rev-lib: Make it easy to do rename tracking (take 2)
[git.git] / revision.c
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"
9
10 static char *path_name(struct name_path *path, const char *name)
11 {
12         struct name_path *p;
13         char *n, *m;
14         int nlen = strlen(name);
15         int len = nlen + 1;
16
17         for (p = path; p; p = p->up) {
18                 if (p->elem_len)
19                         len += p->elem_len + 1;
20         }
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] = '/';
29                 }
30         }
31         return n;
32 }
33
34 struct object_list **add_object(struct object *obj,
35                                        struct object_list **p,
36                                        struct name_path *path,
37                                        const char *name)
38 {
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;
45 }
46
47 static void mark_blob_uninteresting(struct blob *blob)
48 {
49         if (blob->object.flags & UNINTERESTING)
50                 return;
51         blob->object.flags |= UNINTERESTING;
52 }
53
54 void mark_tree_uninteresting(struct tree *tree)
55 {
56         struct object *obj = &tree->object;
57         struct tree_entry_list *entry;
58
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;
76         }
77 }
78
79 void mark_parents_uninteresting(struct commit *commit)
80 {
81         struct commit_list *parents = commit->parents;
82
83         while (parents) {
84                 struct commit *commit = parents->item;
85                 if (!(commit->object.flags & UNINTERESTING)) {
86                         commit->object.flags |= UNINTERESTING;
87
88                         /*
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..
95                          */
96                         if (commit->parents)
97                                 mark_parents_uninteresting(commit);
98                 }
99
100                 /*
101                  * A missing commit is ok iff its parent is marked
102                  * uninteresting.
103                  *
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.
107                  */
108                 if (!has_sha1_file(commit->object.sha1))
109                         commit->object.parsed = 1;
110                 parents = parents->next;
111         }
112 }
113
114 static void add_pending_object(struct rev_info *revs, struct object *obj, const char *name)
115 {
116         add_object(obj, &revs->pending_objects, NULL, name);
117 }
118
119 static struct commit *get_commit_reference(struct rev_info *revs, const char *name, const unsigned char *sha1, unsigned int flags)
120 {
121         struct object *object;
122
123         object = parse_object(sha1);
124         if (!object)
125                 die("bad object %s", name);
126
127         /*
128          * Tag object? Look what it points to..
129          */
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));
138         }
139
140         /*
141          * Commit object? Just return it, we'll do all the complex
142          * reachability crud.
143          */
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;
152                 }
153                 return commit;
154         }
155
156         /*
157          * Tree object? Either mark it uniniteresting, or add it
158          * to the list of objects to look at later..
159          */
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;
167                 }
168                 add_pending_object(revs, object, "");
169                 return NULL;
170         }
171
172         /*
173          * Blob object? You know the drill by now..
174          */
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;
182                 }
183                 add_pending_object(revs, object, "");
184                 return NULL;
185         }
186         die("%s is unknown object", name);
187 }
188
189 static int everybody_uninteresting(struct commit_list *orig)
190 {
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;
198         }
199         return 1;
200 }
201
202 static int tree_difference = REV_TREE_SAME;
203
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)
208 {
209         int diff = REV_TREE_DIFFERENT;
210
211         /*
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).
