Pileus Git - ~andy/git/blob - revision.c

   1 #include "cache.h"
   2 #include "tag.h"
   3 #include "blob.h"
   4 #include "tree.h"
   5 #include "commit.h"
   6 #include "diff.h"
   7 #include "refs.h"
   8 #include "revision.h"
   9 #include "graph.h"
  10 #include "grep.h"
  11 #include "reflog-walk.h"
  12 #include "patch-ids.h"
  13 #include "decorate.h"
  14 #include "log-tree.h"
  15 #include "string-list.h"
  16 #include "line-log.h"
  17 #include "mailmap.h"
  18 #include "commit-slab.h"
  19
  20 volatile show_early_output_fn_t show_early_output;
  21
  22 char *path_name(const struct name_path *path, const char *name)
  23 {
  24         const struct name_path *p;
  25         char *n, *m;
  26         int nlen = strlen(name);
  27         int len = nlen + 1;
  28
  29         for (p = path; p; p = p->up) {
  30                 if (p->elem_len)
  31                         len += p->elem_len + 1;
  32         }
  33         n = xmalloc(len);
  34         m = n + len - (nlen + 1);
  35         strcpy(m, name);
  36         for (p = path; p; p = p->up) {
  37                 if (p->elem_len) {
  38                         m -= p->elem_len + 1;
  39                         memcpy(m, p->elem, p->elem_len);
  40                         m[p->elem_len] = '/';
  41                 }
  42         }
  43         return n;
  44 }
  45
  46 static int show_path_component_truncated(FILE *out, const char *name, int len)
  47 {
  48         int cnt;
  49         for (cnt = 0; cnt < len; cnt++) {
  50                 int ch = name[cnt];
  51                 if (!ch || ch == '\n')
  52                         return -1;
  53                 fputc(ch, out);
  54         }
  55         return len;
  56 }
  57
  58 static int show_path_truncated(FILE *out, const struct name_path *path)
  59 {
  60         int emitted, ours;
  61
  62         if (!path)
  63                 return 0;
  64         emitted = show_path_truncated(out, path->up);
  65         if (emitted < 0)
  66                 return emitted;
  67         if (emitted)
  68                 fputc('/', out);
  69         ours = show_path_component_truncated(out, path->elem, path->elem_len);
  70         if (ours < 0)
  71                 return ours;
  72         return ours || emitted;
  73 }
  74
  75 void show_object_with_name(FILE *out, struct object *obj,
  76                            const struct name_path *path, const char *component)
  77 {
  78         struct name_path leaf;
  79         leaf.up = (struct name_path *)path;
  80         leaf.elem = component;
  81         leaf.elem_len = strlen(component);
  82
  83         fprintf(out, "%s ", sha1_to_hex(obj->sha1));
  84         show_path_truncated(out, &leaf);
  85         fputc('\n', out);
  86 }
  87
  88 void add_object(struct object *obj,
  89                 struct object_array *p,
  90                 struct name_path *path,
  91                 const char *name)
  92 {
  93         char *pn = path_name(path, name);
  94         add_object_array(obj, pn, p);
  95         free(pn);
  96 }
  97
  98 static void mark_blob_uninteresting(struct blob *blob)
  99 {
 100         if (!blob)
 101                 return;
 102         if (blob->object.flags & UNINTERESTING)
 103                 return;
 104         blob->object.flags |= UNINTERESTING;
 105 }
 106
 107 void mark_tree_uninteresting(struct tree *tree)
 108 {
 109         struct tree_desc desc;
 110         struct name_entry entry;
 111         struct object *obj = &tree->object;
 112
 113         if (!tree)
 114                 return;
 115         if (obj->flags & UNINTERESTING)
 116                 return;
 117         obj->flags |= UNINTERESTING;
 118         if (!has_sha1_file(obj->sha1))
 119                 return;
 120         if (parse_tree(tree) < 0)
 121                 die("bad tree %s", sha1_to_hex(obj->sha1));
 122
 123         init_tree_desc(&desc, tree->buffer, tree->size);
 124         while (tree_entry(&desc, &entry)) {
 125                 switch (object_type(entry.mode)) {
 126                 case OBJ_TREE:
 127                         mark_tree_uninteresting(lookup_tree(entry.sha1));
 128                         break;
 129                 case OBJ_BLOB:
 130                         mark_blob_uninteresting(lookup_blob(entry.sha1));
 131                         break;
 132                 default:
 133                         /* Subproject commit - not in this repository */
 134                         break;
 135                 }
 136         }
 137
 138         /*
 139          * We don't care about the tree any more
 140          * after it has been marked uninteresting.
 141          */
 142         free_tree_buffer(tree);
 143 }
 144
 145 void mark_parents_uninteresting(struct commit *commit)
 146 {
 147         struct commit_list *parents = NULL, *l;
 148
 149         for (l = commit->parents; l; l = l->next)
 150                 commit_list_insert(l->item, &parents);
 151
 152         while (parents) {
 153                 struct commit *commit = parents->item;
 154                 l = parents;
 155                 parents = parents->next;
 156                 free(l);
 157
 158                 while (commit) {
 159                         /*
 160                          * A missing commit is ok iff its parent is marked
 161                          * uninteresting.
 162                          *
 163                          * We just mark such a thing parsed, so that when
 164                          * it is popped next time around, we won't be trying
 165                          * to parse it and get an error.
 166                          */
 167                         if (!has_sha1_file(commit->object.sha1))
 168                                 commit->object.parsed = 1;
 169
 170                         if (commit->object.flags & UNINTERESTING)
 171                                 break;
 172
 173                         commit->object.flags |= UNINTERESTING;
 174
 175                         /*
 176                          * Normally we haven't parsed the parent
 177                          * yet, so we won't have a parent of a parent
 178                          * here. However, it may turn out that we've
 179                          * reached this commit some other way (where it
 180                          * wasn't uninteresting), in which case we need
 181                          * to mark its parents recursively too..
 182                          */
 183                         if (!commit->parents)
 184                                 break;
 185
 186                         for (l = commit->parents->next; l; l = l->next)
 187                                 commit_list_insert(l->item, &parents);
 188                         commit = commit->parents->item;
 189                 }
 190         }
 191 }
 192
 193 static void add_pending_object_with_mode(struct rev_info *revs,
 194                                          struct object *obj,
 195                                          const char *name, unsigned mode)
 196 {
 197         if (!obj)
 198                 return;
 199         if (revs->no_walk && (obj->flags & UNINTERESTING))
 200                 revs->no_walk = 0;
 201         if (revs->reflog_info && obj->type == OBJ_COMMIT) {
 202                 struct strbuf buf = STRBUF_INIT;
 203                 int len = interpret_branch_name(name, 0, &buf);
 204                 int st;
 205
 206                 if (0 < len && name[len] && buf.len)
 207                         strbuf_addstr(&buf, name + len);
 208                 st = add_reflog_for_walk(revs->reflog_info,
 209                                          (struct commit *)obj,
 210                                          buf.buf[0] ? buf.buf: name);
 211                 strbuf_release(&buf);
 212                 if (st)
 213                         return;
 214         }
 215         add_object_array_with_mode(obj, name, &revs->pending, mode);
 216 }
 217
 218 void add_pending_object(struct rev_info *revs,
 219                         struct object *obj, const char *name)
 220 {
 221         add_pending_object_with_mode(revs, obj, name, S_IFINVALID);
 222 }
 223
 224 void add_head_to_pending(struct rev_info *revs)
 225 {
 226         unsigned char sha1[20];
 227         struct object *obj;
 228         if (get_sha1("HEAD", sha1))
 229                 return;
 230         obj = parse_object(sha1);
 231         if (!obj)
 232                 return;
 233         add_pending_object(revs, obj, "HEAD");
 234 }
 235
 236 static struct object *get_reference(struct rev_info *revs, const char *name,
 237                                     const unsigned char *sha1,
 238                                     unsigned int flags)
 239 {
 240         struct object *object;
 241
 242         object = parse_object(sha1);
 243         if (!object) {
 244                 if (revs->ignore_missing)
 245                         return object;
 246                 die("bad object %s", name);
 247         }
 248         object->flags |= flags;
 249         return object;
 250 }
 251
 252 void add_pending_sha1(struct rev_info *revs, const char *name,
 253                       const unsigned char *sha1, unsigned int flags)
 254 {
 255         struct object *object = get_reference(revs, name, sha1, flags);
 256         add_pending_object(revs, object, name);
 257 }
 258
 259 static struct commit *handle_commit(struct rev_info *revs,
 260                                     struct object *object, const char *name)
 261 {
 262         unsigned long flags = object->flags;
 263
 264         /*
 265          * Tag object? Look what it points to..
 266          */
 267         while (object->type == OBJ_TAG) {
 268                 struct tag *tag = (struct tag *) object;
 269                 if (revs->tag_objects && !(flags & UNINTERESTING))
 270                         add_pending_object(revs, object, tag->tag);
 271                 if (!tag->tagged)
 272                         die("bad tag");
 273                 object = parse_object(tag->tagged->sha1);
 274                 if (!object) {
 275                         if (flags & UNINTERESTING)
 276                                 return NULL;
 277                         die("bad object %s", sha1_to_hex(tag->tagged->sha1));
 278                 }
 279         }
 280
 281         /*
 282          * Commit object? Just return it, we'll do all the complex
 283          * reachability crud.
 284          */
 285         if (object->type == OBJ_COMMIT) {
 286                 struct commit *commit = (struct commit *)object;
 287                 if (parse_commit(commit) < 0)
 288                         die("unable to parse commit %s", name);
 289                 if (flags & UNINTERESTING) {
 290                         commit->object.flags |= UNINTERESTING;
 291                         mark_parents_uninteresting(commit);
 292                         revs->limited = 1;
 293                 }
 294                 if (revs->show_source && !commit->util)
 295                         commit->util = (void *) name;
 296                 return commit;
 297         }
 298
 299         /*
 300          * Tree object? Either mark it uninteresting, or add it
 301          * to the list of objects to look at later..
 302          */
 303         if (object->type == OBJ_TREE) {
 304                 struct tree *tree = (struct tree *)object;
 305                 if (!revs->tree_objects)
 306                         return NULL;
 307                 if (flags & UNINTERESTING) {
 308                         mark_tree_uninteresting(tree);
 309                         return NULL;
 310                 }
 311                 add_pending_object(revs, object, "");
 312                 return NULL;
 313         }
 314
 315         /*
 316          * Blob object? You know the drill by now..
 317          */
 318         if (object->type == OBJ_BLOB) {
 319                 struct blob *blob = (struct blob *)object;
 320                 if (!revs->blob_objects)
 321                         return NULL;
 322                 if (flags & UNINTERESTING) {
 323                         mark_blob_uninteresting(blob);
 324                         return NULL;
 325                 }
 326                 add_pending_object(revs, object, "");
 327                 return NULL;
 328         }
 329         die("%s is unknown object", name);
 330 }
 331
 332 static int everybody_uninteresting(struct commit_list *orig)
 333 {
 334         struct commit_list *list = orig;
 335         while (list) {
 336                 struct commit *commit = list->item;
 337                 list = list->next;
 338                 if (commit->object.flags & UNINTERESTING)
 339                         continue;
 340                 return 0;
 341         }
 342         return 1;
 343 }
 344
 345 /*
 346  * A definition of "relevant" commit that we can use to simplify limited graphs
 347  * by eliminating side branches.
 348  *
 349  * A "relevant" commit is one that is !UNINTERESTING (ie we are including it
 350  * in our list), or that is a specified BOTTOM commit. Then after computing
 351  * a limited list, during processing we can generally ignore boundary merges
 352  * coming from outside the graph, (ie from irrelevant parents), and treat
 353  * those merges as if they were single-parent. TREESAME is defined to consider
 354  * only relevant parents, if any. If we are TREESAME to our on-graph parents,
 355  * we don't care if we were !TREESAME to non-graph parents.
 356  *
 357  * Treating bottom commits as relevant ensures that a limited graph's
 358  * connection to the actual bottom commit is not viewed as a side branch, but
 359  * treated as part of the graph. For example:
 360  *
 361  *   ....Z...A---X---o---o---B
 362  *        .     /
 363  *         W---Y
 364  *
 365  * When computing "A..B", the A-X connection is at least as important as
 366  * Y-X, despite A being flagged UNINTERESTING.
 367  *
 368  * And when computing --ancestry-path "A..B", the A-X connection is more
 369  * important than Y-X, despite both A and Y being flagged UNINTERESTING.
 370  */
 371 static inline int relevant_commit(struct commit *commit)
 372 {
 373         return (commit->object.flags & (UNINTERESTING | BOTTOM)) != UNINTERESTING;
 374 }
 375
 376 /*
 377  * Return a single relevant commit from a parent list. If we are a TREESAME
 378  * commit, and this selects one of our parents, then we can safely simplify to
 379  * that parent.
 380  */
 381 static struct commit *one_relevant_parent(const struct rev_info *revs,
 382                                           struct commit_list *orig)
 383 {
 384         struct commit_list *list = orig;
 385         struct commit *relevant = NULL;
 386
 387         if (!orig)
 388                 return NULL;
 389
 390         /*
 391          * For 1-parent commits, or if first-parent-only, then return that
 392          * first parent (even if not "relevant" by the above definition).
 393          * TREESAME will have been set purely on that parent.
 394          */
 395         if (revs->first_parent_only || !orig->next)
 396                 return orig->item;
 397
 398         /*
 399          * For multi-parent commits, identify a sole relevant parent, if any.
 400          * If we have only one relevant parent, then TREESAME will be set purely
 401          * with regard to that parent, and we can simplify accordingly.
 402          *
 403          * If we have more than one relevant parent, or no relevant parents
 404          * (and multiple irrelevant ones), then we can't select a parent here
 405          * and return NULL.
 406          */
 407         while (list) {
 408                 struct commit *commit = list->item;
 409                 list = list->next;
 410                 if (relevant_commit(commit)) {
 411                         if (relevant)
 412                                 return NULL;
 413                         relevant = commit;
 414                 }
 415         }
 416         return relevant;
 417 }
 418
 419 /*
 420  * The goal is to get REV_TREE_NEW as the result only if the
 421  * diff consists of all '+' (and no other changes), REV_TREE_OLD
 422  * if the whole diff is removal of old data, and otherwise
 423  * REV_TREE_DIFFERENT (of course if the trees are the same we
 424  * want REV_TREE_SAME).
 425  * That means that once we get to REV_TREE_DIFFERENT, we do not
 426  * have to look any further.
