Merge branch 'js/diff-rename-force-stable-sort'
[git] / revision.c
1 #include "cache.h"
2 #include "object-store.h"
3 #include "tag.h"
4 #include "blob.h"
5 #include "tree.h"
6 #include "commit.h"
7 #include "diff.h"
8 #include "refs.h"
9 #include "revision.h"
10 #include "repository.h"
11 #include "graph.h"
12 #include "grep.h"
13 #include "reflog-walk.h"
14 #include "patch-ids.h"
15 #include "decorate.h"
16 #include "log-tree.h"
17 #include "string-list.h"
18 #include "line-log.h"
19 #include "mailmap.h"
20 #include "commit-slab.h"
21 #include "dir.h"
22 #include "cache-tree.h"
23 #include "bisect.h"
24 #include "packfile.h"
25 #include "worktree.h"
26 #include "argv-array.h"
27 #include "commit-reach.h"
28 #include "commit-graph.h"
29 #include "prio-queue.h"
30 #include "hashmap.h"
31
32 volatile show_early_output_fn_t show_early_output;
33
34 static const char *term_bad;
35 static const char *term_good;
36
37 implement_shared_commit_slab(revision_sources, char *);
38
39 void show_object_with_name(FILE *out, struct object *obj, const char *name)
40 {
41         const char *p;
42
43         fprintf(out, "%s ", oid_to_hex(&obj->oid));
44         for (p = name; *p && *p != '\n'; p++)
45                 fputc(*p, out);
46         fputc('\n', out);
47 }
48
49 static void mark_blob_uninteresting(struct blob *blob)
50 {
51         if (!blob)
52                 return;
53         if (blob->object.flags & UNINTERESTING)
54                 return;
55         blob->object.flags |= UNINTERESTING;
56 }
57
58 static void mark_tree_contents_uninteresting(struct repository *r,
59                                              struct tree *tree)
60 {
61         struct tree_desc desc;
62         struct name_entry entry;
63
64         if (parse_tree_gently(tree, 1) < 0)
65                 return;
66
67         init_tree_desc(&desc, tree->buffer, tree->size);
68         while (tree_entry(&desc, &entry)) {
69                 switch (object_type(entry.mode)) {
70                 case OBJ_TREE:
71                         mark_tree_uninteresting(r, lookup_tree(r, &entry.oid));
72                         break;
73                 case OBJ_BLOB:
74                         mark_blob_uninteresting(lookup_blob(r, &entry.oid));
75                         break;
76                 default:
77                         /* Subproject commit - not in this repository */
78                         break;
79                 }
80         }
81
82         /*
83          * We don't care about the tree any more
84          * after it has been marked uninteresting.
85          */
86         free_tree_buffer(tree);
87 }
88
89 void mark_tree_uninteresting(struct repository *r, struct tree *tree)
90 {
91         struct object *obj;
92
93         if (!tree)
94                 return;
95
96         obj = &tree->object;
97         if (obj->flags & UNINTERESTING)
98                 return;
99         obj->flags |= UNINTERESTING;
100         mark_tree_contents_uninteresting(r, tree);
101 }
102
103 struct path_and_oids_entry {
104         struct hashmap_entry ent;
105         char *path;
106         struct oidset trees;
107 };
108
109 static int path_and_oids_cmp(const void *hashmap_cmp_fn_data,
110                              const struct path_and_oids_entry *e1,
111                              const struct path_and_oids_entry *e2,
112                              const void *keydata)
113 {
114         return strcmp(e1->path, e2->path);
115 }
116
117 static void paths_and_oids_init(struct hashmap *map)
118 {
119         hashmap_init(map, (hashmap_cmp_fn) path_and_oids_cmp, NULL, 0);
120 }
121
122 static void paths_and_oids_clear(struct hashmap *map)
123 {
124         struct hashmap_iter iter;
125         struct path_and_oids_entry *entry;
126         hashmap_iter_init(map, &iter);
127
128         while ((entry = (struct path_and_oids_entry *)hashmap_iter_next(&iter))) {
129                 oidset_clear(&entry->trees);
130                 free(entry->path);
131         }
132
133         hashmap_free(map, 1);
134 }
135
136 static void paths_and_oids_insert(struct hashmap *map,
137                                   const char *path,
138                                   const struct object_id *oid)
139 {
140         int hash = strhash(path);
141         struct path_and_oids_entry key;
142         struct path_and_oids_entry *entry;
143
144         hashmap_entry_init(&key, hash);
145
146         /* use a shallow copy for the lookup */
147         key.path = (char *)path;
148         oidset_init(&key.trees, 0);
149
150         if (!(entry = (struct path_and_oids_entry *)hashmap_get(map, &key, NULL))) {
151                 entry = xcalloc(1, sizeof(struct path_and_oids_entry));
152                 hashmap_entry_init(entry, hash);
153                 entry->path = xstrdup(key.path);
154                 oidset_init(&entry->trees, 16);
155                 hashmap_put(map, entry);
156         }
157
158         oidset_insert(&entry->trees, oid);
159 }
160
161 static void add_children_by_path(struct repository *r,
162                                  struct tree *tree,
163                                  struct hashmap *map)
164 {
165         struct tree_desc desc;
166         struct name_entry entry;
167
168         if (!tree)
169                 return;
170
171         if (parse_tree_gently(tree, 1) < 0)
172                 return;
173
174         init_tree_desc(&desc, tree->buffer, tree->size);
175         while (tree_entry(&desc, &entry)) {
176                 switch (object_type(entry.mode)) {
177                 case OBJ_TREE:
178                         paths_and_oids_insert(map, entry.path, &entry.oid);
179
180                         if (tree->object.flags & UNINTERESTING) {
181                                 struct tree *child = lookup_tree(r, &entry.oid);
182                                 if (child)
183                                         child->object.flags |= UNINTERESTING;
184                         }
185                         break;
186                 case OBJ_BLOB:
187                         if (tree->object.flags & UNINTERESTING) {
188                                 struct blob *child = lookup_blob(r, &entry.oid);
189                                 if (child)
190                                         child->object.flags |= UNINTERESTING;
191                         }
192                         break;
193                 default:
194                         /* Subproject commit - not in this repository */
195                         break;
196                 }
197         }
198
199         free_tree_buffer(tree);
200 }
201
202 void mark_trees_uninteresting_sparse(struct repository *r,
203                                      struct oidset *trees)
204 {
205         unsigned has_interesting = 0, has_uninteresting = 0;
206         struct hashmap map;
207         struct hashmap_iter map_iter;
208         struct path_and_oids_entry *entry;
209         struct object_id *oid;
210         struct oidset_iter iter;
211
212         oidset_iter_init(trees, &iter);
213         while ((!has_interesting || !has_uninteresting) &&
214                (oid = oidset_iter_next(&iter))) {
215                 struct tree *tree = lookup_tree(r, oid);
216
217                 if (!tree)
218                         continue;
219
220                 if (tree->object.flags & UNINTERESTING)
221                         has_uninteresting = 1;
222                 else
223                         has_interesting = 1;
224         }
225
226         /* Do not walk unless we have both types of trees. */
227         if (!has_uninteresting || !has_interesting)
228                 return;
229
230         paths_and_oids_init(&map);
231
232         oidset_iter_init(trees, &iter);
233         while ((oid = oidset_iter_next(&iter))) {
234                 struct tree *tree = lookup_tree(r, oid);
235                 add_children_by_path(r, tree, &map);
236         }
237
238         hashmap_iter_init(&map, &map_iter);
239         while ((entry = hashmap_iter_next(&map_iter)))
240                 mark_trees_uninteresting_sparse(r, &entry->trees);
241
242         paths_and_oids_clear(&map);
243 }
244
245 struct commit_stack {
246         struct commit **items;
247         size_t nr, alloc;
248 };
249 #define COMMIT_STACK_INIT { NULL, 0, 0 }
250
251 static void commit_stack_push(struct commit_stack *stack, struct commit *commit)
252 {
253         ALLOC_GROW(stack->items, stack->nr + 1, stack->alloc);
254         stack->items[stack->nr++] = commit;
255 }
256
257 static struct commit *commit_stack_pop(struct commit_stack *stack)
258 {
259         return stack->nr ? stack->items[--stack->nr] : NULL;
260 }
261
262 static void commit_stack_clear(struct commit_stack *stack)
263 {
264         FREE_AND_NULL(stack->items);
265         stack->nr = stack->alloc = 0;
266 }
267
268 static void mark_one_parent_uninteresting(struct commit *commit,
269                                           struct commit_stack *pending)
270 {
271         struct commit_list *l;
272
273         if (commit->object.flags & UNINTERESTING)
274                 return;
275         commit->object.flags |= UNINTERESTING;
276
277         /*
278          * Normally we haven't parsed the parent
279          * yet, so we won't have a parent of a parent
280          * here. However, it may turn out that we've
281          * reached this commit some other way (where it
282          * wasn't uninteresting), in which case we need
283          * to mark its parents recursively too..
284          */
285         for (l = commit->parents; l; l = l->next)
286                 commit_stack_push(pending, l->item);
287 }
288
289 void mark_parents_uninteresting(struct commit *commit)
290 {
291         struct commit_stack pending = COMMIT_STACK_INIT;
292         struct commit_list *l;
293
294         for (l = commit->parents; l; l = l->next)
295                 mark_one_parent_uninteresting(l->item, &pending);
296
297         while (pending.nr > 0)
298                 mark_one_parent_uninteresting(commit_stack_pop(&pending),
299                                               &pending);
300
301         commit_stack_clear(&pending);
302 }
303
304 static void add_pending_object_with_path(struct rev_info *revs,
305                                          struct object *obj,
306                                          const char *name, unsigned mode,
307                                          const char *path)
308 {
309         if (!obj)
310                 return;
311         if (revs->no_walk && (obj->flags & UNINTERESTING))
312                 revs->no_walk = 0;
313         if (revs->reflog_info && obj->type == OBJ_COMMIT) {
314                 struct strbuf buf = STRBUF_INIT;
315                 int len = interpret_branch_name(name, 0, &buf, 0);
316
317                 if (0 < len && name[len] && buf.len)
318                         strbuf_addstr(&buf, name + len);
319                 add_reflog_for_walk(revs->reflog_info,
320                                     (struct commit *)obj,
321                                     buf.buf[0] ? buf.buf: name);
322                 strbuf_release(&buf);
323                 return; /* do not add the commit itself */
324         }
325         add_object_array_with_path(obj, name, &revs->pending, mode, path);
326 }
327
328 static void add_pending_object_with_mode(struct rev_info *revs,
329                                          struct object *obj,
330                                          const char *name, unsigned mode)
331 {
332         add_pending_object_with_path(revs, obj, name, mode, NULL);
333 }
334
335 void add_pending_object(struct rev_info *revs,
336                         struct object *obj, const char *name)
337 {
338         add_pending_object_with_mode(revs, obj, name, S_IFINVALID);
339 }
340
341 void add_head_to_pending(struct rev_info *revs)
342 {
343         struct object_id oid;
344         struct object *obj;
345         if (get_oid("HEAD", &oid))
346                 return;
347         obj = parse_object(revs->repo, &oid);
348         if (!obj)
349                 return;
350         add_pending_object(revs, obj, "HEAD");
351 }
352
353 static struct object *get_reference(struct rev_info *revs, const char *name,
354                                     const struct object_id *oid,
355                                     unsigned int flags)
356 {
357         struct object *object;
358
359         /*
360          * If the repository has commit graphs, repo_parse_commit() avoids
361          * reading the object buffer, so use it whenever possible.
362          */
363         if (oid_object_info(revs->repo, oid, NULL) == OBJ_COMMIT) {
364                 struct commit *c = lookup_commit(revs->repo, oid);
365                 if (!repo_parse_commit(revs->repo, c))
366                         object = (struct object *) c;
367                 else
368                         object = NULL;
369         } else {
370                 object = parse_object(revs->repo, oid);
371         }
372
373         if (!object) {
374                 if (revs->ignore_missing)
375                         return object;
376                 if (revs->exclude_promisor_objects && is_promisor_object(oid))
377                         return NULL;
378                 die("bad object %s", name);
379         }
380         object->flags |= flags;
381         return object;
382 }
383
384 void add_pending_oid(struct rev_info *revs, const char *name,
385                       const struct object_id *oid, unsigned int flags)
386 {
387         struct object *object = get_reference(revs, name, oid, flags);
388         add_pending_object(revs, object, name);
389 }
390
391 static struct commit *handle_commit(struct rev_info *revs,
392                                     struct object_array_entry *entry)
393 {
394         struct object *object = entry->item;
395         const char *name = entry->name;
396         const char *path = entry->path;
397         unsigned int mode = entry->mode;
398         unsigned long flags = object->flags;
399
400         /*
401          * Tag object? Look what it points to..
402          */
403         while (object->type == OBJ_TAG) {
404                 struct tag *tag = (struct tag *) object;
405                 if (revs->tag_objects && !(flags & UNINTERESTING))
406                         add_pending_object(revs, object, tag->tag);
407                 object = parse_object(revs->repo, get_tagged_oid(tag));
408                 if (!object) {
409                         if (revs->ignore_missing_links || (flags & UNINTERESTING))
410                                 return NULL;
411                         if (revs->exclude_promisor_objects &&
412                             is_promisor_object(&tag->tagged->oid))
413                                 return NULL;
414                         die("bad object %s", oid_to_hex(&tag->tagged->oid));
415                 }
416                 object->flags |= flags;
417                 /*
418                  * We'll handle the tagged object by looping or dropping
419                  * through to the non-tag handlers below. Do not
420                  * propagate path data from the tag's pending entry.
421                  */
422                 path = NULL;
423                 mode = 0;
424         }
425
426         /*
427          * Commit object? Just return it, we'll do all the complex
428          * reachability crud.
429          */
430         if (object->type == OBJ_COMMIT) {
431                 struct commit *commit = (struct commit *)object;
432
433                 if (parse_commit(commit) < 0)
434                         die("unable to parse commit %s", name);
435                 if (flags & UNINTERESTING) {
436                         mark_parents_uninteresting(commit);
437
438                         if (!revs->topo_order || !generation_numbers_enabled(the_repository))
439                                 revs->limited = 1;
440                 }
441                 if (revs->sources) {
442                         char **slot = revision_sources_at(revs->sources, commit);
443
444                         if (!*slot)
445                                 *slot = xstrdup(name);
446                 }
447                 return commit;
448         }
449
450         /*
451          * Tree object? Either mark it uninteresting, or add it
452          * to the list of objects to look at later..
453          */
454         if (object->type == OBJ_TREE) {
455                 struct tree *tree = (struct tree *)object;
456                 if (!revs->tree_objects)
457                         return NULL;
458                 if (flags & UNINTERESTING) {
459                         mark_tree_contents_uninteresting(revs->repo, tree);
460                         return NULL;
461                 }
462                 add_pending_object_with_path(revs, object, name, mode, path);
463                 return NULL;
464         }
465
466         /*
467          * Blob object? You know the drill by now..
468          */
469         if (object->type == OBJ_BLOB) {
470                 if (!revs->blob_objects)
471                         return NULL;
472                 if (flags & UNINTERESTING)
473                         return NULL;
474                 add_pending_object_with_path(revs, object, name, mode, path);
475                 return NULL;
476         }
477         die("%s is unknown object", name);
478 }
479
480 static int everybody_uninteresting(struct commit_list *orig,
481                                    struct commit **interesting_cache)
482 {
483         struct commit_list *list = orig;
484
485         if (*interesting_cache) {
486                 struct commit *commit = *interesting_cache;
487                 if (!(commit->object.flags & UNINTERESTING))
488                         return 0;
489         }
490
491         while (list) {
492                 struct commit *commit = list->item;
493                 list = list->next;
494                 if (commit->object.flags & UNINTERESTING)
495                         continue;
496
497                 *interesting_cache = commit;
498                 return 0;
499         }
500         return 1;
501 }
502
503 /*
504  * A definition of "relevant" commit that we can use to simplify limited graphs
505  * by eliminating side branches.
506  *
507  * A "relevant" commit is one that is !UNINTERESTING (ie we are including it
508  * in our list), or that is a specified BOTTOM commit. Then after computing
509  * a limited list, during processing we can generally ignore boundary merges
510  * coming from outside the graph, (ie from irrelevant parents), and treat
511  * those merges as if they were single-parent. TREESAME is defined to consider
512  * only relevant parents, if any. If we are TREESAME to our on-graph parents,
513  * we don't care if we were !TREESAME to non-graph parents.
514  *
515  * Treating bottom commits as relevant ensures that a limited graph's
516  * connection to the actual bottom commit is not viewed as a side branch, but
517  * treated as part of the graph. For example:
518  *
519  *   ....Z...A---X---o---o---B
520  *        .     /
521  *         W---Y
522  *
523  * When computing "A..B", the A-X connection is at least as important as
524  * Y-X, despite A being flagged UNINTERESTING.
525  *
526  * And when computing --ancestry-path "A..B", the A-X connection is more
527  * important than Y-X, despite both A and Y being flagged UNINTERESTING.
528  */
529 static inline int relevant_commit(struct commit *commit)
530 {
531         return (commit->object.flags & (UNINTERESTING | BOTTOM)) != UNINTERESTING;
532 }
533
534 /*
535  * Return a single relevant commit from a parent list. If we are a TREESAME
536  * commit, and this selects one of our parents, then we can safely simplify to
537  * that parent.
538  */
539 static struct commit *one_relevant_parent(const struct rev_info *revs,
540                                           struct commit_list *orig)
541 {
542         struct commit_list *list = orig;
543         struct commit *relevant = NULL;
544
545         if (!orig)
546                 return NULL;
547
548         /*
549          * For 1-parent commits, or if first-parent-only, then return that
550          * first parent (even if not "relevant" by the above definition).
551          * TREESAME will have been set purely on that parent.
552          */
553         if (revs->first_parent_only || !orig->next)
554                 return orig->item;
555
556         /*
557          * For multi-parent commits, identify a sole relevant parent, if any.
558          * If we have only one relevant parent, then TREESAME will be set purely
559          * with regard to that parent, and we can simplify accordingly.
560          *
561          * If we have more than one relevant parent, or no relevant parents
562          * (and multiple irrelevant ones), then we can't select a parent here
563          * and return NULL.
564          */
565         while (list) {
566                 struct commit *commit = list->item;
567                 list = list->next;
568                 if (relevant_commit(commit)) {
569                         if (relevant)
570                                 return NULL;
571                         relevant = commit;
572                 }
573         }
574         return relevant;
575 }
576
577 /*
578  * The goal is to get REV_TREE_NEW as the result only if the
579  * diff consists of all '+' (and no other changes), REV_TREE_OLD
580  * if the whole diff is removal of old data, and otherwise
581  * REV_TREE_DIFFERENT (of course if the trees are the same we
582  * want REV_TREE_SAME).
