git-svnimport: if a limit is specified, respect it
[git/dscho.git] / revision.c
blob12cd0529a5dfac9af01af0f47e895e9cf142c42a
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", &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;
344 * If the commit is uninteresting, don't try to
345 * prune parents - we want the maximal uninteresting
346 * set.
348 * Normally we haven't parsed the parent
349 * yet, so we won't have a parent of a parent
350 * here. However, it may turn out that we've
351 * reached this commit some other way (where it
352 * wasn't uninteresting), in which case we need
353 * to mark its parents recursively too..
355 if (commit->object.flags & UNINTERESTING) {
356 while (parent) {
357 struct commit *p = parent->item;
358 parent = parent->next;
359 parse_commit(p);
360 p->object.flags |= UNINTERESTING;
361 if (p->parents)
362 mark_parents_uninteresting(p);
363 if (p->object.flags & SEEN)
364 continue;
365 p->object.flags |= SEEN;
366 insert_by_date(p, list);
368 return;
372 * Ok, the commit wasn't uninteresting. Try to
373 * simplify the commit history and find the parent
374 * that has no differences in the path set if one exists.
376 if (revs->prune_fn)
377 revs->prune_fn(revs, commit);
379 parent = commit->parents;
380 while (parent) {
381 struct commit *p = parent->item;
383 parent = parent->next;
385 parse_commit(p);
386 if (p->object.flags & SEEN)
387 continue;
388 p->object.flags |= SEEN;
389 insert_by_date(p, list);
393 static void limit_list(struct rev_info *revs)
395 struct commit_list *list = revs->commits;
396 struct commit_list *newlist = NULL;
397 struct commit_list **p = &newlist;
399 while (list) {
400 struct commit_list *entry = list;
401 struct commit *commit = list->item;
402 struct object *obj = &commit->object;
404 list = list->next;
405 free(entry);
407 if (revs->max_age != -1 && (commit->date < revs->max_age))
408 obj->flags |= UNINTERESTING;
409 if (revs->unpacked && has_sha1_pack(obj->sha1))
410 obj->flags |= UNINTERESTING;
411 add_parents_to_list(revs, commit, &list);
412 if (obj->flags & UNINTERESTING) {
413 mark_parents_uninteresting(commit);
414 if (everybody_uninteresting(list))
415 break;
416 continue;
418 if (revs->min_age != -1 && (commit->date > revs->min_age))
419 continue;
420 p = &commit_list_insert(commit, p)->next;
422 revs->commits = newlist;
425 static void add_one_commit(struct commit *commit, struct rev_info *revs)
427 if (!commit || (commit->object.flags & SEEN))
428 return;
429 commit->object.flags |= SEEN;
430 commit_list_insert(commit, &revs->commits);
433 static int all_flags;
434 static struct rev_info *all_revs;
436 static int handle_one_ref(const char *path, const unsigned char *sha1)
438 struct commit *commit = get_commit_reference(all_revs, path, sha1, all_flags);
439 add_one_commit(commit, all_revs);
440 return 0;
443 static void handle_all(struct rev_info *revs, unsigned flags)
445 all_revs = revs;
446 all_flags = flags;
447 for_each_ref(handle_one_ref);
450 void init_revisions(struct rev_info *revs)
452 memset(revs, 0, sizeof(*revs));
453 revs->lifo = 1;
454 revs->dense = 1;
455 revs->prefix = setup_git_directory();
456 revs->max_age = -1;
457 revs->min_age = -1;
458 revs->max_count = -1;
460 revs->prune_fn = NULL;
461 revs->prune_data = NULL;
463 revs->topo_setter = topo_sort_default_setter;
464 revs->topo_getter = topo_sort_default_getter;
468 * Parse revision information, filling in the "rev_info" structure,
469 * and removing the used arguments from the argument list.