 427  */
 428 static int tree_difference = REV_TREE_SAME;
 429
 430 static void file_add_remove(struct diff_options *options,
 431                     int addremove, unsigned mode,
 432                     const unsigned char *sha1,
 433                     int sha1_valid,
 434                     const char *fullpath, unsigned dirty_submodule)
 435 {
 436         int diff = addremove == '+' ? REV_TREE_NEW : REV_TREE_OLD;
 437
 438         tree_difference |= diff;
 439         if (tree_difference == REV_TREE_DIFFERENT)
 440                 DIFF_OPT_SET(options, HAS_CHANGES);
 441 }
 442
 443 static void file_change(struct diff_options *options,
 444                  unsigned old_mode, unsigned new_mode,
 445                  const unsigned char *old_sha1,
 446                  const unsigned char *new_sha1,
 447                  int old_sha1_valid, int new_sha1_valid,
 448                  const char *fullpath,
 449                  unsigned old_dirty_submodule, unsigned new_dirty_submodule)
 450 {
 451         tree_difference = REV_TREE_DIFFERENT;
 452         DIFF_OPT_SET(options, HAS_CHANGES);
 453 }
 454
 455 static int rev_compare_tree(struct rev_info *revs,
 456                             struct commit *parent, struct commit *commit)
 457 {
 458         struct tree *t1 = parent->tree;
 459         struct tree *t2 = commit->tree;
 460
 461         if (!t1)
 462                 return REV_TREE_NEW;
 463         if (!t2)
 464                 return REV_TREE_OLD;
 465
 466         if (revs->simplify_by_decoration) {
 467                 /*
 468                  * If we are simplifying by decoration, then the commit
 469                  * is worth showing if it has a tag pointing at it.
 470                  */
 471                 if (lookup_decoration(&name_decoration, &commit->object))
 472                         return REV_TREE_DIFFERENT;
 473                 /*
 474                  * A commit that is not pointed by a tag is uninteresting
 475                  * if we are not limited by path.  This means that you will
 476                  * see the usual "commits that touch the paths" plus any
 477                  * tagged commit by specifying both --simplify-by-decoration
 478                  * and pathspec.
 479                  */
 480                 if (!revs->prune_data.nr)
 481                         return REV_TREE_SAME;
 482         }
 483
 484         tree_difference = REV_TREE_SAME;
 485         DIFF_OPT_CLR(&revs->pruning, HAS_CHANGES);
 486         if (diff_tree_sha1(t1->object.sha1, t2->object.sha1, "",
 487                            &revs->pruning) < 0)
 488                 return REV_TREE_DIFFERENT;
 489         return tree_difference;
 490 }
 491
 492 static int rev_same_tree_as_empty(struct rev_info *revs, struct commit *commit)
 493 {
 494         int retval;
 495         void *tree;
 496         unsigned long size;
 497         struct tree_desc empty, real;
 498         struct tree *t1 = commit->tree;
 499
 500         if (!t1)
 501                 return 0;
 502
 503         tree = read_object_with_reference(t1->object.sha1, tree_type, &size, NULL);
 504         if (!tree)
 505                 return 0;
 506         init_tree_desc(&real, tree, size);
 507         init_tree_desc(&empty, "", 0);
 508
 509         tree_difference = REV_TREE_SAME;
 510         DIFF_OPT_CLR(&revs->pruning, HAS_CHANGES);
 511         retval = diff_tree(&empty, &real, "", &revs->pruning);
 512         free(tree);
 513
 514         return retval >= 0 && (tree_difference == REV_TREE_SAME);
 515 }
 516
 517 struct treesame_state {
 518         unsigned int nparents;
 519         unsigned char treesame[FLEX_ARRAY];
 520 };
 521
 522 static struct treesame_state *initialise_treesame(struct rev_info *revs, struct commit *commit)
 523 {
 524         unsigned n = commit_list_count(commit->parents);
 525         struct treesame_state *st = xcalloc(1, sizeof(*st) + n);
 526         st->nparents = n;
 527         add_decoration(&revs->treesame, &commit->object, st);
 528         return st;
 529 }
 530
 531 /*
 532  * Must be called immediately after removing the nth_parent from a commit's
 533  * parent list, if we are maintaining the per-parent treesame[] decoration.
 534  * This does not recalculate the master TREESAME flag - update_treesame()
 535  * should be called to update it after a sequence of treesame[] modifications
 536  * that may have affected it.
 537  */
 538 static int compact_treesame(struct rev_info *revs, struct commit *commit, unsigned nth_parent)
 539 {
 540         struct treesame_state *st;
 541         int old_same;
 542
 543         if (!commit->parents) {
 544                 /*
 545                  * Have just removed the only parent from a non-merge.
 546                  * Different handling, as we lack decoration.
 547                  */
 548                 if (nth_parent != 0)
 549                         die("compact_treesame %u", nth_parent);
 550                 old_same = !!(commit->object.flags & TREESAME);
 551                 if (rev_same_tree_as_empty(revs, commit))
 552                         commit->object.flags |= TREESAME;
 553                 else
 554                         commit->object.flags &= ~TREESAME;
 555                 return old_same;
 556         }
 557
 558         st = lookup_decoration(&revs->treesame, &commit->object);
 559         if (!st || nth_parent >= st->nparents)
 560                 die("compact_treesame %u", nth_parent);
 561
 562         old_same = st->treesame[nth_parent];
 563         memmove(st->treesame + nth_parent,
 564                 st->treesame + nth_parent + 1,
 565                 st->nparents - nth_parent - 1);
 566
 567         /*
 568          * If we've just become a non-merge commit, update TREESAME
 569          * immediately, and remove the no-longer-needed decoration.
 570          * If still a merge, defer update until update_treesame().
 571          */
 572         if (--st->nparents == 1) {
 573                 if (commit->parents->next)
 574                         die("compact_treesame parents mismatch");
 575                 if (st->treesame[0] && revs->dense)
 576                         commit->object.flags |= TREESAME;
 577                 else
 578                         commit->object.flags &= ~TREESAME;
 579                 free(add_decoration(&revs->treesame, &commit->object, NULL));
 580         }
 581
 582         return old_same;
 583 }
 584
 585 static unsigned update_treesame(struct rev_info *revs, struct commit *commit)
 586 {
 587         if (commit->parents && commit->parents->next) {
 588                 unsigned n;
 589                 struct treesame_state *st;
 590                 struct commit_list *p;
 591                 unsigned relevant_parents;
 592                 unsigned relevant_change, irrelevant_change;
 593
 594                 st = lookup_decoration(&revs->treesame, &commit->object);
 595                 if (!st)
 596                         die("update_treesame %s", sha1_to_hex(commit->object.sha1));
 597                 relevant_parents = 0;
 598                 relevant_change = irrelevant_change = 0;
 599                 for (p = commit->parents, n = 0; p; n++, p = p->next) {
 600                         if (relevant_commit(p->item)) {
 601                                 relevant_change |= !st->treesame[n];
 602                                 relevant_parents++;
 603                         } else
 604                                 irrelevant_change |= !st->treesame[n];
 605                 }
 606                 if (relevant_parents ? relevant_change : irrelevant_change)
 607                         commit->object.flags &= ~TREESAME;
 608                 else
 609                         commit->object.flags |= TREESAME;
 610         }
 611
 612         return commit->object.flags & TREESAME;
 613 }
 614
 615 static inline int limiting_can_increase_treesame(const struct rev_info *revs)
 616 {
 617         /*
 618          * TREESAME is irrelevant unless prune && dense;
 619          * if simplify_history is set, we can't have a mixture of TREESAME and
 620          *    !TREESAME INTERESTING parents (and we don't have treesame[]
 621          *    decoration anyway);
 622          * if first_parent_only is set, then the TREESAME flag is locked
 623          *    against the first parent (and again we lack treesame[] decoration).
 624          */
 625         return revs->prune && revs->dense &&
 626                !revs->simplify_history &&
 627                !revs->first_parent_only;
 628 }
 629
 630 static void try_to_simplify_commit(struct rev_info *revs, struct commit *commit)
 631 {
 632         struct commit_list **pp, *parent;
 633         struct treesame_state *ts = NULL;
 634         int relevant_change = 0, irrelevant_change = 0;
 635         int relevant_parents, nth_parent;
 636
 637         /*
 638          * If we don't do pruning, everything is interesting
 639          */
 640         if (!revs->prune)
 641                 return;
 642
 643         if (!commit->tree)
 644                 return;
 645
 646         if (!commit->parents) {
 647                 if (rev_same_tree_as_empty(revs, commit))
 648                         commit->object.flags |= TREESAME;
 649                 return;
 650         }
 651
 652         /*
 653          * Normal non-merge commit? If we don't want to make the
 654          * history dense, we consider it always to be a change..
 655          */
 656         if (!revs->dense && !commit->parents->next)
 657                 return;
 658
 659         for (pp = &commit->parents, nth_parent = 0, relevant_parents = 0;
 660              (parent = *pp) != NULL;
 661              pp = &parent->next, nth_parent++) {
 662                 struct commit *p = parent->item;
 663                 if (relevant_commit(p))
 664                         relevant_parents++;
 665
 666                 if (nth_parent == 1) {
 667                         /*
 668                          * This our second loop iteration - so we now know
 669                          * we're dealing with a merge.
 670                          *
 671                          * Do not compare with later parents when we care only about
 672                          * the first parent chain, in order to avoid derailing the
 673                          * traversal to follow a side branch that brought everything
 674                          * in the path we are limited to by the pathspec.
 675                          */
 676                         if (revs->first_parent_only)
 677                                 break;
 678                         /*
 679                          * If this will remain a potentially-simplifiable
 680                          * merge, remember per-parent treesame if needed.
 681                          * Initialise the array with the comparison from our
 682                          * first iteration.
 683                          */
 684                         if (revs->treesame.name &&
 685                             !revs->simplify_history &&
 686                             !(commit->object.flags & UNINTERESTING)) {
 687                                 ts = initialise_treesame(revs, commit);
 688                                 if (!(irrelevant_change || relevant_change))
 689                                         ts->treesame[0] = 1;
 690                         }
 691                 }
 692                 if (parse_commit(p) < 0)
 693                         die("cannot simplify commit %s (because of %s)",
 694                             sha1_to_hex(commit->object.sha1),
 695                             sha1_to_hex(p->object.sha1));
 696                 switch (rev_compare_tree(revs, p, commit)) {
 697                 case REV_TREE_SAME:
 698                         if (!revs->simplify_history || !relevant_commit(p)) {
 699                                 /* Even if a merge with an uninteresting
 700                                  * side branch brought the entire change
 701                                  * we are interested in, we do not want
 702                                  * to lose the other branches of this
 703                                  * merge, so we just keep going.
 704                                  */
 705                                 if (ts)
 706                                         ts->treesame[nth_parent] = 1;
 707                                 continue;
 708                         }
 709                         parent->next = NULL;
 710                         commit->parents = parent;
 711                         commit->object.flags |= TREESAME;
 712                         return;
 713
 714                 case REV_TREE_NEW:
 715                         if (revs->remove_empty_trees &&
 716                             rev_same_tree_as_empty(revs, p)) {
 717                                 /* We are adding all the specified
 718                                  * paths from this parent, so the
 719                                  * history beyond this parent is not
 720                                  * interesting.  Remove its parents
 721                                  * (they are grandparents for us).
 722                                  * IOW, we pretend this parent is a
 723                                  * "root" commit.
 724                                  */
 725                                 if (parse_commit(p) < 0)
 726                                         die("cannot simplify commit %s (invalid %s)",
 727                                             sha1_to_hex(commit->object.sha1),
 728                                             sha1_to_hex(p->object.sha1));
 729                                 p->parents = NULL;
 730                         }
 731                 /* fallthrough */
 732                 case REV_TREE_OLD:
 733                 case REV_TREE_DIFFERENT:
 734                         if (relevant_commit(p))
 735                                 relevant_change = 1;
 736                         else
 737                                 irrelevant_change = 1;
 738                         continue;
 739                 }
 740                 die("bad tree compare for commit %s", sha1_to_hex(commit->object.sha1));
 741         }
 742
 743         /*
 744          * TREESAME is straightforward for single-parent commits. For merge
 745          * commits, it is most useful to define it so that "irrelevant"
 746          * parents cannot make us !TREESAME - if we have any relevant
 747          * parents, then we only consider TREESAMEness with respect to them,
 748          * allowing irrelevant merges from uninteresting branches to be
 749          * simplified away. Only if we have only irrelevant parents do we
 750          * base TREESAME on them. Note that this logic is replicated in
 751          * update_treesame, which should be kept in sync.
 752          */
 753         if (relevant_parents ? !relevant_change : !irrelevant_change)
 754                 commit->object.flags |= TREESAME;
 755 }
 756
 757 static void commit_list_insert_by_date_cached(struct commit *p, struct commit_list **head,
 758                     struct commit_list *cached_base, struct commit_list **cache)
 759 {
 760         struct commit_list *new_entry;
 761
 762         if (cached_base && p->date < cached_base->item->date)
 763                 new_entry = commit_list_insert_by_date(p, &cached_base->next);
 764         else
 765                 new_entry = commit_list_insert_by_date(p, head);
 766
 767         if (cache && (!*cache || p->date < (*cache)->item->date))
 768                 *cache = new_entry;
 769 }
 770
 771 static int add_parents_to_list(struct rev_info *revs, struct commit *commit,
 772                     struct commit_list **list, struct commit_list **cache_ptr)
 773 {
 774         struct commit_list *parent = commit->parents;
 775         unsigned left_flag;
 776         struct commit_list *cached_base = cache_ptr ? *cache_ptr : NULL;
 777
 778         if (commit->object.flags & ADDED)
 779                 return 0;
 780         commit->object.flags |= ADDED;
 781
 782         /*
 783          * If the commit is uninteresting, don't try to
 784          * prune parents - we want the maximal uninteresting
 785          * set.
 786          *
 787          * Normally we haven't parsed the parent
 788          * yet, so we won't have a parent of a parent
 789          * here. However, it may turn out that we've
 790          * reached this commit some other way (where it
 791          * wasn't uninteresting), in which case we need
 792          * to mark its parents recursively too..
 793          */
 794         if (commit->object.flags & UNINTERESTING) {
 795                 while (parent) {
 796                         struct commit *p = parent->item;
 797                         parent = parent->next;
 798                         if (p)
 799                                 p->object.flags |= UNINTERESTING;
 800                         if (parse_commit(p) < 0)
 801                                 continue;
 802                         if (p->parents)
 803                                 mark_parents_uninteresting(p);
 804                         if (p->object.flags & SEEN)
 805                                 continue;
 806                         p->object.flags |= SEEN;
 807                         commit_list_insert_by_date_cached(p, list, cached_base, cache_ptr);
 808                 }
 809                 return 0;
 810         }
 811
 812         /*
 813          * Ok, the commit wasn't uninteresting. Try to
 814          * simplify the commit history and find the parent
 815          * that has no differences in the path set if one exists.
