Sync with 2.19.2
[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
31 volatile show_early_output_fn_t show_early_output;
32
33 static const char *term_bad;
34 static const char *term_good;
35
36 implement_shared_commit_slab(revision_sources, char *);
37
38 void show_object_with_name(FILE *out, struct object *obj, const char *name)
39 {
40         const char *p;
41
42         fprintf(out, "%s ", oid_to_hex(&obj->oid));
43         for (p = name; *p && *p != '\n'; p++)
44                 fputc(*p, out);
45         fputc('\n', out);
46 }
47
48 static void mark_blob_uninteresting(struct blob *blob)
49 {
50         if (!blob)
51                 return;
52         if (blob->object.flags & UNINTERESTING)
53                 return;
54         blob->object.flags |= UNINTERESTING;
55 }
56
57 static void mark_tree_contents_uninteresting(struct repository *r,
58                                              struct tree *tree)
59 {
60         struct tree_desc desc;
61         struct name_entry entry;
62
63         if (parse_tree_gently(tree, 1) < 0)
64                 return;
65
66         init_tree_desc(&desc, tree->buffer, tree->size);
67         while (tree_entry(&desc, &entry)) {
68                 switch (object_type(entry.mode)) {
69                 case OBJ_TREE:
70                         mark_tree_uninteresting(r, lookup_tree(r, entry.oid));
71                         break;
72                 case OBJ_BLOB:
73                         mark_blob_uninteresting(lookup_blob(r, entry.oid));
74                         break;
75                 default:
76                         /* Subproject commit - not in this repository */
77                         break;
78                 }
79         }
80
81         /*
82          * We don't care about the tree any more
83          * after it has been marked uninteresting.
84          */
85         free_tree_buffer(tree);
86 }
87
88 void mark_tree_uninteresting(struct repository *r, struct tree *tree)
89 {
90         struct object *obj;
91
92         if (!tree)
93                 return;
94
95         obj = &tree->object;
96         if (obj->flags & UNINTERESTING)
97                 return;
98         obj->flags |= UNINTERESTING;
99         mark_tree_contents_uninteresting(r, tree);
100 }
101
102 struct commit_stack {
103         struct commit **items;
104         size_t nr, alloc;
105 };
106 #define COMMIT_STACK_INIT { NULL, 0, 0 }
107
108 static void commit_stack_push(struct commit_stack *stack, struct commit *commit)
109 {
110         ALLOC_GROW(stack->items, stack->nr + 1, stack->alloc);
111         stack->items[stack->nr++] = commit;
112 }
113
114 static struct commit *commit_stack_pop(struct commit_stack *stack)
115 {
116         return stack->nr ? stack->items[--stack->nr] : NULL;
117 }
118
119 static void commit_stack_clear(struct commit_stack *stack)
120 {
121         FREE_AND_NULL(stack->items);
122         stack->nr = stack->alloc = 0;
123 }
124
125 static void mark_one_parent_uninteresting(struct commit *commit,
126                                           struct commit_stack *pending)
127 {
128         struct commit_list *l;
129
130         if (commit->object.flags & UNINTERESTING)
131                 return;
132         commit->object.flags |= UNINTERESTING;
133
134         /*
135          * Normally we haven't parsed the parent
136          * yet, so we won't have a parent of a parent
137          * here. However, it may turn out that we've
138          * reached this commit some other way (where it
139          * wasn't uninteresting), in which case we need
140          * to mark its parents recursively too..
141          */
142         for (l = commit->parents; l; l = l->next)
143                 commit_stack_push(pending, l->item);
144 }
145
146 void mark_parents_uninteresting(struct commit *commit)
147 {
148         struct commit_stack pending = COMMIT_STACK_INIT;
149         struct commit_list *l;
150
151         for (l = commit->parents; l; l = l->next)
152                 mark_one_parent_uninteresting(l->item, &pending);
153
154         while (pending.nr > 0)
155                 mark_one_parent_uninteresting(commit_stack_pop(&pending),
156                                               &pending);
157
158         commit_stack_clear(&pending);
159 }
160
161 static void add_pending_object_with_path(struct rev_info *revs,
162                                          struct object *obj,
163                                          const char *name, unsigned mode,
164                                          const char *path)
165 {
166         if (!obj)
167                 return;
168         if (revs->no_walk && (obj->flags & UNINTERESTING))
169                 revs->no_walk = 0;
170         if (revs->reflog_info && obj->type == OBJ_COMMIT) {
171                 struct strbuf buf = STRBUF_INIT;
172                 int len = interpret_branch_name(name, 0, &buf, 0);
173
174                 if (0 < len && name[len] && buf.len)
175                         strbuf_addstr(&buf, name + len);
176                 add_reflog_for_walk(revs->reflog_info,
177                                     (struct commit *)obj,
178                                     buf.buf[0] ? buf.buf: name);
179                 strbuf_release(&buf);
180                 return; /* do not add the commit itself */
181         }
182         add_object_array_with_path(obj, name, &revs->pending, mode, path);
183 }
184
185 static void add_pending_object_with_mode(struct rev_info *revs,
186                                          struct object *obj,
187                                          const char *name, unsigned mode)
188 {
189         add_pending_object_with_path(revs, obj, name, mode, NULL);
190 }
191
192 void add_pending_object(struct rev_info *revs,
193                         struct object *obj, const char *name)
194 {
195         add_pending_object_with_mode(revs, obj, name, S_IFINVALID);
196 }
197
198 void add_head_to_pending(struct rev_info *revs)
199 {
200         struct object_id oid;
201         struct object *obj;
202         if (get_oid("HEAD", &oid))
203                 return;
204         obj = parse_object(revs->repo, &oid);
205         if (!obj)
206                 return;
207         add_pending_object(revs, obj, "HEAD");
208 }
209
210 static struct object *get_reference(struct rev_info *revs, const char *name,
211                                     const struct object_id *oid,
212                                     unsigned int flags)
213 {
214         struct object *object;
215
216         object = parse_object(revs->repo, oid);
217         if (!object) {
218                 if (revs->ignore_missing)
219                         return object;
220                 if (revs->exclude_promisor_objects && is_promisor_object(oid))
221                         return NULL;
222                 die("bad object %s", name);
223         }
224         object->flags |= flags;
225         return object;
226 }
227
228 void add_pending_oid(struct rev_info *revs, const char *name,
229                       const struct object_id *oid, unsigned int flags)
230 {
231         struct object *object = get_reference(revs, name, oid, flags);
232         add_pending_object(revs, object, name);
233 }
234
235 static struct commit *handle_commit(struct rev_info *revs,
236                                     struct object_array_entry *entry)
237 {
238         struct object *object = entry->item;
239         const char *name = entry->name;
240         const char *path = entry->path;
241         unsigned int mode = entry->mode;
242         unsigned long flags = object->flags;
243
244         /*
245          * Tag object? Look what it points to..
246          */
247         while (object->type == OBJ_TAG) {
248                 struct tag *tag = (struct tag *) object;
249                 if (revs->tag_objects && !(flags & UNINTERESTING))
250                         add_pending_object(revs, object, tag->tag);
251                 if (!tag->tagged)
252                         die("bad tag");
253                 object = parse_object(revs->repo, &tag->tagged->oid);
254                 if (!object) {
255                         if (revs->ignore_missing_links || (flags & UNINTERESTING))
256                                 return NULL;
257                         if (revs->exclude_promisor_objects &&
258                             is_promisor_object(&tag->tagged->oid))
259                                 return NULL;
260                         die("bad object %s", oid_to_hex(&tag->tagged->oid));
261                 }
262                 object->flags |= flags;
263                 /*
264                  * We'll handle the tagged object by looping or dropping
265                  * through to the non-tag handlers below. Do not
266                  * propagate path data from the tag's pending entry.
267                  */
268                 path = NULL;
269                 mode = 0;
270         }
271
272         /*
273          * Commit object? Just return it, we'll do all the complex
274          * reachability crud.
275          */
276         if (object->type == OBJ_COMMIT) {
277                 struct commit *commit = (struct commit *)object;
278
279                 if (parse_commit(commit) < 0)
280                         die("unable to parse commit %s", name);
281                 if (flags & UNINTERESTING) {
282                         mark_parents_uninteresting(commit);
283                         revs->limited = 1;
284                 }
285                 if (revs->sources) {
286                         char **slot = revision_sources_at(revs->sources, commit);
287
288                         if (!*slot)
289                                 *slot = xstrdup(name);
290                 }
291                 return commit;
292         }
293
294         /*
295          * Tree object? Either mark it uninteresting, or add it
296          * to the list of objects to look at later..
297          */
298         if (object->type == OBJ_TREE) {
299                 struct tree *tree = (struct tree *)object;
300                 if (!revs->tree_objects)
301                         return NULL;
302                 if (flags & UNINTERESTING) {
303                         mark_tree_contents_uninteresting(revs->repo, tree);
304                         return NULL;
305                 }
306                 add_pending_object_with_path(revs, object, name, mode, path);
307                 return NULL;
308         }
309
310         /*
311          * Blob object? You know the drill by now..
312          */
313         if (object->type == OBJ_BLOB) {
314                 if (!revs->blob_objects)
315                         return NULL;
316                 if (flags & UNINTERESTING)
317                         return NULL;
318                 add_pending_object_with_path(revs, object, name, mode, path);
319                 return NULL;
320         }
321         die("%s is unknown object", name);
322 }
323
324 static int everybody_uninteresting(struct commit_list *orig,
325                                    struct commit **interesting_cache)
326 {
327         struct commit_list *list = orig;
328
329         if (*interesting_cache) {
330                 struct commit *commit = *interesting_cache;
331                 if (!(commit->object.flags & UNINTERESTING))
332                         return 0;
333         }
334
335         while (list) {
336                 struct commit *commit = list->item;
337                 list = list->next;
338                 if (commit->object.flags & UNINTERESTING)
339                         continue;
340
341                 *interesting_cache = commit;
342                 return 0;
343         }
344         return 1;
345 }
346
347 /*
348  * A definition of "relevant" commit that we can use to simplify limited graphs
349  * by eliminating side branches.
350  *
351  * A "relevant" commit is one that is !UNINTERESTING (ie we are including it
352  * in our list), or that is a specified BOTTOM commit. Then after computing
353  * a limited list, during processing we can generally ignore boundary merges
354  * coming from outside the graph, (ie from irrelevant parents), and treat
355  * those merges as if they were single-parent. TREESAME is defined to consider
356  * only relevant parents, if any. If we are TREESAME to our on-graph parents,
357  * we don't care if we were !TREESAME to non-graph parents.
358  *
359  * Treating bottom commits as relevant ensures that a limited graph's
360  * connection to the actual bottom commit is not viewed as a side branch, but
361  * treated as part of the graph. For example:
362  *
363  *   ....Z...A---X---o---o---B
364  *        .     /
365  *         W---Y
366  *
367  * When computing "A..B", the A-X connection is at least as important as
368  * Y-X, despite A being flagged UNINTERESTING.
369  *
370  * And when computing --ancestry-path "A..B", the A-X connection is more
371  * important than Y-X, despite both A and Y being flagged UNINTERESTING.
372  */
373 static inline int relevant_commit(struct commit *commit)
374 {
375         return (commit->object.flags & (UNINTERESTING | BOTTOM)) != UNINTERESTING;
376 }
377
378 /*
379  * Return a single relevant commit from a parent list. If we are a TREESAME
380  * commit, and this selects one of our parents, then we can safely simplify to
381  * that parent.
382  */
383 static struct commit *one_relevant_parent(const struct rev_info *revs,
384                                           struct commit_list *orig)
385 {
386         struct commit_list *list = orig;
387         struct commit *relevant = NULL;
388
389         if (!orig)
390                 return NULL;
391
392         /*
393          * For 1-parent commits, or if first-parent-only, then return that
394          * first parent (even if not "relevant" by the above definition).
395          * TREESAME will have been set purely on that parent.
396          */
397         if (revs->first_parent_only || !orig->next)
398                 return orig->item;
399
400         /*
401          * For multi-parent commits, identify a sole relevant parent, if any.
402          * If we have only one relevant parent, then TREESAME will be set purely
403          * with regard to that parent, and we can simplify accordingly.
404          *
405          * If we have more than one relevant parent, or no relevant parents
406          * (and multiple irrelevant ones), then we can't select a parent here
407          * and return NULL.
408          */
409         while (list) {
410                 struct commit *commit = list->item;
411                 list = list->next;
412                 if (relevant_commit(commit)) {
413                         if (relevant)
414                                 return NULL;
415                         relevant = commit;
416                 }
417         }
418         return relevant;
419 }
420
421 /*
422  * The goal is to get REV_TREE_NEW as the result only if the
423  * diff consists of all '+' (and no other changes), REV_TREE_OLD
424  * if the whole diff is removal of old data, and otherwise
425  * REV_TREE_DIFFERENT (of course if the trees are the same we
426  * want REV_TREE_SAME).
427  *
428  * The only time we care about the distinction is when
429  * remove_empty_trees is in effect, in which case we care only about
430  * whether the whole change is REV_TREE_NEW, or if there's another type
431  * of change. Which means we can stop the diff early in either of these
432  * cases:
433  *
434  *   1. We're not using remove_empty_trees at all.
435  *
436  *   2. We saw anything except REV_TREE_NEW.
437  */
438 static int tree_difference = REV_TREE_SAME;
439
440 static void file_add_remove(struct diff_options *options,
441                     int addremove, unsigned mode,
442                     const struct object_id *oid,
443                     int oid_valid,
444                     const char *fullpath, unsigned dirty_submodule)
445 {
446         int diff = addremove == '+' ? REV_TREE_NEW : REV_TREE_OLD;
447         struct rev_info *revs = options->change_fn_data;
448
449         tree_difference |= diff;
450         if (!revs->remove_empty_trees || tree_difference != REV_TREE_NEW)
451                 options->flags.has_changes = 1;
452 }
453
454 static void file_change(struct diff_options *options,
455                  unsigned old_mode, unsigned new_mode,
456                  const struct object_id *old_oid,
457                  const struct object_id *new_oid,
458                  int old_oid_valid, int new_oid_valid,
459                  const char *fullpath,
460                  unsigned old_dirty_submodule, unsigned new_dirty_submodule)
461 {
462         tree_difference = REV_TREE_DIFFERENT;
463         options->flags.has_changes = 1;
464 }
465
466 static int rev_compare_tree(struct rev_info *revs,
467                             struct commit *parent, struct commit *commit)
468 {
469         struct tree *t1 = get_commit_tree(parent);
470         struct tree *t2 = get_commit_tree(commit);
471
472         if (!t1)
473                 return REV_TREE_NEW;
474         if (!t2)
475                 return REV_TREE_OLD;
476
477         if (revs->simplify_by_decoration) {
478                 /*
479                  * If we are simplifying by decoration, then the commit
480                  * is worth showing if it has a tag pointing at it.
481                  */
482                 if (get_name_decoration(&commit->object))
483                         return REV_TREE_DIFFERENT;
484                 /*
485                  * A commit that is not pointed by a tag is uninteresting
486                  * if we are not limited by path.  This means that you will
487                  * see the usual "commits that touch the paths" plus any
488                  * tagged commit by specifying both --simplify-by-decoration
489                  * and pathspec.
490                  */
491                 if (!revs->prune_data.nr)
492                         return REV_TREE_SAME;
493         }
494
495         tree_difference = REV_TREE_SAME;
496         revs->pruning.flags.has_changes = 0;
497         if (diff_tree_oid(&t1->object.oid, &t2->object.oid, "",
498                            &revs->pruning) < 0)
499                 return REV_TREE_DIFFERENT;
500         return tree_difference;
501 }
502
503 static int rev_same_tree_as_empty(struct rev_info *revs, struct commit *commit)
504 {
505         int retval;
506         struct tree *t1 = get_commit_tree(commit);
507
508         if (!t1)
509                 return 0;
510
511         tree_difference = REV_TREE_SAME;
512         revs->pruning.flags.has_changes = 0;
513         retval = diff_tree_oid(NULL, &t1->object.oid, "", &revs->pruning);
514
515         return retval >= 0 && (tree_difference == REV_TREE_SAME);
516 }
517
518 struct treesame_state {
519         unsigned int nparents;
520         unsigned char treesame[FLEX_ARRAY];
521 };
522
523 static struct treesame_state *initialise_treesame(struct rev_info *revs, struct commit *commit)
524 {
525         unsigned n = commit_list_count(commit->parents);
526         struct treesame_state *st = xcalloc(1, st_add(sizeof(*st), n));
527         st->nparents = n;
528         add_decoration(&revs->treesame, &commit->object, st);
529         return st;
530 }
531
532 /*
533  * Must be called immediately after removing the nth_parent from a commit's
534  * parent list, if we are maintaining the per-parent treesame[] decoration.
535  * This does not recalculate the master TREESAME flag - update_treesame()
536  * should be called to update it after a sequence of treesame[] modifications
537  * that may have affected it.
538  */
539 static int compact_treesame(struct rev_info *revs, struct commit *commit, unsigned nth_parent)
540 {
541         struct treesame_state *st;
542         int old_same;
543
544         if (!commit->parents) {
545                 /*
546                  * Have just removed the only parent from a non-merge.
547                  * Different handling, as we lack decoration.