583  *
584  * The only time we care about the distinction is when
585  * remove_empty_trees is in effect, in which case we care only about
586  * whether the whole change is REV_TREE_NEW, or if there's another type
587  * of change. Which means we can stop the diff early in either of these
588  * cases:
589  *
590  *   1. We're not using remove_empty_trees at all.
591  *
592  *   2. We saw anything except REV_TREE_NEW.
593  */
594 static int tree_difference = REV_TREE_SAME;
595
596 static void file_add_remove(struct diff_options *options,
597                     int addremove, unsigned mode,
598                     const struct object_id *oid,
599                     int oid_valid,
600                     const char *fullpath, unsigned dirty_submodule)
601 {
602         int diff = addremove == '+' ? REV_TREE_NEW : REV_TREE_OLD;
603         struct rev_info *revs = options->change_fn_data;
604
605         tree_difference |= diff;
606         if (!revs->remove_empty_trees || tree_difference != REV_TREE_NEW)
607                 options->flags.has_changes = 1;
608 }
609
610 static void file_change(struct diff_options *options,
611                  unsigned old_mode, unsigned new_mode,
612                  const struct object_id *old_oid,
613                  const struct object_id *new_oid,
614                  int old_oid_valid, int new_oid_valid,
615                  const char *fullpath,
616                  unsigned old_dirty_submodule, unsigned new_dirty_submodule)
617 {
618         tree_difference = REV_TREE_DIFFERENT;
619         options->flags.has_changes = 1;
620 }
621
622 static int rev_compare_tree(struct rev_info *revs,
623                             struct commit *parent, struct commit *commit)
624 {
625         struct tree *t1 = get_commit_tree(parent);
626         struct tree *t2 = get_commit_tree(commit);
627
628         if (!t1)
629                 return REV_TREE_NEW;
630         if (!t2)
631                 return REV_TREE_OLD;
632
633         if (revs->simplify_by_decoration) {
634                 /*
635                  * If we are simplifying by decoration, then the commit
636                  * is worth showing if it has a tag pointing at it.
637                  */
638                 if (get_name_decoration(&commit->object))
639                         return REV_TREE_DIFFERENT;
640                 /*
641                  * A commit that is not pointed by a tag is uninteresting
642                  * if we are not limited by path.  This means that you will
643                  * see the usual "commits that touch the paths" plus any
644                  * tagged commit by specifying both --simplify-by-decoration
645                  * and pathspec.
646                  */
647                 if (!revs->prune_data.nr)
648                         return REV_TREE_SAME;
649         }
650
651         tree_difference = REV_TREE_SAME;
652         revs->pruning.flags.has_changes = 0;
653         if (diff_tree_oid(&t1->object.oid, &t2->object.oid, "",
654                            &revs->pruning) < 0)
655                 return REV_TREE_DIFFERENT;
656         return tree_difference;
657 }
658
659 static int rev_same_tree_as_empty(struct rev_info *revs, struct commit *commit)
660 {
661         int retval;
662         struct tree *t1 = get_commit_tree(commit);
663
664         if (!t1)
665                 return 0;
666
667         tree_difference = REV_TREE_SAME;
668         revs->pruning.flags.has_changes = 0;
669         retval = diff_tree_oid(NULL, &t1->object.oid, "", &revs->pruning);
670
671         return retval >= 0 && (tree_difference == REV_TREE_SAME);
672 }
673
674 struct treesame_state {
675         unsigned int nparents;
676         unsigned char treesame[FLEX_ARRAY];
677 };
678
679 static struct treesame_state *initialise_treesame(struct rev_info *revs, struct commit *commit)
680 {
681         unsigned n = commit_list_count(commit->parents);
682         struct treesame_state *st = xcalloc(1, st_add(sizeof(*st), n));
683         st->nparents = n;
684         add_decoration(&revs->treesame, &commit->object, st);
685         return st;
686 }
687
688 /*
689  * Must be called immediately after removing the nth_parent from a commit's
690  * parent list, if we are maintaining the per-parent treesame[] decoration.
691  * This does not recalculate the master TREESAME flag - update_treesame()
692  * should be called to update it after a sequence of treesame[] modifications
693  * that may have affected it.
694  */
695 static int compact_treesame(struct rev_info *revs, struct commit *commit, unsigned nth_parent)
696 {
697         struct treesame_state *st;
698         int old_same;
699
700         if (!commit->parents) {
701                 /*
702                  * Have just removed the only parent from a non-merge.
703                  * Different handling, as we lack decoration.
704                  */
705                 if (nth_parent != 0)
706                         die("compact_treesame %u", nth_parent);
707                 old_same = !!(commit->object.flags & TREESAME);
708                 if (rev_same_tree_as_empty(revs, commit))
709                         commit->object.flags |= TREESAME;
710                 else
711                         commit->object.flags &= ~TREESAME;
712                 return old_same;
713         }
714
715         st = lookup_decoration(&revs->treesame, &commit->object);
716         if (!st || nth_parent >= st->nparents)
717                 die("compact_treesame %u", nth_parent);
718
719         old_same = st->treesame[nth_parent];
720         memmove(st->treesame + nth_parent,
721                 st->treesame + nth_parent + 1,
722                 st->nparents - nth_parent - 1);
723
724         /*
725          * If we've just become a non-merge commit, update TREESAME
726          * immediately, and remove the no-longer-needed decoration.
727          * If still a merge, defer update until update_treesame().
728          */
729         if (--st->nparents == 1) {
730                 if (commit->parents->next)
731                         die("compact_treesame parents mismatch");
732                 if (st->treesame[0] && revs->dense)
733                         commit->object.flags |= TREESAME;
734                 else
735                         commit->object.flags &= ~TREESAME;
736                 free(add_decoration(&revs->treesame, &commit->object, NULL));
737         }
738
739         return old_same;
740 }
741
742 static unsigned update_treesame(struct rev_info *revs, struct commit *commit)
743 {
744         if (commit->parents && commit->parents->next) {
745                 unsigned n;
746                 struct treesame_state *st;
747                 struct commit_list *p;
748                 unsigned relevant_parents;
749                 unsigned relevant_change, irrelevant_change;
750
751                 st = lookup_decoration(&revs->treesame, &commit->object);
752                 if (!st)
753                         die("update_treesame %s", oid_to_hex(&commit->object.oid));
754                 relevant_parents = 0;
755                 relevant_change = irrelevant_change = 0;
756                 for (p = commit->parents, n = 0; p; n++, p = p->next) {
757                         if (relevant_commit(p->item)) {
758                                 relevant_change |= !st->treesame[n];
759                                 relevant_parents++;
760                         } else
761                                 irrelevant_change |= !st->treesame[n];
762                 }
763                 if (relevant_parents ? relevant_change : irrelevant_change)
764                         commit->object.flags &= ~TREESAME;
765                 else
766                         commit->object.flags |= TREESAME;
767         }
768
769         return commit->object.flags & TREESAME;
770 }
771
772 static inline int limiting_can_increase_treesame(const struct rev_info *revs)
773 {
774         /*
775          * TREESAME is irrelevant unless prune && dense;
776          * if simplify_history is set, we can't have a mixture of TREESAME and
777          *    !TREESAME INTERESTING parents (and we don't have treesame[]
778          *    decoration anyway);
779          * if first_parent_only is set, then the TREESAME flag is locked
780          *    against the first parent (and again we lack treesame[] decoration).
781          */
782         return revs->prune && revs->dense &&
783                !revs->simplify_history &&
784                !revs->first_parent_only;
785 }
786
787 static void try_to_simplify_commit(struct rev_info *revs, struct commit *commit)
788 {
789         struct commit_list **pp, *parent;
790         struct treesame_state *ts = NULL;
791         int relevant_change = 0, irrelevant_change = 0;
792         int relevant_parents, nth_parent;
793
794         /*
795          * If we don't do pruning, everything is interesting
796          */
797         if (!revs->prune)
798                 return;
799
800         if (!get_commit_tree(commit))
801                 return;
802
803         if (!commit->parents) {
804                 if (rev_same_tree_as_empty(revs, commit))
805                         commit->object.flags |= TREESAME;
806                 return;
807         }
808
809         /*
810          * Normal non-merge commit? If we don't want to make the
811          * history dense, we consider it always to be a change..
812          */
813         if (!revs->dense && !commit->parents->next)
814                 return;
815
816         for (pp = &commit->parents, nth_parent = 0, relevant_parents = 0;
817              (parent = *pp) != NULL;
818              pp = &parent->next, nth_parent++) {
819                 struct commit *p = parent->item;
820                 if (relevant_commit(p))
821                         relevant_parents++;
822
823                 if (nth_parent == 1) {
824                         /*
825                          * This our second loop iteration - so we now know
826                          * we're dealing with a merge.
827                          *
828                          * Do not compare with later parents when we care only about
829                          * the first parent chain, in order to avoid derailing the
830                          * traversal to follow a side branch that brought everything
831                          * in the path we are limited to by the pathspec.
832                          */
833                         if (revs->first_parent_only)
834                                 break;
835                         /*
836                          * If this will remain a potentially-simplifiable
837                          * merge, remember per-parent treesame if needed.
838                          * Initialise the array with the comparison from our
839                          * first iteration.
840                          */
841                         if (revs->treesame.name &&
842                             !revs->simplify_history &&
843                             !(commit->object.flags & UNINTERESTING)) {
844                                 ts = initialise_treesame(revs, commit);
845                                 if (!(irrelevant_change || relevant_change))
846                                         ts->treesame[0] = 1;
847                         }
848                 }
849                 if (parse_commit(p) < 0)
850                         die("cannot simplify commit %s (because of %s)",
851                             oid_to_hex(&commit->object.oid),
852                             oid_to_hex(&p->object.oid));
853                 switch (rev_compare_tree(revs, p, commit)) {
854                 case REV_TREE_SAME:
855                         if (!revs->simplify_history || !relevant_commit(p)) {
856                                 /* Even if a merge with an uninteresting
857                                  * side branch brought the entire change
858                                  * we are interested in, we do not want
859                                  * to lose the other branches of this
860                                  * merge, so we just keep going.
861                                  */
862                                 if (ts)
863                                         ts->treesame[nth_parent] = 1;
864                                 continue;
865                         }
866                         parent->next = NULL;
867                         commit->parents = parent;
868                         commit->object.flags |= TREESAME;
869                         return;
870
871                 case REV_TREE_NEW:
872                         if (revs->remove_empty_trees &&
873                             rev_same_tree_as_empty(revs, p)) {
874                                 /* We are adding all the specified
875                                  * paths from this parent, so the
876                                  * history beyond this parent is not
877                                  * interesting.  Remove its parents
878                                  * (they are grandparents for us).
879                                  * IOW, we pretend this parent is a
880                                  * "root" commit.
881                                  */
882                                 if (parse_commit(p) < 0)
883                                         die("cannot simplify commit %s (invalid %s)",
884                                             oid_to_hex(&commit->object.oid),
885                                             oid_to_hex(&p->object.oid));
886                                 p->parents = NULL;
887                         }
888                 /* fallthrough */
889                 case REV_TREE_OLD:
890                 case REV_TREE_DIFFERENT:
891                         if (relevant_commit(p))
892                                 relevant_change = 1;
893                         else
894                                 irrelevant_change = 1;
895                         continue;
896                 }
897                 die("bad tree compare for commit %s", oid_to_hex(&commit->object.oid));
898         }
899
900         /*
901          * TREESAME is straightforward for single-parent commits. For merge
902          * commits, it is most useful to define it so that "irrelevant"
903          * parents cannot make us !TREESAME - if we have any relevant
904          * parents, then we only consider TREESAMEness with respect to them,
905          * allowing irrelevant merges from uninteresting branches to be
906          * simplified away. Only if we have only irrelevant parents do we
907          * base TREESAME on them. Note that this logic is replicated in
908          * update_treesame, which should be kept in sync.
909          */
910         if (relevant_parents ? !relevant_change : !irrelevant_change)
911                 commit->object.flags |= TREESAME;
912 }
913
914 static int process_parents(struct rev_info *revs, struct commit *commit,
915                            struct commit_list **list, struct prio_queue *queue)
916 {
917         struct commit_list *parent = commit->parents;
918         unsigned left_flag;
919
920         if (commit->object.flags & ADDED)
921                 return 0;
922         commit->object.flags |= ADDED;
923
924         if (revs->include_check &&
925             !revs->include_check(commit, revs->include_check_data))
926                 return 0;
927
928         /*
929          * If the commit is uninteresting, don't try to
930          * prune parents - we want the maximal uninteresting
931          * set.
932          *
933          * Normally we haven't parsed the parent
934          * yet, so we won't have a parent of a parent
935          * here. However, it may turn out that we've
936          * reached this commit some other way (where it
937          * wasn't uninteresting), in which case we need
938          * to mark its parents recursively too..
939          */
940         if (commit->object.flags & UNINTERESTING) {
941                 while (parent) {
942                         struct commit *p = parent->item;
943                         parent = parent->next;
944                         if (p)
945                                 p->object.flags |= UNINTERESTING;
946                         if (parse_commit_gently(p, 1) < 0)
947                                 continue;
948                         if (p->parents)
949                                 mark_parents_uninteresting(p);
950                         if (p->object.flags & SEEN)
951                                 continue;
952                         p->object.flags |= SEEN;
953                         if (list)
954                                 commit_list_insert_by_date(p, list);
955                         if (queue)
956                                 prio_queue_put(queue, p);
957                 }
958                 return 0;
959         }
960
961         /*
962          * Ok, the commit wasn't uninteresting. Try to
963          * simplify the commit history and find the parent
964          * that has no differences in the path set if one exists.
965          */
966         try_to_simplify_commit(revs, commit);
967
968         if (revs->no_walk)
969                 return 0;
970
971         left_flag = (commit->object.flags & SYMMETRIC_LEFT);
972
973         for (parent = commit->parents; parent; parent = parent->next) {
974                 struct commit *p = parent->item;
975                 int gently = revs->ignore_missing_links ||
976                              revs->exclude_promisor_objects;
977                 if (parse_commit_gently(p, gently) < 0) {
978                         if (revs->exclude_promisor_objects &&
979                             is_promisor_object(&p->object.oid)) {
980                                 if (revs->first_parent_only)
981                                         break;
982                                 continue;
983                         }
984                         return -1;
985                 }
986                 if (revs->sources) {
987                         char **slot = revision_sources_at(revs->sources, p);
988
989                         if (!*slot)
990                                 *slot = *revision_sources_at(revs->sources, commit);
991                 }
992                 p->object.flags |= left_flag;
993                 if (!(p->object.flags & SEEN)) {
994                         p->object.flags |= SEEN;
995                         if (list)
996                                 commit_list_insert_by_date(p, list);
997                         if (queue)
998                                 prio_queue_put(queue, p);
999                 }
1000                 if (revs->first_parent_only)
1001                         break;
1002         }
1003         return 0;
1004 }
1005
1006 static void cherry_pick_list(struct commit_list *list, struct rev_info *revs)
1007 {
1008         struct commit_list *p;
1009         int left_count = 0, right_count = 0;
1010         int left_first;
1011         struct patch_ids ids;
1012         unsigned cherry_flag;
1013
1014         /* First count the commits on the left and on the right */
1015         for (p = list; p; p = p->next) {
1016                 struct commit *commit = p->item;
1017                 unsigned flags = commit->object.flags;
1018                 if (flags & BOUNDARY)
1019                         ;
1020                 else if (flags & SYMMETRIC_LEFT)
1021                         left_count++;
1022                 else
1023                         right_count++;
1024         }
1025
1026         if (!left_count || !right_count)
1027                 return;
1028
1029         left_first = left_count < right_count;
1030         init_patch_ids(revs->repo, &ids);
1031         ids.diffopts.pathspec = revs->diffopt.pathspec;
1032
1033         /* Compute patch-ids for one side */
1034         for (p = list; p; p = p->next) {
1035                 struct commit *commit = p->item;
1036                 unsigned flags = commit->object.flags;
1037
1038                 if (flags & BOUNDARY)
1039                         continue;
1040                 /*
1041                  * If we have fewer left, left_first is set and we omit
1042                  * commits on the right branch in this loop.  If we have
1043                  * fewer right, we skip the left ones.
1044                  */
1045                 if (left_first != !!(flags & SYMMETRIC_LEFT))
1046                         continue;
1047                 add_commit_patch_id(commit, &ids);
1048         }
1049
1050         /* either cherry_mark or cherry_pick are true */
1051         cherry_flag = revs->cherry_mark ? PATCHSAME : SHOWN;
1052
1053         /* Check the other side */
1054         for (p = list; p; p = p->next) {
1055                 struct commit *commit = p->item;
1056                 struct patch_id *id;
1057                 unsigned flags = commit->object.flags;
1058
1059                 if (flags & BOUNDARY)
1060                         continue;
1061                 /*
1062                  * If we have fewer left, left_first is set and we omit
1063                  * commits on the left branch in this loop.
1064                  */
1065                 if (left_first == !!(flags & SYMMETRIC_LEFT))
1066                         continue;
1067
1068                 /*
1069                  * Have we seen the same patch id?
1070                  */
1071                 id = has_commit_patch_id(commit, &ids);
1072                 if (!id)
1073                         continue;
1074
1075                 commit->object.flags |= cherry_flag;
1076                 id->commit->object.flags |= cherry_flag;
1077         }
1078
1079         free_patch_ids(&ids);
1080 }
1081
1082 /* How many extra uninteresting commits we want to see.. */
1083 #define SLOP 5
1084
1085 static int still_interesting(struct commit_list *src, timestamp_t date, int slop,
1086                              struct commit **interesting_cache)
1087 {
1088         /*
1089          * No source list at all? We're definitely done..
1090          */
1091         if (!src)
1092                 return 0;
1093
1094         /*
1095          * Does the destination list contain entries with a date
1096          * before the source list? Definitely _not_ done.
1097          */
1098         if (date <= src->item->date)
1099                 return SLOP;
1100
1101         /*
1102          * Does the source list still have interesting commits in
1103          * it? Definitely not done..
1104          */
1105         if (!everybody_uninteresting(src, interesting_cache))
1106                 return SLOP;
1107
1108         /* Ok, we're closing in.. */
1109         return slop-1;
1110 }
1111
1112 /*
1113  * "rev-list --ancestry-path A..B" computes commits that are ancestors
1114  * of B but not ancestors of A but further limits the result to those
1115  * that are descendants of A.  This takes the list of bottom commits and
1116  * the result of "A..B" without --ancestry-path, and limits the latter
1117  * further to the ones that can reach one of the commits in "bottom".
