cmd_diff(): use an object_array for holding trees
[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 #include "graph.h"
10 #include "grep.h"
11 #include "reflog-walk.h"
12 #include "patch-ids.h"
13 #include "decorate.h"
14 #include "log-tree.h"
15 #include "string-list.h"
16 #include "mailmap.h"
17
18 volatile show_early_output_fn_t show_early_output;
19
20 char *path_name(const struct name_path *path, const char *name)
21 {
22         const struct name_path *p;
23         char *n, *m;
24         int nlen = strlen(name);
25         int len = nlen + 1;
26
27         for (p = path; p; p = p->up) {
28                 if (p->elem_len)
29                         len += p->elem_len + 1;
30         }
31         n = xmalloc(len);
32         m = n + len - (nlen + 1);
33         strcpy(m, name);
34         for (p = path; p; p = p->up) {
35                 if (p->elem_len) {
36                         m -= p->elem_len + 1;
37                         memcpy(m, p->elem, p->elem_len);
38                         m[p->elem_len] = '/';
39                 }
40         }
41         return n;
42 }
43
44 static int show_path_component_truncated(FILE *out, const char *name, int len)
45 {
46         int cnt;
47         for (cnt = 0; cnt < len; cnt++) {
48                 int ch = name[cnt];
49                 if (!ch || ch == '\n')
50                         return -1;
51                 fputc(ch, out);
52         }
53         return len;
54 }
55
56 static int show_path_truncated(FILE *out, const struct name_path *path)
57 {
58         int emitted, ours;
59
60         if (!path)
61                 return 0;
62         emitted = show_path_truncated(out, path->up);
63         if (emitted < 0)
64                 return emitted;
65         if (emitted)
66                 fputc('/', out);
67         ours = show_path_component_truncated(out, path->elem, path->elem_len);
68         if (ours < 0)
69                 return ours;
70         return ours || emitted;
71 }
72
73 void show_object_with_name(FILE *out, struct object *obj, const struct name_path *path, const char *component)
74 {
75         struct name_path leaf;
76         leaf.up = (struct name_path *)path;
77         leaf.elem = component;
78         leaf.elem_len = strlen(component);
79
80         fprintf(out, "%s ", sha1_to_hex(obj->sha1));
81         show_path_truncated(out, &leaf);
82         fputc('\n', out);
83 }
84
85 void add_object(struct object *obj,
86                 struct object_array *p,
87                 struct name_path *path,
88                 const char *name)
89 {
90         add_object_array(obj, path_name(path, name), p);
91 }
92
93 static void mark_blob_uninteresting(struct blob *blob)
94 {
95         if (!blob)
96                 return;
97         if (blob->object.flags & UNINTERESTING)
98                 return;
99         blob->object.flags |= UNINTERESTING;
100 }
101
102 void mark_tree_uninteresting(struct tree *tree)
103 {
104         struct tree_desc desc;
105         struct name_entry entry;
106         struct object *obj = &tree->object;
107
108         if (!tree)
109                 return;
110         if (obj->flags & UNINTERESTING)
111                 return;
112         obj->flags |= UNINTERESTING;
113         if (!has_sha1_file(obj->sha1))
114                 return;
115         if (parse_tree(tree) < 0)
116                 die("bad tree %s", sha1_to_hex(obj->sha1));
117
118         init_tree_desc(&desc, tree->buffer, tree->size);
119         while (tree_entry(&desc, &entry)) {
120                 switch (object_type(entry.mode)) {
121                 case OBJ_TREE:
122                         mark_tree_uninteresting(lookup_tree(entry.sha1));
123                         break;
124                 case OBJ_BLOB:
125                         mark_blob_uninteresting(lookup_blob(entry.sha1));
126                         break;
127                 default:
128                         /* Subproject commit - not in this repository */
129                         break;
130                 }
131         }
132
133         /*
134          * We don't care about the tree any more
135          * after it has been marked uninteresting.
136          */
137         free(tree->buffer);
138         tree->buffer = NULL;
139 }
140
141 void mark_parents_uninteresting(struct commit *commit)
142 {
143         struct commit_list *parents = NULL, *l;
144
145         for (l = commit->parents; l; l = l->next)
146                 commit_list_insert(l->item, &parents);
147
148         while (parents) {
149                 struct commit *commit = parents->item;
150                 l = parents;
151                 parents = parents->next;
152                 free(l);
153
154                 while (commit) {
155                         /*
156                          * A missing commit is ok iff its parent is marked
157                          * uninteresting.
158                          *
159                          * We just mark such a thing parsed, so that when
160                          * it is popped next time around, we won't be trying
161                          * to parse it and get an error.
162                          */
163                         if (!has_sha1_file(commit->object.sha1))
164                                 commit->object.parsed = 1;
165
166                         if (commit->object.flags & UNINTERESTING)
167                                 break;
168
169                         commit->object.flags |= UNINTERESTING;
170
171                         /*
172                          * Normally we haven't parsed the parent
173                          * yet, so we won't have a parent of a parent
174                          * here. However, it may turn out that we've
175                          * reached this commit some other way (where it
176                          * wasn't uninteresting), in which case we need
177                          * to mark its parents recursively too..
178                          */
179                         if (!commit->parents)
180                                 break;
181
182                         for (l = commit->parents->next; l; l = l->next)
183                                 commit_list_insert(l->item, &parents);
184                         commit = commit->parents->item;
185                 }
186         }
187 }
188
189 static void add_pending_object_with_mode(struct rev_info *revs, struct object *obj, const char *name, unsigned mode)
190 {
191         if (!obj)
192                 return;
193         if (revs->no_walk && (obj->flags & UNINTERESTING))
194                 revs->no_walk = 0;
195         if (revs->reflog_info && obj->type == OBJ_COMMIT) {
196                 struct strbuf buf = STRBUF_INIT;
197                 int len = interpret_branch_name(name, &buf);
198                 int st;
199
200                 if (0 < len && name[len] && buf.len)
201                         strbuf_addstr(&buf, name + len);
202                 st = add_reflog_for_walk(revs->reflog_info,
203                                          (struct commit *)obj,
204                                          buf.buf[0] ? buf.buf: name);
205                 strbuf_release(&buf);
206                 if (st)
207                         return;
208         }
209         add_object_array_with_mode(obj, name, &revs->pending, mode);
210 }
211
212 void add_pending_object(struct rev_info *revs, struct object *obj, const char *name)
213 {
214         add_pending_object_with_mode(revs, obj, name, S_IFINVALID);
215 }
216
217 void add_head_to_pending(struct rev_info *revs)
218 {
219         unsigned char sha1[20];
220         struct object *obj;
221         if (get_sha1("HEAD", sha1))
222                 return;
223         obj = parse_object(sha1);
224         if (!obj)
225                 return;
226         add_pending_object(revs, obj, "HEAD");
227 }
228
229 static struct object *get_reference(struct rev_info *revs, const char *name, const unsigned char *sha1, unsigned int flags)
230 {
231         struct object *object;
232
233         object = parse_object(sha1);
234         if (!object) {
235                 if (revs->ignore_missing)
236                         return object;
237                 die("bad object %s", name);
238         }
239         object->flags |= flags;
240         return object;
241 }
242
243 void add_pending_sha1(struct rev_info *revs, const char *name,
244                       const unsigned char *sha1, unsigned int flags)
245 {
246         struct object *object = get_reference(revs, name, sha1, flags);
247         add_pending_object(revs, object, name);
248 }
249
250 static struct commit *handle_commit(struct rev_info *revs, struct object *object, const char *name)
251 {
252         unsigned long flags = object->flags;
253
254         /*
255          * Tag object? Look what it points to..
256          */
257         while (object->type == OBJ_TAG) {
258                 struct tag *tag = (struct tag *) object;
259                 if (revs->tag_objects && !(flags & UNINTERESTING))
260                         add_pending_object(revs, object, tag->tag);
261                 if (!tag->tagged)
262                         die("bad tag");
263                 object = parse_object(tag->tagged->sha1);
264                 if (!object) {
265                         if (flags & UNINTERESTING)
266                                 return NULL;
267                         die("bad object %s", sha1_to_hex(tag->tagged->sha1));
268                 }
269         }
270
271         /*
272          * Commit object? Just return it, we'll do all the complex
273          * reachability crud.
274          */
275         if (object->type == OBJ_COMMIT) {
276                 struct commit *commit = (struct commit *)object;
277                 if (parse_commit(commit) < 0)
278                         die("unable to parse commit %s", name);
279                 if (flags & UNINTERESTING) {
280                         commit->object.flags |= UNINTERESTING;
281                         mark_parents_uninteresting(commit);
282                         revs->limited = 1;
283                 }
284                 if (revs->show_source && !commit->util)
285                         commit->util = (void *) name;
286                 return commit;
287         }
288
289         /*
290          * Tree object? Either mark it uninteresting, or add it
291          * to the list of objects to look at later..
292          */
293         if (object->type == OBJ_TREE) {
294                 struct tree *tree = (struct tree *)object;
295                 if (!revs->tree_objects)
296                         return NULL;
297                 if (flags & UNINTERESTING) {
298                         mark_tree_uninteresting(tree);
299                         return NULL;
300                 }
301                 add_pending_object(revs, object, "");
302                 return NULL;
303         }
304
305         /*
306          * Blob object? You know the drill by now..
307          */
308         if (object->type == OBJ_BLOB) {
309                 struct blob *blob = (struct blob *)object;
310                 if (!revs->blob_objects)
311                         return NULL;
312                 if (flags & UNINTERESTING) {
313                         mark_blob_uninteresting(blob);
314                         return NULL;
315                 }
316                 add_pending_object(revs, object, "");
317                 return NULL;
318         }
319         die("%s is unknown object", name);
320 }
321
322 static int everybody_uninteresting(struct commit_list *orig)
323 {
324         struct commit_list *list = orig;
325         while (list) {
326                 struct commit *commit = list->item;
327                 list = list->next;
328                 if (commit->object.flags & UNINTERESTING)
329                         continue;
330                 return 0;
331         }
332         return 1;
333 }
334
335 /*
336  * The goal is to get REV_TREE_NEW as the result only if the
337  * diff consists of all '+' (and no other changes), REV_TREE_OLD
338  * if the whole diff is removal of old data, and otherwise
339  * REV_TREE_DIFFERENT (of course if the trees are the same we
340  * want REV_TREE_SAME).
341  * That means that once we get to REV_TREE_DIFFERENT, we do not
342  * have to look any further.
343  */
344 static int tree_difference = REV_TREE_SAME;
345
346 static void file_add_remove(struct diff_options *options,
347                     int addremove, unsigned mode,
348                     const unsigned char *sha1,
349                     int sha1_valid,
350                     const char *fullpath, unsigned dirty_submodule)
351 {
352         int diff = addremove == '+' ? REV_TREE_NEW : REV_TREE_OLD;
353
354         tree_difference |= diff;
355         if (tree_difference == REV_TREE_DIFFERENT)
356                 DIFF_OPT_SET(options, HAS_CHANGES);
357 }
358
359 static void file_change(struct diff_options *options,
360                  unsigned old_mode, unsigned new_mode,
361                  const unsigned char *old_sha1,
362                  const unsigned char *new_sha1,
363                  int old_sha1_valid, int new_sha1_valid,
364                  const char *fullpath,
365                  unsigned old_dirty_submodule, unsigned new_dirty_submodule)
366 {
367         tree_difference = REV_TREE_DIFFERENT;
368         DIFF_OPT_SET(options, HAS_CHANGES);
369 }
370
371 static int rev_compare_tree(struct rev_info *revs, struct commit *parent, struct commit *commit)
372 {
373         struct tree *t1 = parent->tree;
374         struct tree *t2 = commit->tree;
375
376         if (!t1)
377                 return REV_TREE_NEW;
378         if (!t2)
379                 return REV_TREE_OLD;
380
381         if (revs->simplify_by_decoration) {
382                 /*
383                  * If we are simplifying by decoration, then the commit
384                  * is worth showing if it has a tag pointing at it.
385                  */
386                 if (lookup_decoration(&name_decoration, &commit->object))
387                         return REV_TREE_DIFFERENT;
388                 /*
389                  * A commit that is not pointed by a tag is uninteresting
390                  * if we are not limited by path.  This means that you will
391                  * see the usual "commits that touch the paths" plus any
392                  * tagged commit by specifying both --simplify-by-decoration
393                  * and pathspec.