548                  */
549                 if (nth_parent != 0)
550                         die("compact_treesame %u", nth_parent);
551                 old_same = !!(commit->object.flags & TREESAME);
552                 if (rev_same_tree_as_empty(revs, commit))
553                         commit->object.flags |= TREESAME;
554                 else
555                         commit->object.flags &= ~TREESAME;
556                 return old_same;
557         }
558
559         st = lookup_decoration(&revs->treesame, &commit->object);
560         if (!st || nth_parent >= st->nparents)
561                 die("compact_treesame %u", nth_parent);
562
563         old_same = st->treesame[nth_parent];
564         memmove(st->treesame + nth_parent,
565                 st->treesame + nth_parent + 1,
566                 st->nparents - nth_parent - 1);
567
568         /*
569          * If we've just become a non-merge commit, update TREESAME
570          * immediately, and remove the no-longer-needed decoration.
571          * If still a merge, defer update until update_treesame().
572          */
573         if (--st->nparents == 1) {
574                 if (commit->parents->next)
575                         die("compact_treesame parents mismatch");
576                 if (st->treesame[0] && revs->dense)
577                         commit->object.flags |= TREESAME;
578                 else
579                         commit->object.flags &= ~TREESAME;
580                 free(add_decoration(&revs->treesame, &commit->object, NULL));
581         }
582
583         return old_same;
584 }
585
586 static unsigned update_treesame(struct rev_info *revs, struct commit *commit)
587 {
588         if (commit->parents && commit->parents->next) {
589                 unsigned n;
590                 struct treesame_state *st;
591                 struct commit_list *p;
592                 unsigned relevant_parents;
593                 unsigned relevant_change, irrelevant_change;
594
595                 st = lookup_decoration(&revs->treesame, &commit->object);
596                 if (!st)
597                         die("update_treesame %s", oid_to_hex(&commit->object.oid));
598                 relevant_parents = 0;
599                 relevant_change = irrelevant_change = 0;
600                 for (p = commit->parents, n = 0; p; n++, p = p->next) {
601                         if (relevant_commit(p->item)) {
602                                 relevant_change |= !st->treesame[n];
603                                 relevant_parents++;
604                         } else
605                                 irrelevant_change |= !st->treesame[n];
606                 }
607                 if (relevant_parents ? relevant_change : irrelevant_change)
608                         commit->object.flags &= ~TREESAME;
609                 else
610                         commit->object.flags |= TREESAME;
611         }
612
613         return commit->object.flags & TREESAME;
614 }
615
616 static inline int limiting_can_increase_treesame(const struct rev_info *revs)
617 {
618         /*
619          * TREESAME is irrelevant unless prune && dense;
620          * if simplify_history is set, we can't have a mixture of TREESAME and
621          *    !TREESAME INTERESTING parents (and we don't have treesame[]
622          *    decoration anyway);
623          * if first_parent_only is set, then the TREESAME flag is locked
624          *    against the first parent (and again we lack treesame[] decoration).
625          */
626         return revs->prune && revs->dense &&
627                !revs->simplify_history &&
628                !revs->first_parent_only;
629 }
630
631 static void try_to_simplify_commit(struct rev_info *revs, struct commit *commit)
632 {
633         struct commit_list **pp, *parent;
634         struct treesame_state *ts = NULL;
635         int relevant_change = 0, irrelevant_change = 0;
636         int relevant_parents, nth_parent;
637
638         /*
639          * If we don't do pruning, everything is interesting
640          */
641         if (!revs->prune)
642                 return;
643
644         if (!get_commit_tree(commit))
645                 return;
646
647         if (!commit->parents) {
648                 if (rev_same_tree_as_empty(revs, commit))
649                         commit->object.flags |= TREESAME;
650                 return;
651         }
652
653         /*
654          * Normal non-merge commit? If we don't want to make the
655          * history dense, we consider it always to be a change..
656          */
657         if (!revs->dense && !commit->parents->next)
658                 return;
659
660         for (pp = &commit->parents, nth_parent = 0, relevant_parents = 0;
661              (parent = *pp) != NULL;
662              pp = &parent->next, nth_parent++) {
663                 struct commit *p = parent->item;
664                 if (relevant_commit(p))
665                         relevant_parents++;
666
667                 if (nth_parent == 1) {
668                         /*
669                          * This our second loop iteration - so we now know
670                          * we're dealing with a merge.
671                          *
672                          * Do not compare with later parents when we care only about
673                          * the first parent chain, in order to avoid derailing the
674                          * traversal to follow a side branch that brought everything
675                          * in the path we are limited to by the pathspec.
676                          */
677                         if (revs->first_parent_only)
678                                 break;
679                         /*
680                          * If this will remain a potentially-simplifiable
681                          * merge, remember per-parent treesame if needed.
682                          * Initialise the array with the comparison from our
683                          * first iteration.
684                          */
685                         if (revs->treesame.name &&
686                             !revs->simplify_history &&
687                             !(commit->object.flags & UNINTERESTING)) {
688                                 ts = initialise_treesame(revs, commit);
689                                 if (!(irrelevant_change || relevant_change))
690                                         ts->treesame[0] = 1;
691                         }
692                 }
693                 if (parse_commit(p) < 0)
694                         die("cannot simplify commit %s (because of %s)",
695                             oid_to_hex(&commit->object.oid),
696                             oid_to_hex(&p->object.oid));
697                 switch (rev_compare_tree(revs, p, commit)) {
698                 case REV_TREE_SAME:
699                         if (!revs->simplify_history || !relevant_commit(p)) {
700                                 /* Even if a merge with an uninteresting
701                                  * side branch brought the entire change
702                                  * we are interested in, we do not want
703                                  * to lose the other branches of this
704                                  * merge, so we just keep going.
705                                  */
706                                 if (ts)
707                                         ts->treesame[nth_parent] = 1;
708                                 continue;
709                         }
710                         parent->next = NULL;
711                         commit->parents = parent;
712                         commit->object.flags |= TREESAME;
713                         return;
714
715                 case REV_TREE_NEW:
716                         if (revs->remove_empty_trees &&
717                             rev_same_tree_as_empty(revs, p)) {
718                                 /* We are adding all the specified
719                                  * paths from this parent, so the
720                                  * history beyond this parent is not
721                                  * interesting.  Remove its parents
722                                  * (they are grandparents for us).
723                                  * IOW, we pretend this parent is a
724                                  * "root" commit.
725                                  */
726                                 if (parse_commit(p) < 0)
727                                         die("cannot simplify commit %s (invalid %s)",
728                                             oid_to_hex(&commit->object.oid),
729                                             oid_to_hex(&p->object.oid));
730                                 p->parents = NULL;
731                         }
732                 /* fallthrough */
733                 case REV_TREE_OLD:
734                 case REV_TREE_DIFFERENT:
735                         if (relevant_commit(p))
736                                 relevant_change = 1;
737                         else
738                                 irrelevant_change = 1;
739                         continue;
740                 }
741                 die("bad tree compare for commit %s", oid_to_hex(&commit->object.oid));
742         }
743
744         /*
745          * TREESAME is straightforward for single-parent commits. For merge
746          * commits, it is most useful to define it so that "irrelevant"
747          * parents cannot make us !TREESAME - if we have any relevant
748          * parents, then we only consider TREESAMEness with respect to them,
749          * allowing irrelevant merges from uninteresting branches to be
750          * simplified away. Only if we have only irrelevant parents do we
751          * base TREESAME on them. Note that this logic is replicated in
752          * update_treesame, which should be kept in sync.
753          */
754         if (relevant_parents ? !relevant_change : !irrelevant_change)
755                 commit->object.flags |= TREESAME;
756 }
757
758 static void commit_list_insert_by_date_cached(struct commit *p, struct commit_list **head,
759                     struct commit_list *cached_base, struct commit_list **cache)
760 {
761         struct commit_list *new_entry;
762
763         if (cached_base && p->date < cached_base->item->date)
764                 new_entry = commit_list_insert_by_date(p, &cached_base->next);
765         else
766                 new_entry = commit_list_insert_by_date(p, head);
767
768         if (cache && (!*cache || p->date < (*cache)->item->date))
769                 *cache = new_entry;
770 }
771
772 static int process_parents(struct rev_info *revs, struct commit *commit,
773                            struct commit_list **list, struct commit_list **cache_ptr)
774 {
775         struct commit_list *parent = commit->parents;
776         unsigned left_flag;
777         struct commit_list *cached_base = cache_ptr ? *cache_ptr : NULL;
778
779         if (commit->object.flags & ADDED)
780                 return 0;
781         commit->object.flags |= ADDED;
782
783         if (revs->include_check &&
784             !revs->include_check(commit, revs->include_check_data))
785                 return 0;
786
787         /*
788          * If the commit is uninteresting, don't try to
789          * prune parents - we want the maximal uninteresting
790          * set.
791          *
792          * Normally we haven't parsed the parent
793          * yet, so we won't have a parent of a parent
794          * here. However, it may turn out that we've
795          * reached this commit some other way (where it
796          * wasn't uninteresting), in which case we need
797          * to mark its parents recursively too..
798          */
799         if (commit->object.flags & UNINTERESTING) {
800                 while (parent) {
801                         struct commit *p = parent->item;
802                         parent = parent->next;
803                         if (p)
804                                 p->object.flags |= UNINTERESTING;
805                         if (parse_commit_gently(p, 1) < 0)
806                                 continue;
807                         if (p->parents)
808                                 mark_parents_uninteresting(p);
809                         if (p->object.flags & SEEN)
810                                 continue;
811                         p->object.flags |= SEEN;
812                         if (list)
813                                 commit_list_insert_by_date_cached(p, list, cached_base, cache_ptr);
814                 }
815                 return 0;
816         }
817
818         /*
819          * Ok, the commit wasn't uninteresting. Try to
820          * simplify the commit history and find the parent
821          * that has no differences in the path set if one exists.
822          */
823         try_to_simplify_commit(revs, commit);
824
825         if (revs->no_walk)
826                 return 0;
827
828         left_flag = (commit->object.flags & SYMMETRIC_LEFT);
829
830         for (parent = commit->parents; parent; parent = parent->next) {
831                 struct commit *p = parent->item;
832                 int gently = revs->ignore_missing_links ||
833                              revs->exclude_promisor_objects;
834                 if (parse_commit_gently(p, gently) < 0) {
835                         if (revs->exclude_promisor_objects &&
836                             is_promisor_object(&p->object.oid)) {
837                                 if (revs->first_parent_only)
838                                         break;
839                                 continue;
840                         }
841                         return -1;
842                 }
843                 if (revs->sources) {
844                         char **slot = revision_sources_at(revs->sources, p);
845
846                         if (!*slot)
847                                 *slot = *revision_sources_at(revs->sources, commit);
848                 }
849                 p->object.flags |= left_flag;
850                 if (!(p->object.flags & SEEN)) {
851                         p->object.flags |= SEEN;
852                         if (list)
853                                 commit_list_insert_by_date_cached(p, list, cached_base, cache_ptr);
854                 }
855                 if (revs->first_parent_only)
856                         break;
857         }
858         return 0;
859 }
860
861 static void cherry_pick_list(struct commit_list *list, struct rev_info *revs)
862 {
863         struct commit_list *p;
864         int left_count = 0, right_count = 0;
865         int left_first;
866         struct patch_ids ids;
867         unsigned cherry_flag;
868
869         /* First count the commits on the left and on the right */
870         for (p = list; p; p = p->next) {
871                 struct commit *commit = p->item;
872                 unsigned flags = commit->object.flags;
873                 if (flags & BOUNDARY)
874                         ;
875                 else if (flags & SYMMETRIC_LEFT)
876                         left_count++;
877                 else
878                         right_count++;
879         }
880
881         if (!left_count || !right_count)
882                 return;
883
884         left_first = left_count < right_count;
885         init_patch_ids(revs->repo, &ids);
886         ids.diffopts.pathspec = revs->diffopt.pathspec;
887
888         /* Compute patch-ids for one side */
889         for (p = list; p; p = p->next) {
890                 struct commit *commit = p->item;
891                 unsigned flags = commit->object.flags;
892
893                 if (flags & BOUNDARY)
894                         continue;
895                 /*
896                  * If we have fewer left, left_first is set and we omit
897                  * commits on the right branch in this loop.  If we have
898                  * fewer right, we skip the left ones.
899                  */
900                 if (left_first != !!(flags & SYMMETRIC_LEFT))
901                         continue;
902                 add_commit_patch_id(commit, &ids);
903         }
904
905         /* either cherry_mark or cherry_pick are true */
906         cherry_flag = revs->cherry_mark ? PATCHSAME : SHOWN;
907
908         /* Check the other side */
909         for (p = list; p; p = p->next) {
910                 struct commit *commit = p->item;
911                 struct patch_id *id;
912                 unsigned flags = commit->object.flags;
913
914                 if (flags & BOUNDARY)
915                         continue;
916                 /*
917                  * If we have fewer left, left_first is set and we omit
918                  * commits on the left branch in this loop.
919                  */
920                 if (left_first == !!(flags & SYMMETRIC_LEFT))
921                         continue;
922
923                 /*
924                  * Have we seen the same patch id?
925                  */
926                 id = has_commit_patch_id(commit, &ids);
927                 if (!id)
928                         continue;
929
930                 commit->object.flags |= cherry_flag;
931                 id->commit->object.flags |= cherry_flag;
932         }
933
934         free_patch_ids(&ids);
935 }
936
937 /* How many extra uninteresting commits we want to see.. */
938 #define SLOP 5
939
940 static int still_interesting(struct commit_list *src, timestamp_t date, int slop,
941                              struct commit **interesting_cache)
942 {
943         /*
944          * No source list at all? We're definitely done..
945          */
946         if (!src)
947                 return 0;
948
949         /*
950          * Does the destination list contain entries with a date
951          * before the source list? Definitely _not_ done.
952          */
953         if (date <= src->item->date)
954                 return SLOP;
955
956         /*
957          * Does the source list still have interesting commits in
958          * it? Definitely not done..
959          */
960         if (!everybody_uninteresting(src, interesting_cache))
961                 return SLOP;
962
963         /* Ok, we're closing in.. */
964         return slop-1;
965 }
966
967 /*
968  * "rev-list --ancestry-path A..B" computes commits that are ancestors
969  * of B but not ancestors of A but further limits the result to those
970  * that are descendants of A.  This takes the list of bottom commits and
971  * the result of "A..B" without --ancestry-path, and limits the latter
972  * further to the ones that can reach one of the commits in "bottom".
973  */
974 static void limit_to_ancestry(struct commit_list *bottom, struct commit_list *list)
975 {
976         struct commit_list *p;
977         struct commit_list *rlist = NULL;
978         int made_progress;
979
980         /*
981          * Reverse the list so that it will be likely that we would
982          * process parents before children.
983          */
984         for (p = list; p; p = p->next)
985                 commit_list_insert(p->item, &rlist);
986
987         for (p = bottom; p; p = p->next)
988                 p->item->object.flags |= TMP_MARK;
989
990         /*
991          * Mark the ones that can reach bottom commits in "list",
992          * in a bottom-up fashion.
993          */
994         do {
995                 made_progress = 0;
996                 for (p = rlist; p; p = p->next) {
997                         struct commit *c = p->item;
998                         struct commit_list *parents;
999                         if (c->object.flags & (TMP_MARK | UNINTERESTING))
1000                                 continue;
1001                         for (parents = c->parents;
1002                              parents;
1003                              parents = parents->next) {
1004                                 if (!(parents->item->object.flags & TMP_MARK))
1005                                         continue;
1006                                 c->object.flags |= TMP_MARK;
1007                                 made_progress = 1;
1008                                 break;
1009                         }
1010                 }
1011         } while (made_progress);
1012
1013         /*
1014          * NEEDSWORK: decide if we want to remove parents that are
1015          * not marked with TMP_MARK from commit->parents for commits
1016          * in the resulting list.  We may not want to do that, though.
1017          */
1018
1019         /*
1020          * The ones that are not marked with TMP_MARK are uninteresting
1021          */
1022         for (p = list; p; p = p->next) {
1023                 struct commit *c = p->item;
1024                 if (c->object.flags & TMP_MARK)
1025                         continue;
1026                 c->object.flags |= UNINTERESTING;
1027         }
1028
1029         /* We are done with the TMP_MARK */
1030         for (p = list; p; p = p->next)
1031                 p->item->object.flags &= ~TMP_MARK;
1032         for (p = bottom; p; p = p->next)
1033                 p->item->object.flags &= ~TMP_MARK;
1034         free_commit_list(rlist);
1035 }
1036
1037 /*
1038  * Before walking the history, keep the set of "negative" refs the
1039  * caller has asked to exclude.
1040  *
1041  * This is used to compute "rev-list --ancestry-path A..B", as we need
1042  * to filter the result of "A..B" further to the ones that can actually
1043  * reach A.