 816          */
 817         try_to_simplify_commit(revs, commit);
 818
 819         if (revs->no_walk)
 820                 return 0;
 821
 822         left_flag = (commit->object.flags & SYMMETRIC_LEFT);
 823
 824         for (parent = commit->parents; parent; parent = parent->next) {
 825                 struct commit *p = parent->item;
 826
 827                 if (parse_commit(p) < 0)
 828                         return -1;
 829                 if (revs->show_source && !p->util)
 830                         p->util = commit->util;
 831                 p->object.flags |= left_flag;
 832                 if (!(p->object.flags & SEEN)) {
 833                         p->object.flags |= SEEN;
 834                         commit_list_insert_by_date_cached(p, list, cached_base, cache_ptr);
 835                 }
 836                 if (revs->first_parent_only)
 837                         break;
 838         }
 839         return 0;
 840 }
 841
 842 static void cherry_pick_list(struct commit_list *list, struct rev_info *revs)
 843 {
 844         struct commit_list *p;
 845         int left_count = 0, right_count = 0;
 846         int left_first;
 847         struct patch_ids ids;
 848         unsigned cherry_flag;
 849
 850         /* First count the commits on the left and on the right */
 851         for (p = list; p; p = p->next) {
 852                 struct commit *commit = p->item;
 853                 unsigned flags = commit->object.flags;
 854                 if (flags & BOUNDARY)
 855                         ;
 856                 else if (flags & SYMMETRIC_LEFT)
 857                         left_count++;
 858                 else
 859                         right_count++;
 860         }
 861
 862         if (!left_count || !right_count)
 863                 return;
 864
 865         left_first = left_count < right_count;
 866         init_patch_ids(&ids);
 867         ids.diffopts.pathspec = revs->diffopt.pathspec;
 868
 869         /* Compute patch-ids for one side */
 870         for (p = list; p; p = p->next) {
 871                 struct commit *commit = p->item;
 872                 unsigned flags = commit->object.flags;
 873
 874                 if (flags & BOUNDARY)
 875                         continue;
 876                 /*
 877                  * If we have fewer left, left_first is set and we omit
 878                  * commits on the right branch in this loop.  If we have
 879                  * fewer right, we skip the left ones.
 880                  */
 881                 if (left_first != !!(flags & SYMMETRIC_LEFT))
 882                         continue;
 883                 commit->util = add_commit_patch_id(commit, &ids);
 884         }
 885
 886         /* either cherry_mark or cherry_pick are true */
 887         cherry_flag = revs->cherry_mark ? PATCHSAME : SHOWN;
 888
 889         /* Check the other side */
 890         for (p = list; p; p = p->next) {
 891                 struct commit *commit = p->item;
 892                 struct patch_id *id;
 893                 unsigned flags = commit->object.flags;
 894
 895                 if (flags & BOUNDARY)
 896                         continue;
 897                 /*
 898                  * If we have fewer left, left_first is set and we omit
 899                  * commits on the left branch in this loop.
 900                  */
 901                 if (left_first == !!(flags & SYMMETRIC_LEFT))
 902                         continue;
 903
 904                 /*
 905                  * Have we seen the same patch id?
 906                  */
 907                 id = has_commit_patch_id(commit, &ids);
 908                 if (!id)
 909                         continue;
 910                 id->seen = 1;
 911                 commit->object.flags |= cherry_flag;
 912         }
 913
 914         /* Now check the original side for seen ones */
 915         for (p = list; p; p = p->next) {
 916                 struct commit *commit = p->item;
 917                 struct patch_id *ent;
 918
 919                 ent = commit->util;
 920                 if (!ent)
 921                         continue;
 922                 if (ent->seen)
 923                         commit->object.flags |= cherry_flag;
 924                 commit->util = NULL;
 925         }
 926
 927         free_patch_ids(&ids);
 928 }
 929
 930 /* How many extra uninteresting commits we want to see.. */
 931 #define SLOP 5
 932
 933 static int still_interesting(struct commit_list *src, unsigned long date, int slop)
 934 {
 935         /*
 936          * No source list at all? We're definitely done..
 937          */
 938         if (!src)
 939                 return 0;
 940
 941         /*
 942          * Does the destination list contain entries with a date
 943          * before the source list? Definitely _not_ done.
 944          */
 945         if (date <= src->item->date)
 946                 return SLOP;
 947
 948         /*
 949          * Does the source list still have interesting commits in
 950          * it? Definitely not done..
 951          */
 952         if (!everybody_uninteresting(src))
 953                 return SLOP;
 954
 955         /* Ok, we're closing in.. */
 956         return slop-1;
 957 }
 958
 959 /*
 960  * "rev-list --ancestry-path A..B" computes commits that are ancestors
 961  * of B but not ancestors of A but further limits the result to those
 962  * that are descendants of A.  This takes the list of bottom commits and
 963  * the result of "A..B" without --ancestry-path, and limits the latter
 964  * further to the ones that can reach one of the commits in "bottom".
 965  */
 966 static void limit_to_ancestry(struct commit_list *bottom, struct commit_list *list)
 967 {
 968         struct commit_list *p;
 969         struct commit_list *rlist = NULL;
 970         int made_progress;
 971
 972         /*
 973          * Reverse the list so that it will be likely that we would
 974          * process parents before children.
 975          */
 976         for (p = list; p; p = p->next)
 977                 commit_list_insert(p->item, &rlist);
 978
 979         for (p = bottom; p; p = p->next)
 980                 p->item->object.flags |= TMP_MARK;
 981
 982         /*
 983          * Mark the ones that can reach bottom commits in "list",
 984          * in a bottom-up fashion.
 985          */
 986         do {
 987                 made_progress = 0;
 988                 for (p = rlist; p; p = p->next) {
 989                         struct commit *c = p->item;
 990                         struct commit_list *parents;
 991                         if (c->object.flags & (TMP_MARK | UNINTERESTING))
 992                                 continue;
 993                         for (parents = c->parents;
 994                              parents;
 995                              parents = parents->next) {
 996                                 if (!(parents->item->object.flags & TMP_MARK))
 997                                         continue;
 998                                 c->object.flags |= TMP_MARK;
 999                                 made_progress = 1;
1000                                 break;
1001                         }
1002                 }
1003         } while (made_progress);
1004
1005         /*
1006          * NEEDSWORK: decide if we want to remove parents that are
1007          * not marked with TMP_MARK from commit->parents for commits
1008          * in the resulting list.  We may not want to do that, though.
1009          */
1010
1011         /*
1012          * The ones that are not marked with TMP_MARK are uninteresting
1013          */
1014         for (p = list; p; p = p->next) {
1015                 struct commit *c = p->item;
1016                 if (c->object.flags & TMP_MARK)
1017                         continue;
1018                 c->object.flags |= UNINTERESTING;
1019         }
1020
1021         /* We are done with the TMP_MARK */
1022         for (p = list; p; p = p->next)
1023                 p->item->object.flags &= ~TMP_MARK;
1024         for (p = bottom; p; p = p->next)
1025                 p->item->object.flags &= ~TMP_MARK;
1026         free_commit_list(rlist);
1027 }
1028
1029 /*
1030  * Before walking the history, keep the set of "negative" refs the
1031  * caller has asked to exclude.
1032  *
1033  * This is used to compute "rev-list --ancestry-path A..B", as we need
1034  * to filter the result of "A..B" further to the ones that can actually
1035  * reach A.
1036  */
1037 static struct commit_list *collect_bottom_commits(struct commit_list *list)
1038 {
1039         struct commit_list *elem, *bottom = NULL;
1040         for (elem = list; elem; elem = elem->next)
1041                 if (elem->item->object.flags & BOTTOM)
1042                         commit_list_insert(elem->item, &bottom);
1043         return bottom;
1044 }
1045
1046 /* Assumes either left_only or right_only is set */
1047 static void limit_left_right(struct commit_list *list, struct rev_info *revs)
1048 {
1049         struct commit_list *p;
1050
1051         for (p = list; p; p = p->next) {
1052                 struct commit *commit = p->item;
1053
1054                 if (revs->right_only) {
1055                         if (commit->object.flags & SYMMETRIC_LEFT)
1056                                 commit->object.flags |= SHOWN;
1057                 } else  /* revs->left_only is set */
1058                         if (!(commit->object.flags & SYMMETRIC_LEFT))
1059                                 commit->object.flags |= SHOWN;
1060         }
1061 }
1062
1063 static int limit_list(struct rev_info *revs)
1064 {
1065         int slop = SLOP;
1066         unsigned long date = ~0ul;
1067         struct commit_list *list = revs->commits;
1068         struct commit_list *newlist = NULL;
1069         struct commit_list **p = &newlist;
1070         struct commit_list *bottom = NULL;
1071
1072         if (revs->ancestry_path) {
1073                 bottom = collect_bottom_commits(list);
1074                 if (!bottom)
1075                         die("--ancestry-path given but there are no bottom commits");
1076         }
1077
1078         while (list) {
1079                 struct commit_list *entry = list;
1080                 struct commit *commit = list->item;
1081                 struct object *obj = &commit->object;
1082                 show_early_output_fn_t show;
1083
1084                 list = list->next;
1085                 free(entry);
1086
1087                 if (revs->max_age != -1 && (commit->date < revs->max_age))
1088                         obj->flags |= UNINTERESTING;
1089                 if (add_parents_to_list(revs, commit, &list, NULL) < 0)
1090                         return -1;
1091                 if (obj->flags & UNINTERESTING) {
1092                         mark_parents_uninteresting(commit);
1093                         if (revs->show_all)
1094                                 p = &commit_list_insert(commit, p)->next;
1095                         slop = still_interesting(list, date, slop);
1096                         if (slop)
1097                                 continue;
1098                         /* If showing all, add the whole pending list to the end */
1099                         if (revs->show_all)
1100                                 *p = list;
1101                         break;
1102                 }
1103                 if (revs->min_age != -1 && (commit->date > revs->min_age))
1104                         continue;
1105                 date = commit->date;
1106                 p = &commit_list_insert(commit, p)->next;
1107
1108                 show = show_early_output;
1109                 if (!show)
1110                         continue;
1111
1112                 show(revs, newlist);
1113                 show_early_output = NULL;
1114         }
1115         if (revs->cherry_pick || revs->cherry_mark)
1116                 cherry_pick_list(newlist, revs);
1117
1118         if (revs->left_only || revs->right_only)
1119                 limit_left_right(newlist, revs);
1120
1121         if (bottom) {
1122                 limit_to_ancestry(bottom, newlist);
1123                 free_commit_list(bottom);
1124         }
1125
1126         /*
1127          * Check if any commits have become TREESAME by some of their parents
1128          * becoming UNINTERESTING.
1129          */
1130         if (limiting_can_increase_treesame(revs))
1131                 for (list = newlist; list; list = list->next) {
1132                         struct commit *c = list->item;
1133                         if (c->object.flags & (UNINTERESTING | TREESAME))
1134                                 continue;
1135                         update_treesame(revs, c);
1136                 }
1137
1138         revs->commits = newlist;
1139         return 0;
1140 }
1141
1142 /*
1143  * Add an entry to refs->cmdline with the specified information.
1144  * *name is copied.