394                  */
395                 if (!revs->prune_data.nr)
396                         return REV_TREE_SAME;
397         }
398
399         tree_difference = REV_TREE_SAME;
400         DIFF_OPT_CLR(&revs->pruning, HAS_CHANGES);
401         if (diff_tree_sha1(t1->object.sha1, t2->object.sha1, "",
402                            &revs->pruning) < 0)
403                 return REV_TREE_DIFFERENT;
404         return tree_difference;
405 }
406
407 static int rev_same_tree_as_empty(struct rev_info *revs, struct commit *commit)
408 {
409         int retval;
410         void *tree;
411         unsigned long size;
412         struct tree_desc empty, real;
413         struct tree *t1 = commit->tree;
414
415         if (!t1)
416                 return 0;
417
418         tree = read_object_with_reference(t1->object.sha1, tree_type, &size, NULL);
419         if (!tree)
420                 return 0;
421         init_tree_desc(&real, tree, size);
422         init_tree_desc(&empty, "", 0);
423
424         tree_difference = REV_TREE_SAME;
425         DIFF_OPT_CLR(&revs->pruning, HAS_CHANGES);
426         retval = diff_tree(&empty, &real, "", &revs->pruning);
427         free(tree);
428
429         return retval >= 0 && (tree_difference == REV_TREE_SAME);
430 }
431
432 static void try_to_simplify_commit(struct rev_info *revs, struct commit *commit)
433 {
434         struct commit_list **pp, *parent;
435         int tree_changed = 0, tree_same = 0, nth_parent = 0;
436
437         /*
438          * If we don't do pruning, everything is interesting
439          */
440         if (!revs->prune)
441                 return;
442
443         if (!commit->tree)
444                 return;
445
446         if (!commit->parents) {
447                 if (rev_same_tree_as_empty(revs, commit))
448                         commit->object.flags |= TREESAME;
449                 return;
450         }
451
452         /*
453          * Normal non-merge commit? If we don't want to make the
454          * history dense, we consider it always to be a change..
455          */
456         if (!revs->dense && !commit->parents->next)
457                 return;
458
459         pp = &commit->parents;
460         while ((parent = *pp) != NULL) {
461                 struct commit *p = parent->item;
462
463                 /*
464                  * Do not compare with later parents when we care only about
465                  * the first parent chain, in order to avoid derailing the
466                  * traversal to follow a side branch that brought everything
467                  * in the path we are limited to by the pathspec.
468                  */
469                 if (revs->first_parent_only && nth_parent++)
470                         break;
471                 if (parse_commit(p) < 0)
472                         die("cannot simplify commit %s (because of %s)",
473                             sha1_to_hex(commit->object.sha1),
474                             sha1_to_hex(p->object.sha1));
475                 switch (rev_compare_tree(revs, p, commit)) {
476                 case REV_TREE_SAME:
477                         tree_same = 1;
478                         if (!revs->simplify_history || (p->object.flags & UNINTERESTING)) {
479                                 /* Even if a merge with an uninteresting
480                                  * side branch brought the entire change
481                                  * we are interested in, we do not want
482                                  * to lose the other branches of this
483                                  * merge, so we just keep going.
484                                  */
485                                 pp = &parent->next;
486                                 continue;
487                         }
488                         parent->next = NULL;
489                         commit->parents = parent;
490                         commit->object.flags |= TREESAME;
491                         return;
492
493                 case REV_TREE_NEW:
494                         if (revs->remove_empty_trees &&
495                             rev_same_tree_as_empty(revs, p)) {
496                                 /* We are adding all the specified
497                                  * paths from this parent, so the
498                                  * history beyond this parent is not
499                                  * interesting.  Remove its parents
500                                  * (they are grandparents for us).
501                                  * IOW, we pretend this parent is a
502                                  * "root" commit.
503                                  */
504                                 if (parse_commit(p) < 0)
505                                         die("cannot simplify commit %s (invalid %s)",
506                                             sha1_to_hex(commit->object.sha1),
507                                             sha1_to_hex(p->object.sha1));
508                                 p->parents = NULL;
509                         }
510                 /* fallthrough */
511                 case REV_TREE_OLD:
512                 case REV_TREE_DIFFERENT:
513                         tree_changed = 1;
514                         pp = &parent->next;
515                         continue;
516                 }
517                 die("bad tree compare for commit %s", sha1_to_hex(commit->object.sha1));
518         }
519         if (tree_changed && !tree_same)
520                 return;
521         commit->object.flags |= TREESAME;
522 }
523
524 static void commit_list_insert_by_date_cached(struct commit *p, struct commit_list **head,
525                     struct commit_list *cached_base, struct commit_list **cache)
526 {
527         struct commit_list *new_entry;
528
529         if (cached_base && p->date < cached_base->item->date)
530                 new_entry = commit_list_insert_by_date(p, &cached_base->next);
531         else
532                 new_entry = commit_list_insert_by_date(p, head);
533
534         if (cache && (!*cache || p->date < (*cache)->item->date))
535                 *cache = new_entry;
536 }
537
538 static int add_parents_to_list(struct rev_info *revs, struct commit *commit,
539                     struct commit_list **list, struct commit_list **cache_ptr)
540 {
541         struct commit_list *parent = commit->parents;
542         unsigned left_flag;
543         struct commit_list *cached_base = cache_ptr ? *cache_ptr : NULL;
544
545         if (commit->object.flags & ADDED)
546                 return 0;
547         commit->object.flags |= ADDED;
548
549         /*
550          * If the commit is uninteresting, don't try to
551          * prune parents - we want the maximal uninteresting
552          * set.
553          *
554          * Normally we haven't parsed the parent
555          * yet, so we won't have a parent of a parent
556          * here. However, it may turn out that we've
557          * reached this commit some other way (where it
558          * wasn't uninteresting), in which case we need
559          * to mark its parents recursively too..
560          */
561         if (commit->object.flags & UNINTERESTING) {
562                 while (parent) {
563                         struct commit *p = parent->item;
564                         parent = parent->next;
565                         if (p)
566                                 p->object.flags |= UNINTERESTING;
567                         if (parse_commit(p) < 0)
568                                 continue;
569                         if (p->parents)
570                                 mark_parents_uninteresting(p);
571                         if (p->object.flags & SEEN)
572                                 continue;
573                         p->object.flags |= SEEN;
574                         commit_list_insert_by_date_cached(p, list, cached_base, cache_ptr);
575                 }
576                 return 0;
577         }
578
579         /*
580          * Ok, the commit wasn't uninteresting. Try to
581          * simplify the commit history and find the parent
582          * that has no differences in the path set if one exists.
583          */
584         try_to_simplify_commit(revs, commit);
585
586         if (revs->no_walk)
587                 return 0;
588
589         left_flag = (commit->object.flags & SYMMETRIC_LEFT);
590
591         for (parent = commit->parents; parent; parent = parent->next) {
592                 struct commit *p = parent->item;
593
594                 if (parse_commit(p) < 0)
595                         return -1;
596                 if (revs->show_source && !p->util)
597                         p->util = commit->util;
598                 p->object.flags |= left_flag;
599                 if (!(p->object.flags & SEEN)) {
600                         p->object.flags |= SEEN;
601                         commit_list_insert_by_date_cached(p, list, cached_base, cache_ptr);
602                 }
603                 if (revs->first_parent_only)
604                         break;
605         }
606         return 0;
607 }
608
609 static void cherry_pick_list(struct commit_list *list, struct rev_info *revs)
610 {
611         struct commit_list *p;
612         int left_count = 0, right_count = 0;
613         int left_first;
614         struct patch_ids ids;
615         unsigned cherry_flag;
616
617         /* First count the commits on the left and on the right */
618         for (p = list; p; p = p->next) {
619                 struct commit *commit = p->item;
620                 unsigned flags = commit->object.flags;
621                 if (flags & BOUNDARY)
622                         ;
623                 else if (flags & SYMMETRIC_LEFT)
624                         left_count++;
625                 else
626                         right_count++;
627         }
628
629         if (!left_count || !right_count)
630                 return;
631
632         left_first = left_count < right_count;
633         init_patch_ids(&ids);
634         ids.diffopts.pathspec = revs->diffopt.pathspec;
635
636         /* Compute patch-ids for one side */
637         for (p = list; p; p = p->next) {
638                 struct commit *commit = p->item;
639                 unsigned flags = commit->object.flags;
640
641                 if (flags & BOUNDARY)
642                         continue;
643                 /*
644                  * If we have fewer left, left_first is set and we omit
645                  * commits on the right branch in this loop.  If we have
646                  * fewer right, we skip the left ones.
647                  */
648                 if (left_first != !!(flags & SYMMETRIC_LEFT))
649                         continue;
650                 commit->util = add_commit_patch_id(commit, &ids);
651         }
652
653         /* either cherry_mark or cherry_pick are true */
654         cherry_flag = revs->cherry_mark ? PATCHSAME : SHOWN;
655
656         /* Check the other side */
657         for (p = list; p; p = p->next) {
658                 struct commit *commit = p->item;
659                 struct patch_id *id;
660                 unsigned flags = commit->object.flags;
661
662                 if (flags & BOUNDARY)
663                         continue;
664                 /*
665                  * If we have fewer left, left_first is set and we omit
666                  * commits on the left branch in this loop.
667                  */
668                 if (left_first == !!(flags & SYMMETRIC_LEFT))
669                         continue;
670
671                 /*
672                  * Have we seen the same patch id?
673                  */
674                 id = has_commit_patch_id(commit, &ids);
675                 if (!id)
676                         continue;
677                 id->seen = 1;
678                 commit->object.flags |= cherry_flag;
679         }
680
681         /* Now check the original side for seen ones */
682         for (p = list; p; p = p->next) {
683                 struct commit *commit = p->item;
684                 struct patch_id *ent;
685
686                 ent = commit->util;
687                 if (!ent)
688                         continue;
689                 if (ent->seen)
690                         commit->object.flags |= cherry_flag;
691                 commit->util = NULL;
692         }
693
694         free_patch_ids(&ids);
695 }
696
697 /* How many extra uninteresting commits we want to see.. */
698 #define SLOP 5
699
700 static int still_interesting(struct commit_list *src, unsigned long date, int slop)
701 {
702         /*
703          * No source list at all? We're definitely done..
704          */
705         if (!src)
706                 return 0;
707
708         /*
709          * Does the destination list contain entries with a date
710          * before the source list? Definitely _not_ done.
711          */
712         if (date <= src->item->date)
713                 return SLOP;
714
715         /*
716          * Does the source list still have interesting commits in
717          * it? Definitely not done..
718          */
719         if (!everybody_uninteresting(src))
720                 return SLOP;
721
722         /* Ok, we're closing in.. */
723         return slop-1;
724 }
725
726 /*
727  * "rev-list --ancestry-path A..B" computes commits that are ancestors
728  * of B but not ancestors of A but further limits the result to those
729  * that are descendants of A.  This takes the list of bottom commits and
730  * the result of "A..B" without --ancestry-path, and limits the latter
731  * further to the ones that can reach one of the commits in "bottom".
732  */
733 static void limit_to_ancestry(struct commit_list *bottom, struct commit_list *list)
734 {
735         struct commit_list *p;
736         struct commit_list *rlist = NULL;
737         int made_progress;
738
739         /*
740          * Reverse the list so that it will be likely that we would
741          * process parents before children.
742          */
743         for (p = list; p; p = p->next)
744                 commit_list_insert(p->item, &rlist);
745
746         for (p = bottom; p; p = p->next)
747                 p->item->object.flags |= TMP_MARK;
748
749         /*
750          * Mark the ones that can reach bottom commits in "list",
751          * in a bottom-up fashion.
752          */
753         do {
754                 made_progress = 0;
755                 for (p = rlist; p; p = p->next) {
756                         struct commit *c = p->item;
757                         struct commit_list *parents;
758                         if (c->object.flags & (TMP_MARK | UNINTERESTING))
759                                 continue;
760                         for (parents = c->parents;
761                              parents;
762                              parents = parents->next) {
763                                 if (!(parents->item->object.flags & TMP_MARK))
764                                         continue;
765                                 c->object.flags |= TMP_MARK;
766                                 made_progress = 1;
767                                 break;
768                         }
769                 }
770         } while (made_progress);
771
772         /*
773          * NEEDSWORK: decide if we want to remove parents that are
774          * not marked with TMP_MARK from commit->parents for commits
775          * in the resulting list.  We may not want to do that, though.