1118  */
1119 static void limit_to_ancestry(struct commit_list *bottom, struct commit_list *list)
1120 {
1121         struct commit_list *p;
1122         struct commit_list *rlist = NULL;
1123         int made_progress;
1124
1125         /*
1126          * Reverse the list so that it will be likely that we would
1127          * process parents before children.
1128          */
1129         for (p = list; p; p = p->next)
1130                 commit_list_insert(p->item, &rlist);
1131
1132         for (p = bottom; p; p = p->next)
1133                 p->item->object.flags |= TMP_MARK;
1134
1135         /*
1136          * Mark the ones that can reach bottom commits in "list",
1137          * in a bottom-up fashion.
1138          */
1139         do {
1140                 made_progress = 0;
1141                 for (p = rlist; p; p = p->next) {
1142                         struct commit *c = p->item;
1143                         struct commit_list *parents;
1144                         if (c->object.flags & (TMP_MARK | UNINTERESTING))
1145                                 continue;
1146                         for (parents = c->parents;
1147                              parents;
1148                              parents = parents->next) {
1149                                 if (!(parents->item->object.flags & TMP_MARK))
1150                                         continue;
1151                                 c->object.flags |= TMP_MARK;
1152                                 made_progress = 1;
1153                                 break;
1154                         }
1155                 }
1156         } while (made_progress);
1157
1158         /*
1159          * NEEDSWORK: decide if we want to remove parents that are
1160          * not marked with TMP_MARK from commit->parents for commits
1161          * in the resulting list.  We may not want to do that, though.
1162          */
1163
1164         /*
1165          * The ones that are not marked with TMP_MARK are uninteresting
1166          */
1167         for (p = list; p; p = p->next) {
1168                 struct commit *c = p->item;
1169                 if (c->object.flags & TMP_MARK)
1170                         continue;
1171                 c->object.flags |= UNINTERESTING;
1172         }
1173
1174         /* We are done with the TMP_MARK */
1175         for (p = list; p; p = p->next)
1176                 p->item->object.flags &= ~TMP_MARK;
1177         for (p = bottom; p; p = p->next)
1178                 p->item->object.flags &= ~TMP_MARK;
1179         free_commit_list(rlist);
1180 }
1181
1182 /*
1183  * Before walking the history, keep the set of "negative" refs the
1184  * caller has asked to exclude.
1185  *
1186  * This is used to compute "rev-list --ancestry-path A..B", as we need
1187  * to filter the result of "A..B" further to the ones that can actually
1188  * reach A.
1189  */
1190 static struct commit_list *collect_bottom_commits(struct commit_list *list)
1191 {
1192         struct commit_list *elem, *bottom = NULL;
1193         for (elem = list; elem; elem = elem->next)
1194                 if (elem->item->object.flags & BOTTOM)
1195                         commit_list_insert(elem->item, &bottom);
1196         return bottom;
1197 }
1198
1199 /* Assumes either left_only or right_only is set */
1200 static void limit_left_right(struct commit_list *list, struct rev_info *revs)
1201 {
1202         struct commit_list *p;
1203
1204         for (p = list; p; p = p->next) {
1205                 struct commit *commit = p->item;
1206
1207                 if (revs->right_only) {
1208                         if (commit->object.flags & SYMMETRIC_LEFT)
1209                                 commit->object.flags |= SHOWN;
1210                 } else  /* revs->left_only is set */
1211                         if (!(commit->object.flags & SYMMETRIC_LEFT))
1212                                 commit->object.flags |= SHOWN;
1213         }
1214 }
1215
1216 static int limit_list(struct rev_info *revs)
1217 {
1218         int slop = SLOP;
1219         timestamp_t date = TIME_MAX;
1220         struct commit_list *list = revs->commits;
1221         struct commit_list *newlist = NULL;
1222         struct commit_list **p = &newlist;
1223         struct commit_list *bottom = NULL;
1224         struct commit *interesting_cache = NULL;
1225
1226         if (revs->ancestry_path) {
1227                 bottom = collect_bottom_commits(list);
1228                 if (!bottom)
1229                         die("--ancestry-path given but there are no bottom commits");
1230         }
1231
1232         while (list) {
1233                 struct commit *commit = pop_commit(&list);
1234                 struct object *obj = &commit->object;
1235                 show_early_output_fn_t show;
1236
1237                 if (commit == interesting_cache)
1238                         interesting_cache = NULL;
1239
1240                 if (revs->max_age != -1 && (commit->date < revs->max_age))
1241                         obj->flags |= UNINTERESTING;
1242                 if (process_parents(revs, commit, &list, NULL) < 0)
1243                         return -1;
1244                 if (obj->flags & UNINTERESTING) {
1245                         mark_parents_uninteresting(commit);
1246                         slop = still_interesting(list, date, slop, &interesting_cache);
1247                         if (slop)
1248                                 continue;
1249                         break;
1250                 }
1251                 if (revs->min_age != -1 && (commit->date > revs->min_age))
1252                         continue;
1253                 date = commit->date;
1254                 p = &commit_list_insert(commit, p)->next;
1255
1256                 show = show_early_output;
1257                 if (!show)
1258                         continue;
1259
1260                 show(revs, newlist);
1261                 show_early_output = NULL;
1262         }
1263         if (revs->cherry_pick || revs->cherry_mark)
1264                 cherry_pick_list(newlist, revs);
1265
1266         if (revs->left_only || revs->right_only)
1267                 limit_left_right(newlist, revs);
1268
1269         if (bottom) {
1270                 limit_to_ancestry(bottom, newlist);
1271                 free_commit_list(bottom);
1272         }
1273
1274         /*
1275          * Check if any commits have become TREESAME by some of their parents
1276          * becoming UNINTERESTING.
1277          */
1278         if (limiting_can_increase_treesame(revs))
1279                 for (list = newlist; list; list = list->next) {
1280                         struct commit *c = list->item;
1281                         if (c->object.flags & (UNINTERESTING | TREESAME))
1282                                 continue;
1283                         update_treesame(revs, c);
1284                 }
1285
1286         revs->commits = newlist;
1287         return 0;
1288 }
1289
1290 /*
1291  * Add an entry to refs->cmdline with the specified information.
1292  * *name is copied.
1293  */
1294 static void add_rev_cmdline(struct rev_info *revs,
1295                             struct object *item,
1296                             const char *name,
1297                             int whence,
1298                             unsigned flags)
1299 {
1300         struct rev_cmdline_info *info = &revs->cmdline;
1301         unsigned int nr = info->nr;
1302
1303         ALLOC_GROW(info->rev, nr + 1, info->alloc);
1304         info->rev[nr].item = item;
1305         info->rev[nr].name = xstrdup(name);
1306         info->rev[nr].whence = whence;
1307         info->rev[nr].flags = flags;
1308         info->nr++;
1309 }
1310
1311 static void add_rev_cmdline_list(struct rev_info *revs,
1312                                  struct commit_list *commit_list,
1313                                  int whence,
1314                                  unsigned flags)
1315 {
1316         while (commit_list) {
1317                 struct object *object = &commit_list->item->object;
1318                 add_rev_cmdline(revs, object, oid_to_hex(&object->oid),
1319                                 whence, flags);
1320                 commit_list = commit_list->next;
1321         }
1322 }
1323
1324 struct all_refs_cb {
1325         int all_flags;
1326         int warned_bad_reflog;
1327         struct rev_info *all_revs;
1328         const char *name_for_errormsg;
1329         struct worktree *wt;
1330 };
1331
1332 int ref_excluded(struct string_list *ref_excludes, const char *path)
1333 {
1334         struct string_list_item *item;
1335
1336         if (!ref_excludes)
1337                 return 0;
1338         for_each_string_list_item(item, ref_excludes) {
1339                 if (!wildmatch(item->string, path, 0))
1340                         return 1;
1341         }
1342         return 0;
1343 }
1344
1345 static int handle_one_ref(const char *path, const struct object_id *oid,
1346                           int flag, void *cb_data)
1347 {
1348         struct all_refs_cb *cb = cb_data;
1349         struct object *object;
1350
1351         if (ref_excluded(cb->all_revs->ref_excludes, path))
1352             return 0;
1353
1354         object = get_reference(cb->all_revs, path, oid, cb->all_flags);
1355         add_rev_cmdline(cb->all_revs, object, path, REV_CMD_REF, cb->all_flags);
1356         add_pending_oid(cb->all_revs, path, oid, cb->all_flags);
1357         return 0;
1358 }
1359
1360 static void init_all_refs_cb(struct all_refs_cb *cb, struct rev_info *revs,
1361         unsigned flags)
1362 {
1363         cb->all_revs = revs;
1364         cb->all_flags = flags;
1365         revs->rev_input_given = 1;
1366         cb->wt = NULL;
1367 }
1368
1369 void clear_ref_exclusion(struct string_list **ref_excludes_p)
1370 {
1371         if (*ref_excludes_p) {
1372                 string_list_clear(*ref_excludes_p, 0);
1373                 free(*ref_excludes_p);
1374         }
1375         *ref_excludes_p = NULL;
1376 }
1377
1378 void add_ref_exclusion(struct string_list **ref_excludes_p, const char *exclude)
1379 {
1380         if (!*ref_excludes_p) {
1381                 *ref_excludes_p = xcalloc(1, sizeof(**ref_excludes_p));
1382                 (*ref_excludes_p)->strdup_strings = 1;
1383         }
1384         string_list_append(*ref_excludes_p, exclude);
1385 }
1386
1387 static void handle_refs(struct ref_store *refs,
1388                         struct rev_info *revs, unsigned flags,
1389                         int (*for_each)(struct ref_store *, each_ref_fn, void *))
1390 {
1391         struct all_refs_cb cb;
1392
1393         if (!refs) {
1394                 /* this could happen with uninitialized submodules */
1395                 return;
1396         }
1397
1398         init_all_refs_cb(&cb, revs, flags);
1399         for_each(refs, handle_one_ref, &cb);
1400 }
1401
1402 static void handle_one_reflog_commit(struct object_id *oid, void *cb_data)
1403 {
1404         struct all_refs_cb *cb = cb_data;
1405         if (!is_null_oid(oid)) {
1406                 struct object *o = parse_object(cb->all_revs->repo, oid);
1407                 if (o) {
1408                         o->flags |= cb->all_flags;
1409                         /* ??? CMDLINEFLAGS ??? */
1410                         add_pending_object(cb->all_revs, o, "");
1411                 }
1412                 else if (!cb->warned_bad_reflog) {
1413                         warning("reflog of '%s' references pruned commits",
1414                                 cb->name_for_errormsg);
1415                         cb->warned_bad_reflog = 1;
1416                 }
1417         }
1418 }
1419
1420 static int handle_one_reflog_ent(struct object_id *ooid, struct object_id *noid,
1421                 const char *email, timestamp_t timestamp, int tz,
1422                 const char *message, void *cb_data)
1423 {
1424         handle_one_reflog_commit(ooid, cb_data);
1425         handle_one_reflog_commit(noid, cb_data);
1426         return 0;
1427 }
1428
1429 static int handle_one_reflog(const char *refname_in_wt,
1430                              const struct object_id *oid,
1431                              int flag, void *cb_data)
1432 {
1433         struct all_refs_cb *cb = cb_data;
1434         struct strbuf refname = STRBUF_INIT;
1435
1436         cb->warned_bad_reflog = 0;
1437         strbuf_worktree_ref(cb->wt, &refname, refname_in_wt);
1438         cb->name_for_errormsg = refname.buf;
1439         refs_for_each_reflog_ent(get_main_ref_store(the_repository),
1440                                  refname.buf,
1441                                  handle_one_reflog_ent, cb_data);
1442         strbuf_release(&refname);
1443         return 0;
1444 }
1445
1446 static void add_other_reflogs_to_pending(struct all_refs_cb *cb)
1447 {
1448         struct worktree **worktrees, **p;
1449
1450         worktrees = get_worktrees(0);
1451         for (p = worktrees; *p; p++) {
1452                 struct worktree *wt = *p;
1453
1454                 if (wt->is_current)
1455                         continue;
1456
1457                 cb->wt = wt;
1458                 refs_for_each_reflog(get_worktree_ref_store(wt),
1459                                      handle_one_reflog,
1460                                      cb);
1461         }
1462         free_worktrees(worktrees);
1463 }
1464
1465 void add_reflogs_to_pending(struct rev_info *revs, unsigned flags)
1466 {
1467         struct all_refs_cb cb;
1468
1469         cb.all_revs = revs;
1470         cb.all_flags = flags;
1471         cb.wt = NULL;
1472         for_each_reflog(handle_one_reflog, &cb);
1473
1474         if (!revs->single_worktree)
1475                 add_other_reflogs_to_pending(&cb);
1476 }
1477
1478 static void add_cache_tree(struct cache_tree *it, struct rev_info *revs,
1479                            struct strbuf *path, unsigned int flags)
1480 {
1481         size_t baselen = path->len;
1482         int i;
1483
1484         if (it->entry_count >= 0) {
1485                 struct tree *tree = lookup_tree(revs->repo, &it->oid);
1486                 tree->object.flags |= flags;
1487                 add_pending_object_with_path(revs, &tree->object, "",
1488                                              040000, path->buf);
1489         }
1490
1491         for (i = 0; i < it->subtree_nr; i++) {
1492                 struct cache_tree_sub *sub = it->down[i];
1493                 strbuf_addf(path, "%s%s", baselen ? "/" : "", sub->name);
1494                 add_cache_tree(sub->cache_tree, revs, path, flags);
1495                 strbuf_setlen(path, baselen);
1496         }
1497
1498 }
1499
1500 static void do_add_index_objects_to_pending(struct rev_info *revs,
1501                                             struct index_state *istate,
1502                                             unsigned int flags)
1503 {
1504         int i;
1505
1506         for (i = 0; i < istate->cache_nr; i++) {
1507                 struct cache_entry *ce = istate->cache[i];
1508                 struct blob *blob;
1509
1510                 if (S_ISGITLINK(ce->ce_mode))
1511                         continue;
1512
1513                 blob = lookup_blob(revs->repo, &ce->oid);
1514                 if (!blob)
1515                         die("unable to add index blob to traversal");
1516                 blob->object.flags |= flags;
1517                 add_pending_object_with_path(revs, &blob->object, "",
1518                                              ce->ce_mode, ce->name);
1519         }
1520
1521         if (istate->cache_tree) {
1522                 struct strbuf path = STRBUF_INIT;
1523                 add_cache_tree(istate->cache_tree, revs, &path, flags);
1524                 strbuf_release(&path);
1525         }
1526 }
1527
1528 void add_index_objects_to_pending(struct rev_info *revs, unsigned int flags)
1529 {
1530         struct worktree **worktrees, **p;
1531
1532         repo_read_index(revs->repo);
1533         do_add_index_objects_to_pending(revs, revs->repo->index, flags);
1534
1535         if (revs->single_worktree)
1536                 return;
1537
1538         worktrees = get_worktrees(0);
1539         for (p = worktrees; *p; p++) {
1540                 struct worktree *wt = *p;
1541                 struct index_state istate = { NULL };
1542
1543                 if (wt->is_current)
1544                         continue; /* current index already taken care of */
1545
1546                 if (read_index_from(&istate,
1547                                     worktree_git_path(wt, "index"),
1548                                     get_worktree_git_dir(wt)) > 0)
1549                         do_add_index_objects_to_pending(revs, &istate, flags);
1550                 discard_index(&istate);
1551         }
1552         free_worktrees(worktrees);
1553 }
1554
1555 struct add_alternate_refs_data {
1556         struct rev_info *revs;
1557         unsigned int flags;
1558 };
1559
1560 static void add_one_alternate_ref(const struct object_id *oid,
1561                                   void *vdata)
1562 {
1563         const char *name = ".alternate";
1564         struct add_alternate_refs_data *data = vdata;
1565         struct object *obj;
1566
1567         obj = get_reference(data->revs, name, oid, data->flags);
1568         add_rev_cmdline(data->revs, obj, name, REV_CMD_REV, data->flags);
1569         add_pending_object(data->revs, obj, name);
1570 }
1571
1572 static void add_alternate_refs_to_pending(struct rev_info *revs,
1573                                           unsigned int flags)
1574 {
1575         struct add_alternate_refs_data data;
1576         data.revs = revs;
1577         data.flags = flags;
1578         for_each_alternate_ref(add_one_alternate_ref, &data);
1579 }
1580
1581 static int add_parents_only(struct rev_info *revs, const char *arg_, int flags,
1582                             int exclude_parent)
1583 {
1584         struct object_id oid;
1585         struct object *it;
1586         struct commit *commit;
1587         struct commit_list *parents;
1588         int parent_number;
1589         const char *arg = arg_;
1590
1591         if (*arg == '^') {
1592                 flags ^= UNINTERESTING | BOTTOM;
1593                 arg++;
1594         }
1595         if (get_oid_committish(arg, &oid))
1596                 return 0;
1597         while (1) {
1598                 it = get_reference(revs, arg, &oid, 0);
1599                 if (!it && revs->ignore_missing)
1600                         return 0;
1601                 if (it->type != OBJ_TAG)
1602                         break;
1603                 if (!((struct tag*)it)->tagged)
1604                         return 0;
1605                 oidcpy(&oid, &((struct tag*)it)->tagged->oid);
