (!strcmp(arg, "--no-merges")) {
621 revs->no_merges = 1;
622 continue;
624 if (!strcmp(arg, "--boundary")) {
625 revs->boundary = 1;
626 continue;
628 if (!strcmp(arg, "--objects")) {
629 revs->tag_objects = 1;
630 revs->tree_objects = 1;
631 revs->blob_objects = 1;
632 continue;
634 if (!strcmp(arg, "--objects-edge")) {
635 revs->tag_objects = 1;
636 revs->tree_objects = 1;
637 revs->blob_objects = 1;
638 revs->edge_hint = 1;
639 continue;
641 if (!strcmp(arg, "--unpacked")) {
642 revs->unpacked = 1;
643 continue;
645 *unrecognized++ = arg;
646 left++;
647 continue;
649 dotdot = strstr(arg, "..");
650 if (dotdot) {
651 unsigned char from_sha1[20];
652 const char *next = dotdot + 2;
653 const char *this = arg;
654 *dotdot = 0;
655 if (!*next)
656 next = "HEAD";
657 if (dotdot == arg)
658 this = "HEAD";
659 if (!get_sha1(this, from_sha1) &&
660 !get_sha1(next, sha1)) {
661 struct commit *exclude;
662 struct commit *include;
664 exclude = get_commit_reference(revs, this, from_sha1, flags ^ UNINTERESTING);
665 include = get_commit_reference(revs, next, sha1, flags);
666 if (!exclude || !include)
667 die("Invalid revision range %s..%s", arg, next);
668 add_one_commit(exclude, revs);
669 add_one_commit(include, revs);
670 continue;
672 *dotdot = '.';
674 local_flags = 0;
675 if (*arg == '^') {
676 local_flags = UNINTERESTING;
677 arg++;
679 if (get_sha1(arg, sha1) < 0) {
680 struct stat st;
681 int j;
683 if (seen_dashdash || local_flags)
684 die("bad revision '%s'", arg);
686 /* If we didn't have a "--", all filenames must exist */
687 for (j = i; j < argc; j++) {
688 if (lstat(argv[j], &st) < 0)
689 die("'%s': %s", argv[j], strerror(errno));
691 revs->prune_data = get_pathspec(revs->prefix, argv + i);
692 break;
694 commit = get_commit_reference(revs, arg, sha1, flags ^ local_flags);
695 add_one_commit(commit, revs);
697 if (def && !revs->commits) {
698 unsigned char sha1[20];
699 struct commit *commit;
700 if (get_sha1(def, sha1) < 0)
701 die("bad default revision '%s'", def);
702 commit = get_commit_reference(revs, def, sha1, 0);
703 add_one_commit(commit, revs);
706 if (revs->topo_order || revs->unpacked)
707 revs->limited = 1;
709 if (revs->prune_data) {
710 diff_tree_setup_paths(revs->prune_data);
711 revs->prune_fn = try_to_simplify_commit;
714 return left;
717 void prepare_revision_walk(struct rev_info *revs)
719 sort_by_date(&revs->commits);
720 if (revs->limited)
721 limit_list(revs);
722 if (revs->topo_order)
723 sort_in_topological_order_fn(&revs->commits, revs->lifo,
724 revs->topo_setter,
725 revs->topo_getter);
728 static int rewrite_one(struct rev_info *revs, struct commit **pp)
730 for (;;) {
731 struct commit *p = *pp;
732 if (!revs->limited)
733 add_parents_to_list(revs, p, &revs->commits);
734 if (p->object.flags & (TREECHANGE | UNINTERESTING))
735 return 0;
736 if (!p->parents)
737 return -1;
738 *pp = p->parents->item;
742 static void rewrite_parents(struct rev_info *revs, struct commit *commit)
744 struct commit_list **pp = &commit->parents;
745 while (*pp) {
746 struct commit_list *parent = *pp;
747 if (rewrite_one(revs, &parent->item) < 0) {
748 *pp = parent->next;
749 continue;
751 pp = &parent->next;
755 struct commit *get_revision(struct rev_info *revs)
757 struct commit_list *list = revs->commits;
759 if (!list)
760 return NULL;
762 /* Check the max_count ... */
763 switch (revs->max_count) {
764 case -1:
765 break;
766 case 0:
767 return NULL;
768 default:
769 revs->max_count--;
772 do {
773 struct commit *commit = revs->commits->item;
775 revs->commits = revs->commits->next;
778 * If we haven't done the list limiting, we need to look at
779 * the parents here. We also need to do the date-based limiting
780 * that we'd otherwise have done in limit_list().
782 if (!revs->limited) {
783 if ((revs->unpacked &&
784 has_sha1_pack(commit->object.sha1)) ||
785 (revs->max_age != -1 &&
786 (commit->date < revs->max_age)))
787 continue;
788 add_parents_to_list(revs, commit, &revs->commits);
790 if (commit->object.flags & SHOWN)
791 continue;
792 if (!(commit->object.flags & BOUNDARY) &&
793 (commit->object.flags & UNINTERESTING))
794 continue;
795 if (revs->min_age != -1 && (commit->date > revs->min_age))
796 continue;
797 if (revs->no_merges &&
798 commit->parents && commit->parents->next)
799 continue;
800 if (revs->prune_fn && revs->dense) {
801 if (!(commit->object.flags & TREECHANGE))
802 continue;
803 if (revs->parents)
804 rewrite_parents(revs, commit);
806 commit->object.flags |= SHOWN;
807 return commit;
808 } while (revs->commits);
809 return NULL;