776          */
777
778         /*
779          * The ones that are not marked with TMP_MARK are uninteresting
780          */
781         for (p = list; p; p = p->next) {
782                 struct commit *c = p->item;
783                 if (c->object.flags & TMP_MARK)
784                         continue;
785                 c->object.flags |= UNINTERESTING;
786         }
787
788         /* We are done with the TMP_MARK */
789         for (p = list; p; p = p->next)
790                 p->item->object.flags &= ~TMP_MARK;
791         for (p = bottom; p; p = p->next)
792                 p->item->object.flags &= ~TMP_MARK;
793         free_commit_list(rlist);
794 }
795
796 /*
797  * Before walking the history, keep the set of "negative" refs the
798  * caller has asked to exclude.
799  *
800  * This is used to compute "rev-list --ancestry-path A..B", as we need
801  * to filter the result of "A..B" further to the ones that can actually
802  * reach A.
803  */
804 static struct commit_list *collect_bottom_commits(struct rev_info *revs)
805 {
806         struct commit_list *bottom = NULL;
807         int i;
808         for (i = 0; i < revs->cmdline.nr; i++) {
809                 struct rev_cmdline_entry *elem = &revs->cmdline.rev[i];
810                 if ((elem->flags & UNINTERESTING) &&
811                     elem->item->type == OBJ_COMMIT)
812                         commit_list_insert((struct commit *)elem->item, &bottom);
813         }
814         return bottom;
815 }
816
817 /* Assumes either left_only or right_only is set */
818 static void limit_left_right(struct commit_list *list, struct rev_info *revs)
819 {
820         struct commit_list *p;
821
822         for (p = list; p; p = p->next) {
823                 struct commit *commit = p->item;
824
825                 if (revs->right_only) {
826                         if (commit->object.flags & SYMMETRIC_LEFT)
827                                 commit->object.flags |= SHOWN;
828                 } else  /* revs->left_only is set */
829                         if (!(commit->object.flags & SYMMETRIC_LEFT))
830                                 commit->object.flags |= SHOWN;
831         }
832 }
833
834 static int limit_list(struct rev_info *revs)
835 {
836         int slop = SLOP;
837         unsigned long date = ~0ul;
838         struct commit_list *list = revs->commits;
839         struct commit_list *newlist = NULL;
840         struct commit_list **p = &newlist;
841         struct commit_list *bottom = NULL;
842
843         if (revs->ancestry_path) {
844                 bottom = collect_bottom_commits(revs);
845                 if (!bottom)
846                         die("--ancestry-path given but there are no bottom commits");
847         }
848
849         while (list) {
850                 struct commit_list *entry = list;
851                 struct commit *commit = list->item;
852                 struct object *obj = &commit->object;
853                 show_early_output_fn_t show;
854
855                 list = list->next;
856                 free(entry);
857
858                 if (revs->max_age != -1 && (commit->date < revs->max_age))
859                         obj->flags |= UNINTERESTING;
860                 if (add_parents_to_list(revs, commit, &list, NULL) < 0)
861                         return -1;
862                 if (obj->flags & UNINTERESTING) {
863                         mark_parents_uninteresting(commit);
864                         if (revs->show_all)
865                                 p = &commit_list_insert(commit, p)->next;
866                         slop = still_interesting(list, date, slop);
867                         if (slop)
868                                 continue;
869                         /* If showing all, add the whole pending list to the end */
870                         if (revs->show_all)
871                                 *p = list;
872                         break;
873                 }
874                 if (revs->min_age != -1 && (commit->date > revs->min_age))
875                         continue;
876                 date = commit->date;
877                 p = &commit_list_insert(commit, p)->next;
878
879                 show = show_early_output;
880                 if (!show)
881                         continue;
882
883                 show(revs, newlist);
884                 show_early_output = NULL;
885         }
886         if (revs->cherry_pick || revs->cherry_mark)
887                 cherry_pick_list(newlist, revs);
888
889         if (revs->left_only || revs->right_only)
890                 limit_left_right(newlist, revs);
891
892         if (bottom) {
893                 limit_to_ancestry(bottom, newlist);
894                 free_commit_list(bottom);
895         }
896
897         revs->commits = newlist;
898         return 0;
899 }
900
901 /*
902  * Add an entry to refs->cmdline with the specified information.
903  * *name is copied.
904  */
905 static void add_rev_cmdline(struct rev_info *revs,
906                             struct object *item,
907                             const char *name,
908                             int whence,
909                             unsigned flags)
910 {
911         struct rev_cmdline_info *info = &revs->cmdline;
912         int nr = info->nr;
913
914         ALLOC_GROW(info->rev, nr + 1, info->alloc);
915         info->rev[nr].item = item;
916         info->rev[nr].name = xstrdup(name);
917         info->rev[nr].whence = whence;
918         info->rev[nr].flags = flags;
919         info->nr++;
920 }
921
922 struct all_refs_cb {
923         int all_flags;
924         int warned_bad_reflog;
925         struct rev_info *all_revs;
926         const char *name_for_errormsg;
927 };
928
929 static int handle_one_ref(const char *path, const unsigned char *sha1, int flag, void *cb_data)
930 {
931         struct all_refs_cb *cb = cb_data;
932         struct object *object = get_reference(cb->all_revs, path, sha1,
933                                               cb->all_flags);
934         add_rev_cmdline(cb->all_revs, object, path, REV_CMD_REF, cb->all_flags);
935         add_pending_sha1(cb->all_revs, path, sha1, cb->all_flags);
936         return 0;
937 }
938
939 static void init_all_refs_cb(struct all_refs_cb *cb, struct rev_info *revs,
940         unsigned flags)
941 {
942         cb->all_revs = revs;
943         cb->all_flags = flags;
944 }
945
946 static void handle_refs(const char *submodule, struct rev_info *revs, unsigned flags,
947                 int (*for_each)(const char *, each_ref_fn, void *))
948 {
949         struct all_refs_cb cb;
950         init_all_refs_cb(&cb, revs, flags);
951         for_each(submodule, handle_one_ref, &cb);
952 }
953
954 static void handle_one_reflog_commit(unsigned char *sha1, void *cb_data)
955 {
956         struct all_refs_cb *cb = cb_data;
957         if (!is_null_sha1(sha1)) {
958                 struct object *o = parse_object(sha1);
959                 if (o) {
960                         o->flags |= cb->all_flags;
961                         /* ??? CMDLINEFLAGS ??? */
962                         add_pending_object(cb->all_revs, o, "");
963                 }
964                 else if (!cb->warned_bad_reflog) {
965                         warning("reflog of '%s' references pruned commits",
966                                 cb->name_for_errormsg);
967                         cb->warned_bad_reflog = 1;
968                 }
969         }
970 }
971
972 static int handle_one_reflog_ent(unsigned char *osha1, unsigned char *nsha1,
973                 const char *email, unsigned long timestamp, int tz,
974                 const char *message, void *cb_data)
975 {
976         handle_one_reflog_commit(osha1, cb_data);
977         handle_one_reflog_commit(nsha1, cb_data);
978         return 0;
979 }
980
981 static int handle_one_reflog(const char *path, const unsigned char *sha1, int flag, void *cb_data)
982 {
983         struct all_refs_cb *cb = cb_data;
984         cb->warned_bad_reflog = 0;
985         cb->name_for_errormsg = path;
986         for_each_reflog_ent(path, handle_one_reflog_ent, cb_data);
987         return 0;
988 }
989
990 static void handle_reflog(struct rev_info *revs, unsigned flags)
991 {
992         struct all_refs_cb cb;
993         cb.all_revs = revs;
994         cb.all_flags = flags;
995         for_each_reflog(handle_one_reflog, &cb);
996 }
997
998 static int add_parents_only(struct rev_info *revs, const char *arg_, int flags)
999 {
1000         unsigned char sha1[20];
1001         struct object *it;
1002         struct commit *commit;
1003         struct commit_list *parents;
1004         const char *arg = arg_;
1005
1006         if (*arg == '^') {
1007                 flags ^= UNINTERESTING;
1008                 arg++;
1009         }
1010         if (get_sha1_committish(arg, sha1))
1011                 return 0;
1012         while (1) {
1013                 it = get_reference(revs, arg, sha1, 0);
1014                 if (!it && revs->ignore_missing)
1015                         return 0;
1016                 if (it->type != OBJ_TAG)
1017                         break;
1018                 if (!((struct tag*)it)->tagged)
1019                         return 0;
1020                 hashcpy(sha1, ((struct tag*)it)->tagged->sha1);
1021         }
1022         if (it->type != OBJ_COMMIT)
1023                 return 0;
1024         commit = (struct commit *)it;
1025         for (parents = commit->parents; parents; parents = parents->next) {
1026                 it = &parents->item->object;
1027                 it->flags |= flags;
1028                 add_rev_cmdline(revs, it, arg_, REV_CMD_PARENTS_ONLY, flags);
1029                 add_pending_object(revs, it, arg);
1030         }
1031         return 1;
1032 }
1033
1034 void init_revisions(struct rev_info *revs, const char *prefix)
1035 {
1036         memset(revs, 0, sizeof(*revs));
1037
1038         revs->abbrev = DEFAULT_ABBREV;
1039         revs->ignore_merges = 1;
1040         revs->simplify_history = 1;
1041         DIFF_OPT_SET(&revs->pruning, RECURSIVE);
1042         DIFF_OPT_SET(&revs->pruning, QUICK);
1043         revs->pruning.add_remove = file_add_remove;
1044         revs->pruning.change = file_change;
1045         revs->lifo = 1;
1046         revs->dense = 1;
1047         revs->prefix = prefix;
1048         revs->max_age = -1;
1049         revs->min_age = -1;
1050         revs->skip_count = -1;
1051         revs->max_count = -1;
1052         revs->max_parents = -1;
1053
1054         revs->commit_format = CMIT_FMT_DEFAULT;
1055
1056         init_grep_defaults();
1057         grep_init(&revs->grep_filter, prefix);
1058         revs->grep_filter.status_only = 1;
1059         revs->grep_filter.regflags = REG_NEWLINE;
1060
1061         diff_setup(&revs->diffopt);
1062         if (prefix && !revs->diffopt.prefix) {
1063                 revs->diffopt.prefix = prefix;
1064                 revs->diffopt.prefix_length = strlen(prefix);
1065         }
1066
1067         revs->notes_opt.use_default_notes = -1;
1068 }
1069
1070 static void add_pending_commit_list(struct rev_info *revs,
1071                                     struct commit_list *commit_list,
1072                                     unsigned int flags)
1073 {
1074         while (commit_list) {
1075                 struct object *object = &commit_list->item->object;
1076                 object->flags |= flags;
1077                 add_pending_object(revs, object, sha1_to_hex(object->sha1));
1078                 commit_list = commit_list->next;
1079         }
1080 }
1081
1082 static void prepare_show_merge(struct rev_info *revs)
1083 {
1084         struct commit_list *bases;
1085         struct commit *head, *other;
1086         unsigned char sha1[20];
1087         const char **prune = NULL;
1088         int i, prune_num = 1; /* counting terminating NULL */
1089
1090         if (get_sha1("HEAD", sha1))
1091                 die("--merge without HEAD?");
1092         head = lookup_commit_or_die(sha1, "HEAD");
1093         if (get_sha1("MERGE_HEAD", sha1))
1094                 die("--merge without MERGE_HEAD?");
1095         other = lookup_commit_or_die(sha1, "MERGE_HEAD");
1096         add_pending_object(revs, &head->object, "HEAD");
1097         add_pending_object(revs, &other->object, "MERGE_HEAD");
1098         bases = get_merge_bases(head, other, 1);
1099         add_pending_commit_list(revs, bases, UNINTERESTING);
1100         free_commit_list(bases);
1101         head->object.flags |= SYMMETRIC_LEFT;
1102
1103         if (!active_nr)
1104                 read_cache();
1105         for (i = 0; i < active_nr; i++) {
1106                 struct cache_entry *ce = active_cache[i];
1107                 if (!ce_stage(ce))
1108                         continue;
1109                 if (ce_path_match(ce, &revs->prune_data)) {
1110                         prune_num++;
1111                         prune = xrealloc(prune, sizeof(*prune) * prune_num);
1112                         prune[prune_num-2] = ce->name;
1113                         prune[prune_num-1] = NULL;
1114                 }
1115                 while ((i+1 < active_nr) &&
1116                        ce_same_name(ce, active_cache[i+1]))
1117                         i++;
1118         }
1119         free_pathspec(&revs->prune_data);
1120         init_pathspec(&revs->prune_data, prune);
1121         revs->limited = 1;
1122 }
1123
1124 int handle_revision_arg(const char *arg_, struct rev_info *revs, int flags, unsigned revarg_opt)
1125 {
1126         struct object_context oc;
1127         char *dotdot;
1128         struct object *object;
1129         unsigned char sha1[20];
1130         int local_flags;
1131         const char *arg = arg_;
1132         int cant_be_filename = revarg_opt & REVARG_CANNOT_BE_FILENAME;
1133         unsigned get_sha1_flags = 0;
1134
1135         dotdot = strstr(arg, "..");
1136         if (dotdot) {
1137                 unsigned char from_sha1[20];
1138                 const char *next = dotdot + 2;
1139                 const char *this = arg;
1140                 int symmetric = *next == '.';
1141                 unsigned int flags_exclude = flags ^ UNINTERESTING;
1142                 static const char head_by_default[] = "HEAD";
1143                 unsigned int a_flags;
1144
1145                 *dotdot = 0;
1146                 next += symmetric;
1147
1148                 if (!*next)
1149                         next = head_by_default;
1150                 if (dotdot == arg)
1151                         this = head_by_default;
1152                 if (this == head_by_default && next == head_by_default &&
1153                     !symmetric) {
1154                         /*
1155                          * Just ".."?  That is not a range but the
1156                          * pathspec for the parent directory.