1606         }
1607         if (it->type != OBJ_COMMIT)
1608                 return 0;
1609         commit = (struct commit *)it;
1610         if (exclude_parent &&
1611             exclude_parent > commit_list_count(commit->parents))
1612                 return 0;
1613         for (parents = commit->parents, parent_number = 1;
1614              parents;
1615              parents = parents->next, parent_number++) {
1616                 if (exclude_parent && parent_number != exclude_parent)
1617                         continue;
1618
1619                 it = &parents->item->object;
1620                 it->flags |= flags;
1621                 add_rev_cmdline(revs, it, arg_, REV_CMD_PARENTS_ONLY, flags);
1622                 add_pending_object(revs, it, arg);
1623         }
1624         return 1;
1625 }
1626
1627 void repo_init_revisions(struct repository *r,
1628                          struct rev_info *revs,
1629                          const char *prefix)
1630 {
1631         memset(revs, 0, sizeof(*revs));
1632
1633         revs->repo = r;
1634         revs->abbrev = DEFAULT_ABBREV;
1635         revs->ignore_merges = 1;
1636         revs->simplify_history = 1;
1637         revs->pruning.repo = r;
1638         revs->pruning.flags.recursive = 1;
1639         revs->pruning.flags.quick = 1;
1640         revs->pruning.add_remove = file_add_remove;
1641         revs->pruning.change = file_change;
1642         revs->pruning.change_fn_data = revs;
1643         revs->sort_order = REV_SORT_IN_GRAPH_ORDER;
1644         revs->dense = 1;
1645         revs->prefix = prefix;
1646         revs->max_age = -1;
1647         revs->min_age = -1;
1648         revs->skip_count = -1;
1649         revs->max_count = -1;
1650         revs->max_parents = -1;
1651         revs->expand_tabs_in_log = -1;
1652
1653         revs->commit_format = CMIT_FMT_DEFAULT;
1654         revs->expand_tabs_in_log_default = 8;
1655
1656         init_grep_defaults(revs->repo);
1657         grep_init(&revs->grep_filter, revs->repo, prefix);
1658         revs->grep_filter.status_only = 1;
1659
1660         repo_diff_setup(revs->repo, &revs->diffopt);
1661         if (prefix && !revs->diffopt.prefix) {
1662                 revs->diffopt.prefix = prefix;
1663                 revs->diffopt.prefix_length = strlen(prefix);
1664         }
1665
1666         revs->notes_opt.use_default_notes = -1;
1667 }
1668
1669 static void add_pending_commit_list(struct rev_info *revs,
1670                                     struct commit_list *commit_list,
1671                                     unsigned int flags)
1672 {
1673         while (commit_list) {
1674                 struct object *object = &commit_list->item->object;
1675                 object->flags |= flags;
1676                 add_pending_object(revs, object, oid_to_hex(&object->oid));
1677                 commit_list = commit_list->next;
1678         }
1679 }
1680
1681 static void prepare_show_merge(struct rev_info *revs)
1682 {
1683         struct commit_list *bases;
1684         struct commit *head, *other;
1685         struct object_id oid;
1686         const char **prune = NULL;
1687         int i, prune_num = 1; /* counting terminating NULL */
1688         struct index_state *istate = revs->repo->index;
1689
1690         if (get_oid("HEAD", &oid))
1691                 die("--merge without HEAD?");
1692         head = lookup_commit_or_die(&oid, "HEAD");
1693         if (get_oid("MERGE_HEAD", &oid))
1694                 die("--merge without MERGE_HEAD?");
1695         other = lookup_commit_or_die(&oid, "MERGE_HEAD");
1696         add_pending_object(revs, &head->object, "HEAD");
1697         add_pending_object(revs, &other->object, "MERGE_HEAD");
1698         bases = get_merge_bases(head, other);
1699         add_rev_cmdline_list(revs, bases, REV_CMD_MERGE_BASE, UNINTERESTING | BOTTOM);
1700         add_pending_commit_list(revs, bases, UNINTERESTING | BOTTOM);
1701         free_commit_list(bases);
1702         head->object.flags |= SYMMETRIC_LEFT;
1703
1704         if (!istate->cache_nr)
1705                 repo_read_index(revs->repo);
1706         for (i = 0; i < istate->cache_nr; i++) {
1707                 const struct cache_entry *ce = istate->cache[i];
1708                 if (!ce_stage(ce))
1709                         continue;
1710                 if (ce_path_match(istate, ce, &revs->prune_data, NULL)) {
1711                         prune_num++;
1712                         REALLOC_ARRAY(prune, prune_num);
1713                         prune[prune_num-2] = ce->name;
1714                         prune[prune_num-1] = NULL;
1715                 }
1716                 while ((i+1 < istate->cache_nr) &&
1717                        ce_same_name(ce, istate->cache[i+1]))
1718                         i++;
1719         }
1720         clear_pathspec(&revs->prune_data);
1721         parse_pathspec(&revs->prune_data, PATHSPEC_ALL_MAGIC & ~PATHSPEC_LITERAL,
1722                        PATHSPEC_PREFER_FULL | PATHSPEC_LITERAL_PATH, "", prune);
1723         revs->limited = 1;
1724 }
1725
1726 static int dotdot_missing(const char *arg, char *dotdot,
1727                           struct rev_info *revs, int symmetric)
1728 {
1729         if (revs->ignore_missing)
1730                 return 0;
1731         /* de-munge so we report the full argument */
1732         *dotdot = '.';
1733         die(symmetric
1734             ? "Invalid symmetric difference expression %s"
1735             : "Invalid revision range %s", arg);
1736 }
1737
1738 static int handle_dotdot_1(const char *arg, char *dotdot,
1739                            struct rev_info *revs, int flags,
1740                            int cant_be_filename,
1741                            struct object_context *a_oc,
1742                            struct object_context *b_oc)
1743 {
1744         const char *a_name, *b_name;
1745         struct object_id a_oid, b_oid;
1746         struct object *a_obj, *b_obj;
1747         unsigned int a_flags, b_flags;
1748         int symmetric = 0;
1749         unsigned int flags_exclude = flags ^ (UNINTERESTING | BOTTOM);
1750         unsigned int oc_flags = GET_OID_COMMITTISH | GET_OID_RECORD_PATH;
1751
1752         a_name = arg;
1753         if (!*a_name)
1754                 a_name = "HEAD";
1755
1756         b_name = dotdot + 2;
1757         if (*b_name == '.') {
1758                 symmetric = 1;
1759                 b_name++;
1760         }
1761         if (!*b_name)
1762                 b_name = "HEAD";
1763
1764         if (get_oid_with_context(revs->repo, a_name, oc_flags, &a_oid, a_oc) ||
1765             get_oid_with_context(revs->repo, b_name, oc_flags, &b_oid, b_oc))
1766                 return -1;
1767
1768         if (!cant_be_filename) {
1769                 *dotdot = '.';
1770                 verify_non_filename(revs->prefix, arg);
1771                 *dotdot = '\0';
1772         }
1773
1774         a_obj = parse_object(revs->repo, &a_oid);
1775         b_obj = parse_object(revs->repo, &b_oid);
1776         if (!a_obj || !b_obj)
1777                 return dotdot_missing(arg, dotdot, revs, symmetric);
1778
1779         if (!symmetric) {
1780                 /* just A..B */
1781                 b_flags = flags;
1782                 a_flags = flags_exclude;
1783         } else {
1784                 /* A...B -- find merge bases between the two */
1785                 struct commit *a, *b;
1786                 struct commit_list *exclude;
1787
1788                 a = lookup_commit_reference(revs->repo, &a_obj->oid);
1789                 b = lookup_commit_reference(revs->repo, &b_obj->oid);
1790                 if (!a || !b)
1791                         return dotdot_missing(arg, dotdot, revs, symmetric);
1792
1793                 exclude = get_merge_bases(a, b);
1794                 add_rev_cmdline_list(revs, exclude, REV_CMD_MERGE_BASE,
1795                                      flags_exclude);
1796                 add_pending_commit_list(revs, exclude, flags_exclude);
1797                 free_commit_list(exclude);
1798
1799                 b_flags = flags;
1800                 a_flags = flags | SYMMETRIC_LEFT;
1801         }
1802
1803         a_obj->flags |= a_flags;
1804         b_obj->flags |= b_flags;
1805         add_rev_cmdline(revs, a_obj, a_name, REV_CMD_LEFT, a_flags);
1806         add_rev_cmdline(revs, b_obj, b_name, REV_CMD_RIGHT, b_flags);
1807         add_pending_object_with_path(revs, a_obj, a_name, a_oc->mode, a_oc->path);
1808         add_pending_object_with_path(revs, b_obj, b_name, b_oc->mode, b_oc->path);
1809         return 0;
1810 }
1811
1812 static int handle_dotdot(const char *arg,
1813                          struct rev_info *revs, int flags,
1814                          int cant_be_filename)
1815 {
1816         struct object_context a_oc, b_oc;
1817         char *dotdot = strstr(arg, "..");
1818         int ret;
1819
1820         if (!dotdot)
1821                 return -1;
1822
1823         memset(&a_oc, 0, sizeof(a_oc));
1824         memset(&b_oc, 0, sizeof(b_oc));
1825
1826         *dotdot = '\0';
1827         ret = handle_dotdot_1(arg, dotdot, revs, flags, cant_be_filename,
1828                               &a_oc, &b_oc);
1829         *dotdot = '.';
1830
1831         free(a_oc.path);
1832         free(b_oc.path);
1833
1834         return ret;
1835 }
1836
1837 int handle_revision_arg(const char *arg_, struct rev_info *revs, int flags, unsigned revarg_opt)
1838 {
1839         struct object_context oc;
1840         char *mark;
1841         struct object *object;
1842         struct object_id oid;
1843         int local_flags;
1844         const char *arg = arg_;
1845         int cant_be_filename = revarg_opt & REVARG_CANNOT_BE_FILENAME;
1846         unsigned get_sha1_flags = GET_OID_RECORD_PATH;
1847
1848         flags = flags & UNINTERESTING ? flags | BOTTOM : flags & ~BOTTOM;
1849
1850         if (!cant_be_filename && !strcmp(arg, "..")) {
1851                 /*
1852                  * Just ".."?  That is not a range but the
1853                  * pathspec for the parent directory.
1854                  */
1855                 return -1;
1856         }
1857
1858         if (!handle_dotdot(arg, revs, flags, revarg_opt))
1859                 return 0;
1860
1861         mark = strstr(arg, "^@");
1862         if (mark && !mark[2]) {
1863                 *mark = 0;
1864                 if (add_parents_only(revs, arg, flags, 0))
1865                         return 0;
1866                 *mark = '^';
1867         }
1868         mark = strstr(arg, "^!");
1869         if (mark && !mark[2]) {
1870                 *mark = 0;
1871                 if (!add_parents_only(revs, arg, flags ^ (UNINTERESTING | BOTTOM), 0))
1872                         *mark = '^';
1873         }
1874         mark = strstr(arg, "^-");
1875         if (mark) {
1876                 int exclude_parent = 1;
1877
1878                 if (mark[2]) {
1879                         char *end;
1880                         exclude_parent = strtoul(mark + 2, &end, 10);
1881                         if (*end != '\0' || !exclude_parent)
1882                                 return -1;
1883                 }
1884
1885                 *mark = 0;
1886                 if (!add_parents_only(revs, arg, flags ^ (UNINTERESTING | BOTTOM), exclude_parent))
1887                         *mark = '^';
1888         }
1889
1890         local_flags = 0;
1891         if (*arg == '^') {
1892                 local_flags = UNINTERESTING | BOTTOM;
1893                 arg++;
1894         }
1895
1896         if (revarg_opt & REVARG_COMMITTISH)
1897                 get_sha1_flags |= GET_OID_COMMITTISH;
1898
1899         if (get_oid_with_context(revs->repo, arg, get_sha1_flags, &oid, &oc))
1900                 return revs->ignore_missing ? 0 : -1;
1901         if (!cant_be_filename)
1902                 verify_non_filename(revs->prefix, arg);
1903         object = get_reference(revs, arg, &oid, flags ^ local_flags);
1904         if (!object)
1905                 return revs->ignore_missing ? 0 : -1;
1906         add_rev_cmdline(revs, object, arg_, REV_CMD_REV, flags ^ local_flags);
1907         add_pending_object_with_path(revs, object, arg, oc.mode, oc.path);
1908         free(oc.path);
1909         return 0;
1910 }
1911
1912 static void read_pathspec_from_stdin(struct strbuf *sb,
1913                                      struct argv_array *prune)
1914 {
1915         while (strbuf_getline(sb, stdin) != EOF)
1916                 argv_array_push(prune, sb->buf);
1917 }
1918
1919 static void read_revisions_from_stdin(struct rev_info *revs,
1920                                       struct argv_array *prune)
1921 {
1922         struct strbuf sb;
1923         int seen_dashdash = 0;
1924         int save_warning;
1925
1926         save_warning = warn_on_object_refname_ambiguity;
1927         warn_on_object_refname_ambiguity = 0;
1928
1929         strbuf_init(&sb, 1000);
1930         while (strbuf_getline(&sb, stdin) != EOF) {
1931                 int len = sb.len;
1932                 if (!len)
1933                         break;
1934                 if (sb.buf[0] == '-') {
1935                         if (len == 2 && sb.buf[1] == '-') {
1936                                 seen_dashdash = 1;
1937                                 break;
1938                         }
1939                         die("options not supported in --stdin mode");
1940                 }
1941                 if (handle_revision_arg(sb.buf, revs, 0,
1942                                         REVARG_CANNOT_BE_FILENAME))
1943                         die("bad revision '%s'", sb.buf);
1944         }
1945         if (seen_dashdash)
1946                 read_pathspec_from_stdin(&sb, prune);
1947
1948         strbuf_release(&sb);
1949         warn_on_object_refname_ambiguity = save_warning;
1950 }
1951
1952 static void add_grep(struct rev_info *revs, const char *ptn, enum grep_pat_token what)
1953 {
1954         append_grep_pattern(&revs->grep_filter, ptn, "command line", 0, what);
1955 }
1956
1957 static void add_header_grep(struct rev_info *revs, enum grep_header_field field, const char *pattern)
1958 {
1959         append_header_grep_pattern(&revs->grep_filter, field, pattern);
1960 }
1961
1962 static void add_message_grep(struct rev_info *revs, const char *pattern)
1963 {
1964         add_grep(revs, pattern, GREP_PATTERN_BODY);
1965 }
1966
1967 static int handle_revision_opt(struct rev_info *revs, int argc, const char **argv,
1968                                int *unkc, const char **unkv,
1969                                const struct setup_revision_opt* opt)
1970 {
1971         const char *arg = argv[0];
1972         const char *optarg;
1973         int argcount;
1974         const unsigned hexsz = the_hash_algo->hexsz;
1975
1976         /* pseudo revision arguments */
1977         if (!strcmp(arg, "--all") || !strcmp(arg, "--branches") ||
1978             !strcmp(arg, "--tags") || !strcmp(arg, "--remotes") ||
1979             !strcmp(arg, "--reflog") || !strcmp(arg, "--not") ||
1980             !strcmp(arg, "--no-walk") || !strcmp(arg, "--do-walk") ||
1981             !strcmp(arg, "--bisect") || starts_with(arg, "--glob=") ||
1982             !strcmp(arg, "--indexed-objects") ||
1983             !strcmp(arg, "--alternate-refs") ||
1984             starts_with(arg, "--exclude=") ||
1985             starts_with(arg, "--branches=") || starts_with(arg, "--tags=") ||
1986             starts_with(arg, "--remotes=") || starts_with(arg, "--no-walk="))
1987         {
1988                 unkv[(*unkc)++] = arg;
1989                 return 1;
1990         }
1991
1992         if ((argcount = parse_long_opt("max-count", argv, &optarg))) {
1993                 revs->max_count = atoi(optarg);
1994                 revs->no_walk = 0;
1995                 return argcount;
1996         } else if ((argcount = parse_long_opt("skip", argv, &optarg))) {
1997                 revs->skip_count = atoi(optarg);
1998                 return argcount;
1999         } else if ((*arg == '-') && isdigit(arg[1])) {
2000                 /* accept -<digit>, like traditional "head" */
2001                 if (strtol_i(arg + 1, 10, &revs->max_count) < 0 ||
2002                     revs->max_count < 0)
2003                         die("'%s': not a non-negative integer", arg + 1);
2004                 revs->no_walk = 0;
2005         } else if (!strcmp(arg, "-n")) {
2006                 if (argc <= 1)
2007                         return error("-n requires an argument");
2008                 revs->max_count = atoi(argv[1]);
2009                 revs->no_walk = 0;
2010                 return 2;
2011         } else if (skip_prefix(arg, "-n", &optarg)) {
2012                 revs->max_count = atoi(optarg);
2013                 revs->no_walk = 0;
2014         } else if ((argcount = parse_long_opt("max-age", argv, &optarg))) {
2015                 revs->max_age = atoi(optarg);
2016                 return argcount;
2017         } else if ((argcount = parse_long_opt("since", argv, &optarg))) {
2018                 revs->max_age = approxidate(optarg);
2019                 return argcount;
2020         } else if ((argcount = parse_long_opt("after", argv, &optarg))) {
2021                 revs->max_age = approxidate(optarg);
2022                 return argcount;
2023         } else if ((argcount = parse_long_opt("min-age", argv, &optarg))) {
2024                 revs->min_age = atoi(optarg);
2025                 return argcount;
2026         } else if ((argcount = parse_long_opt("before", argv, &optarg))) {
2027                 revs->min_age = approxidate(optarg);
2028                 return argcount;
2029         } else if ((argcount = parse_long_opt("until", argv, &optarg))) {
2030                 revs->min_age = approxidate(optarg);
2031                 return argcount;
2032         } else if (!strcmp(arg, "--first-parent")) {
2033                 revs->first_parent_only = 1;
2034         } else if (!strcmp(arg, "--ancestry-path")) {
2035                 revs->ancestry_path = 1;
2036                 revs->simplify_history = 0;
2037                 revs->limited = 1;
2038         } else if (!strcmp(arg, "-g") || !strcmp(arg, "--walk-reflogs")) {
2039                 init_reflog_walk(&revs->reflog_info);
2040         } else if (!strcmp(arg, "--default")) {
2041                 if (argc <= 1)
2042                         return error("bad --default argument");
2043                 revs->def = argv[1];
2044                 return 2;
2045         } else if (!strcmp(arg, "--merge")) {
2046                 revs->show_merge = 1;
2047         } else if (!strcmp(arg, "--topo-order")) {
2048                 revs->sort_order = REV_SORT_IN_GRAPH_ORDER;
2049                 revs->topo_order = 1;