1145  */
1146 static void add_rev_cmdline(struct rev_info *revs,
1147                             struct object *item,
1148                             const char *name,
1149                             int whence,
1150                             unsigned flags)
1151 {
1152         struct rev_cmdline_info *info = &revs->cmdline;
1153         int nr = info->nr;
1154
1155         ALLOC_GROW(info->rev, nr + 1, info->alloc);
1156         info->rev[nr].item = item;
1157         info->rev[nr].name = xstrdup(name);
1158         info->rev[nr].whence = whence;
1159         info->rev[nr].flags = flags;
1160         info->nr++;
1161 }
1162
1163 static void add_rev_cmdline_list(struct rev_info *revs,
1164                                  struct commit_list *commit_list,
1165                                  int whence,
1166                                  unsigned flags)
1167 {
1168         while (commit_list) {
1169                 struct object *object = &commit_list->item->object;
1170                 add_rev_cmdline(revs, object, sha1_to_hex(object->sha1),
1171                                 whence, flags);
1172                 commit_list = commit_list->next;
1173         }
1174 }
1175
1176 struct all_refs_cb {
1177         int all_flags;
1178         int warned_bad_reflog;
1179         struct rev_info *all_revs;
1180         const char *name_for_errormsg;
1181 };
1182
1183 int ref_excluded(struct string_list *ref_excludes, const char *path)
1184 {
1185         struct string_list_item *item;
1186
1187         if (!ref_excludes)
1188                 return 0;
1189         for_each_string_list_item(item, ref_excludes) {
1190                 if (!fnmatch(item->string, path, 0))
1191                         return 1;
1192         }
1193         return 0;
1194 }
1195
1196 static int handle_one_ref(const char *path, const unsigned char *sha1, int flag, void *cb_data)
1197 {
1198         struct all_refs_cb *cb = cb_data;
1199         struct object *object;
1200
1201         if (ref_excluded(cb->all_revs->ref_excludes, path))
1202             return 0;
1203
1204         object = get_reference(cb->all_revs, path, sha1, cb->all_flags);
1205         add_rev_cmdline(cb->all_revs, object, path, REV_CMD_REF, cb->all_flags);
1206         add_pending_sha1(cb->all_revs, path, sha1, cb->all_flags);
1207         return 0;
1208 }
1209
1210 static void init_all_refs_cb(struct all_refs_cb *cb, struct rev_info *revs,
1211         unsigned flags)
1212 {
1213         cb->all_revs = revs;
1214         cb->all_flags = flags;
1215 }
1216
1217 void clear_ref_exclusion(struct string_list **ref_excludes_p)
1218 {
1219         if (*ref_excludes_p) {
1220                 string_list_clear(*ref_excludes_p, 0);
1221                 free(*ref_excludes_p);
1222         }
1223         *ref_excludes_p = NULL;
1224 }
1225
1226 void add_ref_exclusion(struct string_list **ref_excludes_p, const char *exclude)
1227 {
1228         if (!*ref_excludes_p) {
1229                 *ref_excludes_p = xcalloc(1, sizeof(**ref_excludes_p));
1230                 (*ref_excludes_p)->strdup_strings = 1;
1231         }
1232         string_list_append(*ref_excludes_p, exclude);
1233 }
1234
1235 static void handle_refs(const char *submodule, struct rev_info *revs, unsigned flags,
1236                 int (*for_each)(const char *, each_ref_fn, void *))
1237 {
1238         struct all_refs_cb cb;
1239         init_all_refs_cb(&cb, revs, flags);
1240         for_each(submodule, handle_one_ref, &cb);
1241 }
1242
1243 static void handle_one_reflog_commit(unsigned char *sha1, void *cb_data)
1244 {
1245         struct all_refs_cb *cb = cb_data;
1246         if (!is_null_sha1(sha1)) {
1247                 struct object *o = parse_object(sha1);
1248                 if (o) {
1249                         o->flags |= cb->all_flags;
1250                         /* ??? CMDLINEFLAGS ??? */
1251                         add_pending_object(cb->all_revs, o, "");
1252                 }
1253                 else if (!cb->warned_bad_reflog) {
1254                         warning("reflog of '%s' references pruned commits",
1255                                 cb->name_for_errormsg);
1256                         cb->warned_bad_reflog = 1;
1257                 }
1258         }
1259 }
1260
1261 static int handle_one_reflog_ent(unsigned char *osha1, unsigned char *nsha1,
1262                 const char *email, unsigned long timestamp, int tz,
1263                 const char *message, void *cb_data)
1264 {
1265         handle_one_reflog_commit(osha1, cb_data);
1266         handle_one_reflog_commit(nsha1, cb_data);
1267         return 0;
1268 }
1269
1270 static int handle_one_reflog(const char *path, const unsigned char *sha1, int flag, void *cb_data)
1271 {
1272         struct all_refs_cb *cb = cb_data;
1273         cb->warned_bad_reflog = 0;
1274         cb->name_for_errormsg = path;
1275         for_each_reflog_ent(path, handle_one_reflog_ent, cb_data);
1276         return 0;
1277 }
1278
1279 static void handle_reflog(struct rev_info *revs, unsigned flags)
1280 {
1281         struct all_refs_cb cb;
1282         cb.all_revs = revs;
1283         cb.all_flags = flags;
1284         for_each_reflog(handle_one_reflog, &cb);
1285 }
1286
1287 static int add_parents_only(struct rev_info *revs, const char *arg_, int flags)
1288 {
1289         unsigned char sha1[20];
1290         struct object *it;
1291         struct commit *commit;
1292         struct commit_list *parents;
1293         const char *arg = arg_;
1294
1295         if (*arg == '^') {
1296                 flags ^= UNINTERESTING | BOTTOM;
1297                 arg++;
1298         }
1299         if (get_sha1_committish(arg, sha1))
1300                 return 0;
1301         while (1) {
1302                 it = get_reference(revs, arg, sha1, 0);
1303                 if (!it && revs->ignore_missing)
1304                         return 0;
1305                 if (it->type != OBJ_TAG)
1306                         break;
1307                 if (!((struct tag*)it)->tagged)
1308                         return 0;
1309                 hashcpy(sha1, ((struct tag*)it)->tagged->sha1);
1310         }
1311         if (it->type != OBJ_COMMIT)
1312                 return 0;
1313         commit = (struct commit *)it;
1314         for (parents = commit->parents; parents; parents = parents->next) {
1315                 it = &parents->item->object;
1316                 it->flags |= flags;
1317                 add_rev_cmdline(revs, it, arg_, REV_CMD_PARENTS_ONLY, flags);
1318                 add_pending_object(revs, it, arg);
1319         }
1320         return 1;
1321 }
1322
1323 void init_revisions(struct rev_info *revs, const char *prefix)
1324 {
1325         memset(revs, 0, sizeof(*revs));
1326
1327         revs->abbrev = DEFAULT_ABBREV;
1328         revs->ignore_merges = 1;
1329         revs->simplify_history = 1;
1330         DIFF_OPT_SET(&revs->pruning, RECURSIVE);
1331         DIFF_OPT_SET(&revs->pruning, QUICK);
1332         revs->pruning.add_remove = file_add_remove;
1333         revs->pruning.change = file_change;
1334         revs->sort_order = REV_SORT_IN_GRAPH_ORDER;
1335         revs->dense = 1;
1336         revs->prefix = prefix;
1337         revs->max_age = -1;
1338         revs->min_age = -1;
1339         revs->skip_count = -1;
1340         revs->max_count = -1;
1341         revs->max_parents = -1;
1342
1343         revs->commit_format = CMIT_FMT_DEFAULT;
1344
1345         init_grep_defaults();
1346         grep_init(&revs->grep_filter, prefix);
1347         revs->grep_filter.status_only = 1;
1348         revs->grep_filter.regflags = REG_NEWLINE;
1349
1350         diff_setup(&revs->diffopt);
1351         if (prefix && !revs->diffopt.prefix) {
1352                 revs->diffopt.prefix = prefix;
1353                 revs->diffopt.prefix_length = strlen(prefix);
1354         }
1355
1356         revs->notes_opt.use_default_notes = -1;
1357 }
1358
1359 static void add_pending_commit_list(struct rev_info *revs,
1360                                     struct commit_list *commit_list,
1361                                     unsigned int flags)
1362 {
1363         while (commit_list) {
1364                 struct object *object = &commit_list->item->object;
1365                 object->flags |= flags;
1366                 add_pending_object(revs, object, sha1_to_hex(object->sha1));
1367                 commit_list = commit_list->next;
1368         }
1369 }
1370
1371 static void prepare_show_merge(struct rev_info *revs)
1372 {
1373         struct commit_list *bases;
1374         struct commit *head, *other;
1375         unsigned char sha1[20];
1376         const char **prune = NULL;
1377         int i, prune_num = 1; /* counting terminating NULL */
1378
1379         if (get_sha1("HEAD", sha1))
1380                 die("--merge without HEAD?");
1381         head = lookup_commit_or_die(sha1, "HEAD");
1382         if (get_sha1("MERGE_HEAD", sha1))
1383                 die("--merge without MERGE_HEAD?");
1384         other = lookup_commit_or_die(sha1, "MERGE_HEAD");
1385         add_pending_object(revs, &head->object, "HEAD");
1386         add_pending_object(revs, &other->object, "MERGE_HEAD");
1387         bases = get_merge_bases(head, other, 1);
1388         add_rev_cmdline_list(revs, bases, REV_CMD_MERGE_BASE, UNINTERESTING | BOTTOM);
1389         add_pending_commit_list(revs, bases, UNINTERESTING | BOTTOM);
1390         free_commit_list(bases);
1391         head->object.flags |= SYMMETRIC_LEFT;
1392
1393         if (!active_nr)
1394                 read_cache();
1395         for (i = 0; i < active_nr; i++) {
1396                 const struct cache_entry *ce = active_cache[i];
1397                 if (!ce_stage(ce))
1398                         continue;
1399                 if (ce_path_match(ce, &revs->prune_data)) {
1400                         prune_num++;
1401                         prune = xrealloc(prune, sizeof(*prune) * prune_num);
1402                         prune[prune_num-2] = ce->name;
1403                         prune[prune_num-1] = NULL;
1404                 }
1405                 while ((i+1 < active_nr) &&
1406                        ce_same_name(ce, active_cache[i+1]))
1407                         i++;
1408         }
1409         free_pathspec(&revs->prune_data);
1410         parse_pathspec(&revs->prune_data, PATHSPEC_ALL_MAGIC & ~PATHSPEC_LITERAL,
1411                        PATHSPEC_PREFER_FULL | PATHSPEC_LITERAL_PATH, "", prune);
1412         revs->limited = 1;
1413 }
1414
1415 int handle_revision_arg(const char *arg_, struct rev_info *revs, int flags, unsigned revarg_opt)
1416 {
1417         struct object_context oc;
1418         char *dotdot;
1419         struct object *object;
1420         unsigned char sha1[20];
1421         int local_flags;
1422         const char *arg = arg_;
1423         int cant_be_filename = revarg_opt & REVARG_CANNOT_BE_FILENAME;
1424         unsigned get_sha1_flags = 0;
1425
1426         flags = flags & UNINTERESTING ? flags | BOTTOM : flags & ~BOTTOM;
1427
1428         dotdot = strstr(arg, "..");
1429         if (dotdot) {
1430                 unsigned char from_sha1[20];
1431                 const char *next = dotdot + 2;
1432                 const char *this = arg;
1433                 int symmetric = *next == '.';
1434                 unsigned int flags_exclude = flags ^ (UNINTERESTING | BOTTOM);
1435                 static const char head_by_default[] = "HEAD";
1436                 unsigned int a_flags;
1437
1438                 *dotdot = 0;
1439                 next += symmetric;
1440
1441                 if (!*next)
1442                         next = head_by_default;
1443                 if (dotdot == arg)
1444                         this = head_by_default;
1445                 if (this == head_by_default && next == head_by_default &&
1446                     !symmetric) {
1447                         /*
1448                          * Just ".."?  That is not a range but the
1449                          * pathspec for the parent directory.
1450                          */
1451                         if (!cant_be_filename) {
1452                                 *dotdot = '.';
1453                                 return -1;
1454                         }
1455                 }
1456                 if (!get_sha1_committish(this, from_sha1) &&
1457                     !get_sha1_committish(next, sha1)) {
1458                         struct object *a_obj, *b_obj;
1459
1460                         if (!cant_be_filename) {
1461                                 *dotdot = '.';
1462                                 verify_non_filename(revs->prefix, arg);
1463                         }
1464
1465                         a_obj = parse_object(from_sha1);
1466                         b_obj = parse_object(sha1);
1467                         if (!a_obj || !b_obj) {
1468                         missing:
1469                                 if (revs->ignore_missing)
1470                                         return 0;
1471                                 die(symmetric
1472                                     ? "Invalid symmetric difference expression %s"
1473                                     : "Invalid revision range %s", arg);
1474                         }
1475
1476                         if (!symmetric) {
1477                                 /* just A..B */
1478                                 a_flags = flags_exclude;
1479                         } else {
1480                                 /* A...B -- find merge bases between the two */
1481                                 struct commit *a, *b;
1482                                 struct commit_list *exclude;
1483
1484                                 a = (a_obj->type == OBJ_COMMIT
1485                                      ? (struct commit *)a_obj
1486                                      : lookup_commit_reference(a_obj->sha1));
1487                                 b = (b_obj->type == OBJ_COMMIT
1488                                      ? (struct commit *)b_obj
1489                                      : lookup_commit_reference(b_obj->sha1));
1490                                 if (!a || !b)
1491                                         goto missing;
1492                                 exclude = get_merge_bases(a, b, 1);
1493                                 add_rev_cmdline_list(revs, exclude,
1494                                                      REV_CMD_MERGE_BASE,
1495                                                      flags_exclude);
1496                                 add_pending_commit_list(revs, exclude,
1497                                                         flags_exclude);
1498                                 free_commit_list(exclude);
1499
1500                                 a_flags = flags | SYMMETRIC_LEFT;
1501                         }
1502
1503                         a_obj->flags |= a_flags;
1504                         b_obj->flags |= flags;
1505                         add_rev_cmdline(revs, a_obj, this,