1157                          */
1158                         if (!cant_be_filename) {
1159                                 *dotdot = '.';
1160                                 return -1;
1161                         }
1162                 }
1163                 if (!get_sha1_committish(this, from_sha1) &&
1164                     !get_sha1_committish(next, sha1)) {
1165                         struct commit *a, *b;
1166                         struct commit_list *exclude;
1167
1168                         a = lookup_commit_reference(from_sha1);
1169                         b = lookup_commit_reference(sha1);
1170                         if (!a || !b) {
1171                                 if (revs->ignore_missing)
1172                                         return 0;
1173                                 die(symmetric ?
1174                                     "Invalid symmetric difference expression %s...%s" :
1175                                     "Invalid revision range %s..%s",
1176                                     arg, next);
1177                         }
1178
1179                         if (!cant_be_filename) {
1180                                 *dotdot = '.';
1181                                 verify_non_filename(revs->prefix, arg);
1182                         }
1183
1184                         if (symmetric) {
1185                                 exclude = get_merge_bases(a, b, 1);
1186                                 add_pending_commit_list(revs, exclude,
1187                                                         flags_exclude);
1188                                 free_commit_list(exclude);
1189                                 a_flags = flags | SYMMETRIC_LEFT;
1190                         } else
1191                                 a_flags = flags_exclude;
1192                         a->object.flags |= a_flags;
1193                         b->object.flags |= flags;
1194                         add_rev_cmdline(revs, &a->object, this,
1195                                         REV_CMD_LEFT, a_flags);
1196                         add_rev_cmdline(revs, &b->object, next,
1197                                         REV_CMD_RIGHT, flags);
1198                         add_pending_object(revs, &a->object, this);
1199                         add_pending_object(revs, &b->object, next);
1200                         return 0;
1201                 }
1202                 *dotdot = '.';
1203         }
1204         dotdot = strstr(arg, "^@");
1205         if (dotdot && !dotdot[2]) {
1206                 *dotdot = 0;
1207                 if (add_parents_only(revs, arg, flags))
1208                         return 0;
1209                 *dotdot = '^';
1210         }
1211         dotdot = strstr(arg, "^!");
1212         if (dotdot && !dotdot[2]) {
1213                 *dotdot = 0;
1214                 if (!add_parents_only(revs, arg, flags ^ UNINTERESTING))
1215                         *dotdot = '^';
1216         }
1217
1218         local_flags = 0;
1219         if (*arg == '^') {
1220                 local_flags = UNINTERESTING;
1221                 arg++;
1222         }
1223
1224         if (revarg_opt & REVARG_COMMITTISH)
1225                 get_sha1_flags = GET_SHA1_COMMITTISH;
1226
1227         if (get_sha1_with_context(arg, get_sha1_flags, sha1, &oc))
1228                 return revs->ignore_missing ? 0 : -1;
1229         if (!cant_be_filename)
1230                 verify_non_filename(revs->prefix, arg);
1231         object = get_reference(revs, arg, sha1, flags ^ local_flags);
1232         add_rev_cmdline(revs, object, arg_, REV_CMD_REV, flags ^ local_flags);
1233         add_pending_object_with_mode(revs, object, arg, oc.mode);
1234         return 0;
1235 }
1236
1237 struct cmdline_pathspec {
1238         int alloc;
1239         int nr;
1240         const char **path;
1241 };
1242
1243 static void append_prune_data(struct cmdline_pathspec *prune, const char **av)
1244 {
1245         while (*av) {
1246                 ALLOC_GROW(prune->path, prune->nr+1, prune->alloc);
1247                 prune->path[prune->nr++] = *(av++);
1248         }
1249 }
1250
1251 static void read_pathspec_from_stdin(struct rev_info *revs, struct strbuf *sb,
1252                                      struct cmdline_pathspec *prune)
1253 {
1254         while (strbuf_getwholeline(sb, stdin, '\n') != EOF) {
1255                 int len = sb->len;
1256                 if (len && sb->buf[len - 1] == '\n')
1257                         sb->buf[--len] = '\0';
1258                 ALLOC_GROW(prune->path, prune->nr+1, prune->alloc);
1259                 prune->path[prune->nr++] = xstrdup(sb->buf);
1260         }
1261 }
1262
1263 static void read_revisions_from_stdin(struct rev_info *revs,
1264                                       struct cmdline_pathspec *prune)
1265 {
1266         struct strbuf sb;
1267         int seen_dashdash = 0;
1268
1269         strbuf_init(&sb, 1000);
1270         while (strbuf_getwholeline(&sb, stdin, '\n') != EOF) {
1271                 int len = sb.len;
1272                 if (len && sb.buf[len - 1] == '\n')
1273                         sb.buf[--len] = '\0';
1274                 if (!len)
1275                         break;
1276                 if (sb.buf[0] == '-') {
1277                         if (len == 2 && sb.buf[1] == '-') {
1278                                 seen_dashdash = 1;
1279                                 break;
1280                         }
1281                         die("options not supported in --stdin mode");
1282                 }
1283                 if (handle_revision_arg(xstrdup(sb.buf), revs, 0,
1284                                         REVARG_CANNOT_BE_FILENAME))
1285                         die("bad revision '%s'", sb.buf);
1286         }
1287         if (seen_dashdash)
1288                 read_pathspec_from_stdin(revs, &sb, prune);
1289         strbuf_release(&sb);
1290 }
1291
1292 static void add_grep(struct rev_info *revs, const char *ptn, enum grep_pat_token what)
1293 {
1294         append_grep_pattern(&revs->grep_filter, ptn, "command line", 0, what);
1295 }
1296
1297 static void add_header_grep(struct rev_info *revs, enum grep_header_field field, const char *pattern)
1298 {
1299         append_header_grep_pattern(&revs->grep_filter, field, pattern);
1300 }
1301
1302 static void add_message_grep(struct rev_info *revs, const char *pattern)
1303 {
1304         add_grep(revs, pattern, GREP_PATTERN_BODY);
1305 }
1306
1307 static int handle_revision_opt(struct rev_info *revs, int argc, const char **argv,
1308                                int *unkc, const char **unkv)
1309 {
1310         const char *arg = argv[0];
1311         const char *optarg;
1312         int argcount;
1313
1314         /* pseudo revision arguments */
1315         if (!strcmp(arg, "--all") || !strcmp(arg, "--branches") ||
1316             !strcmp(arg, "--tags") || !strcmp(arg, "--remotes") ||
1317             !strcmp(arg, "--reflog") || !strcmp(arg, "--not") ||
1318             !strcmp(arg, "--no-walk") || !strcmp(arg, "--do-walk") ||
1319             !strcmp(arg, "--bisect") || !prefixcmp(arg, "--glob=") ||
1320             !prefixcmp(arg, "--branches=") || !prefixcmp(arg, "--tags=") ||
1321             !prefixcmp(arg, "--remotes=") || !prefixcmp(arg, "--no-walk="))
1322         {
1323                 unkv[(*unkc)++] = arg;
1324                 return 1;
1325         }
1326
1327         if ((argcount = parse_long_opt("max-count", argv, &optarg))) {
1328                 revs->max_count = atoi(optarg);
1329                 revs->no_walk = 0;
1330                 return argcount;
1331         } else if ((argcount = parse_long_opt("skip", argv, &optarg))) {
1332                 revs->skip_count = atoi(optarg);
1333                 return argcount;
1334         } else if ((*arg == '-') && isdigit(arg[1])) {
1335         /* accept -<digit>, like traditional "head" */
1336                 revs->max_count = atoi(arg + 1);
1337                 revs->no_walk = 0;
1338         } else if (!strcmp(arg, "-n")) {
1339                 if (argc <= 1)
1340                         return error("-n requires an argument");
1341                 revs->max_count = atoi(argv[1]);
1342                 revs->no_walk = 0;
1343                 return 2;
1344         } else if (!prefixcmp(arg, "-n")) {
1345                 revs->max_count = atoi(arg + 2);
1346                 revs->no_walk = 0;
1347         } else if ((argcount = parse_long_opt("max-age", argv, &optarg))) {
1348                 revs->max_age = atoi(optarg);
1349                 return argcount;
1350         } else if ((argcount = parse_long_opt("since", argv, &optarg))) {
1351                 revs->max_age = approxidate(optarg);
1352                 return argcount;
1353         } else if ((argcount = parse_long_opt("after", argv, &optarg))) {
1354                 revs->max_age = approxidate(optarg);
1355                 return argcount;
1356         } else if ((argcount = parse_long_opt("min-age", argv, &optarg))) {
1357                 revs->min_age = atoi(optarg);
1358                 return argcount;
1359         } else if ((argcount = parse_long_opt("before", argv, &optarg))) {
1360                 revs->min_age = approxidate(optarg);
1361                 return argcount;
1362         } else if ((argcount = parse_long_opt("until", argv, &optarg))) {
1363                 revs->min_age = approxidate(optarg);
1364                 return argcount;
1365         } else if (!strcmp(arg, "--first-parent")) {
1366                 revs->first_parent_only = 1;
1367         } else if (!strcmp(arg, "--ancestry-path")) {
1368                 revs->ancestry_path = 1;
1369                 revs->simplify_history = 0;
1370                 revs->limited = 1;
1371         } else if (!strcmp(arg, "-g") || !strcmp(arg, "--walk-reflogs")) {
1372                 init_reflog_walk(&revs->reflog_info);
1373         } else if (!strcmp(arg, "--default")) {
1374                 if (argc <= 1)
1375                         return error("bad --default argument");
1376                 revs->def = argv[1];
1377                 return 2;
1378         } else if (!strcmp(arg, "--merge")) {
1379                 revs->show_merge = 1;
1380         } else if (!strcmp(arg, "--topo-order")) {
1381                 revs->lifo = 1;
1382                 revs->topo_order = 1;
1383         } else if (!strcmp(arg, "--simplify-merges")) {
1384                 revs->simplify_merges = 1;
1385                 revs->topo_order = 1;
1386                 revs->rewrite_parents = 1;
1387                 revs->simplify_history = 0;
1388                 revs->limited = 1;
1389         } else if (!strcmp(arg, "--simplify-by-decoration")) {
1390                 revs->simplify_merges = 1;
1391                 revs->topo_order = 1;
1392                 revs->rewrite_parents = 1;
1393                 revs->simplify_history = 0;
1394                 revs->simplify_by_decoration = 1;
1395                 revs->limited = 1;
1396                 revs->prune = 1;
1397                 load_ref_decorations(DECORATE_SHORT_REFS);
1398         } else if (!strcmp(arg, "--date-order")) {
1399                 revs->lifo = 0;
1400                 revs->topo_order = 1;
1401         } else if (!prefixcmp(arg, "--early-output")) {
1402                 int count = 100;
1403                 switch (arg[14]) {
1404                 case '=':
1405                         count = atoi(arg+15);
1406                         /* Fallthrough */
1407                 case 0:
1408                         revs->topo_order = 1;
1409                        revs->early_output = count;
1410                 }
1411         } else if (!strcmp(arg, "--parents")) {
1412                 revs->rewrite_parents = 1;
1413                 revs->print_parents = 1;
1414         } else if (!strcmp(arg, "--dense")) {
1415                 revs->dense = 1;
1416         } else if (!strcmp(arg, "--sparse")) {
1417                 revs->dense = 0;
1418         } else if (!strcmp(arg, "--show-all")) {
1419                 revs->show_all = 1;
1420         } else if (!strcmp(arg, "--remove-empty")) {
1421                 revs->remove_empty_trees = 1;
1422         } else if (!strcmp(arg, "--merges")) {
1423                 revs->min_parents = 2;
1424         } else if (!strcmp(arg, "--no-merges")) {
1425                 revs->max_parents = 1;
1426         } else if (!prefixcmp(arg, "--min-parents=")) {
1427                 revs->min_parents = atoi(arg+14);
1428         } else if (!prefixcmp(arg, "--no-min-parents")) {
1429                 revs->min_parents = 0;
1430         } else if (!prefixcmp(arg, "--max-parents=")) {
1431                 revs->max_parents = atoi(arg+14);