217          */
218         if (addremove == '+') {
219                 diff = tree_difference;
220                 if (diff != REV_TREE_SAME)
221                         return;
222                 diff = REV_TREE_NEW;
223         }
224         tree_difference = diff;
225 }
226
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)
232 {
233         tree_difference = REV_TREE_DIFFERENT;
234 }
235
236 static struct diff_options diff_opt = {
237         .recursive = 1,
238         .add_remove = file_add_remove,
239         .change = file_change,
240 };
241
242 int rev_compare_tree(struct tree *t1, struct tree *t2)
243 {
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;
252 }
253
254 int rev_same_tree_as_empty(struct tree *t1)
255 {
256         int retval;
257         void *tree;
258         struct tree_desc empty, real;
259
260         if (!t1)
261                 return 0;
262
263         tree = read_object_with_reference(t1->object.sha1, "tree", &real.size, NULL);
264         if (!tree)
265                 return 0;
266         real.buf = tree;
267
268         empty.buf = "";
269         empty.size = 0;
270
271         tree_difference = 0;
272         retval = diff_tree(&empty, &real, "", &diff_opt);
273         free(tree);
274
275         return retval >= 0 && !tree_difference;
276 }
277
278 static void try_to_simplify_commit(struct rev_info *revs, struct commit *commit)
279 {
280         struct commit_list **pp, *parent;
281         int tree_changed = 0;
282
283         if (!commit->tree)
284                 return;
285
286         if (!commit->parents) {
287                 if (!rev_same_tree_as_empty(commit->tree))
288                         commit->object.flags |= TREECHANGE;
289                 return;
290         }
291
292         pp = &commit->parents;
293         while ((parent = *pp) != NULL) {
294                 struct commit *p = parent->item;
295
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.
305                                  */
306                                 pp = &parent->next;
307                                 continue;
308                         }
309                         parent->next = NULL;
310                         commit->parents = parent;
311                         return;
312
313                 case REV_TREE_NEW:
314                         if (revs->remove_empty_trees &&
315                             rev_same_tree_as_empty(p->tree)) {
316                                 *pp = parent->next;
317                                 continue;
318                         }
319                 /* fallthrough */
320                 case REV_TREE_DIFFERENT:
321                         tree_changed = 1;
322                         pp = &parent->next;
323                         continue;
324                 }
325                 die("bad tree compare for commit %s", sha1_to_hex(commit->object.sha1));
326         }
327         if (tree_changed)
328                 commit->object.flags |= TREECHANGE;
329 }
330
331 static void add_parents_to_list(struct rev_info *revs, struct commit *commit, struct commit_list **list)
332 {
333         struct commit_list *parent = commit->parents;
334
335         /*
336          * If the commit is uninteresting, don't try to
337          * prune parents - we want the maximal uninteresting
338          * set.
339          *
340          * Normally we haven't parsed the parent
341          * yet, so we won't have a parent of a parent
342          * here. However, it may turn out that we've
343          * reached this commit some other way (where it
344          * wasn't uninteresting), in which case we need
345          * to mark its parents recursively too..
346          */
347         if (commit->object.flags & UNINTERESTING) {
348                 while (parent) {
349                         struct commit *p = parent->item;
350                         parent = parent->next;
351                         parse_commit(p);
352                         p->object.flags |= UNINTERESTING;
353                         if (p->parents)
354                                 mark_parents_uninteresting(p);
355                         if (p->object.flags & SEEN)
356                                 continue;
357                         p->object.flags |= SEEN;
358                         insert_by_date(p, list);
359                 }
360                 return;
361         }
362
363         /*
364          * Ok, the commit wasn't uninteresting. Try to
365          * simplify the commit history and find the parent
366          * that has no differences in the path set if one exists.