1044  */
1045 static struct commit_list *collect_bottom_commits(struct commit_list *list)
1046 {
1047         struct commit_list *elem, *bottom = NULL;
1048         for (elem = list; elem; elem = elem->next)
1049                 if (elem->item->object.flags & BOTTOM)
1050                         commit_list_insert(elem->item, &bottom);
1051         return bottom;
1052 }
1053
1054 /* Assumes either left_only or right_only is set */
1055 static void limit_left_right(struct commit_list *list, struct rev_info *revs)
1056 {
1057         struct commit_list *p;
1058
1059         for (p = list; p; p = p->next) {
1060                 struct commit *commit = p->item;
1061
1062                 if (revs->right_only) {
1063                         if (commit->object.flags & SYMMETRIC_LEFT)
1064                                 commit->object.flags |= SHOWN;
1065                 } else  /* revs->left_only is set */
1066                         if (!(commit->object.flags & SYMMETRIC_LEFT))
1067                                 commit->object.flags |= SHOWN;
1068         }
1069 }
1070
1071 static int limit_list(struct rev_info *revs)
1072 {
1073         int slop = SLOP;
1074         timestamp_t date = TIME_MAX;
1075         struct commit_list *list = revs->commits;
1076         struct commit_list *newlist = NULL;
1077         struct commit_list **p = &newlist;
1078         struct commit_list *bottom = NULL;
1079         struct commit *interesting_cache = NULL;
1080
1081         if (revs->ancestry_path) {
1082                 bottom = collect_bottom_commits(list);
1083                 if (!bottom)
1084                         die("--ancestry-path given but there are no bottom commits");
1085         }
1086
1087         while (list) {
1088                 struct commit *commit = pop_commit(&list);
1089                 struct object *obj = &commit->object;
1090                 show_early_output_fn_t show;
1091
1092                 if (commit == interesting_cache)
1093                         interesting_cache = NULL;
1094
1095                 if (revs->max_age != -1 && (commit->date < revs->max_age))
1096                         obj->flags |= UNINTERESTING;
1097                 if (process_parents(revs, commit, &list, NULL) < 0)
1098                         return -1;
1099                 if (obj->flags & UNINTERESTING) {
1100                         mark_parents_uninteresting(commit);
1101                         slop = still_interesting(list, date, slop, &interesting_cache);
1102                         if (slop)
1103                                 continue;
1104                         break;
1105                 }
1106                 if (revs->min_age != -1 && (commit->date > revs->min_age))
1107                         continue;
1108                 date = commit->date;
1109                 p = &commit_list_insert(commit, p)->next;
1110
1111                 show = show_early_output;
1112                 if (!show)
1113                         continue;
1114
1115                 show(revs, newlist);
1116                 show_early_output = NULL;
1117         }
1118         if (revs->cherry_pick || revs->cherry_mark)
1119                 cherry_pick_list(newlist, revs);
1120
1121         if (revs->left_only || revs->right_only)
1122                 limit_left_right(newlist, revs);
1123
1124         if (bottom) {
1125                 limit_to_ancestry(bottom, newlist);
1126                 free_commit_list(bottom);
1127         }
1128
1129         /*
1130          * Check if any commits have become TREESAME by some of their parents
1131          * becoming UNINTERESTING.
1132          */
1133         if (limiting_can_increase_treesame(revs))
1134                 for (list = newlist; list; list = list->next) {
1135                         struct commit *c = list->item;
1136                         if (c->object.flags & (UNINTERESTING | TREESAME))
1137                                 continue;
1138                         update_treesame(revs, c);
1139                 }
1140
1141         revs->commits = newlist;
1142         return 0;
1143 }
1144
1145 /*
1146  * Add an entry to refs->cmdline with the specified information.
1147  * *name is copied.
1148  */
1149 static void add_rev_cmdline(struct rev_info *revs,
1150                             struct object *item,
1151                             const char *name,
1152                             int whence,
1153                             unsigned flags)
1154 {
1155         struct rev_cmdline_info *info = &revs->cmdline;
1156         unsigned int nr = info->nr;
1157
1158         ALLOC_GROW(info->rev, nr + 1, info->alloc);
1159         info->rev[nr].item = item;
1160         info->rev[nr].name = xstrdup(name);
1161         info->rev[nr].whence = whence;
1162         info->rev[nr].flags = flags;
1163         info->nr++;
1164 }
1165
1166 static void add_rev_cmdline_list(struct rev_info *revs,
1167                                  struct commit_list *commit_list,
1168                                  int whence,
1169                                  unsigned flags)
1170 {
1171         while (commit_list) {
1172                 struct object *object = &commit_list->item->object;
1173                 add_rev_cmdline(revs, object, oid_to_hex(&object->oid),
1174                                 whence, flags);
1175                 commit_list = commit_list->next;
1176         }
1177 }
1178
1179 struct all_refs_cb {
1180         int all_flags;
1181         int warned_bad_reflog;
1182         struct rev_info *all_revs;
1183         const char *name_for_errormsg;
1184         struct worktree *wt;
1185 };
1186
1187 int ref_excluded(struct string_list *ref_excludes, const char *path)
1188 {
1189         struct string_list_item *item;
1190
1191         if (!ref_excludes)
1192                 return 0;
1193         for_each_string_list_item(item, ref_excludes) {
1194                 if (!wildmatch(item->string, path, 0))
1195                         return 1;
1196         }
1197         return 0;
1198 }
1199
1200 static int handle_one_ref(const char *path, const struct object_id *oid,
1201                           int flag, void *cb_data)
1202 {
1203         struct all_refs_cb *cb = cb_data;
1204         struct object *object;
1205
1206         if (ref_excluded(cb->all_revs->ref_excludes, path))
1207             return 0;
1208
1209         object = get_reference(cb->all_revs, path, oid, cb->all_flags);
1210         add_rev_cmdline(cb->all_revs, object, path, REV_CMD_REF, cb->all_flags);
1211         add_pending_oid(cb->all_revs, path, oid, cb->all_flags);
1212         return 0;
1213 }
1214
1215 static void init_all_refs_cb(struct all_refs_cb *cb, struct rev_info *revs,
1216         unsigned flags)
1217 {
1218         cb->all_revs = revs;
1219         cb->all_flags = flags;
1220         revs->rev_input_given = 1;
1221         cb->wt = NULL;
1222 }
1223
1224 void clear_ref_exclusion(struct string_list **ref_excludes_p)
1225 {
1226         if (*ref_excludes_p) {
1227                 string_list_clear(*ref_excludes_p, 0);
1228                 free(*ref_excludes_p);
1229         }
1230         *ref_excludes_p = NULL;
1231 }
1232
1233 void add_ref_exclusion(struct string_list **ref_excludes_p, const char *exclude)
1234 {
1235         if (!*ref_excludes_p) {
1236                 *ref_excludes_p = xcalloc(1, sizeof(**ref_excludes_p));
1237                 (*ref_excludes_p)->strdup_strings = 1;
1238         }
1239         string_list_append(*ref_excludes_p, exclude);
1240 }
1241
1242 static void handle_refs(struct ref_store *refs,
1243                         struct rev_info *revs, unsigned flags,
1244                         int (*for_each)(struct ref_store *, each_ref_fn, void *))
1245 {
1246         struct all_refs_cb cb;
1247
1248         if (!refs) {
1249                 /* this could happen with uninitialized submodules */
1250                 return;
1251         }
1252
1253         init_all_refs_cb(&cb, revs, flags);
1254         for_each(refs, handle_one_ref, &cb);
1255 }
1256
1257 static void handle_one_reflog_commit(struct object_id *oid, void *cb_data)
1258 {
1259         struct all_refs_cb *cb = cb_data;
1260         if (!is_null_oid(oid)) {
1261                 struct object *o = parse_object(cb->all_revs->repo, oid);
1262                 if (o) {
1263                         o->flags |= cb->all_flags;
1264                         /* ??? CMDLINEFLAGS ??? */
1265                         add_pending_object(cb->all_revs, o, "");
1266                 }
1267                 else if (!cb->warned_bad_reflog) {
1268                         warning("reflog of '%s' references pruned commits",
1269                                 cb->name_for_errormsg);
1270                         cb->warned_bad_reflog = 1;
1271                 }
1272         }
1273 }
1274
1275 static int handle_one_reflog_ent(struct object_id *ooid, struct object_id *noid,
1276                 const char *email, timestamp_t timestamp, int tz,
1277                 const char *message, void *cb_data)
1278 {
1279         handle_one_reflog_commit(ooid, cb_data);
1280         handle_one_reflog_commit(noid, cb_data);
1281         return 0;
1282 }
1283
1284 static int handle_one_reflog(const char *refname_in_wt,
1285                              const struct object_id *oid,
1286                              int flag, void *cb_data)
1287 {
1288         struct all_refs_cb *cb = cb_data;
1289         struct strbuf refname = STRBUF_INIT;
1290
1291         cb->warned_bad_reflog = 0;
1292         strbuf_worktree_ref(cb->wt, &refname, refname_in_wt);
1293         cb->name_for_errormsg = refname.buf;
1294         refs_for_each_reflog_ent(get_main_ref_store(the_repository),
1295                                  refname.buf,
1296                                  handle_one_reflog_ent, cb_data);
1297         strbuf_release(&refname);
1298         return 0;
1299 }
1300
1301 static void add_other_reflogs_to_pending(struct all_refs_cb *cb)
1302 {
1303         struct worktree **worktrees, **p;
1304
1305         worktrees = get_worktrees(0);
1306         for (p = worktrees; *p; p++) {
1307                 struct worktree *wt = *p;
1308
1309                 if (wt->is_current)
1310                         continue;
1311
1312                 cb->wt = wt;
1313                 refs_for_each_reflog(get_worktree_ref_store(wt),
1314                                      handle_one_reflog,
1315                                      cb);
1316         }
1317         free_worktrees(worktrees);
1318 }
1319
1320 void add_reflogs_to_pending(struct rev_info *revs, unsigned flags)
1321 {
1322         struct all_refs_cb cb;
1323
1324         cb.all_revs = revs;
1325         cb.all_flags = flags;
1326         cb.wt = NULL;
1327         for_each_reflog(handle_one_reflog, &cb);
1328
1329         if (!revs->single_worktree)
1330                 add_other_reflogs_to_pending(&cb);
1331 }
1332
1333 static void add_cache_tree(struct cache_tree *it, struct rev_info *revs,
1334                            struct strbuf *path, unsigned int flags)
1335 {
1336         size_t baselen = path->len;
1337         int i;
1338
1339         if (it->entry_count >= 0) {
1340                 struct tree *tree = lookup_tree(revs->repo, &it->oid);
1341                 tree->object.flags |= flags;
1342                 add_pending_object_with_path(revs, &tree->object, "",
1343                                              040000, path->buf);
1344         }
1345
1346         for (i = 0; i < it->subtree_nr; i++) {
1347                 struct cache_tree_sub *sub = it->down[i];
1348                 strbuf_addf(path, "%s%s", baselen ? "/" : "", sub->name);
1349                 add_cache_tree(sub->cache_tree, revs, path, flags);
1350                 strbuf_setlen(path, baselen);
1351         }
1352
1353 }
1354
1355 static void do_add_index_objects_to_pending(struct rev_info *revs,
1356                                             struct index_state *istate,
1357                                             unsigned int flags)
1358 {
1359         int i;
1360
1361         for (i = 0; i < istate->cache_nr; i++) {
1362                 struct cache_entry *ce = istate->cache[i];
1363                 struct blob *blob;
1364
1365                 if (S_ISGITLINK(ce->ce_mode))
1366                         continue;
1367
1368                 blob = lookup_blob(revs->repo, &ce->oid);
1369                 if (!blob)
1370                         die("unable to add index blob to traversal");
1371                 blob->object.flags |= flags;
1372                 add_pending_object_with_path(revs, &blob->object, "",
1373                                              ce->ce_mode, ce->name);
1374         }
1375
1376         if (istate->cache_tree) {
1377                 struct strbuf path = STRBUF_INIT;
1378                 add_cache_tree(istate->cache_tree, revs, &path, flags);
1379                 strbuf_release(&path);
1380         }
1381 }
1382
1383 void add_index_objects_to_pending(struct rev_info *revs, unsigned int flags)
1384 {
1385         struct worktree **worktrees, **p;
1386
1387         read_index(revs->repo->index);
1388         do_add_index_objects_to_pending(revs, revs->repo->index, flags);
1389
1390         if (revs->single_worktree)
1391                 return;
1392
1393         worktrees = get_worktrees(0);
1394         for (p = worktrees; *p; p++) {
1395                 struct worktree *wt = *p;
1396                 struct index_state istate = { NULL };
1397
1398                 if (wt->is_current)
1399                         continue; /* current index already taken care of */
1400
1401                 if (read_index_from(&istate,
1402                                     worktree_git_path(wt, "index"),
1403                                     get_worktree_git_dir(wt)) > 0)
1404                         do_add_index_objects_to_pending(revs, &istate, flags);
1405                 discard_index(&istate);
1406         }
1407         free_worktrees(worktrees);
1408 }
1409
1410 static int add_parents_only(struct rev_info *revs, const char *arg_, int flags,
1411                             int exclude_parent)
1412 {
1413         struct object_id oid;
1414         struct object *it;
1415         struct commit *commit;
1416         struct commit_list *parents;
1417         int parent_number;
1418         const char *arg = arg_;
1419
1420         if (*arg == '^') {
1421                 flags ^= UNINTERESTING | BOTTOM;
1422                 arg++;
1423         }
1424         if (get_oid_committish(arg, &oid))
1425                 return 0;
1426         while (1) {
1427                 it = get_reference(revs, arg, &oid, 0);
1428                 if (!it && revs->ignore_missing)
1429                         return 0;
1430                 if (it->type != OBJ_TAG)
1431                         break;
1432                 if (!((struct tag*)it)->tagged)
1433                         return 0;
1434                 oidcpy(&oid, &((struct tag*)it)->tagged->oid);
1435         }
1436         if (it->type != OBJ_COMMIT)
1437                 return 0;
1438         commit = (struct commit *)it;
1439         if (exclude_parent &&
1440             exclude_parent > commit_list_count(commit->parents))
1441                 return 0;
1442         for (parents = commit->parents, parent_number = 1;
1443              parents;
1444              parents = parents->next, parent_number++) {
1445                 if (exclude_parent && parent_number != exclude_parent)
1446                         continue;
1447
1448                 it = &parents->item->object;
1449                 it->flags |= flags;
1450                 add_rev_cmdline(revs, it, arg_, REV_CMD_PARENTS_ONLY, flags);
1451                 add_pending_object(revs, it, arg);
1452         }
1453         return 1;
1454 }
1455
1456 void repo_init_revisions(struct repository *r,
1457                          struct rev_info *revs,
1458                          const char *prefix)
1459 {
1460         memset(revs, 0, sizeof(*revs));
1461
1462         revs->repo = r;
1463         revs->abbrev = DEFAULT_ABBREV;
1464         revs->ignore_merges = 1;
1465         revs->simplify_history = 1;
1466         revs->pruning.flags.recursive = 1;
1467         revs->pruning.flags.quick = 1;
1468         revs->pruning.add_remove = file_add_remove;
1469         revs->pruning.change = file_change;
1470         revs->pruning.change_fn_data = revs;
1471         revs->sort_order = REV_SORT_IN_GRAPH_ORDER;
1472         revs->dense = 1;
1473         revs->prefix = prefix;
1474         revs->max_age = -1;
1475         revs->min_age = -1;
1476         revs->skip_count = -1;
1477         revs->max_count = -1;
1478         revs->max_parents = -1;
1479         revs->expand_tabs_in_log = -1;
1480
1481         revs->commit_format = CMIT_FMT_DEFAULT;
1482         revs->expand_tabs_in_log_default = 8;
1483
1484         init_grep_defaults(revs->repo);
1485         grep_init(&revs->grep_filter, revs->repo, prefix);
1486         revs->grep_filter.status_only = 1;
1487
1488         repo_diff_setup(revs->repo, &revs->diffopt);
1489         if (prefix && !revs->diffopt.prefix) {
1490                 revs->diffopt.prefix = prefix;
1491                 revs->diffopt.prefix_length = strlen(prefix);
1492         }
1493
1494         revs->notes_opt.use_default_notes = -1;
1495 }
1496
1497 static void add_pending_commit_list(struct rev_info *revs,
1498                                     struct commit_list *commit_list,
1499                                     unsigned int flags)
1500 {
1501         while (commit_list) {
1502                 struct object *object = &commit_list->item->object;
1503                 object->flags |= flags;
1504                 add_pending_object(revs, object, oid_to_hex(&object->oid));
1505                 commit_list = commit_list->next;
1506         }
1507 }
1508
1509 static void prepare_show_merge(struct rev_info *revs)
1510 {
1511         struct commit_list *bases;
1512         struct commit *head, *other;
1513         struct object_id oid;
1514         const char **prune = NULL;
1515         int i, prune_num = 1; /* counting terminating NULL */
1516         struct index_state *istate = revs->repo->index;
1517
1518         if (get_oid("HEAD", &oid))
1519                 die("--merge without HEAD?");
1520         head = lookup_commit_or_die(&oid, "HEAD");
1521         if (get_oid("MERGE_HEAD", &oid))
1522                 die("--merge without MERGE_HEAD?");
1523         other = lookup_commit_or_die(&oid, "MERGE_HEAD");
1524         add_pending_object(revs, &head->object, "HEAD");
1525         add_pending_object(revs, &other->object, "MERGE_HEAD");
1526         bases = get_merge_bases(head, other);
1527         add_rev_cmdline_list(revs, bases, REV_CMD_MERGE_BASE, UNINTERESTING | BOTTOM);
1528         add_pending_commit_list(revs, bases, UNINTERESTING | BOTTOM);
1529         free_commit_list(bases);
1530         head->object.flags |= SYMMETRIC_LEFT;
1531
1532         if (!istate->cache_nr)
1533                 read_index(istate);
1534         for (i = 0; i < istate->cache_nr; i++) {
1535                 const struct cache_entry *ce = istate->cache[i];
1536                 if (!ce_stage(ce))
1537                         continue;
1538                 if (ce_path_match(istate, ce, &revs->prune_data, NULL)) {
1539                         prune_num++;
1540                         REALLOC_ARRAY(prune, prune_num);
1541                         prune[prune_num-2] = ce->name;
1542                         prune[prune_num-1] = NULL;
1543                 }
1544                 while ((i+1 < istate->cache_nr) &&
1545                        ce_same_name(ce, istate->cache[i+1]))
1546                         i++;
1547         }
1548         clear_pathspec(&revs->prune_data);
1549         parse_pathspec(&revs->prune_data, PATHSPEC_ALL_MAGIC & ~PATHSPEC_LITERAL,
1550                        PATHSPEC_PREFER_FULL | PATHSPEC_LITERAL_PATH, "", prune);
1551         revs->limited = 1;
1552 }
1553
1554 static int dotdot_missing(const char *arg, char *dotdot,
1555                           struct rev_info *revs, int symmetric)
1556 {
1557         if (revs->ignore_missing)
1558                 return 0;