1506                                         REV_CMD_LEFT, a_flags);
1507                         add_rev_cmdline(revs, b_obj, next,
1508                                         REV_CMD_RIGHT, flags);
1509                         add_pending_object(revs, a_obj, this);
1510                         add_pending_object(revs, b_obj, next);
1511                         return 0;
1512                 }
1513                 *dotdot = '.';
1514         }
1515         dotdot = strstr(arg, "^@");
1516         if (dotdot && !dotdot[2]) {
1517                 *dotdot = 0;
1518                 if (add_parents_only(revs, arg, flags))
1519                         return 0;
1520                 *dotdot = '^';
1521         }
1522         dotdot = strstr(arg, "^!");
1523         if (dotdot && !dotdot[2]) {
1524                 *dotdot = 0;
1525                 if (!add_parents_only(revs, arg, flags ^ (UNINTERESTING | BOTTOM)))
1526                         *dotdot = '^';
1527         }
1528
1529         local_flags = 0;
1530         if (*arg == '^') {
1531                 local_flags = UNINTERESTING | BOTTOM;
1532                 arg++;
1533         }
1534
1535         if (revarg_opt & REVARG_COMMITTISH)
1536                 get_sha1_flags = GET_SHA1_COMMITTISH;
1537
1538         if (get_sha1_with_context(arg, get_sha1_flags, sha1, &oc))
1539                 return revs->ignore_missing ? 0 : -1;
1540         if (!cant_be_filename)
1541                 verify_non_filename(revs->prefix, arg);
1542         object = get_reference(revs, arg, sha1, flags ^ local_flags);
1543         add_rev_cmdline(revs, object, arg_, REV_CMD_REV, flags ^ local_flags);
1544         add_pending_object_with_mode(revs, object, arg, oc.mode);
1545         return 0;
1546 }
1547
1548 struct cmdline_pathspec {
1549         int alloc;
1550         int nr;
1551         const char **path;
1552 };
1553
1554 static void append_prune_data(struct cmdline_pathspec *prune, const char **av)
1555 {
1556         while (*av) {
1557                 ALLOC_GROW(prune->path, prune->nr + 1, prune->alloc);
1558                 prune->path[prune->nr++] = *(av++);
1559         }
1560 }
1561
1562 static void read_pathspec_from_stdin(struct rev_info *revs, struct strbuf *sb,
1563                                      struct cmdline_pathspec *prune)
1564 {
1565         while (strbuf_getwholeline(sb, stdin, '\n') != EOF) {
1566                 int len = sb->len;
1567                 if (len && sb->buf[len - 1] == '\n')
1568                         sb->buf[--len] = '\0';
1569                 ALLOC_GROW(prune->path, prune->nr + 1, prune->alloc);
1570                 prune->path[prune->nr++] = xstrdup(sb->buf);
1571         }
1572 }
1573
1574 static void read_revisions_from_stdin(struct rev_info *revs,
1575                                       struct cmdline_pathspec *prune)
1576 {
1577         struct strbuf sb;
1578         int seen_dashdash = 0;
1579
1580         strbuf_init(&sb, 1000);
1581         while (strbuf_getwholeline(&sb, stdin, '\n') != EOF) {
1582                 int len = sb.len;
1583                 if (len && sb.buf[len - 1] == '\n')
1584                         sb.buf[--len] = '\0';
1585                 if (!len)
1586                         break;
1587                 if (sb.buf[0] == '-') {
1588                         if (len == 2 && sb.buf[1] == '-') {
1589                                 seen_dashdash = 1;
1590                                 break;
1591                         }
1592                         die("options not supported in --stdin mode");
1593                 }
1594                 if (handle_revision_arg(sb.buf, revs, 0,
1595                                         REVARG_CANNOT_BE_FILENAME))
1596                         die("bad revision '%s'", sb.buf);
1597         }
1598         if (seen_dashdash)
1599                 read_pathspec_from_stdin(revs, &sb, prune);
1600         strbuf_release(&sb);
1601 }
1602
1603 static void add_grep(struct rev_info *revs, const char *ptn, enum grep_pat_token what)
1604 {
1605         append_grep_pattern(&revs->grep_filter, ptn, "command line", 0, what);
1606 }
1607
1608 static void add_header_grep(struct rev_info *revs, enum grep_header_field field, const char *pattern)
1609 {
1610         append_header_grep_pattern(&revs->grep_filter, field, pattern);
1611 }
1612
1613 static void add_message_grep(struct rev_info *revs, const char *pattern)
1614 {
1615         add_grep(revs, pattern, GREP_PATTERN_BODY);
1616 }
1617
1618 static int handle_revision_opt(struct rev_info *revs, int argc, const char **argv,
1619                                int *unkc, const char **unkv)
1620 {
1621         const char *arg = argv[0];
1622         const char *optarg;
1623         int argcount;
1624
1625         /* pseudo revision arguments */
1626         if (!strcmp(arg, "--all") || !strcmp(arg, "--branches") ||
1627             !strcmp(arg, "--tags") || !strcmp(arg, "--remotes") ||
1628             !strcmp(arg, "--reflog") || !strcmp(arg, "--not") ||
1629             !strcmp(arg, "--no-walk") || !strcmp(arg, "--do-walk") ||
1630             !strcmp(arg, "--bisect") || !prefixcmp(arg, "--glob=") ||
1631             !prefixcmp(arg, "--branches=") || !prefixcmp(arg, "--tags=") ||
1632             !prefixcmp(arg, "--remotes=") || !prefixcmp(arg, "--no-walk="))
1633         {
1634                 unkv[(*unkc)++] = arg;
1635                 return 1;
1636         }
1637
1638         if ((argcount = parse_long_opt("max-count", argv, &optarg))) {
1639                 revs->max_count = atoi(optarg);
1640                 revs->no_walk = 0;
1641                 return argcount;
1642         } else if ((argcount = parse_long_opt("skip", argv, &optarg))) {
1643                 revs->skip_count = atoi(optarg);
1644                 return argcount;
1645         } else if ((*arg == '-') && isdigit(arg[1])) {
1646         /* accept -<digit>, like traditional "head" */
1647                 revs->max_count = atoi(arg + 1);
1648                 revs->no_walk = 0;
1649         } else if (!strcmp(arg, "-n")) {
1650                 if (argc <= 1)
1651                         return error("-n requires an argument");
1652                 revs->max_count = atoi(argv[1]);
1653                 revs->no_walk = 0;
1654                 return 2;
1655         } else if (!prefixcmp(arg, "-n")) {
1656                 revs->max_count = atoi(arg + 2);
1657                 revs->no_walk = 0;
1658         } else if ((argcount = parse_long_opt("max-age", argv, &optarg))) {
1659                 revs->max_age = atoi(optarg);
1660                 return argcount;
1661         } else if ((argcount = parse_long_opt("since", argv, &optarg))) {
1662                 revs->max_age = approxidate(optarg);
1663                 return argcount;
1664         } else if ((argcount = parse_long_opt("after", argv, &optarg))) {
1665                 revs->max_age = approxidate(optarg);
1666                 return argcount;
1667         } else if ((argcount = parse_long_opt("min-age", argv, &optarg))) {
1668                 revs->min_age = atoi(optarg);
1669                 return argcount;
1670         } else if ((argcount = parse_long_opt("before", argv, &optarg))) {
1671                 revs->min_age = approxidate(optarg);
1672                 return argcount;
1673         } else if ((argcount = parse_long_opt("until", argv, &optarg))) {
1674                 revs->min_age = approxidate(optarg);
1675                 return argcount;
1676         } else if (!strcmp(arg, "--first-parent")) {
1677                 revs->first_parent_only = 1;
1678         } else if (!strcmp(arg, "--ancestry-path")) {
1679                 revs->ancestry_path = 1;
1680                 revs->simplify_history = 0;
1681                 revs->limited = 1;
1682         } else if (!strcmp(arg, "-g") || !strcmp(arg, "--walk-reflogs")) {
1683                 init_reflog_walk(&revs->reflog_info);
1684         } else if (!strcmp(arg, "--default")) {
1685                 if (argc <= 1)
1686                         return error("bad --default argument");
1687                 revs->def = argv[1];
1688                 return 2;
1689         } else if (!strcmp(arg, "--merge")) {
1690                 revs->show_merge = 1;
1691         } else if (!strcmp(arg, "--topo-order")) {
1692                 revs->sort_order = REV_SORT_IN_GRAPH_ORDER;
1693                 revs->topo_order = 1;
1694         } else if (!strcmp(arg, "--simplify-merges")) {
1695                 revs->simplify_merges = 1;
1696                 revs->topo_order = 1;
1697                 revs->rewrite_parents = 1;
1698                 revs->simplify_history = 0;
1699                 revs->limited = 1;
1700         } else if (!strcmp(arg, "--simplify-by-decoration")) {
1701                 revs->simplify_merges = 1;
1702                 revs->topo_order = 1;
1703                 revs->rewrite_parents = 1;
1704                 revs->simplify_history = 0;
1705                 revs->simplify_by_decoration = 1;
1706                 revs->limited = 1;
1707                 revs->prune = 1;
1708                 load_ref_decorations(DECORATE_SHORT_REFS);
1709         } else if (!strcmp(arg, "--date-order")) {
1710                 revs->sort_order = REV_SORT_BY_COMMIT_DATE;
1711                 revs->topo_order = 1;
1712         } else if (!strcmp(arg, "--author-date-order")) {
1713                 revs->sort_order = REV_SORT_BY_AUTHOR_DATE;
1714                 revs->topo_order = 1;
1715         } else if (!prefixcmp(arg, "--early-output")) {
1716                 int count = 100;
1717                 switch (arg[14]) {
1718                 case '=':
1719                         count = atoi(arg+15);
1720                         /* Fallthrough */
1721                 case 0:
1722                         revs->topo_order = 1;
1723                        revs->early_output = count;
1724                 }
1725         } else if (!strcmp(arg, "--parents")) {
1726                 revs->rewrite_parents = 1;
1727                 revs->print_parents = 1;
1728         } else if (!strcmp(arg, "--dense")) {
1729                 revs->dense = 1;
1730         } else if (!strcmp(arg, "--sparse")) {
1731                 revs->dense = 0;
1732         } else if (!strcmp(arg, "--show-all")) {
1733                 revs->show_all = 1;
1734         } else if (!strcmp(arg, "--remove-empty")) {
1735                 revs->remove_empty_trees = 1;
1736         } else if (!strcmp(arg, "--merges")) {
1737                 revs->min_parents = 2;
1738         } else if (!strcmp(arg, "--no-merges")) {
1739                 revs->max_parents = 1;
1740         } else if (!prefixcmp(arg, "--min-parents=")) {
1741                 revs->min_parents = atoi(arg+14);
1742         } else if (!prefixcmp(arg, "--no-min-parents")) {
1743                 revs->min_parents = 0;
1744         } else if (!prefixcmp(arg, "--max-parents=")) {
1745                 revs->max_parents = atoi(arg+14);
1746         } else if (!prefixcmp(arg, "--no-max-parents")) {
1747                 revs->max_parents = -1;
1748         } else if (!strcmp(arg, "--boundary")) {
1749                 revs->boundary = 1;
1750         } else if (!strcmp(arg, "--left-right")) {
1751                 revs->left_right = 1;
1752         } else if (!strcmp(arg, "--left-only")) {
1753                 if (revs->right_only)
1754                         die("--left-only is incompatible with --right-only"
1755                             " or --cherry");
1756                 revs->left_only = 1;
1757         } else if (!strcmp(arg, "--right-only")) {
1758                 if (revs->left_only)
1759                         die("--right-only is incompatible with --left-only");
1760                 revs->right_only = 1;
1761         } else if (!strcmp(arg, "--cherry")) {
1762                 if (revs->left_only)
1763                         die("--cherry is incompatible with --left-only");
1764                 revs->cherry_mark = 1;
1765                 revs->right_only = 1;
1766                 revs->max_parents = 1;
1767                 revs->limited = 1;
1768         } else if (!strcmp(arg, "--count")) {
1769                 revs->count = 1;
1770         } else if (!strcmp(arg, "--cherry-mark")) {
1771                 if (revs->cherry_pick)
1772                         die("--cherry-mark is incompatible with --cherry-pick");
1773                 revs->cherry_mark = 1;
1774                 revs->limited = 1; /* needs limit_list() */
1775         } else if (!strcmp(arg, "--cherry-pick")) {
1776                 if (revs->cherry_mark)
1777                         die("--cherry-pick is incompatible with --cherry-mark");
1778                 revs->cherry_pick = 1;
1779                 revs->limited = 1;
1780         } else if (!strcmp(arg, "--objects")) {
1781                 revs->tag_objects = 1;
1782                 revs->tree_objects = 1;
1783                 revs->blob_objects = 1;
1784         } else if (!strcmp(arg, "--objects-edge")) {
1785                 revs->tag_objects = 1;
1786                 revs->tree_objects = 1;
1787                 revs->blob_objects = 1;
1788                 revs->edge_hint = 1;
1789         } else if (!strcmp(arg, "--verify-objects")) {
1790                 revs->tag_objects = 1;
1791                 revs->tree_objects = 1;
1792                 revs->blob_objects = 1;
1793                 revs->verify_objects = 1;
1794         } else if (!strcmp(arg, "--unpacked")) {
1795                 revs->unpacked = 1;
1796         } else if (!prefixcmp(arg, "--unpacked=")) {
1797                 die("--unpacked=<packfile> no longer supported.");
1798         } else if (!strcmp(arg, "-r")) {
1799                 revs->diff = 1;
1800                 DIFF_OPT_SET(&revs->diffopt, RECURSIVE);
1801         } else if (!strcmp(arg, "-t")) {
1802                 revs->diff = 1;
1803                 DIFF_OPT_SET(&revs->diffopt, RECURSIVE);
1804                 DIFF_OPT_SET(&revs->diffopt, TREE_IN_RECURSIVE);
1805         } else if (!strcmp(arg, "-m")) {
1806                 revs->ignore_merges = 0;
1807         } else if (!strcmp(arg, "-c")) {
1808                 revs->diff = 1;
1809                 revs->dense_combined_merges = 0;
1810                 revs->combine_merges = 1;
1811         } else if (!strcmp(arg, "--cc")) {
1812                 revs->diff = 1;
1813                 revs->dense_combined_merges = 1;
1814                 revs->combine_merges = 1;
1815         } else if (!strcmp(arg, "-v")) {
1816                 revs->verbose_header = 1;
1817         } else if (!strcmp(arg, "--pretty")) {
1818                 revs->verbose_header = 1;
1819                 revs->pretty_given = 1;
1820                 get_commit_format(arg+8, revs);
1821         } else if (!prefixcmp(arg, "--pretty=") || !prefixcmp(arg, "--format=")) {
1822                 /*
1823                  * Detached form ("--pretty X" as opposed to "--pretty=X")
1824                  * not allowed, since the argument is optional.