367          */
368         if (revs->prune_fn)
369                 revs->prune_fn(revs, commit);
370
371         parent = commit->parents;
372         while (parent) {
373                 struct commit *p = parent->item;
374
375                 parent = parent->next;
376
377                 parse_commit(p);
378                 if (p->object.flags & SEEN)
379                         continue;
380                 p->object.flags |= SEEN;
381                 insert_by_date(p, list);
382         }
383 }
384
385 static void limit_list(struct rev_info *revs)
386 {
387         struct commit_list *list = revs->commits;
388         struct commit_list *newlist = NULL;
389         struct commit_list **p = &newlist;
390
391         while (list) {
392                 struct commit_list *entry = list;
393                 struct commit *commit = list->item;
394                 struct object *obj = &commit->object;
395
396                 list = list->next;
397                 free(entry);
398
399                 if (revs->max_age != -1 && (commit->date < revs->max_age))
400                         obj->flags |= UNINTERESTING;
401                 if (revs->unpacked && has_sha1_pack(obj->sha1))
402                         obj->flags |= UNINTERESTING;
403                 add_parents_to_list(revs, commit, &list);
404                 if (obj->flags & UNINTERESTING) {
405                         mark_parents_uninteresting(commit);
406                         if (everybody_uninteresting(list))
407                                 break;
408                         continue;
409                 }
410                 if (revs->min_age != -1 && (commit->date > revs->min_age))
411                         continue;
412                 p = &commit_list_insert(commit, p)->next;
413         }
414         revs->commits = newlist;
415 }
416
417 static void add_one_commit(struct commit *commit, struct rev_info *revs)
418 {
419         if (!commit || (commit->object.flags & SEEN))
420                 return;
421         commit->object.flags |= SEEN;
422         commit_list_insert(commit, &revs->commits);
423 }
424
425 static int all_flags;
426 static struct rev_info *all_revs;
427
428 static int handle_one_ref(const char *path, const unsigned char *sha1)
429 {
430         struct commit *commit = get_commit_reference(all_revs, path, sha1, all_flags);
431         add_one_commit(commit, all_revs);
432         return 0;
433 }
434
435 static void handle_all(struct rev_info *revs, unsigned flags)
436 {
437         all_revs = revs;
438         all_flags = flags;
439         for_each_ref(handle_one_ref);
440 }
441
442 void init_revisions(struct rev_info *revs)
443 {
444         memset(revs, 0, sizeof(*revs));
445         revs->lifo = 1;
446         revs->dense = 1;
447         revs->prefix = setup_git_directory();
448         revs->max_age = -1;
449         revs->min_age = -1;
450         revs->max_count = -1;
451
452         revs->prune_fn = NULL;
453         revs->prune_data = NULL;
454
455         revs->topo_setter = topo_sort_default_setter;
456         revs->topo_getter = topo_sort_default_getter;
457 }
458
459 /*
460  * Parse revision information, filling in the "rev_info" structure,
461  * and removing the used arguments from the argument list.
462  *
463  * Returns the number of arguments left that weren't recognized
464  * (which are also moved to the head of the argument list)
465  */
466 int setup_revisions(int argc, const char **argv, struct rev_info *revs, const char *def)
467 {
468         int i, flags, seen_dashdash;
469         const char **unrecognized = argv + 1;
470         int left = 1;
471
472         init_revisions(revs);
473
474         /* First, search for "--" */
475         seen_dashdash = 0;
476         for (i = 1; i < argc; i++) {
477                 const char *arg = argv[i];
478                 if (strcmp(arg, "--"))
479                         continue;
480                 argv[i] = NULL;
481                 argc = i;
482                 revs->prune_data = get_pathspec(revs->prefix, argv + i + 1);
483                 seen_dashdash = 1;
484                 break;
485         }
486
487         flags = 0;
488         for (i = 1; i < argc; i++) {
489                 struct commit *commit;
490                 const char *arg = argv[i];
491                 unsigned char sha1[20];
492                 char *dotdot;
493                 int local_flags;
494
495                 if (*arg == '-') {
496                         if (!strncmp(arg, "--max-count=", 12)) {
497                                 revs->max_count = atoi(arg + 12);
498                                 continue;
499                         }
500                         /* accept -<digit>, like traditilnal "head" */
501                         if ((*arg == '-') && isdigit(arg[1])) {
502                                 revs->max_count = atoi(arg + 1);
503                                 continue;
504                         }
505                         if (!strcmp(arg, "-n")) {
506                                 if (argc <= i + 1)
507                                         die("-n requires an argument");
508                                 revs->max_count = atoi(argv[++i]);
509                                 continue;
510                         }
511                         if (!strncmp(arg,"-n",2)) {