2050         } else if (!strcmp(arg, "--simplify-merges")) {
2051                 revs->simplify_merges = 1;
2052                 revs->topo_order = 1;
2053                 revs->rewrite_parents = 1;
2054                 revs->simplify_history = 0;
2055                 revs->limited = 1;
2056         } else if (!strcmp(arg, "--simplify-by-decoration")) {
2057                 revs->simplify_merges = 1;
2058                 revs->topo_order = 1;
2059                 revs->rewrite_parents = 1;
2060                 revs->simplify_history = 0;
2061                 revs->simplify_by_decoration = 1;
2062                 revs->limited = 1;
2063                 revs->prune = 1;
2064         } else if (!strcmp(arg, "--date-order")) {
2065                 revs->sort_order = REV_SORT_BY_COMMIT_DATE;
2066                 revs->topo_order = 1;
2067         } else if (!strcmp(arg, "--author-date-order")) {
2068                 revs->sort_order = REV_SORT_BY_AUTHOR_DATE;
2069                 revs->topo_order = 1;
2070         } else if (!strcmp(arg, "--early-output")) {
2071                 revs->early_output = 100;
2072                 revs->topo_order = 1;
2073         } else if (skip_prefix(arg, "--early-output=", &optarg)) {
2074                 if (strtoul_ui(optarg, 10, &revs->early_output) < 0)
2075                         die("'%s': not a non-negative integer", optarg);
2076                 revs->topo_order = 1;
2077         } else if (!strcmp(arg, "--parents")) {
2078                 revs->rewrite_parents = 1;
2079                 revs->print_parents = 1;
2080         } else if (!strcmp(arg, "--dense")) {
2081                 revs->dense = 1;
2082         } else if (!strcmp(arg, "--sparse")) {
2083                 revs->dense = 0;
2084         } else if (!strcmp(arg, "--in-commit-order")) {
2085                 revs->tree_blobs_in_commit_order = 1;
2086         } else if (!strcmp(arg, "--remove-empty")) {
2087                 revs->remove_empty_trees = 1;
2088         } else if (!strcmp(arg, "--merges")) {
2089                 revs->min_parents = 2;
2090         } else if (!strcmp(arg, "--no-merges")) {
2091                 revs->max_parents = 1;
2092         } else if (skip_prefix(arg, "--min-parents=", &optarg)) {
2093                 revs->min_parents = atoi(optarg);
2094         } else if (!strcmp(arg, "--no-min-parents")) {
2095                 revs->min_parents = 0;
2096         } else if (skip_prefix(arg, "--max-parents=", &optarg)) {
2097                 revs->max_parents = atoi(optarg);
2098         } else if (!strcmp(arg, "--no-max-parents")) {
2099                 revs->max_parents = -1;
2100         } else if (!strcmp(arg, "--boundary")) {
2101                 revs->boundary = 1;
2102         } else if (!strcmp(arg, "--left-right")) {
2103                 revs->left_right = 1;
2104         } else if (!strcmp(arg, "--left-only")) {
2105                 if (revs->right_only)
2106                         die("--left-only is incompatible with --right-only"
2107                             " or --cherry");
2108                 revs->left_only = 1;
2109         } else if (!strcmp(arg, "--right-only")) {
2110                 if (revs->left_only)
2111                         die("--right-only is incompatible with --left-only");
2112                 revs->right_only = 1;
2113         } else if (!strcmp(arg, "--cherry")) {
2114                 if (revs->left_only)
2115                         die("--cherry is incompatible with --left-only");
2116                 revs->cherry_mark = 1;
2117                 revs->right_only = 1;
2118                 revs->max_parents = 1;
2119                 revs->limited = 1;
2120         } else if (!strcmp(arg, "--count")) {
2121                 revs->count = 1;
2122         } else if (!strcmp(arg, "--cherry-mark")) {
2123                 if (revs->cherry_pick)
2124                         die("--cherry-mark is incompatible with --cherry-pick");
2125                 revs->cherry_mark = 1;
2126                 revs->limited = 1; /* needs limit_list() */
2127         } else if (!strcmp(arg, "--cherry-pick")) {
2128                 if (revs->cherry_mark)
2129                         die("--cherry-pick is incompatible with --cherry-mark");
2130                 revs->cherry_pick = 1;
2131                 revs->limited = 1;
2132         } else if (!strcmp(arg, "--objects")) {
2133                 revs->tag_objects = 1;
2134                 revs->tree_objects = 1;
2135                 revs->blob_objects = 1;
2136         } else if (!strcmp(arg, "--objects-edge")) {
2137                 revs->tag_objects = 1;
2138                 revs->tree_objects = 1;
2139                 revs->blob_objects = 1;
2140                 revs->edge_hint = 1;
2141         } else if (!strcmp(arg, "--objects-edge-aggressive")) {
2142                 revs->tag_objects = 1;
2143                 revs->tree_objects = 1;
2144                 revs->blob_objects = 1;
2145                 revs->edge_hint = 1;
2146                 revs->edge_hint_aggressive = 1;
2147         } else if (!strcmp(arg, "--verify-objects")) {
2148                 revs->tag_objects = 1;
2149                 revs->tree_objects = 1;
2150                 revs->blob_objects = 1;
2151                 revs->verify_objects = 1;
2152         } else if (!strcmp(arg, "--unpacked")) {
2153                 revs->unpacked = 1;
2154         } else if (starts_with(arg, "--unpacked=")) {
2155                 die("--unpacked=<packfile> no longer supported.");
2156         } else if (!strcmp(arg, "-r")) {
2157                 revs->diff = 1;
2158                 revs->diffopt.flags.recursive = 1;
2159         } else if (!strcmp(arg, "-t")) {
2160                 revs->diff = 1;
2161                 revs->diffopt.flags.recursive = 1;
2162                 revs->diffopt.flags.tree_in_recursive = 1;
2163         } else if (!strcmp(arg, "-m")) {
2164                 revs->ignore_merges = 0;
2165         } else if (!strcmp(arg, "-c")) {
2166                 revs->diff = 1;
2167                 revs->dense_combined_merges = 0;
2168                 revs->combine_merges = 1;
2169         } else if (!strcmp(arg, "--combined-all-paths")) {
2170                 revs->diff = 1;
2171                 revs->combined_all_paths = 1;
2172         } else if (!strcmp(arg, "--cc")) {
2173                 revs->diff = 1;
2174                 revs->dense_combined_merges = 1;
2175                 revs->combine_merges = 1;
2176         } else if (!strcmp(arg, "-v")) {
2177                 revs->verbose_header = 1;
2178         } else if (!strcmp(arg, "--pretty")) {
2179                 revs->verbose_header = 1;
2180                 revs->pretty_given = 1;
2181                 get_commit_format(NULL, revs);
2182         } else if (skip_prefix(arg, "--pretty=", &optarg) ||
2183                    skip_prefix(arg, "--format=", &optarg)) {
2184                 /*
2185                  * Detached form ("--pretty X" as opposed to "--pretty=X")
2186                  * not allowed, since the argument is optional.
2187                  */
2188                 revs->verbose_header = 1;
2189                 revs->pretty_given = 1;
2190                 get_commit_format(optarg, revs);
2191         } else if (!strcmp(arg, "--expand-tabs")) {
2192                 revs->expand_tabs_in_log = 8;
2193         } else if (!strcmp(arg, "--no-expand-tabs")) {
2194                 revs->expand_tabs_in_log = 0;
2195         } else if (skip_prefix(arg, "--expand-tabs=", &arg)) {
2196                 int val;
2197                 if (strtol_i(arg, 10, &val) < 0 || val < 0)
2198                         die("'%s': not a non-negative integer", arg);
2199                 revs->expand_tabs_in_log = val;
2200         } else if (!strcmp(arg, "--show-notes") || !strcmp(arg, "--notes")) {
2201                 revs->show_notes = 1;
2202                 revs->show_notes_given = 1;
2203                 revs->notes_opt.use_default_notes = 1;
2204         } else if (!strcmp(arg, "--show-signature")) {
2205                 revs->show_signature = 1;
2206         } else if (!strcmp(arg, "--no-show-signature")) {
2207                 revs->show_signature = 0;
2208         } else if (!strcmp(arg, "--show-linear-break")) {
2209                 revs->break_bar = "                    ..........";
2210                 revs->track_linear = 1;
2211                 revs->track_first_time = 1;
2212         } else if (skip_prefix(arg, "--show-linear-break=", &optarg)) {
2213                 revs->break_bar = xstrdup(optarg);
2214                 revs->track_linear = 1;
2215                 revs->track_first_time = 1;
2216         } else if (skip_prefix(arg, "--show-notes=", &optarg) ||
2217                    skip_prefix(arg, "--notes=", &optarg)) {
2218                 struct strbuf buf = STRBUF_INIT;
2219                 revs->show_notes = 1;
2220                 revs->show_notes_given = 1;
2221                 if (starts_with(arg, "--show-notes=") &&
2222                     revs->notes_opt.use_default_notes < 0)
2223                         revs->notes_opt.use_default_notes = 1;
2224                 strbuf_addstr(&buf, optarg);
2225                 expand_notes_ref(&buf);
2226                 string_list_append(&revs->notes_opt.extra_notes_refs,
2227                                    strbuf_detach(&buf, NULL));
2228         } else if (!strcmp(arg, "--no-notes")) {
2229                 revs->show_notes = 0;
2230                 revs->show_notes_given = 1;
2231                 revs->notes_opt.use_default_notes = -1;
2232                 /* we have been strdup'ing ourselves, so trick
2233                  * string_list into free()ing strings */
2234                 revs->notes_opt.extra_notes_refs.strdup_strings = 1;
2235                 string_list_clear(&revs->notes_opt.extra_notes_refs, 0);
2236                 revs->notes_opt.extra_notes_refs.strdup_strings = 0;
2237         } else if (!strcmp(arg, "--standard-notes")) {
2238                 revs->show_notes_given = 1;
2239                 revs->notes_opt.use_default_notes = 1;
2240         } else if (!strcmp(arg, "--no-standard-notes")) {
2241                 revs->notes_opt.use_default_notes = 0;
2242         } else if (!strcmp(arg, "--oneline")) {
2243                 revs->verbose_header = 1;
2244                 get_commit_format("oneline", revs);
2245                 revs->pretty_given = 1;
2246                 revs->abbrev_commit = 1;
2247         } else if (!strcmp(arg, "--graph")) {
2248                 revs->topo_order = 1;
2249                 revs->rewrite_parents = 1;
2250                 revs->graph = graph_init(revs);
2251         } else if (!strcmp(arg, "--root")) {
2252                 revs->show_root_diff = 1;
2253         } else if (!strcmp(arg, "--no-commit-id")) {
2254                 revs->no_commit_id = 1;
2255         } else if (!strcmp(arg, "--always")) {
2256                 revs->always_show_header = 1;
2257         } else if (!strcmp(arg, "--no-abbrev")) {
2258                 revs->abbrev = 0;
2259         } else if (!strcmp(arg, "--abbrev")) {
2260                 revs->abbrev = DEFAULT_ABBREV;
2261         } else if (skip_prefix(arg, "--abbrev=", &optarg)) {
2262                 revs->abbrev = strtoul(optarg, NULL, 10);
2263                 if (revs->abbrev < MINIMUM_ABBREV)
2264                         revs->abbrev = MINIMUM_ABBREV;
2265                 else if (revs->abbrev > hexsz)
2266                         revs->abbrev = hexsz;
2267         } else if (!strcmp(arg, "--abbrev-commit")) {
2268                 revs->abbrev_commit = 1;
2269                 revs->abbrev_commit_given = 1;
2270         } else if (!strcmp(arg, "--no-abbrev-commit")) {
2271                 revs->abbrev_commit = 0;
2272         } else if (!strcmp(arg, "--full-diff")) {
2273                 revs->diff = 1;
2274                 revs->full_diff = 1;
2275         } else if (!strcmp(arg, "--full-history")) {
2276                 revs->simplify_history = 0;
2277         } else if (!strcmp(arg, "--relative-date")) {
2278                 revs->date_mode.type = DATE_RELATIVE;
2279                 revs->date_mode_explicit = 1;
2280         } else if ((argcount = parse_long_opt("date", argv, &optarg))) {
2281                 parse_date_format(optarg, &revs->date_mode);
2282                 revs->date_mode_explicit = 1;
2283                 return argcount;
2284         } else if (!strcmp(arg, "--log-size")) {
2285                 revs->show_log_size = 1;
2286         }
2287         /*
2288          * Grepping the commit log
2289          */
2290         else if ((argcount = parse_long_opt("author", argv, &optarg))) {
2291                 add_header_grep(revs, GREP_HEADER_AUTHOR, optarg);
2292                 return argcount;
2293         } else if ((argcount = parse_long_opt("committer", argv, &optarg))) {
2294                 add_header_grep(revs, GREP_HEADER_COMMITTER, optarg);
2295                 return argcount;
2296         } else if ((argcount = parse_long_opt("grep-reflog", argv, &optarg))) {
2297                 add_header_grep(revs, GREP_HEADER_REFLOG, optarg);
2298                 return argcount;
2299         } else if ((argcount = parse_long_opt("grep", argv, &optarg))) {
2300                 add_message_grep(revs, optarg);
2301                 return argcount;
2302         } else if (!strcmp(arg, "--grep-debug")) {
2303                 revs->grep_filter.debug = 1;
2304         } else if (!strcmp(arg, "--basic-regexp")) {
2305                 revs->grep_filter.pattern_type_option = GREP_PATTERN_TYPE_BRE;
2306         } else if (!strcmp(arg, "--extended-regexp") || !strcmp(arg, "-E")) {
2307                 revs->grep_filter.pattern_type_option = GREP_PATTERN_TYPE_ERE;
2308         } else if (!strcmp(arg, "--regexp-ignore-case") || !strcmp(arg, "-i")) {
2309                 revs->grep_filter.ignore_case = 1;
2310                 revs->diffopt.pickaxe_opts |= DIFF_PICKAXE_IGNORE_CASE;
2311         } else if (!strcmp(arg, "--fixed-strings") || !strcmp(arg, "-F")) {
2312                 revs->grep_filter.pattern_type_option = GREP_PATTERN_TYPE_FIXED;
2313         } else if (!strcmp(arg, "--perl-regexp") || !strcmp(arg, "-P")) {
2314                 revs->grep_filter.pattern_type_option = GREP_PATTERN_TYPE_PCRE;
2315         } else if (!strcmp(arg, "--all-match")) {
2316                 revs->grep_filter.all_match = 1;
2317         } else if (!strcmp(arg, "--invert-grep")) {
2318                 revs->invert_grep = 1;
2319         } else if ((argcount = parse_long_opt("encoding", argv, &optarg))) {
2320                 if (strcmp(optarg, "none"))
2321                         git_log_output_encoding = xstrdup(optarg);
2322                 else
2323                         git_log_output_encoding = "";
2324                 return argcount;
2325         } else if (!strcmp(arg, "--reverse")) {
2326                 revs->reverse ^= 1;
2327         } else if (!strcmp(arg, "--children")) {
2328                 revs->children.name = "children";
2329                 revs->limited = 1;
2330         } else if (!strcmp(arg, "--ignore-missing")) {
2331                 revs->ignore_missing = 1;
2332         } else if (opt && opt->allow_exclude_promisor_objects &&
2333                    !strcmp(arg, "--exclude-promisor-objects")) {
2334                 if (fetch_if_missing)
2335                         BUG("exclude_promisor_objects can only be used when fetch_if_missing is 0");
2336                 revs->exclude_promisor_objects = 1;
2337         } else {
2338                 int opts = diff_opt_parse(&revs->diffopt, argv, argc, revs->prefix);
2339                 if (!opts)
2340                         unkv[(*unkc)++] = arg;
2341                 return opts;
2342         }
2343         if (revs->graph && revs->track_linear)
2344                 die("--show-linear-break and --graph are incompatible");
2345
2346         return 1;
2347 }
2348
2349 void parse_revision_opt(struct rev_info *revs, struct parse_opt_ctx_t *ctx,
2350                         const struct option *options,
2351                         const char * const usagestr[])
2352 {
2353         int n = handle_revision_opt(revs, ctx->argc, ctx->argv,
2354                                     &ctx->cpidx, ctx->out, NULL);
2355         if (n <= 0) {
2356                 error("unknown option `%s'", ctx->argv[0]);
2357                 usage_with_options(usagestr, options);
2358         }
2359         ctx->argv += n;
2360         ctx->argc -= n;
2361 }
2362
2363 static int for_each_bisect_ref(struct ref_store *refs, each_ref_fn fn,
2364                                void *cb_data, const char *term)
2365 {
2366         struct strbuf bisect_refs = STRBUF_INIT;
2367         int status;
2368         strbuf_addf(&bisect_refs, "refs/bisect/%s", term);
2369         status = refs_for_each_fullref_in(refs, bisect_refs.buf, fn, cb_data, 0);
2370         strbuf_release(&bisect_refs);
2371         return status;
2372 }
2373
2374 static int for_each_bad_bisect_ref(struct ref_store *refs, each_ref_fn fn, void *cb_data)
2375 {
2376         return for_each_bisect_ref(refs, fn, cb_data, term_bad);
2377 }
2378
2379 static int for_each_good_bisect_ref(struct ref_store *refs, each_ref_fn fn, void *cb_data)
2380 {
2381         return for_each_bisect_ref(refs, fn, cb_data, term_good);
2382 }
2383
2384 static int handle_revision_pseudo_opt(const char *submodule,
2385                                 struct rev_info *revs,
2386                                 int argc, const char **argv, int *flags)
2387 {
2388         const char *arg = argv[0];
2389         const char *optarg;
2390         struct ref_store *refs;
2391         int argcount;
2392
2393         if (submodule) {
2394                 /*
2395                  * We need some something like get_submodule_worktrees()
2396                  * before we can go through all worktrees of a submodule,
2397                  * .e.g with adding all HEADs from --all, which is not
2398                  * supported right now, so stick to single worktree.
2399                  */
2400                 if (!revs->single_worktree)
2401                         BUG("--single-worktree cannot be used together with submodule");
2402                 refs = get_submodule_ref_store(submodule);
2403         } else
2404                 refs = get_main_ref_store(revs->repo);
2405
2406         /*
2407          * NOTE!
2408          *
2409          * Commands like "git shortlog" will not accept the options below
2410          * unless parse_revision_opt queues them (as opposed to erroring
2411          * out).
2412          *
2413          * When implementing your new pseudo-option, remember to
2414          * register it in the list at the top of handle_revision_opt.