1825                  */
1826                 revs->verbose_header = 1;
1827                 revs->pretty_given = 1;
1828                 get_commit_format(arg+9, revs);
1829         } else if (!strcmp(arg, "--show-notes") || !strcmp(arg, "--notes")) {
1830                 revs->show_notes = 1;
1831                 revs->show_notes_given = 1;
1832                 revs->notes_opt.use_default_notes = 1;
1833         } else if (!strcmp(arg, "--show-signature")) {
1834                 revs->show_signature = 1;
1835         } else if (!prefixcmp(arg, "--show-notes=") ||
1836                    !prefixcmp(arg, "--notes=")) {
1837                 struct strbuf buf = STRBUF_INIT;
1838                 revs->show_notes = 1;
1839                 revs->show_notes_given = 1;
1840                 if (!prefixcmp(arg, "--show-notes")) {
1841                         if (revs->notes_opt.use_default_notes < 0)
1842                                 revs->notes_opt.use_default_notes = 1;
1843                         strbuf_addstr(&buf, arg+13);
1844                 }
1845                 else
1846                         strbuf_addstr(&buf, arg+8);
1847                 expand_notes_ref(&buf);
1848                 string_list_append(&revs->notes_opt.extra_notes_refs,
1849                                    strbuf_detach(&buf, NULL));
1850         } else if (!strcmp(arg, "--no-notes")) {
1851                 revs->show_notes = 0;
1852                 revs->show_notes_given = 1;
1853                 revs->notes_opt.use_default_notes = -1;
1854                 /* we have been strdup'ing ourselves, so trick
1855                  * string_list into free()ing strings */
1856                 revs->notes_opt.extra_notes_refs.strdup_strings = 1;
1857                 string_list_clear(&revs->notes_opt.extra_notes_refs, 0);
1858                 revs->notes_opt.extra_notes_refs.strdup_strings = 0;
1859         } else if (!strcmp(arg, "--standard-notes")) {
1860                 revs->show_notes_given = 1;
1861                 revs->notes_opt.use_default_notes = 1;
1862         } else if (!strcmp(arg, "--no-standard-notes")) {
1863                 revs->notes_opt.use_default_notes = 0;
1864         } else if (!strcmp(arg, "--oneline")) {
1865                 revs->verbose_header = 1;
1866                 get_commit_format("oneline", revs);
1867                 revs->pretty_given = 1;
1868                 revs->abbrev_commit = 1;
1869         } else if (!strcmp(arg, "--graph")) {
1870                 revs->topo_order = 1;
1871                 revs->rewrite_parents = 1;
1872                 revs->graph = graph_init(revs);
1873         } else if (!strcmp(arg, "--root")) {
1874                 revs->show_root_diff = 1;
1875         } else if (!strcmp(arg, "--no-commit-id")) {
1876                 revs->no_commit_id = 1;
1877         } else if (!strcmp(arg, "--always")) {
1878                 revs->always_show_header = 1;
1879         } else if (!strcmp(arg, "--no-abbrev")) {
1880                 revs->abbrev = 0;
1881         } else if (!strcmp(arg, "--abbrev")) {
1882                 revs->abbrev = DEFAULT_ABBREV;
1883         } else if (!prefixcmp(arg, "--abbrev=")) {
1884                 revs->abbrev = strtoul(arg + 9, NULL, 10);
1885                 if (revs->abbrev < MINIMUM_ABBREV)
1886                         revs->abbrev = MINIMUM_ABBREV;
1887                 else if (revs->abbrev > 40)
1888                         revs->abbrev = 40;
1889         } else if (!strcmp(arg, "--abbrev-commit")) {
1890                 revs->abbrev_commit = 1;
1891                 revs->abbrev_commit_given = 1;
1892         } else if (!strcmp(arg, "--no-abbrev-commit")) {
1893                 revs->abbrev_commit = 0;
1894         } else if (!strcmp(arg, "--full-diff")) {
1895                 revs->diff = 1;
1896                 revs->full_diff = 1;
1897         } else if (!strcmp(arg, "--full-history")) {
1898                 revs->simplify_history = 0;
1899         } else if (!strcmp(arg, "--relative-date")) {
1900                 revs->date_mode = DATE_RELATIVE;
1901                 revs->date_mode_explicit = 1;
1902         } else if ((argcount = parse_long_opt("date", argv, &optarg))) {
1903                 revs->date_mode = parse_date_format(optarg);
1904                 revs->date_mode_explicit = 1;
1905                 return argcount;
1906         } else if (!strcmp(arg, "--log-size")) {
1907                 revs->show_log_size = 1;
1908         }
1909         /*
1910          * Grepping the commit log
1911          */
1912         else if ((argcount = parse_long_opt("author", argv, &optarg))) {
1913                 add_header_grep(revs, GREP_HEADER_AUTHOR, optarg);
1914                 return argcount;
1915         } else if ((argcount = parse_long_opt("committer", argv, &optarg))) {
1916                 add_header_grep(revs, GREP_HEADER_COMMITTER, optarg);
1917                 return argcount;
1918         } else if ((argcount = parse_long_opt("grep-reflog", argv, &optarg))) {
1919                 add_header_grep(revs, GREP_HEADER_REFLOG, optarg);
1920                 return argcount;
1921         } else if ((argcount = parse_long_opt("grep", argv, &optarg))) {
1922                 add_message_grep(revs, optarg);
1923                 return argcount;
1924         } else if (!strcmp(arg, "--grep-debug")) {
1925                 revs->grep_filter.debug = 1;
1926         } else if (!strcmp(arg, "--basic-regexp")) {
1927                 grep_set_pattern_type_option(GREP_PATTERN_TYPE_BRE, &revs->grep_filter);
1928         } else if (!strcmp(arg, "--extended-regexp") || !strcmp(arg, "-E")) {
1929                 grep_set_pattern_type_option(GREP_PATTERN_TYPE_ERE, &revs->grep_filter);
1930         } else if (!strcmp(arg, "--regexp-ignore-case") || !strcmp(arg, "-i")) {
1931                 revs->grep_filter.regflags |= REG_ICASE;
1932                 DIFF_OPT_SET(&revs->diffopt, PICKAXE_IGNORE_CASE);
1933         } else if (!strcmp(arg, "--fixed-strings") || !strcmp(arg, "-F")) {
1934                 grep_set_pattern_type_option(GREP_PATTERN_TYPE_FIXED, &revs->grep_filter);
1935         } else if (!strcmp(arg, "--perl-regexp")) {
1936                 grep_set_pattern_type_option(GREP_PATTERN_TYPE_PCRE, &revs->grep_filter);
1937         } else if (!strcmp(arg, "--all-match")) {
1938                 revs->grep_filter.all_match = 1;
1939         } else if ((argcount = parse_long_opt("encoding", argv, &optarg))) {
1940                 if (strcmp(optarg, "none"))
1941                         git_log_output_encoding = xstrdup(optarg);
1942                 else
1943                         git_log_output_encoding = "";
1944                 return argcount;
1945         } else if (!strcmp(arg, "--reverse")) {
1946                 revs->reverse ^= 1;
1947         } else if (!strcmp(arg, "--children")) {
1948                 revs->children.name = "children";
1949                 revs->limited = 1;
1950         } else if (!strcmp(arg, "--ignore-missing")) {
1951                 revs->ignore_missing = 1;
1952         } else {
1953                 int opts = diff_opt_parse(&revs->diffopt, argv, argc);
1954                 if (!opts)
1955                         unkv[(*unkc)++] = arg;
1956                 return opts;
1957         }
1958
1959         return 1;
1960 }
1961
1962 void parse_revision_opt(struct rev_info *revs, struct parse_opt_ctx_t *ctx,
1963                         const struct option *options,
1964                         const char * const usagestr[])
1965 {
1966         int n = handle_revision_opt(revs, ctx->argc, ctx->argv,
1967                                     &ctx->cpidx, ctx->out);
1968         if (n <= 0) {
1969                 error("unknown option `%s'", ctx->argv[0]);
1970                 usage_with_options(usagestr, options);
1971         }
1972         ctx->argv += n;
1973         ctx->argc -= n;
1974 }
1975
1976 static int for_each_bad_bisect_ref(const char *submodule, each_ref_fn fn, void *cb_data)
1977 {
1978         return for_each_ref_in_submodule(submodule, "refs/bisect/bad", fn, cb_data);
1979 }
1980
1981 static int for_each_good_bisect_ref(const char *submodule, each_ref_fn fn, void *cb_data)
1982 {
1983         return for_each_ref_in_submodule(submodule, "refs/bisect/good", fn, cb_data);
1984 }
1985
1986 static int handle_revision_pseudo_opt(const char *submodule,
1987                                 struct rev_info *revs,
1988                                 int argc, const char **argv, int *flags)
1989 {
1990         const char *arg = argv[0];
1991         const char *optarg;
1992         int argcount;
1993
1994         /*
1995          * NOTE!
1996          *
1997          * Commands like "git shortlog" will not accept the options below
1998          * unless parse_revision_opt queues them (as opposed to erroring
1999          * out).
2000          *
2001          * When implementing your new pseudo-option, remember to
2002          * register it in the list at the top of handle_revision_opt.
2003          */
2004         if (!strcmp(arg, "--all")) {
2005                 handle_refs(submodule, revs, *flags, for_each_ref_submodule);
2006                 handle_refs(submodule, revs, *flags, head_ref_submodule);
2007                 clear_ref_exclusion(&revs->ref_excludes);
2008         } else if (!strcmp(arg, "--branches")) {
2009                 handle_refs(submodule, revs, *flags, for_each_branch_ref_submodule);
2010                 clear_ref_exclusion(&revs->ref_excludes);
2011         } else if (!strcmp(arg, "--bisect")) {
2012                 handle_refs(submodule, revs, *flags, for_each_bad_bisect_ref);
2013                 handle_refs(submodule, revs, *flags ^ (UNINTERESTING | BOTTOM), for_each_good_bisect_ref);
2014                 revs->bisect = 1;
2015         } else if (!strcmp(arg, "--tags")) {
2016                 handle_refs(submodule, revs, *flags, for_each_tag_ref_submodule);
2017                 clear_ref_exclusion(&revs->ref_excludes);
2018         } else if (!strcmp(arg, "--remotes")) {
2019                 handle_refs(submodule, revs, *flags, for_each_remote_ref_submodule);
2020                 clear_ref_exclusion(&revs->ref_excludes);
2021         } else if ((argcount = parse_long_opt("glob", argv, &optarg))) {
2022                 struct all_refs_cb cb;
2023                 init_all_refs_cb(&cb, revs, *flags);
2024                 for_each_glob_ref(handle_one_ref, optarg, &cb);
2025                 clear_ref_exclusion(&revs->ref_excludes);
2026                 return argcount;
2027         } else if ((argcount = parse_long_opt("exclude", argv, &optarg))) {
2028                 add_ref_exclusion(&revs->ref_excludes, optarg);
2029                 return argcount;
2030         } else if (!prefixcmp(arg, "--branches=")) {
2031                 struct all_refs_cb cb;
2032                 init_all_refs_cb(&cb, revs, *flags);
2033                 for_each_glob_ref_in(handle_one_ref, arg + 11, "refs/heads/", &cb);
2034                 clear_ref_exclusion(&revs->ref_excludes);
2035         } else if (!prefixcmp(arg, "--tags=")) {
2036                 struct all_refs_cb cb;
2037                 init_all_refs_cb(&cb, revs, *flags);
2038                 for_each_glob_ref_in(handle_one_ref, arg + 7, "refs/tags/", &cb);
2039                 clear_ref_exclusion(&revs->ref_excludes);
2040         } else if (!prefixcmp(arg, "--remotes=")) {
2041                 struct all_refs_cb cb;
2042                 init_all_refs_cb(&cb, revs, *flags);
2043                 for_each_glob_ref_in(handle_one_ref, arg + 10, "refs/remotes/", &cb);
2044                 clear_ref_exclusion(&revs->ref_excludes);
2045         } else if (!strcmp(arg, "--reflog")) {
2046                 handle_reflog(revs, *flags);
2047         } else if (!strcmp(arg, "--not")) {
2048                 *flags ^= UNINTERESTING | BOTTOM;
2049         } else if (!strcmp(arg, "--no-walk")) {
2050                 revs->no_walk = REVISION_WALK_NO_WALK_SORTED;
2051         } else if (!prefixcmp(arg, "--no-walk=")) {
2052                 /*
2053                  * Detached form ("--no-walk X" as opposed to "--no-walk=X")
2054                  * not allowed, since the argument is optional.
2055                  */
2056                 if (!strcmp(arg + 10, "sorted"))
2057                         revs->no_walk = REVISION_WALK_NO_WALK_SORTED;
2058                 else if (!strcmp(arg + 10, "unsorted"))
2059                         revs->no_walk = REVISION_WALK_NO_WALK_UNSORTED;
2060                 else
2061                         return error("invalid argument to --no-walk");
2062         } else if (!strcmp(arg, "--do-walk")) {
2063                 revs->no_walk = 0;
2064         } else {
2065                 return 0;
2066         }
2067
2068         return 1;
2069 }
2070
2071 /*
2072  * Parse revision information, filling in the "rev_info" structure,
2073  * and removing the used arguments from the argument list.
2074  *
2075  * Returns the number of arguments left that weren't recognized
2076  * (which are also moved to the head of the argument list)
2077  */
2078 int setup_revisions(int argc, const char **argv, struct rev_info *revs, struct setup_revision_opt *opt)
2079 {
2080         int i, flags, left, seen_dashdash, read_from_stdin, got_rev_arg = 0, revarg_opt;
2081         struct cmdline_pathspec prune_data;
2082         const char *submodule = NULL;
2083
2084         memset(&prune_data, 0, sizeof(prune_data));
2085         if (opt)
2086                 submodule = opt->submodule;
2087
2088         /* First, search for "--" */
2089         if (opt && opt->assume_dashdash) {
2090                 seen_dashdash = 1;
2091         } else {
2092                 seen_dashdash = 0;
2093                 for (i = 1; i < argc; i++) {
2094                         const char *arg = argv[i];
2095                         if (strcmp(arg, "--"))
2096                                 continue;
2097                         argv[i] = NULL;
2098                         argc = i;
2099                         if (argv[i + 1])
2100                                 append_prune_data(&prune_data, argv + i + 1);
2101                         seen_dashdash = 1;
2102                         break;
2103                 }
2104         }
2105
2106         /* Second, deal with arguments and options */
2107         flags = 0;
2108         revarg_opt = opt ? opt->revarg_opt : 0;
2109         if (seen_dashdash)
2110                 revarg_opt |= REVARG_CANNOT_BE_FILENAME;
2111         read_from_stdin = 0;
2112         for (left = i = 1; i < argc; i++) {
2113                 const char *arg = argv[i];
2114                 if (*arg == '-') {
2115                         int opts;
2116
2117                         opts = handle_revision_pseudo_opt(submodule,
2118                                                 revs, argc - i, argv + i,
2119                                                 &flags);
2120                         if (opts > 0) {
2121                                 i += opts - 1;
2122                                 continue;
2123                         }
2124
2125                         if (!strcmp(arg, "--stdin")) {
2126                                 if (revs->disable_stdin) {
2127                                         argv[left++] = arg;
2128                                         continue;
2129                                 }
2130                                 if (read_from_stdin++)
2131                                         die("--stdin given twice?");
2132                                 read_revisions_from_stdin(revs, &prune_data);
2133                                 continue;
2134                         }
2135
2136                         opts = handle_revision_opt(revs, argc - i, argv + i, &left, argv);
2137                         if (opts > 0) {
2138                                 i += opts - 1;
2139                                 continue;
2140                         }
2141                         if (opts < 0)
2142                                 exit(128);
2143                         continue;
2144                 }
2145
2146
2147                 if (handle_revision_arg(arg, revs, flags, revarg_opt)) {
2148                         int j;
2149                         if (seen_dashdash || *arg == '^')
2150                                 die("bad revision '%s'", arg);
2151
2152                         /* If we didn't have a "--":
2153                          * (1) all filenames must exist;
2154                          * (2) all rev-args must not be interpretable
2155                          *     as a valid filename.
2156                          * but the latter we have checked in the main loop.
2157                          */
2158                         for (j = i; j < argc; j++)
2159                                 verify_filename(revs->prefix, argv[j], j == i);
2160
2161                         append_prune_data(&prune_data, argv + i);
2162                         break;
2163                 }
2164                 else
2165                         got_rev_arg = 1;
2166         }
2167
2168         if (prune_data.nr) {
2169                 /*
2170                  * If we need to introduce the magic "a lone ':' means no
2171                  * pathspec whatsoever", here is the place to do so.
2172                  *
2173                  * if (prune_data.nr == 1 && !strcmp(prune_data[0], ":")) {
2174                  *      prune_data.nr = 0;
2175                  *      prune_data.alloc = 0;
2176                  *      free(prune_data.path);
2177                  *      prune_data.path = NULL;
2178                  * } else {
2179                  *      terminate prune_data.alloc with NULL and
2180                  *      call init_pathspec() to set revs->prune_data here.