512                                 revs->max_count = atoi(arg + 2);
513                                 continue;
514                         }
515                         if (!strncmp(arg, "--max-age=", 10)) {
516                                 revs->max_age = atoi(arg + 10);
517                                 revs->limited = 1;
518                                 continue;
519                         }
520                         if (!strncmp(arg, "--min-age=", 10)) {
521                                 revs->min_age = atoi(arg + 10);
522                                 revs->limited = 1;
523                                 continue;
524                         }
525                         if (!strncmp(arg, "--since=", 8)) {
526                                 revs->max_age = approxidate(arg + 8);
527                                 revs->limited = 1;
528                                 continue;
529                         }
530                         if (!strncmp(arg, "--after=", 8)) {
531                                 revs->max_age = approxidate(arg + 8);
532                                 revs->limited = 1;
533                                 continue;
534                         }
535                         if (!strncmp(arg, "--before=", 9)) {
536                                 revs->min_age = approxidate(arg + 9);
537                                 revs->limited = 1;
538                                 continue;
539                         }
540                         if (!strncmp(arg, "--until=", 8)) {
541                                 revs->min_age = approxidate(arg + 8);
542                                 revs->limited = 1;
543                                 continue;
544                         }
545                         if (!strcmp(arg, "--all")) {
546                                 handle_all(revs, flags);
547                                 continue;
548                         }
549                         if (!strcmp(arg, "--not")) {
550                                 flags ^= UNINTERESTING;
551                                 continue;
552                         }
553                         if (!strcmp(arg, "--default")) {
554                                 if (++i >= argc)
555                                         die("bad --default argument");
556                                 def = argv[i];
557                                 continue;
558                         }
559                         if (!strcmp(arg, "--topo-order")) {
560                                 revs->topo_order = 1;
561                                 revs->limited = 1;
562                                 continue;
563                         }
564                         if (!strcmp(arg, "--date-order")) {
565                                 revs->lifo = 0;
566                                 revs->topo_order = 1;
567                                 revs->limited = 1;
568                                 continue;
569                         }
570                         if (!strcmp(arg, "--dense")) {
571                                 revs->dense = 1;
572                                 continue;
573                         }
574                         if (!strcmp(arg, "--sparse")) {
575                                 revs->dense = 0;
576                                 continue;
577                         }
578                         if (!strcmp(arg, "--remove-empty")) {
579                                 revs->remove_empty_trees = 1;
580                                 continue;
581                         }
582                         if (!strncmp(arg, "--no-merges", 11)) {
583                                 revs->no_merges = 1;
584                                 continue;
585                         }
586                         if (!strcmp(arg, "--objects")) {
587                                 revs->tag_objects = 1;
588                                 revs->tree_objects = 1;
589                                 revs->blob_objects = 1;
590                                 continue;
591                         }
592                         if (!strcmp(arg, "--objects-edge")) {
593                                 revs->tag_objects = 1;
594                                 revs->tree_objects = 1;
595                                 revs->blob_objects = 1;
596                                 revs->edge_hint = 1;
597                                 continue;
598                         }
599                         if (!strcmp(arg, "--unpacked")) {
600                                 revs->unpacked = 1;
601                                 revs->limited = 1;
602                                 continue;
603                         }
604                         *unrecognized++ = arg;
605                         left++;
606                         continue;
607                 }
608                 dotdot = strstr(arg, "..");
609                 if (dotdot) {
610                         unsigned char from_sha1[20];
611                         char *next = dotdot + 2;
612                         *dotdot = 0;
613                         if (!*next)
614                                 next = "HEAD";
615                         if (!get_sha1(arg, from_sha1) && !get_sha1(next, sha1)) {