471 * Returns the number of arguments left that weren't recognized
472 * (which are also moved to the head of the argument list)
474 int setup_revisions(int argc, const char **argv, struct rev_info *revs, const char *def)
476 int i, flags, seen_dashdash;
477 const char **unrecognized = argv + 1;
478 int left = 1;
480 init_revisions(revs);
482 /* First, search for "--" */
483 seen_dashdash = 0;
484 for (i = 1; i < argc; i++) {
485 const char *arg = argv[i];
486 if (strcmp(arg, "--"))
487 continue;
488 argv[i] = NULL;
489 argc = i;
490 revs->prune_data = get_pathspec(revs->prefix, argv + i + 1);
491 seen_dashdash = 1;
492 break;
495 flags = 0;
496 for (i = 1; i < argc; i++) {
497 struct commit *commit;
498 const char *arg = argv[i];
499 unsigned char sha1[20];
500 char *dotdot;
501 int local_flags;
503 if (*arg == '-') {
504 if (!strncmp(arg, "--max-count=", 12)) {
505 revs->max_count = atoi(arg + 12);
506 continue;
508 /* accept -<digit>, like traditilnal "head" */
509 if ((*arg == '-') && isdigit(arg[1])) {
510 revs->max_count = atoi(arg + 1);
511 continue;
513 if (!strcmp(arg, "-n")) {
514 if (argc <= i + 1)
515 die("-n requires an argument");
516 revs->max_count = atoi(argv[++i]);
517 continue;
519 if (!strncmp(arg,"-n",2)) {
520 revs->max_count = atoi(arg + 2);
521 continue;
523 if (!strncmp(arg, "--max-age=", 10)) {
524 revs->max_age = atoi(arg + 10);
525 revs->limited = 1;
526 continue;
528 if (!strncmp(arg, "--min-age=", 10)) {
529 revs->min_age = atoi(arg + 10);
530 revs->limited = 1;
531 continue;
533 if (!strncmp(arg, "--since=", 8)) {
534 revs->max_age = approxidate(arg + 8);
535 revs->limited = 1;
536 continue;
538 if (!strncmp(arg, "--after=", 8)) {
539 revs->max_age = approxidate(arg + 8);
540 revs->limited = 1;
541 continue;
543 if (!strncmp(arg, "--before=", 9)) {
544 revs->min_age = approxidate(arg + 9);
545 revs->limited = 1;
546 continue;
548 if (!strncmp(arg, "--until=", 8)) {
549 revs->min_age = approxidate(arg + 8);
550 revs->limited = 1;
551 continue;
553 if (!strcmp(arg, "--all")) {
554 handle_all(revs, flags);
555 continue;
557 if (!strcmp(arg, "--not")) {
558 flags ^= UNINTERESTING;
559 continue;
561 if (!strcmp(arg, "--default")) {
562 if (++i >= argc)
563 die("bad --default argument");
564 def = argv[i];
565 continue;
567 if (!strcmp(arg, "--topo-order")) {
568 revs->topo_order = 1;
569 revs->limited = 1;
570 continue;
572 if (!strcmp(arg, "--date-order")) {
573 revs->lifo = 0;
574 revs->topo_order = 1;
575 revs->limited = 1;
576 continue;
578 if (!strcmp(arg, "--dense")) {
579 revs->dense = 1;
580 continue;
582 if (!strcmp(arg, "--sparse")) {
583 revs->dense = 0;
584 continue;
586 if (!strcmp(arg, "--remove-empty")) {
587 revs->remove_empty_trees = 1;
588 continue;
590 if (!strncmp(arg, "--no-merges", 11)) {
591 revs->no_merges = 1;
592 continue;
594 if (!strcmp(arg, "--objects")) {
595 revs->tag_objects = 1;
596 revs->tree_objects = 1;
597 revs->blob_objects = 1;
598 continue;
600 if (!strcmp(arg, "--objects-edge")) {
601 revs->tag_objects = 1;
602 revs->tree_objects = 1;
603 revs->blob_objects = 1;
604 revs->edge_hint = 1;
605 continue;
607 if (!strcmp(arg, "--unpacked")) {
608 revs->unpacked = 1;
609 revs->limited = 1;
610 continue;
612 *unrecognized++ = arg;
613 left++;
614 continue;
616 dotdot = strstr(arg, "..");
617 if (dotdot) {