2181                  * }
2182                  */
2183                 ALLOC_GROW(prune_data.path, prune_data.nr + 1, prune_data.alloc);
2184                 prune_data.path[prune_data.nr++] = NULL;
2185                 parse_pathspec(&revs->prune_data, 0, 0,
2186                                revs->prefix, prune_data.path);
2187         }
2188
2189         if (revs->def == NULL)
2190                 revs->def = opt ? opt->def : NULL;
2191         if (opt && opt->tweak)
2192                 opt->tweak(revs, opt);
2193         if (revs->show_merge)
2194                 prepare_show_merge(revs);
2195         if (revs->def && !revs->pending.nr && !got_rev_arg) {
2196                 unsigned char sha1[20];
2197                 struct object *object;
2198                 struct object_context oc;
2199                 if (get_sha1_with_context(revs->def, 0, sha1, &oc))
2200                         die("bad default revision '%s'", revs->def);
2201                 object = get_reference(revs, revs->def, sha1, 0);
2202                 add_pending_object_with_mode(revs, object, revs->def, oc.mode);
2203         }
2204
2205         /* Did the user ask for any diff output? Run the diff! */
2206         if (revs->diffopt.output_format & ~DIFF_FORMAT_NO_OUTPUT)
2207                 revs->diff = 1;
2208
2209         /* Pickaxe, diff-filter and rename following need diffs */
2210         if (revs->diffopt.pickaxe ||
2211             revs->diffopt.filter ||
2212             DIFF_OPT_TST(&revs->diffopt, FOLLOW_RENAMES))
2213                 revs->diff = 1;
2214
2215         if (revs->topo_order)
2216                 revs->limited = 1;
2217
2218         if (revs->prune_data.nr) {
2219                 copy_pathspec(&revs->pruning.pathspec, &revs->prune_data);
2220                 /* Can't prune commits with rename following: the paths change.. */
2221                 if (!DIFF_OPT_TST(&revs->diffopt, FOLLOW_RENAMES))
2222                         revs->prune = 1;
2223                 if (!revs->full_diff)
2224                         copy_pathspec(&revs->diffopt.pathspec,
2225                                       &revs->prune_data);
2226         }
2227         if (revs->combine_merges)
2228                 revs->ignore_merges = 0;
2229         revs->diffopt.abbrev = revs->abbrev;
2230
2231         if (revs->line_level_traverse) {
2232                 revs->limited = 1;
2233                 revs->topo_order = 1;
2234         }
2235
2236         diff_setup_done(&revs->diffopt);
2237
2238         grep_commit_pattern_type(GREP_PATTERN_TYPE_UNSPECIFIED,
2239                                  &revs->grep_filter);
2240         compile_grep_patterns(&revs->grep_filter);
2241
2242         if (revs->reverse && revs->reflog_info)
2243                 die("cannot combine --reverse with --walk-reflogs");
2244         if (revs->rewrite_parents && revs->children.name)
2245                 die("cannot combine --parents and --children");
2246
2247         /*
2248          * Limitations on the graph functionality
2249          */
2250         if (revs->reverse && revs->graph)
2251                 die("cannot combine --reverse with --graph");
2252
2253         if (revs->reflog_info && revs->graph)
2254                 die("cannot combine --walk-reflogs with --graph");
2255         if (!revs->reflog_info && revs->grep_filter.use_reflog_filter)
2256                 die("cannot use --grep-reflog without --walk-reflogs");
2257
2258         return left;
2259 }
2260
2261 static void add_child(struct rev_info *revs, struct commit *parent, struct commit *child)
2262 {
2263         struct commit_list *l = xcalloc(1, sizeof(*l));
2264
2265         l->item = child;
2266         l->next = add_decoration(&revs->children, &parent->object, l);
2267 }
2268
2269 static int remove_duplicate_parents(struct rev_info *revs, struct commit *commit)
2270 {
2271         struct treesame_state *ts = lookup_decoration(&revs->treesame, &commit->object);
2272         struct commit_list **pp, *p;
2273         int surviving_parents;
2274
2275         /* Examine existing parents while marking ones we have seen... */
2276         pp = &commit->parents;
2277         surviving_parents = 0;
2278         while ((p = *pp) != NULL) {
2279                 struct commit *parent = p->item;
2280                 if (parent->object.flags & TMP_MARK) {
2281                         *pp = p->next;
2282                         if (ts)
2283                                 compact_treesame(revs, commit, surviving_parents);
2284                         continue;
2285                 }
2286                 parent->object.flags |= TMP_MARK;
2287                 surviving_parents++;
2288                 pp = &p->next;
2289         }
2290         /* clear the temporary mark */
2291         for (p = commit->parents; p; p = p->next) {
2292                 p->item->object.flags &= ~TMP_MARK;
2293         }
2294         /* no update_treesame() - removing duplicates can't affect TREESAME */
2295         return surviving_parents;
2296 }
2297
2298 struct merge_simplify_state {
2299         struct commit *simplified;
2300 };
2301
2302 static struct merge_simplify_state *locate_simplify_state(struct rev_info *revs, struct commit *commit)
2303 {
2304         struct merge_simplify_state *st;
2305
2306         st = lookup_decoration(&revs->merge_simplification, &commit->object);
2307         if (!st) {
2308                 st = xcalloc(1, sizeof(*st));
2309                 add_decoration(&revs->merge_simplification, &commit->object, st);
2310         }
2311         return st;
2312 }
2313
2314 static int mark_redundant_parents(struct rev_info *revs, struct commit *commit)
2315 {
2316         struct commit_list *h = reduce_heads(commit->parents);
2317         int i = 0, marked = 0;
2318         struct commit_list *po, *pn;
2319
2320         /* Want these for sanity-checking only */
2321         int orig_cnt = commit_list_count(commit->parents);
2322         int cnt = commit_list_count(h);
2323
2324         /*
2325          * Not ready to remove items yet, just mark them for now, based
2326          * on the output of reduce_heads(). reduce_heads outputs the reduced
2327          * set in its original order, so this isn't too hard.
2328          */
2329         po = commit->parents;
2330         pn = h;
2331         while (po) {
2332                 if (pn && po->item == pn->item) {
2333                         pn = pn->next;
2334                         i++;
2335                 } else {
2336                         po->item->object.flags |= TMP_MARK;
2337                         marked++;
2338                 }
2339                 po=po->next;
2340         }
2341
2342         if (i != cnt || cnt+marked != orig_cnt)
2343                 die("mark_redundant_parents %d %d %d %d", orig_cnt, cnt, i, marked);
2344
2345         free_commit_list(h);
2346
2347         return marked;
2348 }
2349
2350 static int mark_treesame_root_parents(struct rev_info *revs, struct commit *commit)
2351 {
2352         struct commit_list *p;
2353         int marked = 0;
2354
2355         for (p = commit->parents; p; p = p->next) {
2356                 struct commit *parent = p->item;
2357                 if (!parent->parents && (parent->object.flags & TREESAME)) {
2358                         parent->object.flags |= TMP_MARK;
2359                         marked++;
2360                 }
2361         }
2362
2363         return marked;
2364 }
2365
2366 /*
2367  * Awkward naming - this means one parent we are TREESAME to.
2368  * cf mark_treesame_root_parents: root parents that are TREESAME (to an
2369  * empty tree). Better name suggestions?
2370  */
2371 static int leave_one_treesame_to_parent(struct rev_info *revs, struct commit *commit)
2372 {
2373         struct treesame_state *ts = lookup_decoration(&revs->treesame, &commit->object);
2374         struct commit *unmarked = NULL, *marked = NULL;
2375         struct commit_list *p;
2376         unsigned n;
2377
2378         for (p = commit->parents, n = 0; p; p = p->next, n++) {
2379                 if (ts->treesame[n]) {
2380                         if (p->item->object.flags & TMP_MARK) {
2381                                 if (!marked)
2382                                         marked = p->item;
2383                         } else {
2384                                 if (!unmarked) {
2385                                         unmarked = p->item;
2386                                         break;
2387                                 }
2388                         }
2389                 }
2390         }
2391
2392         /*
2393          * If we are TREESAME to a marked-for-deletion parent, but not to any
2394          * unmarked parents, unmark the first TREESAME parent. This is the
2395          * parent that the default simplify_history==1 scan would have followed,
2396          * and it doesn't make sense to omit that path when asking for a
2397          * simplified full history. Retaining it improves the chances of
2398          * understanding odd missed merges that took an old version of a file.
2399          *
2400          * Example:
2401          *
2402          *   I--------*X       A modified the file, but mainline merge X used
2403          *    \       /        "-s ours", so took the version from I. X is
2404          *     `-*A--'         TREESAME to I and !TREESAME to A.
2405          *
2406          * Default log from X would produce "I". Without this check,
2407          * --full-history --simplify-merges would produce "I-A-X", showing
2408          * the merge commit X and that it changed A, but not making clear that
2409          * it had just taken the I version. With this check, the topology above
2410          * is retained.
2411          *
2412          * Note that it is possible that the simplification chooses a different
2413          * TREESAME parent from the default, in which case this test doesn't
2414          * activate, and we _do_ drop the default parent. Example:
2415          *
2416          *   I------X         A modified the file, but it was reverted in B,
2417          *    \    /          meaning mainline merge X is TREESAME to both
2418          *    *A-*B           parents.
2419          *
2420          * Default log would produce "I" by following the first parent;
2421          * --full-history --simplify-merges will produce "I-A-B". But this is a
2422          * reasonable result - it presents a logical full history leading from
2423          * I to X, and X is not an important merge.
2424          */
2425         if (!unmarked && marked) {
2426                 marked->object.flags &= ~TMP_MARK;
2427                 return 1;
2428         }
2429
2430         return 0;
2431 }
2432
2433 static int remove_marked_parents(struct rev_info *revs, struct commit *commit)
2434 {
2435         struct commit_list **pp, *p;
2436         int nth_parent, removed = 0;
2437
2438         pp = &commit->parents;
2439         nth_parent = 0;
2440         while ((p = *pp) != NULL) {
2441                 struct commit *parent = p->item;
2442                 if (parent->object.flags & TMP_MARK) {
2443                         parent->object.flags &= ~TMP_MARK;
2444                         *pp = p->next;
2445                         free(p);
2446                         removed++;
2447                         compact_treesame(revs, commit, nth_parent);
2448                         continue;
2449                 }
2450                 pp = &p->next;
2451                 nth_parent++;
2452         }
2453
2454         /* Removing parents can only increase TREESAMEness */
2455         if (removed && !(commit->object.flags & TREESAME))
2456                 update_treesame(revs, commit);
2457
2458         return nth_parent;
2459 }
2460
2461 static struct commit_list **simplify_one(struct rev_info *revs, struct commit *commit, struct commit_list **tail)
2462 {
2463         struct commit_list *p;
2464         struct commit *parent;
2465         struct merge_simplify_state *st, *pst;
2466         int cnt;
2467
2468         st = locate_simplify_state(revs, commit);
2469
2470         /*
2471          * Have we handled this one?
2472          */
2473         if (st->simplified)
2474                 return tail;
2475
2476         /*
2477          * An UNINTERESTING commit simplifies to itself, so does a
2478          * root commit.  We do not rewrite parents of such commit
2479          * anyway.
2480          */
2481         if ((commit->object.flags & UNINTERESTING) || !commit->parents) {
2482                 st->simplified = commit;
2483                 return tail;
2484         }
2485
2486         /*
2487          * Do we know what commit all of our parents that matter
2488          * should be rewritten to?  Otherwise we are not ready to
2489          * rewrite this one yet.
2490          */
2491         for (cnt = 0, p = commit->parents; p; p = p->next) {
2492                 pst = locate_simplify_state(revs, p->item);
2493                 if (!pst->simplified) {
2494                         tail = &commit_list_insert(p->item, tail)->next;
2495                         cnt++;
2496                 }
2497                 if (revs->first_parent_only)
2498                         break;
2499         }
2500         if (cnt) {
2501                 tail = &commit_list_insert(commit, tail)->next;
2502                 return tail;
2503         }
2504
2505         /*
2506          * Rewrite our list of parents. Note that this cannot
2507          * affect our TREESAME flags in any way - a commit is
2508          * always TREESAME to its simplification.
2509          */
2510         for (p = commit->parents; p; p = p->next) {
2511                 pst = locate_simplify_state(revs, p->item);
2512                 p->item = pst->simplified;
2513                 if (revs->first_parent_only)
2514                         break;
2515         }
2516
2517         if (revs->first_parent_only)
2518                 cnt = 1;
2519         else
2520                 cnt = remove_duplicate_parents(revs, commit);
2521
2522         /*
2523          * It is possible that we are a merge and one side branch
2524          * does not have any commit that touches the given paths;
2525          * in such a case, the immediate parent from that branch
2526          * will be rewritten to be the merge base.
2527          *
2528          *      o----X          X: the commit we are looking at;
2529          *     /    /           o: a commit that touches the paths;
2530          * ---o----'
2531          *
2532          * Further, a merge of an independent branch that doesn't
2533          * touch the path will reduce to a treesame root parent:
2534          *
2535          *  ----o----X          X: the commit we are looking at;
2536          *          /           o: a commit that touches the paths;
2537          *         r            r: a root commit not touching the paths
2538          *
2539          * Detect and simplify both cases.
2540          */
2541         if (1 < cnt) {
2542                 int marked = mark_redundant_parents(revs, commit);
2543                 marked += mark_treesame_root_parents(revs, commit);
2544                 if (marked)
2545                         marked -= leave_one_treesame_to_parent(revs, commit);
2546                 if (marked)
2547                         cnt = remove_marked_parents(revs, commit);
2548         }
2549
2550         /*
2551          * A commit simplifies to itself if it is a root, if it is
2552          * UNINTERESTING, if it touches the given paths, or if it is a
2553          * merge and its parents don't simplify to one relevant commit
2554          * (the first two cases are already handled at the beginning of
2555          * this function).
2556          *
2557          * Otherwise, it simplifies to what its sole relevant parent
2558          * simplifies to.
2559          */
2560         if (!cnt ||
2561             (commit->object.flags & UNINTERESTING) ||
2562             !(commit->object.flags & TREESAME) ||
2563             (parent = one_relevant_parent(revs, commit->parents)) == NULL)
2564                 st->simplified = commit;
2565         else {
2566                 pst = locate_simplify_state(revs, parent);
2567                 st->simplified = pst->simplified;
2568         }
2569         return tail;
2570 }
2571
2572 static void simplify_merges(struct rev_info *revs)
2573 {
2574         struct commit_list *list, *next;
2575         struct commit_list *yet_to_do, **tail;
2576         struct commit *commit;
2577
2578         if (!revs->prune)
2579                 return;
2580
2581         /* feed the list reversed */
2582         yet_to_do = NULL;
2583         for (list = revs->commits; list; list = next) {
2584                 commit = list->item;
2585                 next = list->next;
2586                 /*
2587                  * Do not free(list) here yet; the original list
2588                  * is used later in this function.