616                                 struct commit *exclude;
617                                 struct commit *include;
618
619                                 exclude = get_commit_reference(revs, arg, from_sha1, flags ^ UNINTERESTING);
620                                 include = get_commit_reference(revs, next, sha1, flags);
621                                 if (!exclude || !include)
622                                         die("Invalid revision range %s..%s", arg, next);
623                                 add_one_commit(exclude, revs);
624                                 add_one_commit(include, revs);
625                                 continue;
626                         }
627                         *dotdot = '.';
628                 }
629                 local_flags = 0;
630                 if (*arg == '^') {
631                         local_flags = UNINTERESTING;
632                         arg++;
633                 }
634                 if (get_sha1(arg, sha1) < 0) {
635                         struct stat st;
636                         int j;
637
638                         if (seen_dashdash || local_flags)
639                                 die("bad revision '%s'", arg);
640
641                         /* If we didn't have a "--", all filenames must exist */
642                         for (j = i; j < argc; j++) {
643                                 if (lstat(argv[j], &st) < 0)
644                                         die("'%s': %s", arg, strerror(errno));
645                         }
646                         revs->prune_data = get_pathspec(revs->prefix, argv + i);
647                         break;
648                 }
649                 commit = get_commit_reference(revs, arg, sha1, flags ^ local_flags);
650                 add_one_commit(commit, revs);
651         }
652         if (def && !revs->commits) {
653                 unsigned char sha1[20];
654                 struct commit *commit;
655                 if (get_sha1(def, sha1) < 0)
656                         die("bad default revision '%s'", def);
657                 commit = get_commit_reference(revs, def, sha1, 0);
658                 add_one_commit(commit, revs);
659         }
660
661         if (revs->prune_data) {
662                 diff_tree_setup_paths(revs->prune_data);
663                 revs->prune_fn = try_to_simplify_commit;
664                 revs->limited = 1;
665         }
666
667         return left;
668 }
669
670 void prepare_revision_walk(struct rev_info *revs)
671 {
672         sort_by_date(&revs->commits);
673         if (revs->limited)
674                 limit_list(revs);
675         if (revs->topo_order)
676                 sort_in_topological_order_fn(&revs->commits, revs->lifo,
677                                              revs->topo_setter,
678                                              revs->topo_getter);
679 }
680
681 static int rewrite_one(struct commit **pp)
682 {
683         for (;;) {
684                 struct commit *p = *pp;
685                 if (p->object.flags & (TREECHANGE | UNINTERESTING))
686                         return 0;
687                 if (!p->parents)
688                         return -1;
689                 *pp = p->parents->item;
690         }
691 }
692
693 static void rewrite_parents(struct commit *commit)
694 {
695         struct commit_list **pp = &commit->parents;
696         while (*pp) {
697                 struct commit_list *parent = *pp;
698                 if (rewrite_one(&parent->item) < 0) {
699                         *pp = parent->next;
700                         continue;
701                 }
702                 pp = &parent->next;
703         }
704 }
705
706 struct commit *get_revision(struct rev_info *revs)
707 {
708         struct commit_list *list = revs->commits;
709
710         if (!list)
711                 return NULL;
712
713         /* Check the max_count ... */
714         switch (revs->max_count) {
715         case -1:
716                 break;
717         case 0:
718                 return NULL;
719         default:
720                 revs->max_count--;
721         }
722
723         do {
724                 struct commit *commit = revs->commits->item;
725
726                 if (commit->object.flags & (UNINTERESTING|SHOWN))
727                         goto next;
728                 if (revs->min_age != -1 && (commit->date > revs->min_age))
729                         goto next;
730                 if (revs->max_age != -1 && (commit->date < revs->max_age))
731                         return NULL;
732                 if (revs->no_merges && commit->parents && commit->parents->next)
733                         goto next;
734                 if (revs->prune_fn && revs->dense) {
735                         if (!(commit->object.flags & TREECHANGE))
736                                 goto next;
737                         rewrite_parents(commit);
738                 }
739                 /* More to go? */
740                 if (revs->max_count)
741                         pop_most_recent_commit(&revs->commits, SEEN);
742                 commit->object.flags |= SHOWN;
743                 return commit;
744 next:
745                 pop_most_recent_commit(&revs->commits, SEEN);
746         } while (revs->commits);
747         return NULL;
748 }