2415          */
2416         if (!strcmp(arg, "--all")) {
2417                 handle_refs(refs, revs, *flags, refs_for_each_ref);
2418                 handle_refs(refs, revs, *flags, refs_head_ref);
2419                 if (!revs->single_worktree) {
2420                         struct all_refs_cb cb;
2421
2422                         init_all_refs_cb(&cb, revs, *flags);
2423                         other_head_refs(handle_one_ref, &cb);
2424                 }
2425                 clear_ref_exclusion(&revs->ref_excludes);
2426         } else if (!strcmp(arg, "--branches")) {
2427                 handle_refs(refs, revs, *flags, refs_for_each_branch_ref);
2428                 clear_ref_exclusion(&revs->ref_excludes);
2429         } else if (!strcmp(arg, "--bisect")) {
2430                 read_bisect_terms(&term_bad, &term_good);
2431                 handle_refs(refs, revs, *flags, for_each_bad_bisect_ref);
2432                 handle_refs(refs, revs, *flags ^ (UNINTERESTING | BOTTOM),
2433                             for_each_good_bisect_ref);
2434                 revs->bisect = 1;
2435         } else if (!strcmp(arg, "--tags")) {
2436                 handle_refs(refs, revs, *flags, refs_for_each_tag_ref);
2437                 clear_ref_exclusion(&revs->ref_excludes);
2438         } else if (!strcmp(arg, "--remotes")) {
2439                 handle_refs(refs, revs, *flags, refs_for_each_remote_ref);
2440                 clear_ref_exclusion(&revs->ref_excludes);
2441         } else if ((argcount = parse_long_opt("glob", argv, &optarg))) {
2442                 struct all_refs_cb cb;
2443                 init_all_refs_cb(&cb, revs, *flags);
2444                 for_each_glob_ref(handle_one_ref, optarg, &cb);
2445                 clear_ref_exclusion(&revs->ref_excludes);
2446                 return argcount;
2447         } else if ((argcount = parse_long_opt("exclude", argv, &optarg))) {
2448                 add_ref_exclusion(&revs->ref_excludes, optarg);
2449                 return argcount;
2450         } else if (skip_prefix(arg, "--branches=", &optarg)) {
2451                 struct all_refs_cb cb;
2452                 init_all_refs_cb(&cb, revs, *flags);
2453                 for_each_glob_ref_in(handle_one_ref, optarg, "refs/heads/", &cb);
2454                 clear_ref_exclusion(&revs->ref_excludes);
2455         } else if (skip_prefix(arg, "--tags=", &optarg)) {
2456                 struct all_refs_cb cb;
2457                 init_all_refs_cb(&cb, revs, *flags);
2458                 for_each_glob_ref_in(handle_one_ref, optarg, "refs/tags/", &cb);
2459                 clear_ref_exclusion(&revs->ref_excludes);
2460         } else if (skip_prefix(arg, "--remotes=", &optarg)) {
2461                 struct all_refs_cb cb;
2462                 init_all_refs_cb(&cb, revs, *flags);
2463                 for_each_glob_ref_in(handle_one_ref, optarg, "refs/remotes/", &cb);
2464                 clear_ref_exclusion(&revs->ref_excludes);
2465         } else if (!strcmp(arg, "--reflog")) {
2466                 add_reflogs_to_pending(revs, *flags);
2467         } else if (!strcmp(arg, "--indexed-objects")) {
2468                 add_index_objects_to_pending(revs, *flags);
2469         } else if (!strcmp(arg, "--alternate-refs")) {
2470                 add_alternate_refs_to_pending(revs, *flags);
2471         } else if (!strcmp(arg, "--not")) {
2472                 *flags ^= UNINTERESTING | BOTTOM;
2473         } else if (!strcmp(arg, "--no-walk")) {
2474                 revs->no_walk = REVISION_WALK_NO_WALK_SORTED;
2475         } else if (skip_prefix(arg, "--no-walk=", &optarg)) {
2476                 /*
2477                  * Detached form ("--no-walk X" as opposed to "--no-walk=X")
2478                  * not allowed, since the argument is optional.
2479                  */
2480                 if (!strcmp(optarg, "sorted"))
2481                         revs->no_walk = REVISION_WALK_NO_WALK_SORTED;
2482                 else if (!strcmp(optarg, "unsorted"))
2483                         revs->no_walk = REVISION_WALK_NO_WALK_UNSORTED;
2484                 else
2485                         return error("invalid argument to --no-walk");
2486         } else if (!strcmp(arg, "--do-walk")) {
2487                 revs->no_walk = 0;
2488         } else if (!strcmp(arg, "--single-worktree")) {
2489                 revs->single_worktree = 1;
2490         } else {
2491                 return 0;
2492         }
2493
2494         return 1;
2495 }
2496
2497 static void NORETURN diagnose_missing_default(const char *def)
2498 {
2499         int flags;
2500         const char *refname;
2501
2502         refname = resolve_ref_unsafe(def, 0, NULL, &flags);
2503         if (!refname || !(flags & REF_ISSYMREF) || (flags & REF_ISBROKEN))
2504                 die(_("your current branch appears to be broken"));
2505
2506         skip_prefix(refname, "refs/heads/", &refname);
2507         die(_("your current branch '%s' does not have any commits yet"),
2508             refname);
2509 }
2510
2511 /*
2512  * Parse revision information, filling in the "rev_info" structure,
2513  * and removing the used arguments from the argument list.
2514  *
2515  * Returns the number of arguments left that weren't recognized
2516  * (which are also moved to the head of the argument list)
2517  */
2518 int setup_revisions(int argc, const char **argv, struct rev_info *revs, struct setup_revision_opt *opt)
2519 {
2520         int i, flags, left, seen_dashdash, got_rev_arg = 0, revarg_opt;
2521         struct argv_array prune_data = ARGV_ARRAY_INIT;
2522         const char *submodule = NULL;
2523         int seen_end_of_options = 0;
2524
2525         if (opt)
2526                 submodule = opt->submodule;
2527
2528         /* First, search for "--" */
2529         if (opt && opt->assume_dashdash) {
2530                 seen_dashdash = 1;
2531         } else {
2532                 seen_dashdash = 0;
2533                 for (i = 1; i < argc; i++) {
2534                         const char *arg = argv[i];
2535                         if (strcmp(arg, "--"))
2536                                 continue;
2537                         argv[i] = NULL;
2538                         argc = i;
2539                         if (argv[i + 1])
2540                                 argv_array_pushv(&prune_data, argv + i + 1);
2541                         seen_dashdash = 1;
2542                         break;
2543                 }
2544         }
2545
2546         /* Second, deal with arguments and options */
2547         flags = 0;
2548         revarg_opt = opt ? opt->revarg_opt : 0;
2549         if (seen_dashdash)
2550                 revarg_opt |= REVARG_CANNOT_BE_FILENAME;
2551         for (left = i = 1; i < argc; i++) {
2552                 const char *arg = argv[i];
2553                 if (!seen_end_of_options && *arg == '-') {
2554                         int opts;
2555
2556                         opts = handle_revision_pseudo_opt(submodule,
2557                                                 revs, argc - i, argv + i,
2558                                                 &flags);
2559                         if (opts > 0) {
2560                                 i += opts - 1;
2561                                 continue;
2562                         }
2563
2564                         if (!strcmp(arg, "--stdin")) {
2565                                 if (revs->disable_stdin) {
2566                                         argv[left++] = arg;
2567                                         continue;
2568                                 }
2569                                 if (revs->read_from_stdin++)
2570                                         die("--stdin given twice?");
2571                                 read_revisions_from_stdin(revs, &prune_data);
2572                                 continue;
2573                         }
2574
2575                         if (!strcmp(arg, "--end-of-options")) {
2576                                 seen_end_of_options = 1;
2577                                 continue;
2578                         }
2579
2580                         opts = handle_revision_opt(revs, argc - i, argv + i,
2581                                                    &left, argv, opt);
2582                         if (opts > 0) {
2583                                 i += opts - 1;
2584                                 continue;
2585                         }
2586                         if (opts < 0)
2587                                 exit(128);
2588                         continue;
2589                 }
2590
2591
2592                 if (handle_revision_arg(arg, revs, flags, revarg_opt)) {
2593                         int j;
2594                         if (seen_dashdash || *arg == '^')
2595                                 die("bad revision '%s'", arg);
2596
2597                         /* If we didn't have a "--":
2598                          * (1) all filenames must exist;
2599                          * (2) all rev-args must not be interpretable
2600                          *     as a valid filename.
2601                          * but the latter we have checked in the main loop.
2602                          */
2603                         for (j = i; j < argc; j++)
2604                                 verify_filename(revs->prefix, argv[j], j == i);
2605
2606                         argv_array_pushv(&prune_data, argv + i);
2607                         break;
2608                 }
2609                 else
2610                         got_rev_arg = 1;
2611         }
2612
2613         if (prune_data.argc) {
2614                 /*
2615                  * If we need to introduce the magic "a lone ':' means no
2616                  * pathspec whatsoever", here is the place to do so.
2617                  *
2618                  * if (prune_data.nr == 1 && !strcmp(prune_data[0], ":")) {
2619                  *      prune_data.nr = 0;
2620                  *      prune_data.alloc = 0;
2621                  *      free(prune_data.path);
2622                  *      prune_data.path = NULL;
2623                  * } else {
2624                  *      terminate prune_data.alloc with NULL and
2625                  *      call init_pathspec() to set revs->prune_data here.
2626                  * }
2627                  */
2628                 parse_pathspec(&revs->prune_data, 0, 0,
2629                                revs->prefix, prune_data.argv);
2630         }
2631         argv_array_clear(&prune_data);
2632
2633         if (revs->def == NULL)
2634                 revs->def = opt ? opt->def : NULL;
2635         if (opt && opt->tweak)
2636                 opt->tweak(revs, opt);
2637         if (revs->show_merge)
2638                 prepare_show_merge(revs);
2639         if (revs->def && !revs->pending.nr && !revs->rev_input_given && !got_rev_arg) {
2640                 struct object_id oid;
2641                 struct object *object;
2642                 struct object_context oc;
2643                 if (get_oid_with_context(revs->repo, revs->def, 0, &oid, &oc))
2644                         diagnose_missing_default(revs->def);
2645                 object = get_reference(revs, revs->def, &oid, 0);
2646                 add_pending_object_with_mode(revs, object, revs->def, oc.mode);
2647         }
2648
2649         /* Did the user ask for any diff output? Run the diff! */
2650         if (revs->diffopt.output_format & ~DIFF_FORMAT_NO_OUTPUT)
2651                 revs->diff = 1;
2652
2653         /* Pickaxe, diff-filter and rename following need diffs */
2654         if ((revs->diffopt.pickaxe_opts & DIFF_PICKAXE_KINDS_MASK) ||
2655             revs->diffopt.filter ||
2656             revs->diffopt.flags.follow_renames)
2657                 revs->diff = 1;
2658
2659         if (revs->diffopt.objfind)
2660                 revs->simplify_history = 0;
2661
2662         if (revs->topo_order && !generation_numbers_enabled(the_repository))
2663                 revs->limited = 1;
2664
2665         if (revs->prune_data.nr) {
2666                 copy_pathspec(&revs->pruning.pathspec, &revs->prune_data);
2667                 /* Can't prune commits with rename following: the paths change.. */
2668                 if (!revs->diffopt.flags.follow_renames)
2669                         revs->prune = 1;
2670                 if (!revs->full_diff)
2671                         copy_pathspec(&revs->diffopt.pathspec,
2672                                       &revs->prune_data);
2673         }
2674         if (revs->combine_merges)
2675                 revs->ignore_merges = 0;
2676         if (revs->combined_all_paths && !revs->combine_merges)
2677                 die("--combined-all-paths makes no sense without -c or --cc");
2678
2679         revs->diffopt.abbrev = revs->abbrev;
2680
2681         if (revs->line_level_traverse) {
2682                 revs->limited = 1;
2683                 revs->topo_order = 1;
2684         }
2685
2686         diff_setup_done(&revs->diffopt);
2687
2688         grep_commit_pattern_type(GREP_PATTERN_TYPE_UNSPECIFIED,
2689                                  &revs->grep_filter);
2690         compile_grep_patterns(&revs->grep_filter);
2691
2692         if (revs->reverse && revs->reflog_info)
2693                 die("cannot combine --reverse with --walk-reflogs");
2694         if (revs->reflog_info && revs->limited)
2695                 die("cannot combine --walk-reflogs with history-limiting options");
2696         if (revs->rewrite_parents && revs->children.name)
2697                 die("cannot combine --parents and --children");
2698
2699         /*
2700          * Limitations on the graph functionality
2701          */
2702         if (revs->reverse && revs->graph)
2703                 die("cannot combine --reverse with --graph");
2704
2705         if (revs->reflog_info && revs->graph)
2706                 die("cannot combine --walk-reflogs with --graph");
2707         if (revs->no_walk && revs->graph)
2708                 die("cannot combine --no-walk with --graph");
2709         if (!revs->reflog_info && revs->grep_filter.use_reflog_filter)
2710                 die("cannot use --grep-reflog without --walk-reflogs");
2711
2712         if (revs->first_parent_only && revs->bisect)
2713                 die(_("--first-parent is incompatible with --bisect"));
2714
2715         if (revs->line_level_traverse &&
2716             (revs->diffopt.output_format & ~(DIFF_FORMAT_PATCH | DIFF_FORMAT_NO_OUTPUT)))
2717                 die(_("-L does not yet support diff formats besides -p and -s"));
2718
2719         if (revs->expand_tabs_in_log < 0)
2720                 revs->expand_tabs_in_log = revs->expand_tabs_in_log_default;
2721
2722         return left;
2723 }
2724
2725 static void add_child(struct rev_info *revs, struct commit *parent, struct commit *child)
2726 {
2727         struct commit_list *l = xcalloc(1, sizeof(*l));
2728
2729         l->item = child;
2730         l->next = add_decoration(&revs->children, &parent->object, l);
2731 }
2732
2733 static int remove_duplicate_parents(struct rev_info *revs, struct commit *commit)
2734 {
2735         struct treesame_state *ts = lookup_decoration(&revs->treesame, &commit->object);
2736         struct commit_list **pp, *p;
2737         int surviving_parents;
2738
2739         /* Examine existing parents while marking ones we have seen... */
2740         pp = &commit->parents;
2741         surviving_parents = 0;
2742         while ((p = *pp) != NULL) {
2743                 struct commit *parent = p->item;
2744                 if (parent->object.flags & TMP_MARK) {
2745                         *pp = p->next;
2746                         if (ts)
2747                                 compact_treesame(revs, commit, surviving_parents);
2748                         continue;
2749                 }
2750                 parent->object.flags |= TMP_MARK;
2751                 surviving_parents++;
2752                 pp = &p->next;
2753         }
2754         /* clear the temporary mark */
2755         for (p = commit->parents; p; p = p->next) {
2756                 p->item->object.flags &= ~TMP_MARK;
2757         }
2758         /* no update_treesame() - removing duplicates can't affect TREESAME */
2759         return surviving_parents;
2760 }
2761
2762 struct merge_simplify_state {
2763         struct commit *simplified;
2764 };
2765
2766 static struct merge_simplify_state *locate_simplify_state(struct rev_info *revs, struct commit *commit)
2767 {
2768         struct merge_simplify_state *st;
2769
2770         st = lookup_decoration(&revs->merge_simplification, &commit->object);
2771         if (!st) {
2772                 st = xcalloc(1, sizeof(*st));
2773                 add_decoration(&revs->merge_simplification, &commit->object, st);
2774         }
2775         return st;
2776 }
2777
2778 static int mark_redundant_parents(struct commit *commit)
2779 {
2780         struct commit_list *h = reduce_heads(commit->parents);
2781         int i = 0, marked = 0;
2782         struct commit_list *po, *pn;
2783
2784         /* Want these for sanity-checking only */
2785         int orig_cnt = commit_list_count(commit->parents);
2786         int cnt = commit_list_count(h);
2787
2788         /*
2789          * Not ready to remove items yet, just mark them for now, based
2790          * on the output of reduce_heads(). reduce_heads outputs the reduced
2791          * set in its original order, so this isn't too hard.
2792          */
2793         po = commit->parents;
2794         pn = h;
2795         while (po) {
2796                 if (pn && po->item == pn->item) {
2797                         pn = pn->next;
2798                         i++;
2799                 } else {
2800                         po->item->object.flags |= TMP_MARK;
2801                         marked++;
2802                 }
2803                 po=po->next;
2804         }
2805
2806         if (i != cnt || cnt+marked != orig_cnt)
2807                 die("mark_redundant_parents %d %d %d %d", orig_cnt, cnt, i, marked);
2808
2809         free_commit_list(h);
2810
2811         return marked;
2812 }
2813
2814 static int mark_treesame_root_parents(struct commit *commit)
2815 {
2816         struct commit_list *p;
2817         int marked = 0;
2818
2819         for (p = commit->parents; p; p = p->next) {
2820                 struct commit *parent = p->item;
2821                 if (!parent->parents && (parent->object.flags & TREESAME)) {
2822                         parent->object.flags |= TMP_MARK;
2823                         marked++;
2824                 }
2825         }
2826
2827         return marked;
2828 }
2829
2830 /*
2831  * Awkward naming - this means one parent we are TREESAME to.
2832  * cf mark_treesame_root_parents: root parents that are TREESAME (to an
2833  * empty tree). Better name suggestions?
2834  */
2835 static int leave_one_treesame_to_parent(struct rev_info *revs, struct commit *commit)
2836 {
2837         struct treesame_state *ts = lookup_decoration(&revs->treesame, &commit->object);
2838         struct commit *unmarked = NULL, *marked = NULL;
2839         struct commit_list *p;
2840         unsigned n;
2841
2842         for (p = commit->parents, n = 0; p; p = p->next, n++) {
2843                 if (ts->treesame[n]) {
2844                         if (p->item->object.flags & TMP_MARK) {
2845                                 if (!marked)
2846                                         marked = p->item;
2847                         } else {
2848                                 if (!unmarked) {
2849                                         unmarked = p->item;
2850                                         break;
2851                                 }
2852                         }
2853                 }
2854         }
2855
2856         /*
2857          * If we are TREESAME to a marked-for-deletion parent, but not to any
2858          * unmarked parents, unmark the first TREESAME parent. This is the
2859          * parent that the default simplify_history==1 scan would have followed,
2860          * and it doesn't make sense to omit that path when asking for a
2861          * simplified full history. Retaining it improves the chances of
2862          * understanding odd missed merges that took an old version of a file.
2863          *
2864          * Example:
2865          *
2866          *   I--------*X       A modified the file, but mainline merge X used
2867          *    \       /        "-s ours", so took the version from I. X is
2868          *     `-*A--'         TREESAME to I and !TREESAME to A.
2869          *
2870          * Default log from X would produce "I". Without this check,
2871          * --full-history --simplify-merges would produce "I-A-X", showing
2872          * the merge commit X and that it changed A, but not making clear that
2873          * it had just taken the I version. With this check, the topology above
2874          * is retained.
2875          *
2876          * Note that it is possible that the simplification chooses a different
2877          * TREESAME parent from the default, in which case this test doesn't
2878          * activate, and we _do_ drop the default parent. Example:
2879          *
2880          *   I------X         A modified the file, but it was reverted in B,
2881          *    \    /          meaning mainline merge X is TREESAME to both
2882          *    *A-*B           parents.
2883          *
2884          * Default log would produce "I" by following the first parent;
2885          * --full-history --simplify-merges will produce "I-A-B". But this is a
2886          * reasonable result - it presents a logical full history leading from
2887          * I to X, and X is not an important merge.