618 unsigned char from_sha1[20];
619 char *next = dotdot + 2;
620 *dotdot = 0;
621 if (!*next)
622 next = "HEAD";
623 if (!get_sha1(arg, from_sha1) && !get_sha1(next, sha1)) {
624 struct commit *exclude;
625 struct commit *include;
627 exclude = get_commit_reference(revs, arg, from_sha1, flags ^ UNINTERESTING);
628 include = get_commit_reference(revs, next, sha1, flags);
629 if (!exclude || !include)
630 die("Invalid revision range %s..%s", arg, next);
631 add_one_commit(exclude, revs);
632 add_one_commit(include, revs);
633 continue;
635 *dotdot = '.';
637 local_flags = 0;
638 if (*arg == '^') {
639 local_flags = UNINTERESTING;
640 arg++;
642 if (get_sha1(arg, sha1) < 0) {
643 struct stat st;
644 int j;
646 if (seen_dashdash || local_flags)
647 die("bad revision '%s'", arg);
649 /* If we didn't have a "--", all filenames must exist */
650 for (j = i; j < argc; j++) {
651 if (lstat(argv[j], &st) < 0)
652 die("'%s': %s", arg, strerror(errno));
654 revs->prune_data = get_pathspec(revs->prefix, argv + i);
655 break;
657 commit = get_commit_reference(revs, arg, sha1, flags ^ local_flags);
658 add_one_commit(commit, revs);
660 if (def && !revs->commits) {
661 unsigned char sha1[20];
662 struct commit *commit;
663 if (get_sha1(def, sha1) < 0)
664 die("bad default revision '%s'", def);
665 commit = get_commit_reference(revs, def, sha1, 0);
666 add_one_commit(commit, revs);
669 if (revs->prune_data) {
670 diff_tree_setup_paths(revs->prune_data);
671 revs->prune_fn = try_to_simplify_commit;
672 revs->limited = 1;
675 return left;
678 void prepare_revision_walk(struct rev_info *revs)
680 sort_by_date(&revs->commits);
681 if (revs->limited)
682 limit_list(revs);
683 if (revs->topo_order)
684 sort_in_topological_order_fn(&revs->commits, revs->lifo,
685 revs->topo_setter,
686 revs->topo_getter);
689 static int rewrite_one(struct commit **pp)
691 for (;;) {
692 struct commit *p = *pp;
693 if (p->object.flags & (TREECHANGE | UNINTERESTING))
694 return 0;
695 if (!p->parents)
696 return -1;
697 *pp = p->parents->item;
701 static void rewrite_parents(struct commit *commit)
703 struct commit_list **pp = &commit->parents;
704 while (*pp) {
705 struct commit_list *parent = *pp;
706 if (rewrite_one(&parent->item) < 0) {
707 *pp = parent->next;
708 continue;
710 pp = &parent->next;
714 struct commit *get_revision(struct rev_info *revs)
716 struct commit_list *list = revs->commits;
718 if (!list)
719 return NULL;
721 /* Check the max_count ... */
722 switch (revs->max_count) {
723 case -1:
724 break;
725 case 0:
726 return NULL;
727 default:
728 revs->max_count--;
731 do {
732 struct commit *commit = revs->commits->item;
734 if (commit->object.flags & (UNINTERESTING|SHOWN))
735 goto next;
736 if (revs->min_age != -1 && (commit->date > revs->min_age))
737 goto next;
738 if (revs->max_age != -1 && (commit->date < revs->max_age))
739 return NULL;
740 if (revs->no_merges && commit->parents && commit->parents->next)
741 goto next;
742 if (revs->prune_fn && revs->dense) {
743 if (!(commit->object.flags & TREECHANGE))
744 goto next;
745 rewrite_parents(commit);
747 /* More to go? */
748 if (revs->max_count)
749 pop_most_recent_commit(&revs->commits, SEEN);
750 commit->object.flags |= SHOWN;
751 return commit;
752 next:
753 pop_most_recent_commit(&revs->commits, SEEN);
754 } while (revs->commits);
755 return NULL;