1559         /* de-munge so we report the full argument */
1560         *dotdot = '.';
1561         die(symmetric
1562             ? "Invalid symmetric difference expression %s"
1563             : "Invalid revision range %s", arg);
1564 }
1565
1566 static int handle_dotdot_1(const char *arg, char *dotdot,
1567                            struct rev_info *revs, int flags,
1568                            int cant_be_filename,
1569                            struct object_context *a_oc,
1570                            struct object_context *b_oc)
1571 {
1572         const char *a_name, *b_name;
1573         struct object_id a_oid, b_oid;
1574         struct object *a_obj, *b_obj;
1575         unsigned int a_flags, b_flags;
1576         int symmetric = 0;
1577         unsigned int flags_exclude = flags ^ (UNINTERESTING | BOTTOM);
1578         unsigned int oc_flags = GET_OID_COMMITTISH | GET_OID_RECORD_PATH;
1579
1580         a_name = arg;
1581         if (!*a_name)
1582                 a_name = "HEAD";
1583
1584         b_name = dotdot + 2;
1585         if (*b_name == '.') {
1586                 symmetric = 1;
1587                 b_name++;
1588         }
1589         if (!*b_name)
1590                 b_name = "HEAD";
1591
1592         if (get_oid_with_context(a_name, oc_flags, &a_oid, a_oc) ||
1593             get_oid_with_context(b_name, oc_flags, &b_oid, b_oc))
1594                 return -1;
1595
1596         if (!cant_be_filename) {
1597                 *dotdot = '.';
1598                 verify_non_filename(revs->prefix, arg);
1599                 *dotdot = '\0';
1600         }
1601
1602         a_obj = parse_object(revs->repo, &a_oid);
1603         b_obj = parse_object(revs->repo, &b_oid);
1604         if (!a_obj || !b_obj)
1605                 return dotdot_missing(arg, dotdot, revs, symmetric);
1606
1607         if (!symmetric) {
1608                 /* just A..B */
1609                 b_flags = flags;
1610                 a_flags = flags_exclude;
1611         } else {
1612                 /* A...B -- find merge bases between the two */
1613                 struct commit *a, *b;
1614                 struct commit_list *exclude;
1615
1616                 a = lookup_commit_reference(revs->repo, &a_obj->oid);
1617                 b = lookup_commit_reference(revs->repo, &b_obj->oid);
1618                 if (!a || !b)
1619                         return dotdot_missing(arg, dotdot, revs, symmetric);
1620
1621                 exclude = get_merge_bases(a, b);
1622                 add_rev_cmdline_list(revs, exclude, REV_CMD_MERGE_BASE,
1623                                      flags_exclude);
1624                 add_pending_commit_list(revs, exclude, flags_exclude);
1625                 free_commit_list(exclude);
1626
1627                 b_flags = flags;
1628                 a_flags = flags | SYMMETRIC_LEFT;
1629         }
1630
1631         a_obj->flags |= a_flags;
1632         b_obj->flags |= b_flags;
1633         add_rev_cmdline(revs, a_obj, a_name, REV_CMD_LEFT, a_flags);
1634         add_rev_cmdline(revs, b_obj, b_name, REV_CMD_RIGHT, b_flags);
1635         add_pending_object_with_path(revs, a_obj, a_name, a_oc->mode, a_oc->path);
1636         add_pending_object_with_path(revs, b_obj, b_name, b_oc->mode, b_oc->path);
1637         return 0;
1638 }
1639
1640 static int handle_dotdot(const char *arg,
1641                          struct rev_info *revs, int flags,
1642                          int cant_be_filename)
1643 {
1644         struct object_context a_oc, b_oc;
1645         char *dotdot = strstr(arg, "..");
1646         int ret;
1647
1648         if (!dotdot)
1649                 return -1;
1650
1651         memset(&a_oc, 0, sizeof(a_oc));
1652         memset(&b_oc, 0, sizeof(b_oc));
1653
1654         *dotdot = '\0';
1655         ret = handle_dotdot_1(arg, dotdot, revs, flags, cant_be_filename,
1656                               &a_oc, &b_oc);
1657         *dotdot = '.';
1658
1659         free(a_oc.path);
1660         free(b_oc.path);
1661
1662         return ret;
1663 }
1664
1665 int handle_revision_arg(const char *arg_, struct rev_info *revs, int flags, unsigned revarg_opt)
1666 {
1667         struct object_context oc;
1668         char *mark;
1669         struct object *object;
1670         struct object_id oid;
1671         int local_flags;
1672         const char *arg = arg_;
1673         int cant_be_filename = revarg_opt & REVARG_CANNOT_BE_FILENAME;
1674         unsigned get_sha1_flags = GET_OID_RECORD_PATH;
1675
1676         flags = flags & UNINTERESTING ? flags | BOTTOM : flags & ~BOTTOM;
1677
1678         if (!cant_be_filename && !strcmp(arg, "..")) {
1679                 /*
1680                  * Just ".."?  That is not a range but the
1681                  * pathspec for the parent directory.
1682                  */
1683                 return -1;
1684         }
1685
1686         if (!handle_dotdot(arg, revs, flags, revarg_opt))
1687                 return 0;
1688
1689         mark = strstr(arg, "^@");
1690         if (mark && !mark[2]) {
1691                 *mark = 0;
1692                 if (add_parents_only(revs, arg, flags, 0))
1693                         return 0;
1694                 *mark = '^';
1695         }
1696         mark = strstr(arg, "^!");
1697         if (mark && !mark[2]) {
1698                 *mark = 0;
1699                 if (!add_parents_only(revs, arg, flags ^ (UNINTERESTING | BOTTOM), 0))
1700                         *mark = '^';
1701         }
1702         mark = strstr(arg, "^-");
1703         if (mark) {
1704                 int exclude_parent = 1;
1705
1706                 if (mark[2]) {
1707                         char *end;
1708                         exclude_parent = strtoul(mark + 2, &end, 10);
1709                         if (*end != '\0' || !exclude_parent)
1710                                 return -1;
1711                 }
1712
1713                 *mark = 0;
1714                 if (!add_parents_only(revs, arg, flags ^ (UNINTERESTING | BOTTOM), exclude_parent))
1715                         *mark = '^';
1716         }
1717
1718         local_flags = 0;
1719         if (*arg == '^') {
1720                 local_flags = UNINTERESTING | BOTTOM;
1721                 arg++;
1722         }
1723
1724         if (revarg_opt & REVARG_COMMITTISH)
1725                 get_sha1_flags |= GET_OID_COMMITTISH;
1726
1727         if (get_oid_with_context(arg, get_sha1_flags, &oid, &oc))
1728                 return revs->ignore_missing ? 0 : -1;
1729         if (!cant_be_filename)
1730                 verify_non_filename(revs->prefix, arg);
1731         object = get_reference(revs, arg, &oid, flags ^ local_flags);
1732         add_rev_cmdline(revs, object, arg_, REV_CMD_REV, flags ^ local_flags);
1733         add_pending_object_with_path(revs, object, arg, oc.mode, oc.path);
1734         free(oc.path);
1735         return 0;
1736 }
1737
1738 static void read_pathspec_from_stdin(struct rev_info *revs, struct strbuf *sb,
1739                                      struct argv_array *prune)
1740 {
1741         while (strbuf_getline(sb, stdin) != EOF)
1742                 argv_array_push(prune, sb->buf);
1743 }
1744
1745 static void read_revisions_from_stdin(struct rev_info *revs,
1746                                       struct argv_array *prune)
1747 {
1748         struct strbuf sb;
1749         int seen_dashdash = 0;
1750         int save_warning;
1751
1752         save_warning = warn_on_object_refname_ambiguity;
1753         warn_on_object_refname_ambiguity = 0;
1754
1755         strbuf_init(&sb, 1000);
1756         while (strbuf_getline(&sb, stdin) != EOF) {
1757                 int len = sb.len;
1758                 if (!len)
1759                         break;
1760                 if (sb.buf[0] == '-') {
1761                         if (len == 2 && sb.buf[1] == '-') {
1762                                 seen_dashdash = 1;
1763                                 break;
1764                         }
1765                         die("options not supported in --stdin mode");
1766                 }
1767                 if (handle_revision_arg(sb.buf, revs, 0,
1768                                         REVARG_CANNOT_BE_FILENAME))
1769                         die("bad revision '%s'", sb.buf);
1770         }
1771         if (seen_dashdash)
1772                 read_pathspec_from_stdin(revs, &sb, prune);
1773
1774         strbuf_release(&sb);
1775         warn_on_object_refname_ambiguity = save_warning;
1776 }
1777
1778 static void add_grep(struct rev_info *revs, const char *ptn, enum grep_pat_token what)
1779 {
1780         append_grep_pattern(&revs->grep_filter, ptn, "command line", 0, what);
1781 }
1782
1783 static void add_header_grep(struct rev_info *revs, enum grep_header_field field, const char *pattern)
1784 {
1785         append_header_grep_pattern(&revs->grep_filter, field, pattern);
1786 }
1787
1788 static void add_message_grep(struct rev_info *revs, const char *pattern)
1789 {
1790         add_grep(revs, pattern, GREP_PATTERN_BODY);
1791 }
1792
1793 static int handle_revision_opt(struct rev_info *revs, int argc, const char **argv,
1794                                int *unkc, const char **unkv)
1795 {
1796         const char *arg = argv[0];
1797         const char *optarg;
1798         int argcount;
1799         const unsigned hexsz = the_hash_algo->hexsz;
1800
1801         /* pseudo revision arguments */
1802         if (!strcmp(arg, "--all") || !strcmp(arg, "--branches") ||
1803             !strcmp(arg, "--tags") || !strcmp(arg, "--remotes") ||
1804             !strcmp(arg, "--reflog") || !strcmp(arg, "--not") ||
1805             !strcmp(arg, "--no-walk") || !strcmp(arg, "--do-walk") ||
1806             !strcmp(arg, "--bisect") || starts_with(arg, "--glob=") ||
1807             !strcmp(arg, "--indexed-objects") ||
1808             starts_with(arg, "--exclude=") ||
1809             starts_with(arg, "--branches=") || starts_with(arg, "--tags=") ||
1810             starts_with(arg, "--remotes=") || starts_with(arg, "--no-walk="))
1811         {
1812                 unkv[(*unkc)++] = arg;
1813                 return 1;
1814         }
1815
1816         if ((argcount = parse_long_opt("max-count", argv, &optarg))) {
1817                 revs->max_count = atoi(optarg);
1818                 revs->no_walk = 0;
1819                 return argcount;
1820         } else if ((argcount = parse_long_opt("skip", argv, &optarg))) {
1821                 revs->skip_count = atoi(optarg);
1822                 return argcount;
1823         } else if ((*arg == '-') && isdigit(arg[1])) {
1824                 /* accept -<digit>, like traditional "head" */
1825                 if (strtol_i(arg + 1, 10, &revs->max_count) < 0 ||
1826                     revs->max_count < 0)
1827                         die("'%s': not a non-negative integer", arg + 1);
1828                 revs->no_walk = 0;
1829         } else if (!strcmp(arg, "-n")) {
1830                 if (argc <= 1)
1831                         return error("-n requires an argument");
1832                 revs->max_count = atoi(argv[1]);
1833                 revs->no_walk = 0;
1834                 return 2;
1835         } else if (skip_prefix(arg, "-n", &optarg)) {
1836                 revs->max_count = atoi(optarg);
1837                 revs->no_walk = 0;
1838         } else if ((argcount = parse_long_opt("max-age", argv, &optarg))) {
1839                 revs->max_age = atoi(optarg);
1840                 return argcount;
1841         } else if ((argcount = parse_long_opt("since", argv, &optarg))) {
1842                 revs->max_age = approxidate(optarg);
1843                 return argcount;
1844         } else if ((argcount = parse_long_opt("after", argv, &optarg))) {
1845                 revs->max_age = approxidate(optarg);
1846                 return argcount;
1847         } else if ((argcount = parse_long_opt("min-age", argv, &optarg))) {
1848                 revs->min_age = atoi(optarg);
1849                 return argcount;
1850         } else if ((argcount = parse_long_opt("before", argv, &optarg))) {
1851                 revs->min_age = approxidate(optarg);
1852                 return argcount;
1853         } else if ((argcount = parse_long_opt("until", argv, &optarg))) {
1854                 revs->min_age = approxidate(optarg);
1855                 return argcount;
1856         } else if (!strcmp(arg, "--first-parent")) {
1857                 revs->first_parent_only = 1;
1858         } else if (!strcmp(arg, "--ancestry-path")) {
1859                 revs->ancestry_path = 1;
1860                 revs->simplify_history = 0;
1861                 revs->limited = 1;
1862         } else if (!strcmp(arg, "-g") || !strcmp(arg, "--walk-reflogs")) {
1863                 init_reflog_walk(&revs->reflog_info);
1864         } else if (!strcmp(arg, "--default")) {
1865                 if (argc <= 1)
1866                         return error("bad --default argument");
1867                 revs->def = argv[1];
1868                 return 2;
1869         } else if (!strcmp(arg, "--merge")) {
1870                 revs->show_merge = 1;
1871         } else if (!strcmp(arg, "--topo-order")) {
1872                 revs->sort_order = REV_SORT_IN_GRAPH_ORDER;
1873                 revs->topo_order = 1;
1874         } else if (!strcmp(arg, "--simplify-merges")) {
1875                 revs->simplify_merges = 1;
1876                 revs->topo_order = 1;
1877                 revs->rewrite_parents = 1;
1878                 revs->simplify_history = 0;
1879                 revs->limited = 1;
1880         } else if (!strcmp(arg, "--simplify-by-decoration")) {
1881                 revs->simplify_merges = 1;
1882                 revs->topo_order = 1;
1883                 revs->rewrite_parents = 1;
1884                 revs->simplify_history = 0;
1885                 revs->simplify_by_decoration = 1;
1886                 revs->limited = 1;
1887                 revs->prune = 1;
1888                 load_ref_decorations(NULL, DECORATE_SHORT_REFS);
1889         } else if (!strcmp(arg, "--date-order")) {
1890                 revs->sort_order = REV_SORT_BY_COMMIT_DATE;
1891                 revs->topo_order = 1;
1892         } else if (!strcmp(arg, "--author-date-order")) {
1893                 revs->sort_order = REV_SORT_BY_AUTHOR_DATE;
1894                 revs->topo_order = 1;
1895         } else if (!strcmp(arg, "--early-output")) {
1896                 revs->early_output = 100;
1897                 revs->topo_order = 1;
1898         } else if (skip_prefix(arg, "--early-output=", &optarg)) {
1899                 if (strtoul_ui(optarg, 10, &revs->early_output) < 0)
1900                         die("'%s': not a non-negative integer", optarg);
1901                 revs->topo_order = 1;
1902         } else if (!strcmp(arg, "--parents")) {
1903                 revs->rewrite_parents = 1;
1904                 revs->print_parents = 1;
1905         } else if (!strcmp(arg, "--dense")) {
1906                 revs->dense = 1;
1907         } else if (!strcmp(arg, "--sparse")) {
1908                 revs->dense = 0;
1909         } else if (!strcmp(arg, "--in-commit-order")) {
1910                 revs->tree_blobs_in_commit_order = 1;
1911         } else if (!strcmp(arg, "--remove-empty")) {
1912                 revs->remove_empty_trees = 1;
1913         } else if (!strcmp(arg, "--merges")) {
1914                 revs->min_parents = 2;
1915         } else if (!strcmp(arg, "--no-merges")) {
1916                 revs->max_parents = 1;
1917         } else if (skip_prefix(arg, "--min-parents=", &optarg)) {
1918                 revs->min_parents = atoi(optarg);
1919         } else if (!strcmp(arg, "--no-min-parents")) {
1920                 revs->min_parents = 0;
1921         } else if (skip_prefix(arg, "--max-parents=", &optarg)) {
1922                 revs->max_parents = atoi(optarg);
1923         } else if (!strcmp(arg, "--no-max-parents")) {
1924                 revs->max_parents = -1;
1925         } else if (!strcmp(arg, "--boundary")) {
1926                 revs->boundary = 1;
1927         } else if (!strcmp(arg, "--left-right")) {
1928                 revs->left_right = 1;
1929         } else if (!strcmp(arg, "--left-only")) {
1930                 if (revs->right_only)
1931                         die("--left-only is incompatible with --right-only"
1932                             " or --cherry");
1933                 revs->left_only = 1;
1934         } else if (!strcmp(arg, "--right-only")) {
1935                 if (revs->left_only)
1936                         die("--right-only is incompatible with --left-only");
1937                 revs->right_only = 1;
1938         } else if (!strcmp(arg, "--cherry")) {
1939                 if (revs->left_only)
1940                         die("--cherry is incompatible with --left-only");
1941                 revs->cherry_mark = 1;
1942                 revs->right_only = 1;
1943                 revs->max_parents = 1;
1944                 revs->limited = 1;
1945         } else if (!strcmp(arg, "--count")) {
1946                 revs->count = 1;
1947         } else if (!strcmp(arg, "--cherry-mark")) {
1948                 if (revs->cherry_pick)
1949                         die("--cherry-mark is incompatible with --cherry-pick");
1950                 revs->cherry_mark = 1;
1951                 revs->limited = 1; /* needs limit_list() */
1952         } else if (!strcmp(arg, "--cherry-pick")) {
1953                 if (revs->cherry_mark)
1954                         die("--cherry-pick is incompatible with --cherry-mark");
1955                 revs->cherry_pick = 1;
1956                 revs->limited = 1;
1957         } else if (!strcmp(arg, "--objects")) {
1958                 revs->tag_objects = 1;
1959                 revs->tree_objects = 1;
1960                 revs->blob_objects = 1;
1961         } else if (!strcmp(arg, "--objects-edge")) {
1962                 revs->tag_objects = 1;
1963                 revs->tree_objects = 1;
1964                 revs->blob_objects = 1;
1965                 revs->edge_hint = 1;
1966         } else if (!strcmp(arg, "--objects-edge-aggressive")) {
1967                 revs->tag_objects = 1;
1968                 revs->tree_objects = 1;
1969                 revs->blob_objects = 1;
1970                 revs->edge_hint = 1;
1971                 revs->edge_hint_aggressive = 1;
1972         } else if (!strcmp(arg, "--verify-objects")) {
1973                 revs->tag_objects = 1;
1974                 revs->tree_objects = 1;
1975                 revs->blob_objects = 1;
1976                 revs->verify_objects = 1;
1977         } else if (!strcmp(arg, "--unpacked")) {
1978                 revs->unpacked = 1;
1979         } else if (starts_with(arg, "--unpacked=")) {
1980                 die("--unpacked=<packfile> no longer supported.");
1981         } else if (!strcmp(arg, "-r")) {
1982                 revs->diff = 1;
1983                 revs->diffopt.flags.recursive = 1;
1984         } else if (!strcmp(arg, "-t")) {
1985                 revs->diff = 1;
1986                 revs->diffopt.flags.recursive = 1;
1987                 revs->diffopt.flags.tree_in_recursive = 1;
1988         } else if (!strcmp(arg, "-m")) {
1989                 revs->ignore_merges = 0;
1990         } else if (!strcmp(arg, "-c")) {
1991                 revs->diff = 1;
1992                 revs->dense_combined_merges = 0;
1993                 revs->combine_merges = 1;
1994         } else if (!strcmp(arg, "--cc")) {
1995                 revs->diff = 1;
1996                 revs->dense_combined_merges = 1;
1997                 revs->combine_merges = 1;
1998         } else if (!strcmp(arg, "-v")) {
1999                 revs->verbose_header = 1;
2000         } else if (!strcmp(arg, "--pretty")) {
2001                 revs->verbose_header = 1;
2002                 revs->pretty_given = 1;
2003                 get_commit_format(NULL, revs);
2004         } else if (skip_prefix(arg, "--pretty=", &optarg) ||
2005                    skip_prefix(arg, "--format=", &optarg)) {
2006                 /*
2007                  * Detached form ("--pretty X" as opposed to "--pretty=X")
2008                  * not allowed, since the argument is optional.