2888          */
2889         if (!unmarked && marked) {
2890                 marked->object.flags &= ~TMP_MARK;
2891                 return 1;
2892         }
2893
2894         return 0;
2895 }
2896
2897 static int remove_marked_parents(struct rev_info *revs, struct commit *commit)
2898 {
2899         struct commit_list **pp, *p;
2900         int nth_parent, removed = 0;
2901
2902         pp = &commit->parents;
2903         nth_parent = 0;
2904         while ((p = *pp) != NULL) {
2905                 struct commit *parent = p->item;
2906                 if (parent->object.flags & TMP_MARK) {
2907                         parent->object.flags &= ~TMP_MARK;
2908                         *pp = p->next;
2909                         free(p);
2910                         removed++;
2911                         compact_treesame(revs, commit, nth_parent);
2912                         continue;
2913                 }
2914                 pp = &p->next;
2915                 nth_parent++;
2916         }
2917
2918         /* Removing parents can only increase TREESAMEness */
2919         if (removed && !(commit->object.flags & TREESAME))
2920                 update_treesame(revs, commit);
2921
2922         return nth_parent;
2923 }
2924
2925 static struct commit_list **simplify_one(struct rev_info *revs, struct commit *commit, struct commit_list **tail)
2926 {
2927         struct commit_list *p;
2928         struct commit *parent;
2929         struct merge_simplify_state *st, *pst;
2930         int cnt;
2931
2932         st = locate_simplify_state(revs, commit);
2933
2934         /*
2935          * Have we handled this one?
2936          */
2937         if (st->simplified)
2938                 return tail;
2939
2940         /*
2941          * An UNINTERESTING commit simplifies to itself, so does a
2942          * root commit.  We do not rewrite parents of such commit
2943          * anyway.
2944          */
2945         if ((commit->object.flags & UNINTERESTING) || !commit->parents) {
2946                 st->simplified = commit;
2947                 return tail;
2948         }
2949
2950         /*
2951          * Do we know what commit all of our parents that matter
2952          * should be rewritten to?  Otherwise we are not ready to
2953          * rewrite this one yet.
2954          */
2955         for (cnt = 0, p = commit->parents; p; p = p->next) {
2956                 pst = locate_simplify_state(revs, p->item);
2957                 if (!pst->simplified) {
2958                         tail = &commit_list_insert(p->item, tail)->next;
2959                         cnt++;
2960                 }
2961                 if (revs->first_parent_only)
2962                         break;
2963         }
2964         if (cnt) {
2965                 tail = &commit_list_insert(commit, tail)->next;
2966                 return tail;
2967         }
2968
2969         /*
2970          * Rewrite our list of parents. Note that this cannot
2971          * affect our TREESAME flags in any way - a commit is
2972          * always TREESAME to its simplification.
2973          */
2974         for (p = commit->parents; p; p = p->next) {
2975                 pst = locate_simplify_state(revs, p->item);
2976                 p->item = pst->simplified;
2977                 if (revs->first_parent_only)
2978                         break;
2979         }
2980
2981         if (revs->first_parent_only)
2982                 cnt = 1;
2983         else
2984                 cnt = remove_duplicate_parents(revs, commit);
2985
2986         /*
2987          * It is possible that we are a merge and one side branch
2988          * does not have any commit that touches the given paths;
2989          * in such a case, the immediate parent from that branch
2990          * will be rewritten to be the merge base.
2991          *
2992          *      o----X          X: the commit we are looking at;
2993          *     /    /           o: a commit that touches the paths;
2994          * ---o----'
2995          *
2996          * Further, a merge of an independent branch that doesn't
2997          * touch the path will reduce to a treesame root parent:
2998          *
2999          *  ----o----X          X: the commit we are looking at;
3000          *          /           o: a commit that touches the paths;
3001          *         r            r: a root commit not touching the paths
3002          *
3003          * Detect and simplify both cases.
3004          */
3005         if (1 < cnt) {
3006                 int marked = mark_redundant_parents(commit);
3007                 marked += mark_treesame_root_parents(commit);
3008                 if (marked)
3009                         marked -= leave_one_treesame_to_parent(revs, commit);
3010                 if (marked)
3011                         cnt = remove_marked_parents(revs, commit);
3012         }
3013
3014         /*
3015          * A commit simplifies to itself if it is a root, if it is
3016          * UNINTERESTING, if it touches the given paths, or if it is a
3017          * merge and its parents don't simplify to one relevant commit
3018          * (the first two cases are already handled at the beginning of
3019          * this function).
3020          *
3021          * Otherwise, it simplifies to what its sole relevant parent
3022          * simplifies to.
3023          */
3024         if (!cnt ||
3025             (commit->object.flags & UNINTERESTING) ||
3026             !(commit->object.flags & TREESAME) ||
3027             (parent = one_relevant_parent(revs, commit->parents)) == NULL)
3028                 st->simplified = commit;
3029         else {
3030                 pst = locate_simplify_state(revs, parent);
3031                 st->simplified = pst->simplified;
3032         }
3033         return tail;
3034 }
3035
3036 static void simplify_merges(struct rev_info *revs)
3037 {
3038         struct commit_list *list, *next;
3039         struct commit_list *yet_to_do, **tail;
3040         struct commit *commit;
3041
3042         if (!revs->prune)
3043                 return;
3044
3045         /* feed the list reversed */
3046         yet_to_do = NULL;
3047         for (list = revs->commits; list; list = next) {
3048                 commit = list->item;
3049                 next = list->next;
3050                 /*
3051                  * Do not free(list) here yet; the original list
3052                  * is used later in this function.
3053                  */
3054                 commit_list_insert(commit, &yet_to_do);
3055         }
3056         while (yet_to_do) {
3057                 list = yet_to_do;
3058                 yet_to_do = NULL;
3059                 tail = &yet_to_do;
3060                 while (list) {
3061                         commit = pop_commit(&list);
3062                         tail = simplify_one(revs, commit, tail);
3063                 }
3064         }
3065
3066         /* clean up the result, removing the simplified ones */
3067         list = revs->commits;
3068         revs->commits = NULL;
3069         tail = &revs->commits;
3070         while (list) {
3071                 struct merge_simplify_state *st;
3072
3073                 commit = pop_commit(&list);
3074                 st = locate_simplify_state(revs, commit);
3075                 if (st->simplified == commit)
3076                         tail = &commit_list_insert(commit, tail)->next;
3077         }
3078 }
3079
3080 static void set_children(struct rev_info *revs)
3081 {
3082         struct commit_list *l;
3083         for (l = revs->commits; l; l = l->next) {
3084                 struct commit *commit = l->item;
3085                 struct commit_list *p;
3086
3087                 for (p = commit->parents; p; p = p->next)
3088                         add_child(revs, p->item, commit);
3089         }
3090 }
3091
3092 void reset_revision_walk(void)
3093 {
3094         clear_object_flags(SEEN | ADDED | SHOWN);
3095 }
3096
3097 static int mark_uninteresting(const struct object_id *oid,
3098                               struct packed_git *pack,
3099                               uint32_t pos,
3100                               void *cb)
3101 {
3102         struct rev_info *revs = cb;
3103         struct object *o = parse_object(revs->repo, oid);
3104         o->flags |= UNINTERESTING | SEEN;
3105         return 0;
3106 }
3107
3108 define_commit_slab(indegree_slab, int);
3109 define_commit_slab(author_date_slab, timestamp_t);
3110
3111 struct topo_walk_info {
3112         uint32_t min_generation;
3113         struct prio_queue explore_queue;
3114         struct prio_queue indegree_queue;
3115         struct prio_queue topo_queue;
3116         struct indegree_slab indegree;
3117         struct author_date_slab author_date;
3118 };
3119
3120 static inline void test_flag_and_insert(struct prio_queue *q, struct commit *c, int flag)
3121 {
3122         if (c->object.flags & flag)
3123                 return;
3124
3125         c->object.flags |= flag;
3126         prio_queue_put(q, c);
3127 }
3128
3129 static void explore_walk_step(struct rev_info *revs)
3130 {
3131         struct topo_walk_info *info = revs->topo_walk_info;
3132         struct commit_list *p;
3133         struct commit *c = prio_queue_get(&info->explore_queue);
3134
3135         if (!c)
3136                 return;
3137
3138         if (parse_commit_gently(c, 1) < 0)
3139                 return;
3140
3141         if (revs->sort_order == REV_SORT_BY_AUTHOR_DATE)
3142                 record_author_date(&info->author_date, c);
3143
3144         if (revs->max_age != -1 && (c->date < revs->max_age))
3145                 c->object.flags |= UNINTERESTING;
3146
3147         if (process_parents(revs, c, NULL, NULL) < 0)
3148                 return;
3149
3150         if (c->object.flags & UNINTERESTING)
3151                 mark_parents_uninteresting(c);
3152
3153         for (p = c->parents; p; p = p->next)
3154                 test_flag_and_insert(&info->explore_queue, p->item, TOPO_WALK_EXPLORED);
3155 }
3156
3157 static void explore_to_depth(struct rev_info *revs,
3158                              uint32_t gen_cutoff)
3159 {
3160         struct topo_walk_info *info = revs->topo_walk_info;
3161         struct commit *c;
3162         while ((c = prio_queue_peek(&info->explore_queue)) &&
3163                c->generation >= gen_cutoff)
3164                 explore_walk_step(revs);
3165 }
3166
3167 static void indegree_walk_step(struct rev_info *revs)
3168 {
3169         struct commit_list *p;
3170         struct topo_walk_info *info = revs->topo_walk_info;
3171         struct commit *c = prio_queue_get(&info->indegree_queue);
3172
3173         if (!c)
3174                 return;
3175
3176         if (parse_commit_gently(c, 1) < 0)
3177                 return;
3178
3179         explore_to_depth(revs, c->generation);
3180
3181         for (p = c->parents; p; p = p->next) {
3182                 struct commit *parent = p->item;
3183                 int *pi = indegree_slab_at(&info->indegree, parent);
3184
3185                 if (*pi)
3186                         (*pi)++;
3187                 else
3188                         *pi = 2;
3189
3190                 test_flag_and_insert(&info->indegree_queue, parent, TOPO_WALK_INDEGREE);
3191
3192                 if (revs->first_parent_only)
3193                         return;
3194         }
3195 }
3196
3197 static void compute_indegrees_to_depth(struct rev_info *revs,
3198                                        uint32_t gen_cutoff)
3199 {
3200         struct topo_walk_info *info = revs->topo_walk_info;
3201         struct commit *c;
3202         while ((c = prio_queue_peek(&info->indegree_queue)) &&
3203                c->generation >= gen_cutoff)
3204                 indegree_walk_step(revs);
3205 }
3206
3207 static void init_topo_walk(struct rev_info *revs)
3208 {
3209         struct topo_walk_info *info;
3210         struct commit_list *list;
3211         revs->topo_walk_info = xmalloc(sizeof(struct topo_walk_info));
3212         info = revs->topo_walk_info;
3213         memset(info, 0, sizeof(struct topo_walk_info));
3214
3215         init_indegree_slab(&info->indegree);
3216         memset(&info->explore_queue, 0, sizeof(info->explore_queue));
3217         memset(&info->indegree_queue, 0, sizeof(info->indegree_queue));
3218         memset(&info->topo_queue, 0, sizeof(info->topo_queue));
3219
3220         switch (revs->sort_order) {
3221         default: /* REV_SORT_IN_GRAPH_ORDER */
3222                 info->topo_queue.compare = NULL;
3223                 break;
3224         case REV_SORT_BY_COMMIT_DATE:
3225                 info->topo_queue.compare = compare_commits_by_commit_date;
3226                 break;
3227         case REV_SORT_BY_AUTHOR_DATE:
3228                 init_author_date_slab(&info->author_date);
3229                 info->topo_queue.compare = compare_commits_by_author_date;
3230                 info->topo_queue.cb_data = &info->author_date;
3231                 break;
3232         }
3233
3234         info->explore_queue.compare = compare_commits_by_gen_then_commit_date;
3235         info->indegree_queue.compare = compare_commits_by_gen_then_commit_date;
3236
3237         info->min_generation = GENERATION_NUMBER_INFINITY;
3238         for (list = revs->commits; list; list = list->next) {
3239                 struct commit *c = list->item;
3240
3241                 if (parse_commit_gently(c, 1))
3242                         continue;
3243
3244                 test_flag_and_insert(&info->explore_queue, c, TOPO_WALK_EXPLORED);
3245                 test_flag_and_insert(&info->indegree_queue, c, TOPO_WALK_INDEGREE);
3246
3247                 if (c->generation < info->min_generation)
3248                         info->min_generation = c->generation;
3249
3250                 *(indegree_slab_at(&info->indegree, c)) = 1;
3251
3252                 if (revs->sort_order == REV_SORT_BY_AUTHOR_DATE)
3253                         record_author_date(&info->author_date, c);
3254         }
3255         compute_indegrees_to_depth(revs, info->min_generation);
3256
3257         for (list = revs->commits; list; list = list->next) {
3258                 struct commit *c = list->item;
3259
3260                 if (*(indegree_slab_at(&info->indegree, c)) == 1)
3261                         prio_queue_put(&info->topo_queue, c);
3262         }
3263
3264         /*
3265          * This is unfortunate; the initial tips need to be shown
3266          * in the order given from the revision traversal machinery.
3267          */
3268         if (revs->sort_order == REV_SORT_IN_GRAPH_ORDER)
3269                 prio_queue_reverse(&info->topo_queue);
3270 }
3271
3272 static struct commit *next_topo_commit(struct rev_info *revs)
3273 {
3274         struct commit *c;
3275         struct topo_walk_info *info = revs->topo_walk_info;
3276
3277         /* pop next off of topo_queue */
3278         c = prio_queue_get(&info->topo_queue);
3279
3280         if (c)
3281                 *(indegree_slab_at(&info->indegree, c)) = 0;
3282
3283         return c;
3284 }
3285
3286 static void expand_topo_walk(struct rev_info *revs, struct commit *commit)
3287 {
3288         struct commit_list *p;
3289         struct topo_walk_info *info = revs->topo_walk_info;
3290         if (process_parents(revs, commit, NULL, NULL) < 0) {
3291                 if (!revs->ignore_missing_links)
3292                         die("Failed to traverse parents of commit %s",
3293                             oid_to_hex(&commit->object.oid));
3294         }
3295
3296         for (p = commit->parents; p; p = p->next) {
3297                 struct commit *parent = p->item;
3298                 int *pi;
3299
3300                 if (parent->object.flags & UNINTERESTING)
3301                         continue;
3302
3303                 if (parse_commit_gently(parent, 1) < 0)
3304                         continue;
3305
3306                 if (parent->generation < info->min_generation) {
3307                         info->min_generation = parent->generation;
3308                         compute_indegrees_to_depth(revs, info->min_generation);
3309                 }
3310
3311                 pi = indegree_slab_at(&info->indegree, parent);
3312
3313                 (*pi)--;
3314                 if (*pi == 1)
3315                         prio_queue_put(&info->topo_queue, parent);
3316
3317                 if (revs->first_parent_only)
3318                         return;
3319         }
3320 }
3321
3322 int prepare_revision_walk(struct rev_info *revs)
3323 {
3324         int i;
3325         struct object_array old_pending;
3326         struct commit_list **next = &revs->commits;
3327
3328         memcpy(&old_pending, &revs->pending, sizeof(old_pending));
3329         revs->pending.nr = 0;
3330         revs->pending.alloc = 0;
3331         revs->pending.objects = NULL;
3332         for (i = 0; i < old_pending.nr; i++) {
3333                 struct object_array_entry *e = old_pending.objects + i;
3334                 struct commit *commit = handle_commit(revs, e);
3335                 if (commit) {
3336                         if (!(commit->object.flags & SEEN)) {
3337                                 commit->object.flags |= SEEN;
3338                                 next = commit_list_append(commit, next);
3339                         }
3340                 }
3341         }
3342         object_array_clear(&old_pending);
3343
3344         /* Signal whether we need per-parent treesame decoration */
3345         if (revs->simplify_merges ||
3346             (revs->limited && limiting_can_increase_treesame(revs)))
3347                 revs->treesame.name = "treesame";
3348
3349         if (revs->exclude_promisor_objects) {
3350                 for_each_packed_object(mark_uninteresting, revs,
3351                                        FOR_EACH_OBJECT_PROMISOR_ONLY);
3352         }
3353
3354         if (revs->no_walk != REVISION_WALK_NO_WALK_UNSORTED)
3355                 commit_list_sort_by_date(&revs->commits);
3356         if (revs->no_walk)
3357                 return 0;
3358         if (revs->limited) {
3359                 if (limit_list(revs) < 0)
3360                         return -1;
3361                 if (revs->topo_order)
3362                         sort_in_topological_order(&revs->commits, revs->sort_order);
3363         } else if (revs->topo_order)
3364                 init_topo_walk(revs);
3365         if (revs->line_level_traverse)
3366                 line_log_filter(revs);
3367         if (revs->simplify_merges)
3368                 simplify_merges(revs);
3369         if (revs->children.name)
3370                 set_children(revs);
3371         return 0;
3372 }
3373
3374 static enum rewrite_result rewrite_one_1(struct rev_info *revs,
3375                                          struct commit **pp,
3376                                          struct prio_queue *queue)
3377 {
3378         for (;;) {
3379                 struct commit *p = *pp;
3380                 if (!revs->limited)
3381                         if (process_parents(revs, p, NULL, queue) < 0)
3382                                 return rewrite_one_error;
3383                 if (p->object.flags & UNINTERESTING)
3384                         return rewrite_one_ok;
3385                 if (!(p->object.flags & TREESAME))
3386                         return rewrite_one_ok;
3387                 if (!p->parents)
3388                         return rewrite_one_noparents;
3389                 if ((p = one_relevant_parent(revs, p->parents)) == NULL)
3390                         return rewrite_one_ok;
3391                 *pp = p;
3392         }
3393 }
3394
3395 static void merge_queue_into_list(struct prio_queue *q, struct commit_list **list)
3396 {
3397         while (q->nr) {
3398                 struct commit *item = prio_queue_peek(q);
3399                 struct commit_list *p = *list;
3400
3401                 if (p && p->item->date >= item->date)
3402                         list = &p->next;
3403                 else {
3404                         p = commit_list_insert(item, list);
3405                         list = &p->next; /* skip newly added item */
3406                         prio_queue_get(q); /* pop item */
3407                 }
3408         }
3409 }
3410
3411 static enum rewrite_result rewrite_one(struct rev_info *revs, struct commit **pp)
3412 {
3413         struct prio_queue queue = { compare_commits_by_commit_date };
3414         enum rewrite_result ret = rewrite_one_1(revs, pp, &queue);
3415         merge_queue_into_list(&queue, &revs->commits);
3416         clear_prio_queue(&queue);
3417         return ret;
3418 }
3419
3420 int rewrite_parents(struct rev_info *revs, struct commit *commit,
3421         rewrite_parent_fn_t rewrite_parent)
3422 {
3423         struct commit_list **pp = &commit->parents;
3424         while (*pp) {
3425                 struct commit_list *parent = *pp;
3426                 switch (rewrite_parent(revs, &parent->item)) {
3427                 case rewrite_one_ok:
3428                         break;
3429                 case rewrite_one_noparents:
3430                         *pp = parent->next;
3431                         continue;
3432                 case rewrite_one_error:
3433                         return -1;
3434                 }
3435                 pp = &parent->next;
3436         }
3437         remove_duplicate_parents(revs, commit);
3438         return 0;
3439 }
3440
3441 static int commit_rewrite_person(struct strbuf *buf, const char *what, struct string_list *mailmap)
3442 {
3443         char *person, *endp;
3444         size_t len, namelen, maillen;
3445         const char *name;
3446         const char *mail;
3447         struct ident_split ident;
3448
3449         person = strstr(buf->buf, what);
3450         if (!person)
3451                 return 0;
3452
3453         person += strlen(what);
3454         endp = strchr(person, '\n');
3455         if (!endp)
3456                 return 0;
3457
3458         len = endp - person;
3459
3460         if (split_ident_line(&ident, person, len))
3461                 return 0;
3462
3463         mail = ident.mail_begin;
3464         maillen = ident.mail_end - ident.mail_begin;
3465         name = ident.name_begin;
3466         namelen = ident.name_end - ident.name_begin;
3467
3468         if (map_user(mailmap, &mail, &maillen, &name, &namelen)) {
3469                 struct strbuf namemail = STRBUF_INIT;
3470
3471                 strbuf_addf(&namemail, "%.*s <%.*s>",
3472                             (int)namelen, name, (int)maillen, mail);
3473
3474                 strbuf_splice(buf, ident.name_begin - buf->buf,
3475                               ident.mail_end - ident.name_begin + 1,
3476                               namemail.buf, namemail.len);
3477
3478                 strbuf_release(&namemail);
3479
3480                 return 1;
3481         }
3482
3483         return 0;
3484 }
3485
3486 static int commit_match(struct commit *commit, struct rev_info *opt)
3487 {
3488         int retval;
3489         const char *encoding;
3490         const char *message;
3491         struct strbuf buf = STRBUF_INIT;
3492
3493         if (!opt->grep_filter.pattern_list && !opt->grep_filter.header_list)
3494                 return 1;
3495
3496         /* Prepend "fake" headers as needed */
3497         if (opt->grep_filter.use_reflog_filter) {
3498                 strbuf_addstr(&buf, "reflog ");
3499                 get_reflog_message(&buf, opt->reflog_info);
3500                 strbuf_addch(&buf, '\n');
3501         }
3502
3503         /*
3504          * We grep in the user's output encoding, under the assumption that it
3505          * is the encoding they are most likely to write their grep pattern
3506          * for. In addition, it means we will match the "notes" encoding below,
3507          * so we will not end up with a buffer that has two different encodings
3508          * in it.