1432         } else if (!prefixcmp(arg, "--no-max-parents")) {
1433                 revs->max_parents = -1;
1434         } else if (!strcmp(arg, "--boundary")) {
1435                 revs->boundary = 1;
1436         } else if (!strcmp(arg, "--left-right")) {
1437                 revs->left_right = 1;
1438         } else if (!strcmp(arg, "--left-only")) {
1439                 if (revs->right_only)
1440                         die("--left-only is incompatible with --right-only"
1441                             " or --cherry");
1442                 revs->left_only = 1;
1443         } else if (!strcmp(arg, "--right-only")) {
1444                 if (revs->left_only)
1445                         die("--right-only is incompatible with --left-only");
1446                 revs->right_only = 1;
1447         } else if (!strcmp(arg, "--cherry")) {
1448                 if (revs->left_only)
1449                         die("--cherry is incompatible with --left-only");
1450                 revs->cherry_mark = 1;
1451                 revs->right_only = 1;
1452                 revs->max_parents = 1;
1453                 revs->limited = 1;
1454         } else if (!strcmp(arg, "--count")) {
1455                 revs->count = 1;
1456         } else if (!strcmp(arg, "--cherry-mark")) {
1457                 if (revs->cherry_pick)
1458                         die("--cherry-mark is incompatible with --cherry-pick");
1459                 revs->cherry_mark = 1;
1460                 revs->limited = 1; /* needs limit_list() */
1461         } else if (!strcmp(arg, "--cherry-pick")) {
1462                 if (revs->cherry_mark)
1463                         die("--cherry-pick is incompatible with --cherry-mark");
1464                 revs->cherry_pick = 1;
1465                 revs->limited = 1;
1466         } else if (!strcmp(arg, "--objects")) {
1467                 revs->tag_objects = 1;
1468                 revs->tree_objects = 1;
1469                 revs->blob_objects = 1;
1470         } else if (!strcmp(arg, "--objects-edge")) {
1471                 revs->tag_objects = 1;
1472                 revs->tree_objects = 1;
1473                 revs->blob_objects = 1;
1474                 revs->edge_hint = 1;
1475         } else if (!strcmp(arg, "--verify-objects")) {
1476                 revs->tag_objects = 1;
1477                 revs->tree_objects = 1;
1478                 revs->blob_objects = 1;
1479                 revs->verify_objects = 1;
1480         } else if (!strcmp(arg, "--unpacked")) {
1481                 revs->unpacked = 1;
1482         } else if (!prefixcmp(arg, "--unpacked=")) {
1483                 die("--unpacked=<packfile> no longer supported.");
1484         } else if (!strcmp(arg, "-r")) {
1485                 revs->diff = 1;
1486                 DIFF_OPT_SET(&revs->diffopt, RECURSIVE);
1487         } else if (!strcmp(arg, "-t")) {
1488                 revs->diff = 1;
1489                 DIFF_OPT_SET(&revs->diffopt, RECURSIVE);
1490                 DIFF_OPT_SET(&revs->diffopt, TREE_IN_RECURSIVE);
1491         } else if (!strcmp(arg, "-m")) {
1492                 revs->ignore_merges = 0;
1493         } else if (!strcmp(arg, "-c")) {
1494                 revs->diff = 1;
1495                 revs->dense_combined_merges = 0;
1496                 revs->combine_merges = 1;
1497         } else if (!strcmp(arg, "--cc")) {
1498                 revs->diff = 1;
1499                 revs->dense_combined_merges = 1;
1500                 revs->combine_merges = 1;
1501         } else if (!strcmp(arg, "-v")) {
1502                 revs->verbose_header = 1;
1503         } else if (!strcmp(arg, "--pretty")) {
1504                 revs->verbose_header = 1;
1505                 revs->pretty_given = 1;
1506                 get_commit_format(arg+8, revs);
1507         } else if (!prefixcmp(arg, "--pretty=") || !prefixcmp(arg, "--format=")) {
1508                 /*
1509                  * Detached form ("--pretty X" as opposed to "--pretty=X")
1510                  * not allowed, since the argument is optional.
1511                  */
1512                 revs->verbose_header = 1;
1513                 revs->pretty_given = 1;
1514                 get_commit_format(arg+9, revs);
1515         } else if (!strcmp(arg, "--show-notes") || !strcmp(arg, "--notes")) {
1516                 revs->show_notes = 1;
1517                 revs->show_notes_given = 1;
1518                 revs->notes_opt.use_default_notes = 1;
1519         } else if (!strcmp(arg, "--show-signature")) {
1520                 revs->show_signature = 1;
1521         } else if (!prefixcmp(arg, "--show-notes=") ||
1522                    !prefixcmp(arg, "--notes=")) {
1523                 struct strbuf buf = STRBUF_INIT;
1524                 revs->show_notes = 1;
1525                 revs->show_notes_given = 1;
1526                 if (!prefixcmp(arg, "--show-notes")) {
1527                         if (revs->notes_opt.use_default_notes < 0)
1528                                 revs->notes_opt.use_default_notes = 1;
1529                         strbuf_addstr(&buf, arg+13);
1530                 }
1531                 else
1532                         strbuf_addstr(&buf, arg+8);
1533                 expand_notes_ref(&buf);
1534                 string_list_append(&revs->notes_opt.extra_notes_refs,
1535                                    strbuf_detach(&buf, NULL));
1536         } else if (!strcmp(arg, "--no-notes")) {
1537                 revs->show_notes = 0;
1538                 revs->show_notes_given = 1;
1539                 revs->notes_opt.use_default_notes = -1;
1540                 /* we have been strdup'ing ourselves, so trick
1541                  * string_list into free()ing strings */
1542                 revs->notes_opt.extra_notes_refs.strdup_strings = 1;
1543                 string_list_clear(&revs->notes_opt.extra_notes_refs, 0);
1544                 revs->notes_opt.extra_notes_refs.strdup_strings = 0;
1545         } else if (!strcmp(arg, "--standard-notes")) {
1546                 revs->show_notes_given = 1;
1547                 revs->notes_opt.use_default_notes = 1;
1548         } else if (!strcmp(arg, "--no-standard-notes")) {
1549                 revs->notes_opt.use_default_notes = 0;
1550         } else if (!strcmp(arg, "--oneline")) {
1551                 revs->verbose_header = 1;
1552                 get_commit_format("oneline", revs);
1553                 revs->pretty_given = 1;
1554                 revs->abbrev_commit = 1;
1555         } else if (!strcmp(arg, "--graph")) {
1556                 revs->topo_order = 1;
1557                 revs->rewrite_parents = 1;
1558                 revs->graph = graph_init(revs);
1559         } else if (!strcmp(arg, "--root")) {
1560                 revs->show_root_diff = 1;
1561         } else if (!strcmp(arg, "--no-commit-id")) {
1562                 revs->no_commit_id = 1;
1563         } else if (!strcmp(arg, "--always")) {
1564                 revs->always_show_header = 1;
1565         } else if (!strcmp(arg, "--no-abbrev")) {
1566                 revs->abbrev = 0;
1567         } else if (!strcmp(arg, "--abbrev")) {
1568                 revs->abbrev = DEFAULT_ABBREV;
1569         } else if (!prefixcmp(arg, "--abbrev=")) {
1570                 revs->abbrev = strtoul(arg + 9, NULL, 10);
1571                 if (revs->abbrev < MINIMUM_ABBREV)
1572                         revs->abbrev = MINIMUM_ABBREV;
1573                 else if (revs->abbrev > 40)
1574                         revs->abbrev = 40;
1575         } else if (!strcmp(arg, "--abbrev-commit")) {
1576                 revs->abbrev_commit = 1;
1577                 revs->abbrev_commit_given = 1;
1578         } else if (!strcmp(arg, "--no-abbrev-commit")) {
1579                 revs->abbrev_commit = 0;
1580         } else if (!strcmp(arg, "--full-diff")) {
1581                 revs->diff = 1;
1582                 revs->full_diff = 1;
1583         } else if (!strcmp(arg, "--full-history")) {
1584                 revs->simplify_history = 0;
1585         } else if (!strcmp(arg, "--relative-date")) {
1586                 revs->date_mode = DATE_RELATIVE;
1587                 revs->date_mode_explicit = 1;
1588         } else if ((argcount = parse_long_opt("date", argv, &optarg))) {
1589                 revs->date_mode = parse_date_format(optarg);
1590                 revs->date_mode_explicit = 1;
1591                 return argcount;
1592         } else if (!strcmp(arg, "--log-size")) {
1593                 revs->show_log_size = 1;
1594         }
1595         /*
1596          * Grepping the commit log
1597          */
1598         else if ((argcount = parse_long_opt("author", argv, &optarg))) {
1599                 add_header_grep(revs, GREP_HEADER_AUTHOR, optarg);
1600                 return argcount;
1601         } else if ((argcount = parse_long_opt("committer", argv, &optarg))) {
1602                 add_header_grep(revs, GREP_HEADER_COMMITTER, optarg);
1603                 return argcount;
1604         } else if ((argcount = parse_long_opt("grep-reflog", argv, &optarg))) {
1605                 add_header_grep(revs, GREP_HEADER_REFLOG, optarg);
1606                 return argcount;
1607         } else if ((argcount = parse_long_opt("grep", argv, &optarg))) {
1608                 add_message_grep(revs, optarg);
1609                 return argcount;
1610         } else if (!strcmp(arg, "--grep-debug")) {
1611                 revs->grep_filter.debug = 1;
1612         } else if (!strcmp(arg, "--basic-regexp")) {
1613                 grep_set_pattern_type_option(GREP_PATTERN_TYPE_BRE, &revs->grep_filter);
1614         } else if (!strcmp(arg, "--extended-regexp") || !strcmp(arg, "-E")) {
1615                 grep_set_pattern_type_option(GREP_PATTERN_TYPE_ERE, &revs->grep_filter);
1616         } else if (!strcmp(arg, "--regexp-ignore-case") || !strcmp(arg, "-i")) {
1617                 revs->grep_filter.regflags |= REG_ICASE;
1618                 DIFF_OPT_SET(&revs->diffopt, PICKAXE_IGNORE_CASE);
1619         } else if (!strcmp(arg, "--fixed-strings") || !strcmp(arg, "-F")) {
1620                 grep_set_pattern_type_option(GREP_PATTERN_TYPE_FIXED, &revs->grep_filter);
1621         } else if (!strcmp(arg, "--perl-regexp")) {
1622                 grep_set_pattern_type_option(GREP_PATTERN_TYPE_PCRE, &revs->grep_filter);
1623         } else if (!strcmp(arg, "--all-match")) {
1624                 revs->grep_filter.all_match = 1;
1625         } else if ((argcount = parse_long_opt("encoding", argv, &optarg))) {
1626                 if (strcmp(optarg, "none"))
1627                         git_log_output_encoding = xstrdup(optarg);
1628                 else
1629                         git_log_output_encoding = "";
1630                 return argcount;
1631         } else if (!strcmp(arg, "--reverse")) {
1632                 revs->reverse ^= 1;
1633         } else if (!strcmp(arg, "--children")) {
1634                 revs->children.name = "children";
1635                 revs->limited = 1;
1636         } else if (!strcmp(arg, "--ignore-missing")) {
1637                 revs->ignore_missing = 1;
1638         } else {
1639                 int opts = diff_opt_parse(&revs->diffopt, argv, argc);
1640                 if (!opts)
1641                         unkv[(*unkc)++] = arg;
1642                 return opts;
1643         }
1644
1645         return 1;
1646 }
1647
1648 void parse_revision_opt(struct rev_info *revs, struct parse_opt_ctx_t *ctx,
1649                         const struct option *options,
1650                         const char * const usagestr[])
1651 {
1652         int n = handle_revision_opt(revs, ctx->argc, ctx->argv,
1653                                     &ctx->cpidx, ctx->out);
1654         if (n <= 0) {
1655                 error("unknown option `%s'", ctx->argv[0]);
1656                 usage_with_options(usagestr, options);
1657         }
1658         ctx->argv += n;
1659         ctx->argc -= n;
1660 }
1661
1662 static int for_each_bad_bisect_ref(const char *submodule, each_ref_fn fn, void *cb_data)
1663 {
1664         return for_each_ref_in_submodule(submodule, "refs/bisect/bad", fn, cb_data);
1665 }
1666
1667 static int for_each_good_bisect_ref(const char *submodule, each_ref_fn fn, void *cb_data)
1668 {
1669         return for_each_ref_in_submodule(submodule, "refs/bisect/good", fn, cb_data);
1670 }
1671
1672 static int handle_revision_pseudo_opt(const char *submodule,
1673                                 struct rev_info *revs,
1674                                 int argc, const char **argv, int *flags)
1675 {
1676         const char *arg = argv[0];
1677         const char *optarg;
1678         int argcount;
1679
1680         /*
1681          * NOTE!