2009                  */
2010                 revs->verbose_header = 1;
2011                 revs->pretty_given = 1;
2012                 get_commit_format(optarg, revs);
2013         } else if (!strcmp(arg, "--expand-tabs")) {
2014                 revs->expand_tabs_in_log = 8;
2015         } else if (!strcmp(arg, "--no-expand-tabs")) {
2016                 revs->expand_tabs_in_log = 0;
2017         } else if (skip_prefix(arg, "--expand-tabs=", &arg)) {
2018                 int val;
2019                 if (strtol_i(arg, 10, &val) < 0 || val < 0)
2020                         die("'%s': not a non-negative integer", arg);
2021                 revs->expand_tabs_in_log = val;
2022         } else if (!strcmp(arg, "--show-notes") || !strcmp(arg, "--notes")) {
2023                 revs->show_notes = 1;
2024                 revs->show_notes_given = 1;
2025                 revs->notes_opt.use_default_notes = 1;
2026         } else if (!strcmp(arg, "--show-signature")) {
2027                 revs->show_signature = 1;
2028         } else if (!strcmp(arg, "--no-show-signature")) {
2029                 revs->show_signature = 0;
2030         } else if (!strcmp(arg, "--show-linear-break")) {
2031                 revs->break_bar = "                    ..........";
2032                 revs->track_linear = 1;
2033                 revs->track_first_time = 1;
2034         } else if (skip_prefix(arg, "--show-linear-break=", &optarg)) {
2035                 revs->break_bar = xstrdup(optarg);
2036                 revs->track_linear = 1;
2037                 revs->track_first_time = 1;
2038         } else if (skip_prefix(arg, "--show-notes=", &optarg) ||
2039                    skip_prefix(arg, "--notes=", &optarg)) {
2040                 struct strbuf buf = STRBUF_INIT;
2041                 revs->show_notes = 1;
2042                 revs->show_notes_given = 1;
2043                 if (starts_with(arg, "--show-notes=") &&
2044                     revs->notes_opt.use_default_notes < 0)
2045                         revs->notes_opt.use_default_notes = 1;
2046                 strbuf_addstr(&buf, optarg);
2047                 expand_notes_ref(&buf);
2048                 string_list_append(&revs->notes_opt.extra_notes_refs,
2049                                    strbuf_detach(&buf, NULL));
2050         } else if (!strcmp(arg, "--no-notes")) {
2051                 revs->show_notes = 0;
2052                 revs->show_notes_given = 1;
2053                 revs->notes_opt.use_default_notes = -1;
2054                 /* we have been strdup'ing ourselves, so trick
2055                  * string_list into free()ing strings */
2056                 revs->notes_opt.extra_notes_refs.strdup_strings = 1;
2057                 string_list_clear(&revs->notes_opt.extra_notes_refs, 0);
2058                 revs->notes_opt.extra_notes_refs.strdup_strings = 0;
2059         } else if (!strcmp(arg, "--standard-notes")) {
2060                 revs->show_notes_given = 1;
2061                 revs->notes_opt.use_default_notes = 1;
2062         } else if (!strcmp(arg, "--no-standard-notes")) {
2063                 revs->notes_opt.use_default_notes = 0;
2064         } else if (!strcmp(arg, "--oneline")) {
2065                 revs->verbose_header = 1;
2066                 get_commit_format("oneline", revs);
2067                 revs->pretty_given = 1;
2068                 revs->abbrev_commit = 1;
2069         } else if (!strcmp(arg, "--graph")) {
2070                 revs->topo_order = 1;
2071                 revs->rewrite_parents = 1;
2072                 revs->graph = graph_init(revs);
2073         } else if (!strcmp(arg, "--root")) {
2074                 revs->show_root_diff = 1;
2075         } else if (!strcmp(arg, "--no-commit-id")) {
2076                 revs->no_commit_id = 1;
2077         } else if (!strcmp(arg, "--always")) {
2078                 revs->always_show_header = 1;
2079         } else if (!strcmp(arg, "--no-abbrev")) {
2080                 revs->abbrev = 0;
2081         } else if (!strcmp(arg, "--abbrev")) {
2082                 revs->abbrev = DEFAULT_ABBREV;
2083         } else if (skip_prefix(arg, "--abbrev=", &optarg)) {
2084                 revs->abbrev = strtoul(optarg, NULL, 10);
2085                 if (revs->abbrev < MINIMUM_ABBREV)
2086                         revs->abbrev = MINIMUM_ABBREV;
2087                 else if (revs->abbrev > hexsz)
2088                         revs->abbrev = hexsz;
2089         } else if (!strcmp(arg, "--abbrev-commit")) {
2090                 revs->abbrev_commit = 1;
2091                 revs->abbrev_commit_given = 1;
2092         } else if (!strcmp(arg, "--no-abbrev-commit")) {
2093                 revs->abbrev_commit = 0;
2094         } else if (!strcmp(arg, "--full-diff")) {
2095                 revs->diff = 1;
2096                 revs->full_diff = 1;
2097         } else if (!strcmp(arg, "--full-history")) {
2098                 revs->simplify_history = 0;
2099         } else if (!strcmp(arg, "--relative-date")) {
2100                 revs->date_mode.type = DATE_RELATIVE;
2101                 revs->date_mode_explicit = 1;
2102         } else if ((argcount = parse_long_opt("date", argv, &optarg))) {
2103                 parse_date_format(optarg, &revs->date_mode);
2104                 revs->date_mode_explicit = 1;
2105                 return argcount;
2106         } else if (!strcmp(arg, "--log-size")) {
2107                 revs->show_log_size = 1;
2108         }
2109         /*
2110          * Grepping the commit log
2111          */
2112         else if ((argcount = parse_long_opt("author", argv, &optarg))) {
2113                 add_header_grep(revs, GREP_HEADER_AUTHOR, optarg);
2114                 return argcount;
2115         } else if ((argcount = parse_long_opt("committer", argv, &optarg))) {
2116                 add_header_grep(revs, GREP_HEADER_COMMITTER, optarg);
2117                 return argcount;
2118         } else if ((argcount = parse_long_opt("grep-reflog", argv, &optarg))) {
2119                 add_header_grep(revs, GREP_HEADER_REFLOG, optarg);
2120                 return argcount;
2121         } else if ((argcount = parse_long_opt("grep", argv, &optarg))) {
2122                 add_message_grep(revs, optarg);
2123                 return argcount;
2124         } else if (!strcmp(arg, "--grep-debug")) {
2125                 revs->grep_filter.debug = 1;
2126         } else if (!strcmp(arg, "--basic-regexp")) {
2127                 revs->grep_filter.pattern_type_option = GREP_PATTERN_TYPE_BRE;
2128         } else if (!strcmp(arg, "--extended-regexp") || !strcmp(arg, "-E")) {
2129                 revs->grep_filter.pattern_type_option = GREP_PATTERN_TYPE_ERE;
2130         } else if (!strcmp(arg, "--regexp-ignore-case") || !strcmp(arg, "-i")) {
2131                 revs->grep_filter.ignore_case = 1;
2132                 revs->diffopt.pickaxe_opts |= DIFF_PICKAXE_IGNORE_CASE;
2133         } else if (!strcmp(arg, "--fixed-strings") || !strcmp(arg, "-F")) {
2134                 revs->grep_filter.pattern_type_option = GREP_PATTERN_TYPE_FIXED;
2135         } else if (!strcmp(arg, "--perl-regexp") || !strcmp(arg, "-P")) {
2136                 revs->grep_filter.pattern_type_option = GREP_PATTERN_TYPE_PCRE;
2137         } else if (!strcmp(arg, "--all-match")) {
2138                 revs->grep_filter.all_match = 1;
2139         } else if (!strcmp(arg, "--invert-grep")) {
2140                 revs->invert_grep = 1;
2141         } else if ((argcount = parse_long_opt("encoding", argv, &optarg))) {
2142                 if (strcmp(optarg, "none"))
2143                         git_log_output_encoding = xstrdup(optarg);
2144                 else
2145                         git_log_output_encoding = "";
2146                 return argcount;
2147         } else if (!strcmp(arg, "--reverse")) {
2148                 revs->reverse ^= 1;
2149         } else if (!strcmp(arg, "--children")) {
2150                 revs->children.name = "children";
2151                 revs->limited = 1;
2152         } else if (!strcmp(arg, "--ignore-missing")) {
2153                 revs->ignore_missing = 1;
2154         } else if (revs->allow_exclude_promisor_objects_opt &&
2155                    !strcmp(arg, "--exclude-promisor-objects")) {
2156                 if (fetch_if_missing)
2157                         BUG("exclude_promisor_objects can only be used when fetch_if_missing is 0");
2158                 revs->exclude_promisor_objects = 1;
2159         } else {
2160                 int opts = diff_opt_parse(&revs->diffopt, argv, argc, revs->prefix);
2161                 if (!opts)
2162                         unkv[(*unkc)++] = arg;
2163                 return opts;
2164         }
2165         if (revs->graph && revs->track_linear)
2166                 die("--show-linear-break and --graph are incompatible");
2167
2168         return 1;
2169 }
2170
2171 void parse_revision_opt(struct rev_info *revs, struct parse_opt_ctx_t *ctx,
2172                         const struct option *options,
2173                         const char * const usagestr[])
2174 {
2175         int n = handle_revision_opt(revs, ctx->argc, ctx->argv,
2176                                     &ctx->cpidx, ctx->out);
2177         if (n <= 0) {
2178                 error("unknown option `%s'", ctx->argv[0]);
2179                 usage_with_options(usagestr, options);
2180         }
2181         ctx->argv += n;
2182         ctx->argc -= n;
2183 }
2184
2185 static int for_each_bisect_ref(struct ref_store *refs, each_ref_fn fn,
2186                                void *cb_data, const char *term)
2187 {
2188         struct strbuf bisect_refs = STRBUF_INIT;
2189         int status;
2190         strbuf_addf(&bisect_refs, "refs/bisect/%s", term);
2191         status = refs_for_each_fullref_in(refs, bisect_refs.buf, fn, cb_data, 0);
2192         strbuf_release(&bisect_refs);
2193         return status;
2194 }
2195
2196 static int for_each_bad_bisect_ref(struct ref_store *refs, each_ref_fn fn, void *cb_data)
2197 {
2198         return for_each_bisect_ref(refs, fn, cb_data, term_bad);
2199 }
2200
2201 static int for_each_good_bisect_ref(struct ref_store *refs, each_ref_fn fn, void *cb_data)
2202 {
2203         return for_each_bisect_ref(refs, fn, cb_data, term_good);
2204 }
2205
2206 static int handle_revision_pseudo_opt(const char *submodule,
2207                                 struct rev_info *revs,
2208                                 int argc, const char **argv, int *flags)
2209 {
2210         const char *arg = argv[0];
2211         const char *optarg;
2212         struct ref_store *refs;
2213         int argcount;
2214
2215         if (submodule) {
2216                 /*
2217                  * We need some something like get_submodule_worktrees()
2218                  * before we can go through all worktrees of a submodule,
2219                  * .e.g with adding all HEADs from --all, which is not
2220                  * supported right now, so stick to single worktree.
2221                  */
2222                 if (!revs->single_worktree)
2223                         BUG("--single-worktree cannot be used together with submodule");
2224                 refs = get_submodule_ref_store(submodule);
2225         } else
2226                 refs = get_main_ref_store(revs->repo);
2227
2228         /*
2229          * NOTE!
2230          *
2231          * Commands like "git shortlog" will not accept the options below
2232          * unless parse_revision_opt queues them (as opposed to erroring
2233          * out).
2234          *
2235          * When implementing your new pseudo-option, remember to
2236          * register it in the list at the top of handle_revision_opt.
2237          */
2238         if (!strcmp(arg, "--all")) {
2239                 handle_refs(refs, revs, *flags, refs_for_each_ref);
2240                 handle_refs(refs, revs, *flags, refs_head_ref);
2241                 if (!revs->single_worktree) {
2242                         struct all_refs_cb cb;
2243
2244                         init_all_refs_cb(&cb, revs, *flags);
2245                         other_head_refs(handle_one_ref, &cb);
2246                 }
2247                 clear_ref_exclusion(&revs->ref_excludes);
2248         } else if (!strcmp(arg, "--branches")) {
2249                 handle_refs(refs, revs, *flags, refs_for_each_branch_ref);
2250                 clear_ref_exclusion(&revs->ref_excludes);
2251         } else if (!strcmp(arg, "--bisect")) {
2252                 read_bisect_terms(&term_bad, &term_good);
2253                 handle_refs(refs, revs, *flags, for_each_bad_bisect_ref);
2254                 handle_refs(refs, revs, *flags ^ (UNINTERESTING | BOTTOM),
2255                             for_each_good_bisect_ref);
2256                 revs->bisect = 1;
2257         } else if (!strcmp(arg, "--tags")) {
2258                 handle_refs(refs, revs, *flags, refs_for_each_tag_ref);
2259                 clear_ref_exclusion(&revs->ref_excludes);
2260         } else if (!strcmp(arg, "--remotes")) {
2261                 handle_refs(refs, revs, *flags, refs_for_each_remote_ref);
2262                 clear_ref_exclusion(&revs->ref_excludes);
2263         } else if ((argcount = parse_long_opt("glob", argv, &optarg))) {
2264                 struct all_refs_cb cb;
2265                 init_all_refs_cb(&cb, revs, *flags);
2266                 for_each_glob_ref(handle_one_ref, optarg, &cb);
2267                 clear_ref_exclusion(&revs->ref_excludes);
2268                 return argcount;
2269         } else if ((argcount = parse_long_opt("exclude", argv, &optarg))) {
2270                 add_ref_exclusion(&revs->ref_excludes, optarg);
2271                 return argcount;
2272         } else if (skip_prefix(arg, "--branches=", &optarg)) {
2273                 struct all_refs_cb cb;
2274                 init_all_refs_cb(&cb, revs, *flags);
2275                 for_each_glob_ref_in(handle_one_ref, optarg, "refs/heads/", &cb);
2276                 clear_ref_exclusion(&revs->ref_excludes);
2277         } else if (skip_prefix(arg, "--tags=", &optarg)) {
2278                 struct all_refs_cb cb;
2279                 init_all_refs_cb(&cb, revs, *flags);
2280                 for_each_glob_ref_in(handle_one_ref, optarg, "refs/tags/", &cb);
2281                 clear_ref_exclusion(&revs->ref_excludes);
2282         } else if (skip_prefix(arg, "--remotes=", &optarg)) {
2283                 struct all_refs_cb cb;
2284                 init_all_refs_cb(&cb, revs, *flags);
2285                 for_each_glob_ref_in(handle_one_ref, optarg, "refs/remotes/", &cb);
2286                 clear_ref_exclusion(&revs->ref_excludes);
2287         } else if (!strcmp(arg, "--reflog")) {
2288                 add_reflogs_to_pending(revs, *flags);
2289         } else if (!strcmp(arg, "--indexed-objects")) {
2290                 add_index_objects_to_pending(revs, *flags);
2291         } else if (!strcmp(arg, "--not")) {
2292                 *flags ^= UNINTERESTING | BOTTOM;
2293         } else if (!strcmp(arg, "--no-walk")) {
2294                 revs->no_walk = REVISION_WALK_NO_WALK_SORTED;
2295         } else if (skip_prefix(arg, "--no-walk=", &optarg)) {
2296                 /*
2297                  * Detached form ("--no-walk X" as opposed to "--no-walk=X")
2298                  * not allowed, since the argument is optional.