3509          */
3510         encoding = get_log_output_encoding();
3511         message = logmsg_reencode(commit, NULL, encoding);
3512
3513         /* Copy the commit to temporary if we are using "fake" headers */
3514         if (buf.len)
3515                 strbuf_addstr(&buf, message);
3516
3517         if (opt->grep_filter.header_list && opt->mailmap) {
3518                 if (!buf.len)
3519                         strbuf_addstr(&buf, message);
3520
3521                 commit_rewrite_person(&buf, "\nauthor ", opt->mailmap);
3522                 commit_rewrite_person(&buf, "\ncommitter ", opt->mailmap);
3523         }
3524
3525         /* Append "fake" message parts as needed */
3526         if (opt->show_notes) {
3527                 if (!buf.len)
3528                         strbuf_addstr(&buf, message);
3529                 format_display_notes(&commit->object.oid, &buf, encoding, 1);
3530         }
3531
3532         /*
3533          * Find either in the original commit message, or in the temporary.
3534          * Note that we cast away the constness of "message" here. It is
3535          * const because it may come from the cached commit buffer. That's OK,
3536          * because we know that it is modifiable heap memory, and that while
3537          * grep_buffer may modify it for speed, it will restore any
3538          * changes before returning.
3539          */
3540         if (buf.len)
3541                 retval = grep_buffer(&opt->grep_filter, buf.buf, buf.len);
3542         else
3543                 retval = grep_buffer(&opt->grep_filter,
3544                                      (char *)message, strlen(message));
3545         strbuf_release(&buf);
3546         unuse_commit_buffer(commit, message);
3547         return opt->invert_grep ? !retval : retval;
3548 }
3549
3550 static inline int want_ancestry(const struct rev_info *revs)
3551 {
3552         return (revs->rewrite_parents || revs->children.name);
3553 }
3554
3555 /*
3556  * Return a timestamp to be used for --since/--until comparisons for this
3557  * commit, based on the revision options.
3558  */
3559 static timestamp_t comparison_date(const struct rev_info *revs,
3560                                    struct commit *commit)
3561 {
3562         return revs->reflog_info ?
3563                 get_reflog_timestamp(revs->reflog_info) :
3564                 commit->date;
3565 }
3566
3567 enum commit_action get_commit_action(struct rev_info *revs, struct commit *commit)
3568 {
3569         if (commit->object.flags & SHOWN)
3570                 return commit_ignore;
3571         if (revs->unpacked && has_object_pack(&commit->object.oid))
3572                 return commit_ignore;
3573         if (commit->object.flags & UNINTERESTING)
3574                 return commit_ignore;
3575         if (revs->min_age != -1 &&
3576             comparison_date(revs, commit) > revs->min_age)
3577                         return commit_ignore;
3578         if (revs->min_parents || (revs->max_parents >= 0)) {
3579                 int n = commit_list_count(commit->parents);
3580                 if ((n < revs->min_parents) ||
3581                     ((revs->max_parents >= 0) && (n > revs->max_parents)))
3582                         return commit_ignore;
3583         }
3584         if (!commit_match(commit, revs))
3585                 return commit_ignore;
3586         if (revs->prune && revs->dense) {
3587                 /* Commit without changes? */
3588                 if (commit->object.flags & TREESAME) {
3589                         int n;
3590                         struct commit_list *p;
3591                         /* drop merges unless we want parenthood */
3592                         if (!want_ancestry(revs))
3593                                 return commit_ignore;
3594                         /*
3595                          * If we want ancestry, then need to keep any merges
3596                          * between relevant commits to tie together topology.
3597                          * For consistency with TREESAME and simplification
3598                          * use "relevant" here rather than just INTERESTING,
3599                          * to treat bottom commit(s) as part of the topology.
3600                          */
3601                         for (n = 0, p = commit->parents; p; p = p->next)
3602                                 if (relevant_commit(p->item))
3603                                         if (++n >= 2)
3604                                                 return commit_show;
3605                         return commit_ignore;
3606                 }
3607         }
3608         return commit_show;
3609 }
3610
3611 define_commit_slab(saved_parents, struct commit_list *);
3612
3613 #define EMPTY_PARENT_LIST ((struct commit_list *)-1)
3614
3615 /*
3616  * You may only call save_parents() once per commit (this is checked
3617  * for non-root commits).
3618  */
3619 static void save_parents(struct rev_info *revs, struct commit *commit)
3620 {
3621         struct commit_list **pp;
3622
3623         if (!revs->saved_parents_slab) {
3624                 revs->saved_parents_slab = xmalloc(sizeof(struct saved_parents));
3625                 init_saved_parents(revs->saved_parents_slab);
3626         }
3627
3628         pp = saved_parents_at(revs->saved_parents_slab, commit);
3629
3630         /*
3631          * When walking with reflogs, we may visit the same commit
3632          * several times: once for each appearance in the reflog.
3633          *
3634          * In this case, save_parents() will be called multiple times.
3635          * We want to keep only the first set of parents.  We need to
3636          * store a sentinel value for an empty (i.e., NULL) parent
3637          * list to distinguish it from a not-yet-saved list, however.
3638          */
3639         if (*pp)
3640                 return;
3641         if (commit->parents)
3642                 *pp = copy_commit_list(commit->parents);
3643         else
3644                 *pp = EMPTY_PARENT_LIST;
3645 }
3646
3647 static void free_saved_parents(struct rev_info *revs)
3648 {
3649         if (revs->saved_parents_slab)
3650                 clear_saved_parents(revs->saved_parents_slab);
3651 }
3652
3653 struct commit_list *get_saved_parents(struct rev_info *revs, const struct commit *commit)
3654 {
3655         struct commit_list *parents;
3656
3657         if (!revs->saved_parents_slab)
3658                 return commit->parents;
3659
3660         parents = *saved_parents_at(revs->saved_parents_slab, commit);
3661         if (parents == EMPTY_PARENT_LIST)
3662                 return NULL;
3663         return parents;
3664 }
3665
3666 enum commit_action simplify_commit(struct rev_info *revs, struct commit *commit)
3667 {
3668         enum commit_action action = get_commit_action(revs, commit);
3669
3670         if (action == commit_show &&
3671             revs->prune && revs->dense && want_ancestry(revs)) {
3672                 /*
3673                  * --full-diff on simplified parents is no good: it
3674                  * will show spurious changes from the commits that
3675                  * were elided.  So we save the parents on the side
3676                  * when --full-diff is in effect.
3677                  */
3678                 if (revs->full_diff)
3679                         save_parents(revs, commit);
3680                 if (rewrite_parents(revs, commit, rewrite_one) < 0)
3681                         return commit_error;
3682         }
3683         return action;
3684 }
3685
3686 static void track_linear(struct rev_info *revs, struct commit *commit)
3687 {
3688         if (revs->track_first_time) {
3689                 revs->linear = 1;
3690                 revs->track_first_time = 0;
3691         } else {
3692                 struct commit_list *p;
3693                 for (p = revs->previous_parents; p; p = p->next)
3694                         if (p->item == NULL || /* first commit */
3695                             oideq(&p->item->object.oid, &commit->object.oid))
3696                                 break;
3697                 revs->linear = p != NULL;
3698         }
3699         if (revs->reverse) {
3700                 if (revs->linear)
3701                         commit->object.flags |= TRACK_LINEAR;
3702         }
3703         free_commit_list(revs->previous_parents);
3704         revs->previous_parents = copy_commit_list(commit->parents);
3705 }
3706
3707 static struct commit *get_revision_1(struct rev_info *revs)
3708 {
3709         while (1) {
3710                 struct commit *commit;
3711
3712                 if (revs->reflog_info)
3713                         commit = next_reflog_entry(revs->reflog_info);
3714                 else if (revs->topo_walk_info)
3715                         commit = next_topo_commit(revs);
3716                 else
3717                         commit = pop_commit(&revs->commits);
3718
3719                 if (!commit)
3720                         return NULL;
3721
3722                 if (revs->reflog_info)
3723                         commit->object.flags &= ~(ADDED | SEEN | SHOWN);
3724
3725                 /*
3726                  * If we haven't done the list limiting, we need to look at
3727                  * the parents here. We also need to do the date-based limiting
3728                  * that we'd otherwise have done in limit_list().
3729                  */
3730                 if (!revs->limited) {
3731                         if (revs->max_age != -1 &&
3732                             comparison_date(revs, commit) < revs->max_age)
3733                                 continue;
3734
3735                         if (revs->reflog_info)
3736                                 try_to_simplify_commit(revs, commit);
3737                         else if (revs->topo_walk_info)
3738                                 expand_topo_walk(revs, commit);
3739                         else if (process_parents(revs, commit, &revs->commits, NULL) < 0) {
3740                                 if (!revs->ignore_missing_links)
3741                                         die("Failed to traverse parents of commit %s",
3742                                                 oid_to_hex(&commit->object.oid));
3743                         }
3744                 }
3745
3746                 switch (simplify_commit(revs, commit)) {
3747                 case commit_ignore:
3748                         continue;
3749                 case commit_error:
3750                         die("Failed to simplify parents of commit %s",
3751                             oid_to_hex(&commit->object.oid));
3752                 default:
3753                         if (revs->track_linear)
3754                                 track_linear(revs, commit);
3755                         return commit;
3756                 }
3757         }
3758 }
3759
3760 /*
3761  * Return true for entries that have not yet been shown.  (This is an
3762  * object_array_each_func_t.)
3763  */
3764 static int entry_unshown(struct object_array_entry *entry, void *cb_data_unused)
3765 {
3766         return !(entry->item->flags & SHOWN);
3767 }
3768
3769 /*
3770  * If array is on the verge of a realloc, garbage-collect any entries
3771  * that have already been shown to try to free up some space.
3772  */
3773 static void gc_boundary(struct object_array *array)
3774 {
3775         if (array->nr == array->alloc)
3776                 object_array_filter(array, entry_unshown, NULL);
3777 }
3778
3779 static void create_boundary_commit_list(struct rev_info *revs)
3780 {
3781         unsigned i;
3782         struct commit *c;
3783         struct object_array *array = &revs->boundary_commits;
3784         struct object_array_entry *objects = array->objects;
3785
3786         /*
3787          * If revs->commits is non-NULL at this point, an error occurred in
3788          * get_revision_1().  Ignore the error and continue printing the
3789          * boundary commits anyway.  (This is what the code has always
3790          * done.)
3791          */
3792         if (revs->commits) {
3793                 free_commit_list(revs->commits);
3794                 revs->commits = NULL;
3795         }
3796
3797         /*
3798          * Put all of the actual boundary commits from revs->boundary_commits
3799          * into revs->commits
3800          */
3801         for (i = 0; i < array->nr; i++) {
3802                 c = (struct commit *)(objects[i].item);
3803                 if (!c)
3804                         continue;
3805                 if (!(c->object.flags & CHILD_SHOWN))
3806                         continue;
3807                 if (c->object.flags & (SHOWN | BOUNDARY))
3808                         continue;
3809                 c->object.flags |= BOUNDARY;
3810                 commit_list_insert(c, &revs->commits);
3811         }
3812
3813         /*
3814          * If revs->topo_order is set, sort the boundary commits
3815          * in topological order
3816          */
3817         sort_in_topological_order(&revs->commits, revs->sort_order);
3818 }
3819
3820 static struct commit *get_revision_internal(struct rev_info *revs)
3821 {
3822         struct commit *c = NULL;
3823         struct commit_list *l;
3824
3825         if (revs->boundary == 2) {
3826                 /*
3827                  * All of the normal commits have already been returned,
3828                  * and we are now returning boundary commits.
3829                  * create_boundary_commit_list() has populated
3830                  * revs->commits with the remaining commits to return.
3831                  */
3832                 c = pop_commit(&revs->commits);
3833                 if (c)
3834                         c->object.flags |= SHOWN;
3835                 return c;
3836         }
3837
3838         /*
3839          * If our max_count counter has reached zero, then we are done. We
3840          * don't simply return NULL because we still might need to show
3841          * boundary commits. But we want to avoid calling get_revision_1, which
3842          * might do a considerable amount of work finding the next commit only
3843          * for us to throw it away.
3844          *
3845          * If it is non-zero, then either we don't have a max_count at all
3846          * (-1), or it is still counting, in which case we decrement.
3847          */
3848         if (revs->max_count) {
3849                 c = get_revision_1(revs);
3850                 if (c) {
3851                         while (revs->skip_count > 0) {
3852                                 revs->skip_count--;
3853                                 c = get_revision_1(revs);
3854                                 if (!c)
3855                                         break;
3856                         }
3857                 }
3858
3859                 if (revs->max_count > 0)
3860                         revs->max_count--;
3861         }
3862
3863         if (c)
3864                 c->object.flags |= SHOWN;
3865
3866         if (!revs->boundary)
3867                 return c;
3868
3869         if (!c) {
3870                 /*
3871                  * get_revision_1() runs out the commits, and
3872                  * we are done computing the boundaries.
3873                  * switch to boundary commits output mode.
3874                  */
3875                 revs->boundary = 2;
3876
3877                 /*
3878                  * Update revs->commits to contain the list of
3879                  * boundary commits.
3880                  */
3881                 create_boundary_commit_list(revs);
3882
3883                 return get_revision_internal(revs);
3884         }
3885
3886         /*
3887          * boundary commits are the commits that are parents of the
3888          * ones we got from get_revision_1() but they themselves are
3889          * not returned from get_revision_1().  Before returning
3890          * 'c', we need to mark its parents that they could be boundaries.
3891          */
3892
3893         for (l = c->parents; l; l = l->next) {
3894                 struct object *p;
3895                 p = &(l->item->object);
3896                 if (p->flags & (CHILD_SHOWN | SHOWN))
3897                         continue;
3898                 p->flags |= CHILD_SHOWN;
3899                 gc_boundary(&revs->boundary_commits);
3900                 add_object_array(p, NULL, &revs->boundary_commits);
3901         }
3902
3903         return c;
3904 }
3905
3906 struct commit *get_revision(struct rev_info *revs)
3907 {
3908         struct commit *c;
3909         struct commit_list *reversed;
3910
3911         if (revs->reverse) {
3912                 reversed = NULL;
3913                 while ((c = get_revision_internal(revs)))
3914                         commit_list_insert(c, &reversed);
3915                 revs->commits = reversed;
3916                 revs->reverse = 0;
3917                 revs->reverse_output_stage = 1;
3918         }
3919
3920         if (revs->reverse_output_stage) {
3921                 c = pop_commit(&revs->commits);
3922                 if (revs->track_linear)
3923                         revs->linear = !!(c && c->object.flags & TRACK_LINEAR);
3924                 return c;
3925         }
3926
3927         c = get_revision_internal(revs);
3928         if (c && revs->graph)
3929                 graph_update(revs->graph, c);
3930         if (!c) {
3931                 free_saved_parents(revs);
3932                 if (revs->previous_parents) {
3933                         free_commit_list(revs->previous_parents);
3934                         revs->previous_parents = NULL;
3935                 }
3936         }
3937         return c;
3938 }
3939
3940 char *get_revision_mark(const struct rev_info *revs, const struct commit *commit)
3941 {
3942         if (commit->object.flags & BOUNDARY)
3943                 return "-";
3944         else if (commit->object.flags & UNINTERESTING)
3945                 return "^";
3946         else if (commit->object.flags & PATCHSAME)
3947                 return "=";
3948         else if (!revs || revs->left_right) {
3949                 if (commit->object.flags & SYMMETRIC_LEFT)
3950                         return "<";
3951                 else
3952                         return ">";
3953         } else if (revs->graph)
3954                 return "*";
3955         else if (revs->cherry_mark)
3956                 return "+";
3957         return "";
3958 }
3959
3960 void put_revision_mark(const struct rev_info *revs, const struct commit *commit)
3961 {
3962         char *mark = get_revision_mark(revs, commit);
3963         if (!strlen(mark))
3964                 return;
3965         fputs(mark, stdout);
3966         putchar(' ');
3967 }