1682          *
1683          * Commands like "git shortlog" will not accept the options below
1684          * unless parse_revision_opt queues them (as opposed to erroring
1685          * out).
1686          *
1687          * When implementing your new pseudo-option, remember to
1688          * register it in the list at the top of handle_revision_opt.
1689          */
1690         if (!strcmp(arg, "--all")) {
1691                 handle_refs(submodule, revs, *flags, for_each_ref_submodule);
1692                 handle_refs(submodule, revs, *flags, head_ref_submodule);
1693         } else if (!strcmp(arg, "--branches")) {
1694                 handle_refs(submodule, revs, *flags, for_each_branch_ref_submodule);
1695         } else if (!strcmp(arg, "--bisect")) {
1696                 handle_refs(submodule, revs, *flags, for_each_bad_bisect_ref);
1697                 handle_refs(submodule, revs, *flags ^ UNINTERESTING, for_each_good_bisect_ref);
1698                 revs->bisect = 1;
1699         } else if (!strcmp(arg, "--tags")) {
1700                 handle_refs(submodule, revs, *flags, for_each_tag_ref_submodule);
1701         } else if (!strcmp(arg, "--remotes")) {
1702                 handle_refs(submodule, revs, *flags, for_each_remote_ref_submodule);
1703         } else if ((argcount = parse_long_opt("glob", argv, &optarg))) {
1704                 struct all_refs_cb cb;
1705                 init_all_refs_cb(&cb, revs, *flags);
1706                 for_each_glob_ref(handle_one_ref, optarg, &cb);
1707                 return argcount;
1708         } else if (!prefixcmp(arg, "--branches=")) {
1709                 struct all_refs_cb cb;
1710                 init_all_refs_cb(&cb, revs, *flags);
1711                 for_each_glob_ref_in(handle_one_ref, arg + 11, "refs/heads/", &cb);
1712         } else if (!prefixcmp(arg, "--tags=")) {
1713                 struct all_refs_cb cb;
1714                 init_all_refs_cb(&cb, revs, *flags);
1715                 for_each_glob_ref_in(handle_one_ref, arg + 7, "refs/tags/", &cb);
1716         } else if (!prefixcmp(arg, "--remotes=")) {
1717                 struct all_refs_cb cb;
1718                 init_all_refs_cb(&cb, revs, *flags);
1719                 for_each_glob_ref_in(handle_one_ref, arg + 10, "refs/remotes/", &cb);
1720         } else if (!strcmp(arg, "--reflog")) {
1721                 handle_reflog(revs, *flags);
1722         } else if (!strcmp(arg, "--not")) {
1723                 *flags ^= UNINTERESTING;
1724         } else if (!strcmp(arg, "--no-walk")) {
1725                 revs->no_walk = REVISION_WALK_NO_WALK_SORTED;
1726         } else if (!prefixcmp(arg, "--no-walk=")) {
1727                 /*
1728                  * Detached form ("--no-walk X" as opposed to "--no-walk=X")
1729                  * not allowed, since the argument is optional.
1730                  */
1731                 if (!strcmp(arg + 10, "sorted"))
1732                         revs->no_walk = REVISION_WALK_NO_WALK_SORTED;
1733                 else if (!strcmp(arg + 10, "unsorted"))
1734                         revs->no_walk = REVISION_WALK_NO_WALK_UNSORTED;
1735                 else
1736                         return error("invalid argument to --no-walk");
1737         } else if (!strcmp(arg, "--do-walk")) {
1738                 revs->no_walk = 0;
1739         } else {
1740                 return 0;
1741         }
1742
1743         return 1;
1744 }
1745
1746 /*
1747  * Parse revision information, filling in the "rev_info" structure,
1748  * and removing the used arguments from the argument list.
1749  *
1750  * Returns the number of arguments left that weren't recognized
1751  * (which are also moved to the head of the argument list)
1752  */
1753 int setup_revisions(int argc, const char **argv, struct rev_info *revs, struct setup_revision_opt *opt)
1754 {
1755         int i, flags, left, seen_dashdash, read_from_stdin, got_rev_arg = 0, revarg_opt;
1756         struct cmdline_pathspec prune_data;
1757         const char *submodule = NULL;
1758
1759         memset(&prune_data, 0, sizeof(prune_data));
1760         if (opt)
1761                 submodule = opt->submodule;
1762
1763         /* First, search for "--" */
1764         if (opt && opt->assume_dashdash) {
1765                 seen_dashdash = 1;
1766         } else {
1767                 seen_dashdash = 0;
1768                 for (i = 1; i < argc; i++) {
1769                         const char *arg = argv[i];
1770                         if (strcmp(arg, "--"))
1771                                 continue;
1772                         argv[i] = NULL;
1773                         argc = i;
1774                         if (argv[i + 1])
1775                                 append_prune_data(&prune_data, argv + i + 1);
1776                         seen_dashdash = 1;
1777                         break;
1778                 }
1779         }
1780
1781         /* Second, deal with arguments and options */
1782         flags = 0;
1783         revarg_opt = opt ? opt->revarg_opt : 0;
1784         if (seen_dashdash)
1785                 revarg_opt |= REVARG_CANNOT_BE_FILENAME;
1786         read_from_stdin = 0;
1787         for (left = i = 1; i < argc; i++) {
1788                 const char *arg = argv[i];
1789                 if (*arg == '-') {
1790                         int opts;
1791
1792                         opts = handle_revision_pseudo_opt(submodule,
1793                                                 revs, argc - i, argv + i,
1794                                                 &flags);
1795                         if (opts > 0) {
1796                                 i += opts - 1;
1797                                 continue;
1798                         }
1799
1800                         if (!strcmp(arg, "--stdin")) {
1801                                 if (revs->disable_stdin) {
1802                                         argv[left++] = arg;
1803                                         continue;
1804                                 }
1805                                 if (read_from_stdin++)
1806                                         die("--stdin given twice?");
1807                                 read_revisions_from_stdin(revs, &prune_data);
1808                                 continue;
1809                         }
1810
1811                         opts = handle_revision_opt(revs, argc - i, argv + i, &left, argv);
1812                         if (opts > 0) {
1813                                 i += opts - 1;
1814                                 continue;
1815                         }
1816                         if (opts < 0)
1817                                 exit(128);
1818                         continue;
1819                 }
1820
1821
1822                 if (handle_revision_arg(arg, revs, flags, revarg_opt)) {
1823                         int j;
1824                         if (seen_dashdash || *arg == '^')
1825                                 die("bad revision '%s'", arg);
1826
1827                         /* If we didn't have a "--":
1828                          * (1) all filenames must exist;
1829                          * (2) all rev-args must not be interpretable
1830                          *     as a valid filename.
1831                          * but the latter we have checked in the main loop.
1832                          */
1833                         for (j = i; j < argc; j++)
1834                                 verify_filename(revs->prefix, argv[j], j == i);
1835
1836                         append_prune_data(&prune_data, argv + i);
1837                         break;
1838                 }
1839                 else
1840                         got_rev_arg = 1;
1841         }
1842
1843         if (prune_data.nr) {
1844                 /*
1845                  * If we need to introduce the magic "a lone ':' means no
1846                  * pathspec whatsoever", here is the place to do so.
1847                  *
1848                  * if (prune_data.nr == 1 && !strcmp(prune_data[0], ":")) {
1849                  *      prune_data.nr = 0;
1850                  *      prune_data.alloc = 0;
1851                  *      free(prune_data.path);
1852                  *      prune_data.path = NULL;
1853                  * } else {
1854                  *      terminate prune_data.alloc with NULL and
1855                  *      call init_pathspec() to set revs->prune_data here.