2299                  */
2300                 if (!strcmp(optarg, "sorted"))
2301                         revs->no_walk = REVISION_WALK_NO_WALK_SORTED;
2302                 else if (!strcmp(optarg, "unsorted"))
2303                         revs->no_walk = REVISION_WALK_NO_WALK_UNSORTED;
2304                 else
2305                         return error("invalid argument to --no-walk");
2306         } else if (!strcmp(arg, "--do-walk")) {
2307                 revs->no_walk = 0;
2308         } else if (!strcmp(arg, "--single-worktree")) {
2309                 revs->single_worktree = 1;
2310         } else {
2311                 return 0;
2312         }
2313
2314         return 1;
2315 }
2316
2317 static void NORETURN diagnose_missing_default(const char *def)
2318 {
2319         int flags;
2320         const char *refname;
2321
2322         refname = resolve_ref_unsafe(def, 0, NULL, &flags);
2323         if (!refname || !(flags & REF_ISSYMREF) || (flags & REF_ISBROKEN))
2324                 die(_("your current branch appears to be broken"));
2325
2326         skip_prefix(refname, "refs/heads/", &refname);
2327         die(_("your current branch '%s' does not have any commits yet"),
2328             refname);
2329 }
2330
2331 /*
2332  * Parse revision information, filling in the "rev_info" structure,
2333  * and removing the used arguments from the argument list.
2334  *
2335  * Returns the number of arguments left that weren't recognized
2336  * (which are also moved to the head of the argument list)
2337  */
2338 int setup_revisions(int argc, const char **argv, struct rev_info *revs, struct setup_revision_opt *opt)
2339 {
2340         int i, flags, left, seen_dashdash, got_rev_arg = 0, revarg_opt;
2341         struct argv_array prune_data = ARGV_ARRAY_INIT;
2342         const char *submodule = NULL;
2343
2344         if (opt)
2345                 submodule = opt->submodule;
2346
2347         /* First, search for "--" */
2348         if (opt && opt->assume_dashdash) {
2349                 seen_dashdash = 1;
2350         } else {
2351                 seen_dashdash = 0;
2352                 for (i = 1; i < argc; i++) {
2353                         const char *arg = argv[i];
2354                         if (strcmp(arg, "--"))
2355                                 continue;
2356                         argv[i] = NULL;
2357                         argc = i;
2358                         if (argv[i + 1])
2359                                 argv_array_pushv(&prune_data, argv + i + 1);
2360                         seen_dashdash = 1;
2361                         break;
2362                 }
2363         }
2364
2365         /* Second, deal with arguments and options */
2366         flags = 0;
2367         revarg_opt = opt ? opt->revarg_opt : 0;
2368         if (seen_dashdash)
2369                 revarg_opt |= REVARG_CANNOT_BE_FILENAME;
2370         for (left = i = 1; i < argc; i++) {
2371                 const char *arg = argv[i];
2372                 if (*arg == '-') {
2373                         int opts;
2374
2375                         opts = handle_revision_pseudo_opt(submodule,
2376                                                 revs, argc - i, argv + i,
2377                                                 &flags);
2378                         if (opts > 0) {
2379                                 i += opts - 1;
2380                                 continue;
2381                         }
2382
2383                         if (!strcmp(arg, "--stdin")) {
2384                                 if (revs->disable_stdin) {
2385                                         argv[left++] = arg;
2386                                         continue;
2387                                 }
2388                                 if (revs->read_from_stdin++)
2389                                         die("--stdin given twice?");
2390                                 read_revisions_from_stdin(revs, &prune_data);
2391                                 continue;
2392                         }
2393
2394                         opts = handle_revision_opt(revs, argc - i, argv + i, &left, argv);
2395                         if (opts > 0) {
2396                                 i += opts - 1;
2397                                 continue;
2398                         }
2399                         if (opts < 0)
2400                                 exit(128);
2401                         continue;
2402                 }
2403
2404
2405                 if (handle_revision_arg(arg, revs, flags, revarg_opt)) {
2406                         int j;
2407                         if (seen_dashdash || *arg == '^')
2408                                 die("bad revision '%s'", arg);
2409
2410                         /* If we didn't have a "--":
2411                          * (1) all filenames must exist;
2412                          * (2) all rev-args must not be interpretable
2413                          *     as a valid filename.
2414                          * but the latter we have checked in the main loop.
2415                          */
2416                         for (j = i; j < argc; j++)
2417                                 verify_filename(revs->prefix, argv[j], j == i);
2418
2419                         argv_array_pushv(&prune_data, argv + i);
2420                         break;
2421                 }
2422                 else
2423                         got_rev_arg = 1;
2424         }
2425
2426         if (prune_data.argc) {
2427                 /*
2428                  * If we need to introduce the magic "a lone ':' means no
2429                  * pathspec whatsoever", here is the place to do so.
2430                  *
2431                  * if (prune_data.nr == 1 && !strcmp(prune_data[0], ":")) {
2432                  *      prune_data.nr = 0;
2433                  *      prune_data.alloc = 0;
2434                  *      free(prune_data.path);
2435                  *      prune_data.path = NULL;
2436                  * } else {
2437                  *      terminate prune_data.alloc with NULL and
2438                  *      call init_pathspec() to set revs->prune_data here.
2439                  * }
2440                  */
2441                 parse_pathspec(&revs->prune_data, 0, 0,
2442                                revs->prefix, prune_data.argv);
2443         }
2444         argv_array_clear(&prune_data);
2445
2446         if (revs->def == NULL)
2447                 revs->def = opt ? opt->def : NULL;
2448         if (opt && opt->tweak)
2449                 opt->tweak(revs, opt);
2450         if (revs->show_merge)
2451                 prepare_show_merge(revs);
2452         if (revs->def && !revs->pending.nr && !revs->rev_input_given && !got_rev_arg) {
2453                 struct object_id oid;
2454                 struct object *object;
2455                 struct object_context oc;
2456                 if (get_oid_with_context(revs->def, 0, &oid, &oc))
2457                         diagnose_missing_default(revs->def);
2458                 object = get_reference(revs, revs->def, &oid, 0);
2459                 add_pending_object_with_mode(revs, object, revs->def, oc.mode);
2460         }
2461
2462         /* Did the user ask for any diff output? Run the diff! */
2463         if (revs->diffopt.output_format & ~DIFF_FORMAT_NO_OUTPUT)
2464                 revs->diff = 1;
2465
2466         /* Pickaxe, diff-filter and rename following need diffs */
2467         if ((revs->diffopt.pickaxe_opts & DIFF_PICKAXE_KINDS_MASK) ||
2468             revs->diffopt.filter ||
2469             revs->diffopt.flags.follow_renames)
2470                 revs->diff = 1;
2471
2472         if (revs->diffopt.objfind)
2473                 revs->simplify_history = 0;
2474
2475         if (revs->topo_order && !generation_numbers_enabled(the_repository))
2476                 revs->limited = 1;
2477
2478         if (revs->prune_data.nr) {
2479                 copy_pathspec(&revs->pruning.pathspec, &revs->prune_data);
2480                 /* Can't prune commits with rename following: the paths change.. */
2481                 if (!revs->diffopt.flags.follow_renames)
2482                         revs->prune = 1;
2483                 if (!revs->full_diff)
2484                         copy_pathspec(&revs->diffopt.pathspec,
2485                                       &revs->prune_data);
2486         }
2487         if (revs->combine_merges)
2488                 revs->ignore_merges = 0;
2489         revs->diffopt.abbrev = revs->abbrev;
2490
2491         if (revs->line_level_traverse) {
2492                 revs->limited = 1;
2493                 revs->topo_order = 1;
2494         }
2495
2496         diff_setup_done(&revs->diffopt);
2497
2498         grep_commit_pattern_type(GREP_PATTERN_TYPE_UNSPECIFIED,
2499                                  &revs->grep_filter);
2500         compile_grep_patterns(&revs->grep_filter);
2501
2502         if (revs->reverse && revs->reflog_info)
2503                 die("cannot combine --reverse with --walk-reflogs");
2504         if (revs->reflog_info && revs->limited)
2505                 die("cannot combine --walk-reflogs with history-limiting options");
2506         if (revs->rewrite_parents && revs->children.name)
2507                 die("cannot combine --parents and --children");
2508
2509         /*
2510          * Limitations on the graph functionality
2511          */
2512         if (revs->reverse && revs->graph)
2513                 die("cannot combine --reverse with --graph");
2514
2515         if (revs->reflog_info && revs->graph)
2516                 die("cannot combine --walk-reflogs with --graph");
2517         if (revs->no_walk && revs->graph)
2518                 die("cannot combine --no-walk with --graph");
2519         if (!revs->reflog_info && revs->grep_filter.use_reflog_filter)
2520                 die("cannot use --grep-reflog without --walk-reflogs");
2521
2522         if (revs->first_parent_only && revs->bisect)
2523                 die(_("--first-parent is incompatible with --bisect"));
2524
2525         if (revs->expand_tabs_in_log < 0)
2526                 revs->expand_tabs_in_log = revs->expand_tabs_in_log_default;
2527
2528         return left;
2529 }
2530
2531 static void add_child(struct rev_info *revs, struct commit *parent, struct commit *child)
2532 {
2533         struct commit_list *l = xcalloc(1, sizeof(*l));
2534
2535         l->item = child;
2536         l->next = add_decoration(&revs->children, &parent->object, l);
2537 }
2538
2539 static int remove_duplicate_parents(struct rev_info *revs, struct commit *commit)
2540 {
2541         struct treesame_state *ts = lookup_decoration(&revs->treesame, &commit->object);
2542         struct commit_list **pp, *p;
2543         int surviving_parents;
2544
2545         /* Examine existing parents while marking ones we have seen... */
2546         pp = &commit->parents;
2547         surviving_parents = 0;
2548         while ((p = *pp) != NULL) {
2549                 struct commit *parent = p->item;
2550                 if (parent->object.flags & TMP_MARK) {
2551                         *pp = p->next;
2552                         if (ts)
2553                                 compact_treesame(revs, commit, surviving_parents);
2554                         continue;
2555                 }
2556                 parent->object.flags |= TMP_MARK;
2557                 surviving_parents++;
2558                 pp = &p->next;
2559         }
2560         /* clear the temporary mark */
2561         for (p = commit->parents; p; p = p->next) {
2562                 p->item->object.flags &= ~TMP_MARK;
2563         }
2564         /* no update_treesame() - removing duplicates can't affect TREESAME */
2565         return surviving_parents;
2566 }
2567
2568 struct merge_simplify_state {
2569         struct commit *simplified;
2570 };
2571
2572 static struct merge_simplify_state *locate_simplify_state(struct rev_info *revs, struct commit *commit)
2573 {
2574         struct merge_simplify_state *st;
2575
2576         st = lookup_decoration(&revs->merge_simplification, &commit->object);
2577         if (!st) {
2578                 st = xcalloc(1, sizeof(*st));
2579                 add_decoration(&revs->merge_simplification, &commit->object, st);
2580         }
2581         return st;
2582 }
2583
2584 static int mark_redundant_parents(struct rev_info *revs, struct commit *commit)
2585 {
2586         struct commit_list *h = reduce_heads(commit->parents);
2587         int i = 0, marked = 0;
2588         struct commit_list *po, *pn;
2589
2590         /* Want these for sanity-checking only */
2591         int orig_cnt = commit_list_count(commit->parents);
2592         int cnt = commit_list_count(h);
2593
2594         /*
2595          * Not ready to remove items yet, just mark them for now, based
2596          * on the output of reduce_heads(). reduce_heads outputs the reduced
2597          * set in its original order, so this isn't too hard.
2598          */
2599         po = commit->parents;
2600         pn = h;
2601         while (po) {
2602                 if (pn && po->item == pn->item) {
2603                         pn = pn->next;
2604                         i++;
2605                 } else {
2606                         po->item->object.flags |= TMP_MARK;
2607                         marked++;
2608                 }
2609                 po=po->next;
2610         }
2611
2612         if (i != cnt || cnt+marked != orig_cnt)
2613                 die("mark_redundant_parents %d %d %d %d", orig_cnt, cnt, i, marked);
2614
2615         free_commit_list(h);
2616
2617         return marked;
2618 }
2619
2620 static int mark_treesame_root_parents(struct rev_info *revs, struct commit *commit)
2621 {
2622         struct commit_list *p;
2623         int marked = 0;
2624
2625         for (p = commit->parents; p; p = p->next) {
2626                 struct commit *parent = p->item;
2627                 if (!parent->parents && (parent->object.flags & TREESAME)) {
2628                         parent->object.flags |= TMP_MARK;
2629                         marked++;
2630                 }
2631         }
2632
2633         return marked;
2634 }
2635
2636 /*
2637  * Awkward naming - this means one parent we are TREESAME to.
2638  * cf mark_treesame_root_parents: root parents that are TREESAME (to an
2639  * empty tree). Better name suggestions?
2640  */
2641 static int leave_one_treesame_to_parent(struct rev_info *revs, struct commit *commit)
2642 {
2643         struct treesame_state *ts = lookup_decoration(&revs->treesame, &commit->object);
2644         struct commit *unmarked = NULL, *marked = NULL;
2645         struct commit_list *p;
2646         unsigned n;
2647
2648         for (p = commit->parents, n = 0; p; p = p->next, n++) {
2649                 if (ts->treesame[n]) {
2650                         if (p->item->object.flags & TMP_MARK) {
2651                                 if (!marked)
2652                                         marked = p->item;
2653                         } else {
2654                                 if (!unmarked) {
2655                                         unmarked = p->item;
2656                                         break;
2657                                 }
2658                         }
2659                 }
2660         }
2661
2662         /*
2663          * If we are TREESAME to a marked-for-deletion parent, but not to any
2664          * unmarked parents, unmark the first TREESAME parent. This is the
2665          * parent that the default simplify_history==1 scan would have followed,
2666          * and it doesn't make sense to omit that path when asking for a
2667          * simplified full history. Retaining it improves the chances of
2668          * understanding odd missed merges that took an old version of a file.
2669          *
2670          * Example:
2671          *
2672          *   I--------*X       A modified the file, but mainline merge X used
2673          *    \       /        "-s ours", so took the version from I. X is
2674          *     `-*A--'         TREESAME to I and !TREESAME to A.
2675          *
2676          * Default log from X would produce "I". Without this check,
2677          * --full-history --simplify-merges would produce "I-A-X", showing
2678          * the merge commit X and that it changed A, but not making clear that
2679          * it had just taken the I version. With this check, the topology above
2680          * is retained.
2681          *
2682          * Note that it is possible that the simplification chooses a different
2683          * TREESAME parent from the default, in which case this test doesn't
2684          * activate, and we _do_ drop the default parent. Example:
2685          *
2686          *   I------X         A modified the file, but it was reverted in B,
2687          *    \    /          meaning mainline merge X is TREESAME to both
2688          *    *A-*B           parents.
2689          *
2690          * Default log would produce "I" by following the first parent;
2691          * --full-history --simplify-merges will produce "I-A-B". But this is a
2692          * reasonable result - it presents a logical full history leading from
2693          * I to X, and X is not an important merge.
2694          */
2695         if (!unmarked && marked) {
2696                 marked->object.flags &= ~TMP_MARK;
2697                 return 1;
2698         }
2699
2700         return 0;
2701 }
2702
2703 static int remove_marked_parents(struct rev_info *revs, struct commit *commit)
2704 {
2705         struct commit_list **pp, *p;
2706         int nth_parent, removed = 0;
2707
2708         pp = &commit->parents;
2709         nth_parent = 0;
2710         while ((p = *pp) != NULL) {
2711                 struct commit *parent = p->item;
2712                 if (parent->object.flags & TMP_MARK) {
2713                         parent->object.flags &= ~TMP_MARK;
2714                         *pp = p->next;
2715                         free(p);
2716                         removed++;
2717                         compact_treesame(revs, commit, nth_parent);
2718                         continue;
2719                 }
2720                 pp = &p->next;
2721                 nth_parent++;
2722         }
2723
2724         /* Removing parents can only increase TREESAMEness */
2725         if (removed && !(commit->object.flags & TREESAME))
2726                 update_treesame(revs, commit);
2727
2728         return nth_parent;
2729 }
2730
2731 static struct commit_list **simplify_one(struct rev_info *revs, struct commit *commit, struct commit_list **tail)
2732 {
2733         struct commit_list *p;
2734         struct commit *parent;
2735         struct merge_simplify_state *st, *pst;
2736         int cnt;
2737
2738         st = locate_simplify_state(revs, commit);
2739
2740         /*
2741          * Have we handled this one?
2742          */
2743         if (st->simplified)
2744                 return tail;
2745
2746         /*
2747          * An UNINTERESTING commit simplifies to itself, so does a
2748          * root commit.  We do not rewrite parents of such commit
2749          * anyway.
2750          */
2751         if ((commit->object.flags & UNINTERESTING) || !commit->parents) {
2752                 st->simplified = commit;
2753                 return tail;
2754         }
2755
2756         /*
2757          * Do we know what commit all of our parents that matter
2758          * should be rewritten to?  Otherwise we are not ready to
2759          * rewrite this one yet.
2760          */
2761         for (cnt = 0, p = commit->parents; p; p = p->next) {
2762                 pst = locate_simplify_state(revs, p->item);
2763                 if (!pst->simplified) {
2764                         tail = &commit_list_insert(p->item, tail)->next;
2765                         cnt++;
2766                 }
2767                 if (revs->first_parent_only)
2768                         break;
2769         }
2770         if (cnt) {
2771                 tail = &commit_list_insert(commit, tail)->next;
2772                 return tail;
2773         }
2774
2775         /*
2776          * Rewrite our list of parents. Note that this cannot
2777          * affect our TREESAME flags in any way - a commit is
2778          * always TREESAME to its simplification.
2779          */
2780         for (p = commit->parents; p; p = p->next) {
2781                 pst = locate_simplify_state(revs, p->item);
2782                 p->item = pst->simplified;
2783                 if (revs->first_parent_only)
2784                         break;
2785         }
2786
2787         if (revs->first_parent_only)
2788                 cnt = 1;
2789         else
2790                 cnt = remove_duplicate_parents(revs, commit);
2791
2792         /*
2793          * It is possible that we are a merge and one side branch
2794          * does not have any commit that touches the given paths;
2795          * in such a case, the immediate parent from that branch
2796          * will be rewritten to be the merge base.
2797          *
2798          *      o----X          X: the commit we are looking at;
2799          *     /    /           o: a commit that touches the paths;
2800          * ---o----'
2801          *
2802          * Further, a merge of an independent branch that doesn't
2803          * touch the path will reduce to a treesame root parent:
2804          *
2805          *  ----o----X          X: the commit we are looking at;
2806          *          /           o: a commit that touches the paths;
2807          *         r            r: a root commit not touching the paths
2808          *
2809          * Detect and simplify both cases.
2810          */
2811         if (1 < cnt) {
2812                 int marked = mark_redundant_parents(revs, commit);
2813                 marked += mark_treesame_root_parents(revs, commit);
2814                 if (marked)
2815                         marked -= leave_one_treesame_to_parent(revs, commit);
2816                 if (marked)
2817                         cnt = remove_marked_parents(revs, commit);
2818         }
2819
2820         /*
2821          * A commit simplifies to itself if it is a root, if it is
2822          * UNINTERESTING, if it touches the given paths, or if it is a
2823          * merge and its parents don't simplify to one relevant commit
2824          * (the first two cases are already handled at the beginning of
2825          * this function).
2826          *
2827          * Otherwise, it simplifies to what its sole relevant parent
2828          * simplifies to.
2829          */
2830         if (!cnt ||
2831             (commit->object.flags & UNINTERESTING) ||
2832             !(commit->object.flags & TREESAME) ||
2833             (parent = one_relevant_parent(revs, commit->parents)) == NULL)
2834                 st->simplified = commit;
2835         else {
2836                 pst = locate_simplify_state(revs, parent);
2837                 st->simplified = pst->simplified;
2838         }
2839         return tail;
2840 }
2841
2842 static void simplify_merges(struct rev_info *revs)
2843 {
2844         struct commit_list *list, *next;
2845         struct commit_list *yet_to_do, **tail;
2846         struct commit *commit;
2847
2848         if (!revs->prune)
2849                 return;
2850
2851         /* feed the list reversed */
2852         yet_to_do = NULL;
2853         for (list = revs->commits; list; list = next) {
2854                 commit = list->item;
2855                 next = list->next;
2856                 /*
2857                  * Do not free(list) here yet; the original list
2858                  * is used later in this function.
2859                  */
2860                 commit_list_insert(commit, &yet_to_do);
2861         }
2862         while (yet_to_do) {
2863                 list = yet_to_do;
2864                 yet_to_do = NULL;
2865                 tail = &yet_to_do;
2866                 while (list) {
2867                         commit = pop_commit(&list);
2868                         tail = simplify_one(revs, commit, tail);
2869                 }
2870         }
2871
2872         /* clean up the result, removing the simplified ones */
2873         list = revs->commits;
2874         revs->commits = NULL;
2875         tail = &revs->commits;
2876         while (list) {
2877                 struct merge_simplify_state *st;
2878
2879                 commit = pop_commit(&list);
2880                 st = locate_simplify_state(revs, commit);
2881                 if (st->simplified == commit)
2882                         tail = &commit_list_insert(commit, tail)->next;
2883         }
2884 }
2885
2886 static void set_children(struct rev_info *revs)
2887 {
2888         struct commit_list *l;
2889         for (l = revs->commits; l; l = l->next) {
2890                 struct commit *commit = l->item;
2891                 struct commit_list *p;
2892
2893                 for (p = commit->parents; p; p = p->next)
2894                         add_child(revs, p->item, commit);
2895         }
2896 }
2897
2898 void reset_revision_walk(void)
2899 {
2900         clear_object_flags(SEEN | ADDED | SHOWN);
2901 }
2902
2903 static int mark_uninteresting(const struct object_id *oid,
2904                               struct packed_git *pack,
2905                               uint32_t pos,
2906                               void *cb)
2907 {
2908         struct rev_info *revs = cb;
2909         struct object *o = parse_object(revs->repo, oid);
2910         o->flags |= UNINTERESTING | SEEN;
2911         return 0;
2912 }
2913
2914 define_commit_slab(indegree_slab, int);
2915 define_commit_slab(author_date_slab, timestamp_t);
2916
2917 struct topo_walk_info {
2918         uint32_t min_generation;
2919         struct prio_queue explore_queue;
2920         struct prio_queue indegree_queue;
2921         struct prio_queue topo_queue;
2922         struct indegree_slab indegree;
2923         struct author_date_slab author_date;
2924 };
2925
2926 static inline void test_flag_and_insert(struct prio_queue *q, struct commit *c, int flag)
2927 {
2928         if (c->object.flags & flag)
2929                 return;
2930
2931         c->object.flags |= flag;
2932         prio_queue_put(q, c);
2933 }
2934
2935 static void explore_walk_step(struct rev_info *revs)
2936 {
2937         struct topo_walk_info *info = revs->topo_walk_info;
2938         struct commit_list *p;
2939         struct commit *c = prio_queue_get(&info->explore_queue);
2940
2941         if (!c)
2942                 return;
2943
2944         if (parse_commit_gently(c, 1) < 0)
2945                 return;
2946
2947         if (revs->sort_order == REV_SORT_BY_AUTHOR_DATE)
2948                 record_author_date(&info->author_date, c);
2949
2950         if (revs->max_age != -1 && (c->date < revs->max_age))
2951                 c->object.flags |= UNINTERESTING;
2952
2953         if (process_parents(revs, c, NULL, NULL) < 0)
2954                 return;
2955
2956         if (c->object.flags & UNINTERESTING)
2957                 mark_parents_uninteresting(c);
2958
2959         for (p = c->parents; p; p = p->next)
2960                 test_flag_and_insert(&info->explore_queue, p->item, TOPO_WALK_EXPLORED);
2961 }
2962
2963 static void explore_to_depth(struct rev_info *revs,
2964                              uint32_t gen_cutoff)
2965 {
2966         struct topo_walk_info *info = revs->topo_walk_info;
2967         struct commit *c;
2968         while ((c = prio_queue_peek(&info->explore_queue)) &&
2969                c->generation >= gen_cutoff)
2970                 explore_walk_step(revs);
2971 }
2972
2973 static void indegree_walk_step(struct rev_info *revs)
2974 {
2975         struct commit_list *p;
2976         struct topo_walk_info *info = revs->topo_walk_info;
2977         struct commit *c = prio_queue_get(&info->indegree_queue);
2978
2979         if (!c)
2980                 return;
2981
2982         if (parse_commit_gently(c, 1) < 0)
2983                 return;
2984
2985         explore_to_depth(revs, c->generation);
2986
2987         for (p = c->parents; p; p = p->next) {
2988                 struct commit *parent = p->item;
2989                 int *pi = indegree_slab_at(&info->indegree, parent);
2990
2991                 if (*pi)
2992                         (*pi)++;
2993                 else
2994                         *pi = 2;
2995
2996                 test_flag_and_insert(&info->indegree_queue, parent, TOPO_WALK_INDEGREE);
2997
2998                 if (revs->first_parent_only)
2999                         return;
3000         }
3001 }
3002
3003 static void compute_indegrees_to_depth(struct rev_info *revs,
3004                                        uint32_t gen_cutoff)
3005 {
3006         struct topo_walk_info *info = revs->topo_walk_info;
3007         struct commit *c;
3008         while ((c = prio_queue_peek(&info->indegree_queue)) &&
3009                c->generation >= gen_cutoff)
3010                 indegree_walk_step(revs);
3011 }
3012
3013 static void init_topo_walk(struct rev_info *revs)
3014 {
3015         struct topo_walk_info *info;
3016         struct commit_list *list;
3017         revs->topo_walk_info = xmalloc(sizeof(struct topo_walk_info));
3018         info = revs->topo_walk_info;
3019         memset(info, 0, sizeof(struct topo_walk_info));
3020
3021         init_indegree_slab(&info->indegree);
3022         memset(&info->explore_queue, 0, sizeof(info->explore_queue));
3023         memset(&info->indegree_queue, 0, sizeof(info->indegree_queue));
3024         memset(&info->topo_queue, 0, sizeof(info->topo_queue));
3025
3026         switch (revs->sort_order) {
3027         default: /* REV_SORT_IN_GRAPH_ORDER */
3028                 info->topo_queue.compare = NULL;
3029                 break;
3030         case REV_SORT_BY_COMMIT_DATE:
3031                 info->topo_queue.compare = compare_commits_by_commit_date;
3032                 break;
3033         case REV_SORT_BY_AUTHOR_DATE:
3034                 init_author_date_slab(&info->author_date);
3035                 info->topo_queue.compare = compare_commits_by_author_date;
3036                 info->topo_queue.cb_data = &info->author_date;
3037                 break;
3038         }
3039
3040         info->explore_queue.compare = compare_commits_by_gen_then_commit_date;
3041         info->indegree_queue.compare = compare_commits_by_gen_then_commit_date;
3042
3043         info->min_generation = GENERATION_NUMBER_INFINITY;
3044         for (list = revs->commits; list; list = list->next) {
3045                 struct commit *c = list->item;
3046
3047                 if (parse_commit_gently(c, 1))
3048                         continue;
3049
3050                 test_flag_and_insert(&info->explore_queue, c, TOPO_WALK_EXPLORED);
3051                 test_flag_and_insert(&info->indegree_queue, c, TOPO_WALK_INDEGREE);
3052
3053                 if (c->generation < info->min_generation)
3054                         info->min_generation = c->generation;
3055
3056                 *(indegree_slab_at(&info->indegree, c)) = 1;
3057
3058                 if (revs->sort_order == REV_SORT_BY_AUTHOR_DATE)
3059                         record_author_date(&info->author_date, c);
3060         }
3061         compute_indegrees_to_depth(revs, info->min_generation);
3062
3063         for (list = revs->commits; list; list = list->next) {
3064                 struct commit *c = list->item;
3065
3066                 if (*(indegree_slab_at(&info->indegree, c)) == 1)
3067                         prio_queue_put(&info->topo_queue, c);
3068         }
3069
3070         /*
3071          * This is unfortunate; the initial tips need to be shown
3072          * in the order given from the revision traversal machinery.
3073          */
3074         if (revs->sort_order == REV_SORT_IN_GRAPH_ORDER)
3075                 prio_queue_reverse(&info->topo_queue);
3076 }
3077
3078 static struct commit *next_topo_commit(struct rev_info *revs)
3079 {
3080         struct commit *c;
3081         struct topo_walk_info *info = revs->topo_walk_info;
3082
3083         /* pop next off of topo_queue */
3084         c = prio_queue_get(&info->topo_queue);
3085
3086         if (c)
3087                 *(indegree_slab_at(&info->indegree, c)) = 0;
3088
3089         return c;
3090 }
3091
3092 static void expand_topo_walk(struct rev_info *revs, struct commit *commit)
3093 {
3094         struct commit_list *p;
3095         struct topo_walk_info *info = revs->topo_walk_info;
3096         if (process_parents(revs, commit, NULL, NULL) < 0) {
3097                 if (!revs->ignore_missing_links)
3098                         die("Failed to traverse parents of commit %s",
3099                             oid_to_hex(&commit->object.oid));
3100         }
3101
3102         for (p = commit->parents; p; p = p->next) {
3103                 struct commit *parent = p->item;
3104                 int *pi;
3105
3106                 if (parse_commit_gently(parent, 1) < 0)
3107                         continue;
3108
3109                 if (parent->generation < info->min_generation) {
3110                         info->min_generation = parent->generation;
3111                         compute_indegrees_to_depth(revs, info->min_generation);
3112                 }
3113
3114                 pi = indegree_slab_at(&info->indegree, parent);
3115
3116                 (*pi)--;
3117                 if (*pi == 1)
3118                         prio_queue_put(&info->topo_queue, parent);
3119
3120                 if (revs->first_parent_only)
3121                         return;
3122         }
3123 }
3124
3125 int prepare_revision_walk(struct rev_info *revs)
3126 {
3127         int i;
3128         struct object_array old_pending;
3129         struct commit_list **next = &revs->commits;
3130
3131         memcpy(&old_pending, &revs->pending, sizeof(old_pending));
3132         revs->pending.nr = 0;
3133         revs->pending.alloc = 0;
3134         revs->pending.objects = NULL;
3135         for (i = 0; i < old_pending.nr; i++) {
3136                 struct object_array_entry *e = old_pending.objects + i;
3137                 struct commit *commit = handle_commit(revs, e);
3138                 if (commit) {
3139                         if (!(commit->object.flags & SEEN)) {
3140                                 commit->object.flags |= SEEN;
3141                                 next = commit_list_append(commit, next);
3142                         }
3143                 }
3144         }
3145         object_array_clear(&old_pending);
3146
3147         /* Signal whether we need per-parent treesame decoration */
3148         if (revs->simplify_merges ||
3149             (revs->limited && limiting_can_increase_treesame(revs)))
3150                 revs->treesame.name = "treesame";
3151
3152         if (revs->exclude_promisor_objects) {
3153                 for_each_packed_object(mark_uninteresting, revs,
3154                                        FOR_EACH_OBJECT_PROMISOR_ONLY);
3155         }
3156
3157         if (revs->no_walk != REVISION_WALK_NO_WALK_UNSORTED)
3158                 commit_list_sort_by_date(&revs->commits);
3159         if (revs->no_walk)
3160                 return 0;
3161         if (revs->limited) {
3162                 if (limit_list(revs) < 0)
3163                         return -1;
3164                 if (revs->topo_order)
3165                         sort_in_topological_order(&revs->commits, revs->sort_order);
3166         } else if (revs->topo_order)
3167                 init_topo_walk(revs);
3168         if (revs->line_level_traverse)
3169                 line_log_filter(revs);
3170         if (revs->simplify_merges)
3171                 simplify_merges(revs);
3172         if (revs->children.name)
3173                 set_children(revs);
3174         return 0;
3175 }
3176
3177 static enum rewrite_result rewrite_one(struct rev_info *revs, struct commit **pp)
3178 {
3179         struct commit_list *cache = NULL;
3180
3181         for (;;) {
3182                 struct commit *p = *pp;
3183                 if (!revs->limited)
3184                         if (process_parents(revs, p, &revs->commits, &cache) < 0)
3185                                 return rewrite_one_error;
3186                 if (p->object.flags & UNINTERESTING)
3187                         return rewrite_one_ok;
3188                 if (!(p->object.flags & TREESAME))
3189                         return rewrite_one_ok;
3190                 if (!p->parents)
3191                         return rewrite_one_noparents;
3192                 if ((p = one_relevant_parent(revs, p->parents)) == NULL)
3193                         return rewrite_one_ok;
3194                 *pp = p;
3195         }
3196 }
3197
3198 int rewrite_parents(struct rev_info *revs, struct commit *commit,
3199         rewrite_parent_fn_t rewrite_parent)
3200 {
3201         struct commit_list **pp = &commit->parents;
3202         while (*pp) {
3203                 struct commit_list *parent = *pp;
3204                 switch (rewrite_parent(revs, &parent->item)) {
3205                 case rewrite_one_ok:
3206                         break;
3207                 case rewrite_one_noparents:
3208                         *pp = parent->next;
3209                         continue;
3210                 case rewrite_one_error:
3211                         return -1;
3212                 }
3213                 pp = &parent->next;
3214         }
3215         remove_duplicate_parents(revs, commit);
3216         return 0;
3217 }
3218
3219 static int commit_rewrite_person(struct strbuf *buf, const char *what, struct string_list *mailmap)
3220 {
3221         char *person, *endp;
3222         size_t len, namelen, maillen;
3223         const char *name;
3224         const char *mail;
3225         struct ident_split ident;
3226
3227         person = strstr(buf->buf, what);
3228         if (!person)
3229                 return 0;
3230
3231         person += strlen(what);
3232         endp = strchr(person, '\n');
3233         if (!endp)
3234                 return 0;
3235
3236         len = endp - person;
3237
3238         if (split_ident_line(&ident, person, len))
3239                 return 0;
3240
3241         mail = ident.mail_begin;
3242         maillen = ident.mail_end - ident.mail_begin;
3243         name = ident.name_begin;
3244         namelen = ident.name_end - ident.name_begin;
3245
3246         if (map_user(mailmap, &mail, &maillen, &name, &namelen)) {
3247                 struct strbuf namemail = STRBUF_INIT;
3248
3249                 strbuf_addf(&namemail, "%.*s <%.*s>",
3250                             (int)namelen, name, (int)maillen, mail);
3251
3252                 strbuf_splice(buf, ident.name_begin - buf->buf,
3253                               ident.mail_end - ident.name_begin + 1,
3254                               namemail.buf, namemail.len);
3255
3256                 strbuf_release(&namemail);
3257
3258                 return 1;
3259         }
3260
3261         return 0;
3262 }
3263
3264 static int commit_match(struct commit *commit, struct rev_info *opt)
3265 {
3266         int retval;
3267         const char *encoding;
3268         const char *message;
3269         struct strbuf buf = STRBUF_INIT;
3270
3271         if (!opt->grep_filter.pattern_list && !opt->grep_filter.header_list)
3272                 return 1;
3273
3274         /* Prepend "fake" headers as needed */
3275         if (opt->grep_filter.use_reflog_filter) {
3276                 strbuf_addstr(&buf, "reflog ");
3277                 get_reflog_message(&buf, opt->reflog_info);
3278                 strbuf_addch(&buf, '\n');
3279         }
3280
3281         /*
3282          * We grep in the user's output encoding, under the assumption that it
3283          * is the encoding they are most likely to write their grep pattern
3284          * for. In addition, it means we will match the "notes" encoding below,
3285          * so we will not end up with a buffer that has two different encodings
3286          * in it.