1856                  * }
1857                  */
1858                 ALLOC_GROW(prune_data.path, prune_data.nr+1, prune_data.alloc);
1859                 prune_data.path[prune_data.nr++] = NULL;
1860                 init_pathspec(&revs->prune_data,
1861                               get_pathspec(revs->prefix, prune_data.path));
1862         }
1863
1864         if (revs->def == NULL)
1865                 revs->def = opt ? opt->def : NULL;
1866         if (opt && opt->tweak)
1867                 opt->tweak(revs, opt);
1868         if (revs->show_merge)
1869                 prepare_show_merge(revs);
1870         if (revs->def && !revs->pending.nr && !got_rev_arg) {
1871                 unsigned char sha1[20];
1872                 struct object *object;
1873                 struct object_context oc;
1874                 if (get_sha1_with_context(revs->def, 0, sha1, &oc))
1875                         die("bad default revision '%s'", revs->def);
1876                 object = get_reference(revs, revs->def, sha1, 0);
1877                 add_pending_object_with_mode(revs, object, revs->def, oc.mode);
1878         }
1879
1880         /* Did the user ask for any diff output? Run the diff! */
1881         if (revs->diffopt.output_format & ~DIFF_FORMAT_NO_OUTPUT)
1882                 revs->diff = 1;
1883
1884         /* Pickaxe, diff-filter and rename following need diffs */
1885         if (revs->diffopt.pickaxe ||
1886             revs->diffopt.filter ||
1887             DIFF_OPT_TST(&revs->diffopt, FOLLOW_RENAMES))
1888                 revs->diff = 1;
1889
1890         if (revs->topo_order)
1891                 revs->limited = 1;
1892
1893         if (revs->prune_data.nr) {
1894                 diff_tree_setup_paths(revs->prune_data.raw, &revs->pruning);
1895                 /* Can't prune commits with rename following: the paths change.. */
1896                 if (!DIFF_OPT_TST(&revs->diffopt, FOLLOW_RENAMES))
1897                         revs->prune = 1;
1898                 if (!revs->full_diff)
1899                         diff_tree_setup_paths(revs->prune_data.raw, &revs->diffopt);
1900         }
1901         if (revs->combine_merges)
1902                 revs->ignore_merges = 0;
1903         revs->diffopt.abbrev = revs->abbrev;
1904         diff_setup_done(&revs->diffopt);
1905
1906         grep_commit_pattern_type(GREP_PATTERN_TYPE_UNSPECIFIED,
1907                                  &revs->grep_filter);
1908         compile_grep_patterns(&revs->grep_filter);
1909
1910         if (revs->reverse && revs->reflog_info)
1911                 die("cannot combine --reverse with --walk-reflogs");
1912         if (revs->rewrite_parents && revs->children.name)
1913                 die("cannot combine --parents and --children");
1914
1915         /*
1916          * Limitations on the graph functionality
1917          */
1918         if (revs->reverse && revs->graph)
1919                 die("cannot combine --reverse with --graph");
1920
1921         if (revs->reflog_info && revs->graph)
1922                 die("cannot combine --walk-reflogs with --graph");
1923         if (!revs->reflog_info && revs->grep_filter.use_reflog_filter)
1924                 die("cannot use --grep-reflog without --walk-reflogs");
1925
1926         return left;
1927 }
1928
1929 static void add_child(struct rev_info *revs, struct commit *parent, struct commit *child)
1930 {
1931         struct commit_list *l = xcalloc(1, sizeof(*l));
1932
1933         l->item = child;
1934         l->next = add_decoration(&revs->children, &parent->object, l);
1935 }
1936
1937 static int remove_duplicate_parents(struct commit *commit)
1938 {
1939         struct commit_list **pp, *p;
1940         int surviving_parents;
1941
1942         /* Examine existing parents while marking ones we have seen... */
1943         pp = &commit->parents;
1944         while ((p = *pp) != NULL) {
1945                 struct commit *parent = p->item;
1946                 if (parent->object.flags & TMP_MARK) {
1947                         *pp = p->next;
1948                         continue;
1949                 }
1950                 parent->object.flags |= TMP_MARK;
1951                 pp = &p->next;
1952         }
1953         /* count them while clearing the temporary mark */
1954         surviving_parents = 0;
1955         for (p = commit->parents; p; p = p->next) {
1956                 p->item->object.flags &= ~TMP_MARK;
1957                 surviving_parents++;
1958         }
1959         return surviving_parents;
1960 }
1961
1962 struct merge_simplify_state {
1963         struct commit *simplified;
1964 };
1965
1966 static struct merge_simplify_state *locate_simplify_state(struct rev_info *revs, struct commit *commit)
1967 {
1968         struct merge_simplify_state *st;
1969
1970         st = lookup_decoration(&revs->merge_simplification, &commit->object);
1971         if (!st) {
1972                 st = xcalloc(1, sizeof(*st));
1973                 add_decoration(&revs->merge_simplification, &commit->object, st);
1974         }
1975         return st;
1976 }
1977
1978 static struct commit_list **simplify_one(struct rev_info *revs, struct commit *commit, struct commit_list **tail)
1979 {
1980         struct commit_list *p;
1981         struct merge_simplify_state *st, *pst;
1982         int cnt;
1983
1984         st = locate_simplify_state(revs, commit);
1985
1986         /*
1987          * Have we handled this one?
1988          */
1989         if (st->simplified)
1990                 return tail;
1991
1992         /*
1993          * An UNINTERESTING commit simplifies to itself, so does a
1994          * root commit.  We do not rewrite parents of such commit
1995          * anyway.
1996          */
1997         if ((commit->object.flags & UNINTERESTING) || !commit->parents) {
1998                 st->simplified = commit;
1999                 return tail;
2000         }
2001
2002         /*
2003          * Do we know what commit all of our parents that matter
2004          * should be rewritten to?  Otherwise we are not ready to
2005          * rewrite this one yet.
2006          */
2007         for (cnt = 0, p = commit->parents; p; p = p->next) {
2008                 pst = locate_simplify_state(revs, p->item);
2009                 if (!pst->simplified) {
2010                         tail = &commit_list_insert(p->item, tail)->next;
2011                         cnt++;
2012                 }
2013                 if (revs->first_parent_only)
2014                         break;
2015         }
2016         if (cnt) {
2017                 tail = &commit_list_insert(commit, tail)->next;
2018                 return tail;
2019         }
2020
2021         /*
2022          * Rewrite our list of parents.
2023          */
2024         for (p = commit->parents; p; p = p->next) {
2025                 pst = locate_simplify_state(revs, p->item);
2026                 p->item = pst->simplified;
2027                 if (revs->first_parent_only)
2028                         break;
2029         }
2030
2031         if (revs->first_parent_only)
2032                 cnt = 1;
2033         else
2034                 cnt = remove_duplicate_parents(commit);
2035
2036         /*
2037          * It is possible that we are a merge and one side branch
2038          * does not have any commit that touches the given paths;
2039          * in such a case, the immediate parents will be rewritten
2040          * to different commits.
2041          *
2042          *      o----X          X: the commit we are looking at;
2043          *     /    /           o: a commit that touches the paths;
2044          * ---o----'
2045          *
2046          * Further reduce the parents by removing redundant parents.
2047          */
2048         if (1 < cnt) {
2049                 struct commit_list *h = reduce_heads(commit->parents);
2050                 cnt = commit_list_count(h);
2051                 free_commit_list(commit->parents);
2052                 commit->parents = h;
2053         }
2054
2055         /*
2056          * A commit simplifies to itself if it is a root, if it is
2057          * UNINTERESTING, if it touches the given paths, or if it is a
2058          * merge and its parents simplifies to more than one commits
2059          * (the first two cases are already handled at the beginning of
2060          * this function).
2061          *
2062          * Otherwise, it simplifies to what its sole parent simplifies to.
2063          */
2064         if (!cnt ||
2065             (commit->object.flags & UNINTERESTING) ||
2066             !(commit->object.flags & TREESAME) ||
2067             (1 < cnt))
2068                 st->simplified = commit;
2069         else {
2070                 pst = locate_simplify_state(revs, commit->parents->item);
2071                 st->simplified = pst->simplified;
2072         }
2073         return tail;
2074 }
2075
2076 static void simplify_merges(struct rev_info *revs)
2077 {
2078         struct commit_list *list, *next;
2079         struct commit_list *yet_to_do, **tail;
2080         struct commit *commit;
2081
2082         if (!revs->prune)
2083                 return;
2084
2085         /* feed the list reversed */
2086         yet_to_do = NULL;
2087         for (list = revs->commits; list; list = next) {
2088                 commit = list->item;
2089                 next = list->next;
2090                 /*
2091                  * Do not free(list) here yet; the original list
2092                  * is used later in this function.
2093                  */
2094                 commit_list_insert(commit, &yet_to_do);
2095         }
2096         while (yet_to_do) {
2097                 list = yet_to_do;
2098                 yet_to_do = NULL;
2099                 tail = &yet_to_do;
2100                 while (list) {
2101                         commit = list->item;
2102                         next = list->next;
2103                         free(list);
2104                         list = next;
2105                         tail = simplify_one(revs, commit, tail);
2106                 }
2107         }
2108
2109         /* clean up the result, removing the simplified ones */
2110         list = revs->commits;
2111         revs->commits = NULL;
2112         tail = &revs->commits;
2113         while (list) {
2114                 struct merge_simplify_state *st;
2115
2116                 commit = list->item;
2117                 next = list->next;
2118                 free(list);
2119                 list = next;
2120                 st = locate_simplify_state(revs, commit);
2121                 if (st->simplified == commit)
2122                         tail = &commit_list_insert(commit, tail)->next;
2123         }
2124 }
2125
2126 static void set_children(struct rev_info *revs)
2127 {
2128         struct commit_list *l;
2129         for (l = revs->commits; l; l = l->next) {
2130                 struct commit *commit = l->item;
2131                 struct commit_list *p;
2132
2133                 for (p = commit->parents; p; p = p->next)
2134                         add_child(revs, p->item, commit);
2135         }
2136 }
2137
2138 void reset_revision_walk(void)
2139 {
2140         clear_object_flags(SEEN | ADDED | SHOWN);
2141 }
2142
2143 int prepare_revision_walk(struct rev_info *revs)
2144 {
2145         int nr = revs->pending.nr;
2146         struct object_array_entry *e, *list;
2147         struct commit_list **next = &revs->commits;
2148
2149         e = list = revs->pending.objects;
2150         revs->pending.nr = 0;
2151         revs->pending.alloc = 0;
2152         revs->pending.objects = NULL;
2153         while (--nr >= 0) {
2154                 struct commit *commit = handle_commit(revs, e->item, e->name);
2155                 if (commit) {
2156                         if (!(commit->object.flags & SEEN)) {
2157                                 commit->object.flags |= SEEN;
2158                                 next = commit_list_append(commit, next);
2159                         }
2160                 }
2161                 e++;
2162         }
2163         if (!revs->leak_pending)
2164                 free(list);
2165
2166         if (revs->no_walk != REVISION_WALK_NO_WALK_UNSORTED)
2167                 commit_list_sort_by_date(&revs->commits);
2168         if (revs->no_walk)
2169                 return 0;
2170         if (revs->limited)
2171                 if (limit_list(revs) < 0)
2172                         return -1;
2173         if (revs->topo_order)
2174                 sort_in_topological_order(&revs->commits, revs->lifo);
2175         if (revs->simplify_merges)
2176                 simplify_merges(revs);
2177         if (revs->children.name)
2178                 set_children(revs);
2179         return 0;
2180 }
2181
2182 enum rewrite_result {
2183         rewrite_one_ok,
2184         rewrite_one_noparents,
2185         rewrite_one_error
2186 };
2187
2188 static enum rewrite_result rewrite_one(struct rev_info *revs, struct commit **pp)
2189 {
2190         struct commit_list *cache = NULL;
2191
2192         for (;;) {
2193                 struct commit *p = *pp;
2194                 if (!revs->limited)
2195                         if (add_parents_to_list(revs, p, &revs->commits, &cache) < 0)
2196                                 return rewrite_one_error;
2197                 if (p->parents && p->parents->next)
2198                         return rewrite_one_ok;
2199                 if (p->object.flags & UNINTERESTING)
2200                         return rewrite_one_ok;
2201                 if (!(p->object.flags & TREESAME))
2202                         return rewrite_one_ok;
2203                 if (!p->parents)
2204                         return rewrite_one_noparents;
2205                 *pp = p->parents->item;
2206         }
2207 }
2208
2209 static int rewrite_parents(struct rev_info *revs, struct commit *commit)
2210 {
2211         struct commit_list **pp = &commit->parents;
2212         while (*pp) {
2213                 struct commit_list *parent = *pp;
2214                 switch (rewrite_one(revs, &parent->item)) {
2215                 case rewrite_one_ok:
2216                         break;
2217                 case rewrite_one_noparents:
2218                         *pp = parent->next;
2219                         continue;
2220                 case rewrite_one_error:
2221                         return -1;
2222                 }
2223                 pp = &parent->next;
2224         }
2225         remove_duplicate_parents(commit);
2226         return 0;
2227 }
2228
2229 static int commit_rewrite_person(struct strbuf *buf, const char *what, struct string_list *mailmap)
2230 {
2231         char *person, *endp;
2232         size_t len, namelen, maillen;
2233         const char *name;
2234         const char *mail;
2235         struct ident_split ident;
2236
2237         person = strstr(buf->buf, what);
2238         if (!person)
2239                 return 0;
2240
2241         person += strlen(what);
2242         endp = strchr(person, '\n');
2243         if (!endp)
2244                 return 0;
2245
2246         len = endp - person;
2247
2248         if (split_ident_line(&ident, person, len))
2249                 return 0;
2250
2251         mail = ident.mail_begin;
2252         maillen = ident.mail_end - ident.mail_begin;
2253         name = ident.name_begin;
2254         namelen = ident.name_end - ident.name_begin;
2255
2256         if (map_user(mailmap, &mail, &maillen, &name, &namelen)) {
2257                 struct strbuf namemail = STRBUF_INIT;
2258
2259                 strbuf_addf(&namemail, "%.*s <%.*s>",
2260                             (int)namelen, name, (int)maillen, mail);
2261
2262                 strbuf_splice(buf, ident.name_begin - buf->buf,
2263                               ident.mail_end - ident.name_begin + 1,
2264                               namemail.buf, namemail.len);
2265
2266                 strbuf_release(&namemail);
2267
2268                 return 1;
2269         }
2270
2271         return 0;
2272 }
2273
2274 static int commit_match(struct commit *commit, struct rev_info *opt)
2275 {
2276         int retval;
2277         const char *encoding;
2278         char *message;
2279         struct strbuf buf = STRBUF_INIT;
2280
2281         if (!opt->grep_filter.pattern_list && !opt->grep_filter.header_list)
2282                 return 1;
2283
2284         /* Prepend "fake" headers as needed */
2285         if (opt->grep_filter.use_reflog_filter) {
2286                 strbuf_addstr(&buf, "reflog ");
2287                 get_reflog_message(&buf, opt->reflog_info);
2288                 strbuf_addch(&buf, '\n');
2289         }
2290
2291         /*
2292          * We grep in the user's output encoding, under the assumption that it
2293          * is the encoding they are most likely to write their grep pattern
2294          * for. In addition, it means we will match the "notes" encoding below,
2295          * so we will not end up with a buffer that has two different encodings
2296          * in it.