3287          */
3288         encoding = get_log_output_encoding();
3289         message = logmsg_reencode(commit, NULL, encoding);
3290
3291         /* Copy the commit to temporary if we are using "fake" headers */
3292         if (buf.len)
3293                 strbuf_addstr(&buf, message);
3294
3295         if (opt->grep_filter.header_list && opt->mailmap) {
3296                 if (!buf.len)
3297                         strbuf_addstr(&buf, message);
3298
3299                 commit_rewrite_person(&buf, "\nauthor ", opt->mailmap);
3300                 commit_rewrite_person(&buf, "\ncommitter ", opt->mailmap);
3301         }
3302
3303         /* Append "fake" message parts as needed */
3304         if (opt->show_notes) {
3305                 if (!buf.len)
3306                         strbuf_addstr(&buf, message);
3307                 format_display_notes(&commit->object.oid, &buf, encoding, 1);
3308         }
3309
3310         /*
3311          * Find either in the original commit message, or in the temporary.
3312          * Note that we cast away the constness of "message" here. It is
3313          * const because it may come from the cached commit buffer. That's OK,
3314          * because we know that it is modifiable heap memory, and that while
3315          * grep_buffer may modify it for speed, it will restore any
3316          * changes before returning.
3317          */
3318         if (buf.len)
3319                 retval = grep_buffer(&opt->grep_filter, buf.buf, buf.len);
3320         else
3321                 retval = grep_buffer(&opt->grep_filter,
3322                                      (char *)message, strlen(message));
3323         strbuf_release(&buf);
3324         unuse_commit_buffer(commit, message);
3325         return opt->invert_grep ? !retval : retval;
3326 }
3327
3328 static inline int want_ancestry(const struct rev_info *revs)
3329 {
3330         return (revs->rewrite_parents || revs->children.name);
3331 }
3332
3333 /*
3334  * Return a timestamp to be used for --since/--until comparisons for this
3335  * commit, based on the revision options.
3336  */
3337 static timestamp_t comparison_date(const struct rev_info *revs,
3338                                    struct commit *commit)
3339 {
3340         return revs->reflog_info ?
3341                 get_reflog_timestamp(revs->reflog_info) :
3342                 commit->date;
3343 }
3344
3345 enum commit_action get_commit_action(struct rev_info *revs, struct commit *commit)
3346 {
3347         if (commit->object.flags & SHOWN)
3348                 return commit_ignore;
3349         if (revs->unpacked && has_object_pack(&commit->object.oid))
3350                 return commit_ignore;
3351         if (commit->object.flags & UNINTERESTING)
3352                 return commit_ignore;
3353         if (revs->min_age != -1 &&
3354             comparison_date(revs, commit) > revs->min_age)
3355                         return commit_ignore;
3356         if (revs->min_parents || (revs->max_parents >= 0)) {
3357                 int n = commit_list_count(commit->parents);
3358                 if ((n < revs->min_parents) ||
3359                     ((revs->max_parents >= 0) && (n > revs->max_parents)))
3360                         return commit_ignore;
3361         }
3362         if (!commit_match(commit, revs))
3363                 return commit_ignore;
3364         if (revs->prune && revs->dense) {
3365                 /* Commit without changes? */
3366                 if (commit->object.flags & TREESAME) {
3367                         int n;
3368                         struct commit_list *p;
3369                         /* drop merges unless we want parenthood */
3370                         if (!want_ancestry(revs))
3371                                 return commit_ignore;
3372                         /*
3373                          * If we want ancestry, then need to keep any merges
3374                          * between relevant commits to tie together topology.
3375                          * For consistency with TREESAME and simplification
3376                          * use "relevant" here rather than just INTERESTING,
3377                          * to treat bottom commit(s) as part of the topology.
3378                          */
3379                         for (n = 0, p = commit->parents; p; p = p->next)
3380                                 if (relevant_commit(p->item))
3381                                         if (++n >= 2)
3382                                                 return commit_show;
3383                         return commit_ignore;
3384                 }
3385         }
3386         return commit_show;
3387 }
3388
3389 define_commit_slab(saved_parents, struct commit_list *);
3390
3391 #define EMPTY_PARENT_LIST ((struct commit_list *)-1)
3392
3393 /*
3394  * You may only call save_parents() once per commit (this is checked
3395  * for non-root commits).
3396  */
3397 static void save_parents(struct rev_info *revs, struct commit *commit)
3398 {
3399         struct commit_list **pp;
3400
3401         if (!revs->saved_parents_slab) {
3402                 revs->saved_parents_slab = xmalloc(sizeof(struct saved_parents));
3403                 init_saved_parents(revs->saved_parents_slab);
3404         }
3405
3406         pp = saved_parents_at(revs->saved_parents_slab, commit);
3407
3408         /*
3409          * When walking with reflogs, we may visit the same commit
3410          * several times: once for each appearance in the reflog.
3411          *
3412          * In this case, save_parents() will be called multiple times.
3413          * We want to keep only the first set of parents.  We need to
3414          * store a sentinel value for an empty (i.e., NULL) parent
3415          * list to distinguish it from a not-yet-saved list, however.
3416          */
3417         if (*pp)
3418                 return;
3419         if (commit->parents)
3420                 *pp = copy_commit_list(commit->parents);
3421         else
3422                 *pp = EMPTY_PARENT_LIST;
3423 }
3424
3425 static void free_saved_parents(struct rev_info *revs)
3426 {
3427         if (revs->saved_parents_slab)
3428                 clear_saved_parents(revs->saved_parents_slab);
3429 }
3430
3431 struct commit_list *get_saved_parents(struct rev_info *revs, const struct commit *commit)
3432 {
3433         struct commit_list *parents;
3434
3435         if (!revs->saved_parents_slab)
3436                 return commit->parents;
3437
3438         parents = *saved_parents_at(revs->saved_parents_slab, commit);
3439         if (parents == EMPTY_PARENT_LIST)
3440                 return NULL;
3441         return parents;
3442 }
3443
3444 enum commit_action simplify_commit(struct rev_info *revs, struct commit *commit)
3445 {
3446         enum commit_action action = get_commit_action(revs, commit);
3447
3448         if (action == commit_show &&
3449             revs->prune && revs->dense && want_ancestry(revs)) {
3450                 /*
3451                  * --full-diff on simplified parents is no good: it
3452                  * will show spurious changes from the commits that
3453                  * were elided.  So we save the parents on the side
3454                  * when --full-diff is in effect.
3455                  */
3456                 if (revs->full_diff)
3457                         save_parents(revs, commit);
3458                 if (rewrite_parents(revs, commit, rewrite_one) < 0)
3459                         return commit_error;
3460         }
3461         return action;
3462 }
3463
3464 static void track_linear(struct rev_info *revs, struct commit *commit)
3465 {
3466         if (revs->track_first_time) {
3467                 revs->linear = 1;
3468                 revs->track_first_time = 0;
3469         } else {
3470                 struct commit_list *p;
3471                 for (p = revs->previous_parents; p; p = p->next)
3472                         if (p->item == NULL || /* first commit */
3473                             oideq(&p->item->object.oid, &commit->object.oid))
3474                                 break;
3475                 revs->linear = p != NULL;
3476         }
3477         if (revs->reverse) {
3478                 if (revs->linear)
3479                         commit->object.flags |= TRACK_LINEAR;
3480         }
3481         free_commit_list(revs->previous_parents);
3482         revs->previous_parents = copy_commit_list(commit->parents);
3483 }
3484
3485 static struct commit *get_revision_1(struct rev_info *revs)
3486 {
3487         while (1) {
3488                 struct commit *commit;
3489
3490                 if (revs->reflog_info)
3491                         commit = next_reflog_entry(revs->reflog_info);
3492                 else if (revs->topo_walk_info)
3493                         commit = next_topo_commit(revs);
3494                 else
3495                         commit = pop_commit(&revs->commits);
3496
3497                 if (!commit)
3498                         return NULL;
3499
3500                 if (revs->reflog_info)
3501                         commit->object.flags &= ~(ADDED | SEEN | SHOWN);
3502
3503                 /*
3504                  * If we haven't done the list limiting, we need to look at
3505                  * the parents here. We also need to do the date-based limiting
3506                  * that we'd otherwise have done in limit_list().
3507                  */
3508                 if (!revs->limited) {
3509                         if (revs->max_age != -1 &&
3510                             comparison_date(revs, commit) < revs->max_age)
3511                                 continue;
3512
3513                         if (revs->reflog_info)
3514                                 try_to_simplify_commit(revs, commit);
3515                         else if (revs->topo_walk_info)
3516                                 expand_topo_walk(revs, commit);
3517                         else if (process_parents(revs, commit, &revs->commits, NULL) < 0) {
3518                                 if (!revs->ignore_missing_links)
3519                                         die("Failed to traverse parents of commit %s",
3520                                                 oid_to_hex(&commit->object.oid));
3521                         }
3522                 }
3523
3524                 switch (simplify_commit(revs, commit)) {
3525                 case commit_ignore:
3526                         continue;
3527                 case commit_error:
3528                         die("Failed to simplify parents of commit %s",
3529                             oid_to_hex(&commit->object.oid));
3530                 default:
3531                         if (revs->track_linear)
3532                                 track_linear(revs, commit);
3533                         return commit;
3534                 }
3535         }
3536 }
3537
3538 /*
3539  * Return true for entries that have not yet been shown.  (This is an
3540  * object_array_each_func_t.)
3541  */
3542 static int entry_unshown(struct object_array_entry *entry, void *cb_data_unused)
3543 {
3544         return !(entry->item->flags & SHOWN);
3545 }
3546
3547 /*
3548  * If array is on the verge of a realloc, garbage-collect any entries
3549  * that have already been shown to try to free up some space.
3550  */
3551 static void gc_boundary(struct object_array *array)
3552 {
3553         if (array->nr == array->alloc)
3554                 object_array_filter(array, entry_unshown, NULL);
3555 }
3556
3557 static void create_boundary_commit_list(struct rev_info *revs)
3558 {
3559         unsigned i;
3560         struct commit *c;
3561         struct object_array *array = &revs->boundary_commits;
3562         struct object_array_entry *objects = array->objects;
3563
3564         /*
3565          * If revs->commits is non-NULL at this point, an error occurred in
3566          * get_revision_1().  Ignore the error and continue printing the
3567          * boundary commits anyway.  (This is what the code has always
3568          * done.)
3569          */
3570         if (revs->commits) {
3571                 free_commit_list(revs->commits);
3572                 revs->commits = NULL;
3573         }
3574
3575         /*
3576          * Put all of the actual boundary commits from revs->boundary_commits
3577          * into revs->commits
3578          */
3579         for (i = 0; i < array->nr; i++) {
3580                 c = (struct commit *)(objects[i].item);
3581                 if (!c)
3582                         continue;
3583                 if (!(c->object.flags & CHILD_SHOWN))
3584                         continue;
3585                 if (c->object.flags & (SHOWN | BOUNDARY))
3586                         continue;
3587                 c->object.flags |= BOUNDARY;
3588                 commit_list_insert(c, &revs->commits);
3589         }
3590
3591         /*
3592          * If revs->topo_order is set, sort the boundary commits
3593          * in topological order
3594          */
3595         sort_in_topological_order(&revs->commits, revs->sort_order);
3596 }
3597
3598 static struct commit *get_revision_internal(struct rev_info *revs)
3599 {
3600         struct commit *c = NULL;
3601         struct commit_list *l;
3602
3603         if (revs->boundary == 2) {
3604                 /*
3605                  * All of the normal commits have already been returned,
3606                  * and we are now returning boundary commits.
3607                  * create_boundary_commit_list() has populated
3608                  * revs->commits with the remaining commits to return.
3609                  */
3610                 c = pop_commit(&revs->commits);
3611                 if (c)
3612                         c->object.flags |= SHOWN;
3613                 return c;
3614         }
3615
3616         /*
3617          * If our max_count counter has reached zero, then we are done. We
3618          * don't simply return NULL because we still might need to show
3619          * boundary commits. But we want to avoid calling get_revision_1, which
3620          * might do a considerable amount of work finding the next commit only
3621          * for us to throw it away.
3622          *
3623          * If it is non-zero, then either we don't have a max_count at all
3624          * (-1), or it is still counting, in which case we decrement.
3625          */
3626         if (revs->max_count) {
3627                 c = get_revision_1(revs);
3628                 if (c) {
3629                         while (revs->skip_count > 0) {
3630                                 revs->skip_count--;
3631                                 c = get_revision_1(revs);
3632                                 if (!c)
3633                                         break;
3634                         }
3635                 }
3636
3637                 if (revs->max_count > 0)
3638                         revs->max_count--;
3639         }
3640
3641         if (c)
3642                 c->object.flags |= SHOWN;
3643
3644         if (!revs->boundary)
3645                 return c;
3646
3647         if (!c) {
3648                 /*
3649                  * get_revision_1() runs out the commits, and
3650                  * we are done computing the boundaries.
3651                  * switch to boundary commits output mode.
3652                  */
3653                 revs->boundary = 2;
3654
3655                 /*
3656                  * Update revs->commits to contain the list of
3657                  * boundary commits.
3658                  */
3659                 create_boundary_commit_list(revs);
3660
3661                 return get_revision_internal(revs);
3662         }
3663
3664         /*
3665          * boundary commits are the commits that are parents of the
3666          * ones we got from get_revision_1() but they themselves are
3667          * not returned from get_revision_1().  Before returning
3668          * 'c', we need to mark its parents that they could be boundaries.
3669          */
3670
3671         for (l = c->parents; l; l = l->next) {
3672                 struct object *p;
3673                 p = &(l->item->object);
3674                 if (p->flags & (CHILD_SHOWN | SHOWN))
3675                         continue;
3676                 p->flags |= CHILD_SHOWN;
3677                 gc_boundary(&revs->boundary_commits);
3678                 add_object_array(p, NULL, &revs->boundary_commits);
3679         }
3680
3681         return c;
3682 }
3683
3684 struct commit *get_revision(struct rev_info *revs)
3685 {
3686         struct commit *c;
3687         struct commit_list *reversed;
3688
3689         if (revs->reverse) {
3690                 reversed = NULL;
3691                 while ((c = get_revision_internal(revs)))
3692                         commit_list_insert(c, &reversed);
3693                 revs->commits = reversed;
3694                 revs->reverse = 0;
3695                 revs->reverse_output_stage = 1;
3696         }
3697
3698         if (revs->reverse_output_stage) {
3699                 c = pop_commit(&revs->commits);
3700                 if (revs->track_linear)
3701                         revs->linear = !!(c && c->object.flags & TRACK_LINEAR);
3702                 return c;
3703         }
3704
3705         c = get_revision_internal(revs);
3706         if (c && revs->graph)
3707                 graph_update(revs->graph, c);
3708         if (!c) {
3709                 free_saved_parents(revs);
3710                 if (revs->previous_parents) {
3711                         free_commit_list(revs->previous_parents);
3712                         revs->previous_parents = NULL;
3713                 }
3714         }
3715         return c;
3716 }
3717
3718 char *get_revision_mark(const struct rev_info *revs, const struct commit *commit)
3719 {
3720         if (commit->object.flags & BOUNDARY)
3721                 return "-";
3722         else if (commit->object.flags & UNINTERESTING)
3723                 return "^";
3724         else if (commit->object.flags & PATCHSAME)
3725                 return "=";
3726         else if (!revs || revs->left_right) {
3727                 if (commit->object.flags & SYMMETRIC_LEFT)
3728                         return "<";
3729                 else
3730                         return ">";
3731         } else if (revs->graph)
3732                 return "*";
3733         else if (revs->cherry_mark)
3734                 return "+";
3735         return "";
3736 }
3737
3738 void put_revision_mark(const struct rev_info *revs, const struct commit *commit)
3739 {
3740         char *mark = get_revision_mark(revs, commit);
3741         if (!strlen(mark))
3742                 return;
3743         fputs(mark, stdout);
3744         putchar(' ');
3745 }