2297          */
2298         encoding = get_log_output_encoding();
2299         message = logmsg_reencode(commit, NULL, encoding);
2300
2301         /* Copy the commit to temporary if we are using "fake" headers */
2302         if (buf.len)
2303                 strbuf_addstr(&buf, message);
2304
2305         if (opt->grep_filter.header_list && opt->mailmap) {
2306                 if (!buf.len)
2307                         strbuf_addstr(&buf, message);
2308
2309                 commit_rewrite_person(&buf, "\nauthor ", opt->mailmap);
2310                 commit_rewrite_person(&buf, "\ncommitter ", opt->mailmap);
2311         }
2312
2313         /* Append "fake" message parts as needed */
2314         if (opt->show_notes) {
2315                 if (!buf.len)
2316                         strbuf_addstr(&buf, message);
2317                 format_display_notes(commit->object.sha1, &buf, encoding, 1);
2318         }
2319
2320         /* Find either in the original commit message, or in the temporary */
2321         if (buf.len)
2322                 retval = grep_buffer(&opt->grep_filter, buf.buf, buf.len);
2323         else
2324                 retval = grep_buffer(&opt->grep_filter,
2325                                      message, strlen(message));
2326         strbuf_release(&buf);
2327         logmsg_free(message, commit);
2328         return retval;
2329 }
2330
2331 static inline int want_ancestry(struct rev_info *revs)
2332 {
2333         return (revs->rewrite_parents || revs->children.name);
2334 }
2335
2336 enum commit_action get_commit_action(struct rev_info *revs, struct commit *commit)
2337 {
2338         if (commit->object.flags & SHOWN)
2339                 return commit_ignore;
2340         if (revs->unpacked && has_sha1_pack(commit->object.sha1))
2341                 return commit_ignore;
2342         if (revs->show_all)
2343                 return commit_show;
2344         if (commit->object.flags & UNINTERESTING)
2345                 return commit_ignore;
2346         if (revs->min_age != -1 && (commit->date > revs->min_age))
2347                 return commit_ignore;
2348         if (revs->min_parents || (revs->max_parents >= 0)) {
2349                 int n = 0;
2350                 struct commit_list *p;
2351                 for (p = commit->parents; p; p = p->next)
2352                         n++;
2353                 if ((n < revs->min_parents) ||
2354                     ((revs->max_parents >= 0) && (n > revs->max_parents)))
2355                         return commit_ignore;
2356         }
2357         if (!commit_match(commit, revs))
2358                 return commit_ignore;
2359         if (revs->prune && revs->dense) {
2360                 /* Commit without changes? */
2361                 if (commit->object.flags & TREESAME) {
2362                         /* drop merges unless we want parenthood */
2363                         if (!want_ancestry(revs))
2364                                 return commit_ignore;
2365                         /* non-merge - always ignore it */
2366                         if (!commit->parents || !commit->parents->next)
2367                                 return commit_ignore;
2368                 }
2369         }
2370         return commit_show;
2371 }
2372
2373 enum commit_action simplify_commit(struct rev_info *revs, struct commit *commit)
2374 {
2375         enum commit_action action = get_commit_action(revs, commit);
2376
2377         if (action == commit_show &&
2378             !revs->show_all &&
2379             revs->prune && revs->dense && want_ancestry(revs)) {
2380                 if (rewrite_parents(revs, commit) < 0)
2381                         return commit_error;
2382         }
2383         return action;
2384 }
2385
2386 static struct commit *get_revision_1(struct rev_info *revs)
2387 {
2388         if (!revs->commits)
2389                 return NULL;
2390
2391         do {
2392                 struct commit_list *entry = revs->commits;
2393                 struct commit *commit = entry->item;
2394
2395                 revs->commits = entry->next;
2396                 free(entry);
2397
2398                 if (revs->reflog_info) {
2399                         fake_reflog_parent(revs->reflog_info, commit);
2400                         commit->object.flags &= ~(ADDED | SEEN | SHOWN);
2401                 }
2402
2403                 /*
2404                  * If we haven't done the list limiting, we need to look at
2405                  * the parents here. We also need to do the date-based limiting
2406                  * that we'd otherwise have done in limit_list().
2407                  */
2408                 if (!revs->limited) {
2409                         if (revs->max_age != -1 &&
2410                             (commit->date < revs->max_age))
2411                                 continue;
2412                         if (add_parents_to_list(revs, commit, &revs->commits, NULL) < 0)
2413                                 die("Failed to traverse parents of commit %s",
2414                                     sha1_to_hex(commit->object.sha1));
2415                 }
2416
2417                 switch (simplify_commit(revs, commit)) {
2418                 case commit_ignore:
2419                         continue;
2420                 case commit_error:
2421                         die("Failed to simplify parents of commit %s",
2422                             sha1_to_hex(commit->object.sha1));
2423                 default:
2424                         return commit;
2425                 }
2426         } while (revs->commits);
2427         return NULL;
2428 }
2429
2430 static void gc_boundary(struct object_array *array)
2431 {
2432         unsigned nr = array->nr;
2433         unsigned alloc = array->alloc;
2434         struct object_array_entry *objects = array->objects;
2435
2436         if (alloc <= nr) {
2437                 unsigned i, j;
2438                 for (i = j = 0; i < nr; i++) {
2439                         if (objects[i].item->flags & SHOWN)
2440                                 continue;
2441                         if (i != j)
2442                                 objects[j] = objects[i];
2443                         j++;
2444                 }
2445                 for (i = j; i < nr; i++)
2446                         objects[i].item = NULL;
2447                 array->nr = j;
2448         }
2449 }
2450
2451 static void create_boundary_commit_list(struct rev_info *revs)
2452 {
2453         unsigned i;
2454         struct commit *c;
2455         struct object_array *array = &revs->boundary_commits;
2456         struct object_array_entry *objects = array->objects;
2457
2458         /*
2459          * If revs->commits is non-NULL at this point, an error occurred in
2460          * get_revision_1().  Ignore the error and continue printing the
2461          * boundary commits anyway.  (This is what the code has always
2462          * done.)
2463          */
2464         if (revs->commits) {
2465                 free_commit_list(revs->commits);
2466                 revs->commits = NULL;
2467         }
2468
2469         /*
2470          * Put all of the actual boundary commits from revs->boundary_commits
2471          * into revs->commits
2472          */
2473         for (i = 0; i < array->nr; i++) {
2474                 c = (struct commit *)(objects[i].item);
2475                 if (!c)
2476                         continue;
2477                 if (!(c->object.flags & CHILD_SHOWN))
2478                         continue;
2479                 if (c->object.flags & (SHOWN | BOUNDARY))
2480                         continue;
2481                 c->object.flags |= BOUNDARY;
2482                 commit_list_insert(c, &revs->commits);
2483         }
2484
2485         /*
2486          * If revs->topo_order is set, sort the boundary commits
2487          * in topological order
2488          */
2489         sort_in_topological_order(&revs->commits, revs->lifo);
2490 }
2491
2492 static struct commit *get_revision_internal(struct rev_info *revs)
2493 {
2494         struct commit *c = NULL;
2495         struct commit_list *l;
2496
2497         if (revs->boundary == 2) {
2498                 /*
2499                  * All of the normal commits have already been returned,
2500                  * and we are now returning boundary commits.
2501                  * create_boundary_commit_list() has populated
2502                  * revs->commits with the remaining commits to return.
2503                  */
2504                 c = pop_commit(&revs->commits);
2505                 if (c)
2506                         c->object.flags |= SHOWN;
2507                 return c;
2508         }
2509
2510         /*
2511          * If our max_count counter has reached zero, then we are done. We
2512          * don't simply return NULL because we still might need to show
2513          * boundary commits. But we want to avoid calling get_revision_1, which
2514          * might do a considerable amount of work finding the next commit only
2515          * for us to throw it away.
2516          *
2517          * If it is non-zero, then either we don't have a max_count at all
2518          * (-1), or it is still counting, in which case we decrement.
2519          */
2520         if (revs->max_count) {
2521                 c = get_revision_1(revs);
2522                 if (c) {
2523                         while (0 < revs->skip_count) {
2524                                 revs->skip_count--;
2525                                 c = get_revision_1(revs);
2526                                 if (!c)
2527                                         break;
2528                         }
2529                 }
2530
2531                 if (revs->max_count > 0)
2532                         revs->max_count--;
2533         }
2534
2535         if (c)
2536                 c->object.flags |= SHOWN;
2537
2538         if (!revs->boundary) {
2539                 return c;
2540         }
2541
2542         if (!c) {
2543                 /*
2544                  * get_revision_1() runs out the commits, and
2545                  * we are done computing the boundaries.
2546                  * switch to boundary commits output mode.
2547                  */
2548                 revs->boundary = 2;
2549
2550                 /*
2551                  * Update revs->commits to contain the list of
2552                  * boundary commits.
2553                  */
2554                 create_boundary_commit_list(revs);
2555
2556                 return get_revision_internal(revs);
2557         }
2558
2559         /*
2560          * boundary commits are the commits that are parents of the
2561          * ones we got from get_revision_1() but they themselves are
2562          * not returned from get_revision_1().  Before returning
2563          * 'c', we need to mark its parents that they could be boundaries.
2564          */
2565
2566         for (l = c->parents; l; l = l->next) {
2567                 struct object *p;
2568                 p = &(l->item->object);
2569                 if (p->flags & (CHILD_SHOWN | SHOWN))
2570                         continue;
2571                 p->flags |= CHILD_SHOWN;
2572                 gc_boundary(&revs->boundary_commits);
2573                 add_object_array(p, NULL, &revs->boundary_commits);
2574         }
2575
2576         return c;
2577 }
2578
2579 struct commit *get_revision(struct rev_info *revs)
2580 {
2581         struct commit *c;
2582         struct commit_list *reversed;
2583
2584         if (revs->reverse) {
2585                 reversed = NULL;
2586                 while ((c = get_revision_internal(revs))) {
2587                         commit_list_insert(c, &reversed);
2588                 }
2589                 revs->commits = reversed;
2590                 revs->reverse = 0;
2591                 revs->reverse_output_stage = 1;
2592         }
2593
2594         if (revs->reverse_output_stage)
2595                 return pop_commit(&revs->commits);
2596
2597         c = get_revision_internal(revs);
2598         if (c && revs->graph)
2599                 graph_update(revs->graph, c);
2600         return c;
2601 }
2602
2603 char *get_revision_mark(const struct rev_info *revs, const struct commit *commit)
2604 {
2605         if (commit->object.flags & BOUNDARY)
2606                 return "-";
2607         else if (commit->object.flags & UNINTERESTING)
2608                 return "^";
2609         else if (commit->object.flags & PATCHSAME)
2610                 return "=";
2611         else if (!revs || revs->left_right) {
2612                 if (commit->object.flags & SYMMETRIC_LEFT)
2613                         return "<";
2614                 else
2615                         return ">";
2616         } else if (revs->graph)
2617                 return "*";
2618         else if (revs->cherry_mark)
2619                 return "+";
2620         return "";
2621 }
2622
2623 void put_revision_mark(const struct rev_info *revs, const struct commit *commit)
2624 {
2625         char *mark = get_revision_mark(revs, commit);
2626         if (!strlen(mark))
2627                 return;
2628         fputs(mark, stdout);
2629         putchar(' ');
2630 }