2589                  */
2590                 commit_list_insert(commit, &yet_to_do);
2591         }
2592         while (yet_to_do) {
2593                 list = yet_to_do;
2594                 yet_to_do = NULL;
2595                 tail = &yet_to_do;
2596                 while (list) {
2597                         commit = list->item;
2598                         next = list->next;
2599                         free(list);
2600                         list = next;
2601                         tail = simplify_one(revs, commit, tail);
2602                 }
2603         }
2604
2605         /* clean up the result, removing the simplified ones */
2606         list = revs->commits;
2607         revs->commits = NULL;
2608         tail = &revs->commits;
2609         while (list) {
2610                 struct merge_simplify_state *st;
2611
2612                 commit = list->item;
2613                 next = list->next;
2614                 free(list);
2615                 list = next;
2616                 st = locate_simplify_state(revs, commit);
2617                 if (st->simplified == commit)
2618                         tail = &commit_list_insert(commit, tail)->next;
2619         }
2620 }
2621
2622 static void set_children(struct rev_info *revs)
2623 {
2624         struct commit_list *l;
2625         for (l = revs->commits; l; l = l->next) {
2626                 struct commit *commit = l->item;
2627                 struct commit_list *p;
2628
2629                 for (p = commit->parents; p; p = p->next)
2630                         add_child(revs, p->item, commit);
2631         }
2632 }
2633
2634 void reset_revision_walk(void)
2635 {
2636         clear_object_flags(SEEN | ADDED | SHOWN);
2637 }
2638
2639 int prepare_revision_walk(struct rev_info *revs)
2640 {
2641         int nr = revs->pending.nr;
2642         struct object_array_entry *e, *list;
2643         struct commit_list **next = &revs->commits;
2644
2645         e = list = revs->pending.objects;
2646         revs->pending.nr = 0;
2647         revs->pending.alloc = 0;
2648         revs->pending.objects = NULL;
2649         while (--nr >= 0) {
2650                 struct commit *commit = handle_commit(revs, e->item, e->name);
2651                 if (commit) {
2652                         if (!(commit->object.flags & SEEN)) {
2653                                 commit->object.flags |= SEEN;
2654                                 next = commit_list_append(commit, next);
2655                         }
2656                 }
2657                 e++;
2658         }
2659         if (!revs->leak_pending)
2660                 free(list);
2661
2662         /* Signal whether we need per-parent treesame decoration */
2663         if (revs->simplify_merges ||
2664             (revs->limited && limiting_can_increase_treesame(revs)))
2665                 revs->treesame.name = "treesame";
2666
2667         if (revs->no_walk != REVISION_WALK_NO_WALK_UNSORTED)
2668                 commit_list_sort_by_date(&revs->commits);
2669         if (revs->no_walk)
2670                 return 0;
2671         if (revs->limited)
2672                 if (limit_list(revs) < 0)
2673                         return -1;
2674         if (revs->topo_order)
2675                 sort_in_topological_order(&revs->commits, revs->sort_order);
2676         if (revs->line_level_traverse)
2677                 line_log_filter(revs);
2678         if (revs->simplify_merges)
2679                 simplify_merges(revs);
2680         if (revs->children.name)
2681                 set_children(revs);
2682         return 0;
2683 }
2684
2685 static enum rewrite_result rewrite_one(struct rev_info *revs, struct commit **pp)
2686 {
2687         struct commit_list *cache = NULL;
2688
2689         for (;;) {
2690                 struct commit *p = *pp;
2691                 if (!revs->limited)
2692                         if (add_parents_to_list(revs, p, &revs->commits, &cache) < 0)
2693                                 return rewrite_one_error;
2694                 if (p->object.flags & UNINTERESTING)
2695                         return rewrite_one_ok;
2696                 if (!(p->object.flags & TREESAME))
2697                         return rewrite_one_ok;
2698                 if (!p->parents)
2699                         return rewrite_one_noparents;
2700                 if ((p = one_relevant_parent(revs, p->parents)) == NULL)
2701                         return rewrite_one_ok;
2702                 *pp = p;
2703         }
2704 }
2705
2706 int rewrite_parents(struct rev_info *revs, struct commit *commit,
2707         rewrite_parent_fn_t rewrite_parent)
2708 {
2709         struct commit_list **pp = &commit->parents;
2710         while (*pp) {
2711                 struct commit_list *parent = *pp;
2712                 switch (rewrite_parent(revs, &parent->item)) {
2713                 case rewrite_one_ok:
2714                         break;
2715                 case rewrite_one_noparents:
2716                         *pp = parent->next;
2717                         continue;
2718                 case rewrite_one_error:
2719                         return -1;
2720                 }
2721                 pp = &parent->next;
2722         }
2723         remove_duplicate_parents(revs, commit);
2724         return 0;
2725 }
2726
2727 static int commit_rewrite_person(struct strbuf *buf, const char *what, struct string_list *mailmap)
2728 {
2729         char *person, *endp;
2730         size_t len, namelen, maillen;
2731         const char *name;
2732         const char *mail;
2733         struct ident_split ident;
2734
2735         person = strstr(buf->buf, what);
2736         if (!person)
2737                 return 0;
2738
2739         person += strlen(what);
2740         endp = strchr(person, '\n');
2741         if (!endp)
2742                 return 0;
2743
2744         len = endp - person;
2745
2746         if (split_ident_line(&ident, person, len))
2747                 return 0;
2748
2749         mail = ident.mail_begin;
2750         maillen = ident.mail_end - ident.mail_begin;
2751         name = ident.name_begin;
2752         namelen = ident.name_end - ident.name_begin;
2753
2754         if (map_user(mailmap, &mail, &maillen, &name, &namelen)) {
2755                 struct strbuf namemail = STRBUF_INIT;
2756
2757                 strbuf_addf(&namemail, "%.*s <%.*s>",
2758                             (int)namelen, name, (int)maillen, mail);
2759
2760                 strbuf_splice(buf, ident.name_begin - buf->buf,
2761                               ident.mail_end - ident.name_begin + 1,
2762                               namemail.buf, namemail.len);
2763
2764                 strbuf_release(&namemail);
2765
2766                 return 1;
2767         }
2768
2769         return 0;
2770 }
2771
2772 static int commit_match(struct commit *commit, struct rev_info *opt)
2773 {
2774         int retval;
2775         const char *encoding;
2776         char *message;
2777         struct strbuf buf = STRBUF_INIT;
2778
2779         if (!opt->grep_filter.pattern_list && !opt->grep_filter.header_list)
2780                 return 1;
2781
2782         /* Prepend "fake" headers as needed */
2783         if (opt->grep_filter.use_reflog_filter) {
2784                 strbuf_addstr(&buf, "reflog ");
2785                 get_reflog_message(&buf, opt->reflog_info);
2786                 strbuf_addch(&buf, '\n');
2787         }
2788
2789         /*
2790          * We grep in the user's output encoding, under the assumption that it
2791          * is the encoding they are most likely to write their grep pattern
2792          * for. In addition, it means we will match the "notes" encoding below,
2793          * so we will not end up with a buffer that has two different encodings
2794          * in it.
2795          */
2796         encoding = get_log_output_encoding();
2797         message = logmsg_reencode(commit, NULL, encoding);
2798
2799         /* Copy the commit to temporary if we are using "fake" headers */
2800         if (buf.len)
2801                 strbuf_addstr(&buf, message);
2802
2803         if (opt->grep_filter.header_list && opt->mailmap) {
2804                 if (!buf.len)
2805                         strbuf_addstr(&buf, message);
2806
2807                 commit_rewrite_person(&buf, "\nauthor ", opt->mailmap);
2808                 commit_rewrite_person(&buf, "\ncommitter ", opt->mailmap);
2809         }
2810
2811         /* Append "fake" message parts as needed */
2812         if (opt->show_notes) {
2813                 if (!buf.len)
2814                         strbuf_addstr(&buf, message);
2815                 format_display_notes(commit->object.sha1, &buf, encoding, 1);
2816         }
2817
2818         /* Find either in the original commit message, or in the temporary */
2819         if (buf.len)
2820                 retval = grep_buffer(&opt->grep_filter, buf.buf, buf.len);
2821         else
2822                 retval = grep_buffer(&opt->grep_filter,
2823                                      message, strlen(message));
2824         strbuf_release(&buf);
2825         logmsg_free(message, commit);
2826         return retval;
2827 }
2828
2829 static inline int want_ancestry(const struct rev_info *revs)
2830 {
2831         return (revs->rewrite_parents || revs->children.name);
2832 }
2833
2834 enum commit_action get_commit_action(struct rev_info *revs, struct commit *commit)
2835 {
2836         if (commit->object.flags & SHOWN)
2837                 return commit_ignore;
2838         if (revs->unpacked && has_sha1_pack(commit->object.sha1))
2839                 return commit_ignore;
2840         if (revs->show_all)
2841                 return commit_show;
2842         if (commit->object.flags & UNINTERESTING)
2843                 return commit_ignore;
2844         if (revs->min_age != -1 && (commit->date > revs->min_age))
2845                 return commit_ignore;
2846         if (revs->min_parents || (revs->max_parents >= 0)) {
2847                 int n = commit_list_count(commit->parents);
2848                 if ((n < revs->min_parents) ||
2849                     ((revs->max_parents >= 0) && (n > revs->max_parents)))
2850                         return commit_ignore;
2851         }
2852         if (!commit_match(commit, revs))
2853                 return commit_ignore;
2854         if (revs->prune && revs->dense) {
2855                 /* Commit without changes? */
2856                 if (commit->object.flags & TREESAME) {
2857                         int n;
2858                         struct commit_list *p;
2859                         /* drop merges unless we want parenthood */
2860                         if (!want_ancestry(revs))
2861                                 return commit_ignore;
2862                         /*
2863                          * If we want ancestry, then need to keep any merges
2864                          * between relevant commits to tie together topology.
2865                          * For consistency with TREESAME and simplification
2866                          * use "relevant" here rather than just INTERESTING,
2867                          * to treat bottom commit(s) as part of the topology.
2868                          */
2869                         for (n = 0, p = commit->parents; p; p = p->next)
2870                                 if (relevant_commit(p->item))
2871                                         if (++n >= 2)
2872                                                 return commit_show;
2873                         return commit_ignore;
2874                 }
2875         }
2876         return commit_show;
2877 }
2878
2879 enum commit_action simplify_commit(struct rev_info *revs, struct commit *commit)
2880 {
2881         enum commit_action action = get_commit_action(revs, commit);
2882
2883         if (action == commit_show &&
2884             !revs->show_all &&
2885             revs->prune && revs->dense && want_ancestry(revs)) {
2886                 /*
2887                  * --full-diff on simplified parents is no good: it
2888                  * will show spurious changes from the commits that
2889                  * were elided.  So we save the parents on the side
2890                  * when --full-diff is in effect.
2891                  */
2892                 if (revs->full_diff)
2893                         save_parents(revs, commit);
2894                 if (rewrite_parents(revs, commit, rewrite_one) < 0)
2895                         return commit_error;
2896         }
2897         return action;
2898 }
2899
2900 static struct commit *get_revision_1(struct rev_info *revs)
2901 {
2902         if (!revs->commits)
2903                 return NULL;
2904
2905         do {
2906                 struct commit_list *entry = revs->commits;
2907                 struct commit *commit = entry->item;
2908
2909                 revs->commits = entry->next;
2910                 free(entry);
2911
2912                 if (revs->reflog_info) {
2913                         save_parents(revs, commit);
2914                         fake_reflog_parent(revs->reflog_info, commit);
2915                         commit->object.flags &= ~(ADDED | SEEN | SHOWN);
2916                 }
2917
2918                 /*
2919                  * If we haven't done the list limiting, we need to look at
2920                  * the parents here. We also need to do the date-based limiting
2921                  * that we'd otherwise have done in limit_list().
2922                  */
2923                 if (!revs->limited) {
2924                         if (revs->max_age != -1 &&
2925                             (commit->date < revs->max_age))
2926                                 continue;
2927                         if (add_parents_to_list(revs, commit, &revs->commits, NULL) < 0)
2928                                 die("Failed to traverse parents of commit %s",
2929                                     sha1_to_hex(commit->object.sha1));
2930                 }
2931
2932                 switch (simplify_commit(revs, commit)) {
2933                 case commit_ignore:
2934                         continue;
2935                 case commit_error:
2936                         die("Failed to simplify parents of commit %s",
2937                             sha1_to_hex(commit->object.sha1));
2938                 default:
2939                         return commit;
2940                 }
2941         } while (revs->commits);
2942         return NULL;
2943 }
2944
2945 /*
2946  * Return true for entries that have not yet been shown.  (This is an
2947  * object_array_each_func_t.)
2948  */
2949 static int entry_unshown(struct object_array_entry *entry, void *cb_data_unused)
2950 {
2951         return !(entry->item->flags & SHOWN);
2952 }
2953
2954 /*
2955  * If array is on the verge of a realloc, garbage-collect any entries
2956  * that have already been shown to try to free up some space.
2957  */
2958 static void gc_boundary(struct object_array *array)
2959 {
2960         if (array->nr == array->alloc)
2961                 object_array_filter(array, entry_unshown, NULL);
2962 }
2963
2964 static void create_boundary_commit_list(struct rev_info *revs)
2965 {
2966         unsigned i;
2967         struct commit *c;
2968         struct object_array *array = &revs->boundary_commits;
2969         struct object_array_entry *objects = array->objects;
2970
2971         /*
2972          * If revs->commits is non-NULL at this point, an error occurred in
2973          * get_revision_1().  Ignore the error and continue printing the
2974          * boundary commits anyway.  (This is what the code has always
2975          * done.)
2976          */
2977         if (revs->commits) {
2978                 free_commit_list(revs->commits);
2979                 revs->commits = NULL;
2980         }
2981
2982         /*
2983          * Put all of the actual boundary commits from revs->boundary_commits
2984          * into revs->commits
2985          */
2986         for (i = 0; i < array->nr; i++) {
2987                 c = (struct commit *)(objects[i].item);
2988                 if (!c)
2989                         continue;
2990                 if (!(c->object.flags & CHILD_SHOWN))
2991                         continue;
2992                 if (c->object.flags & (SHOWN | BOUNDARY))
2993                         continue;
2994                 c->object.flags |= BOUNDARY;
2995                 commit_list_insert(c, &revs->commits);
2996         }
2997
2998         /*
2999          * If revs->topo_order is set, sort the boundary commits
3000          * in topological order
3001          */
3002         sort_in_topological_order(&revs->commits, revs->sort_order);
3003 }
3004
3005 static struct commit *get_revision_internal(struct rev_info *revs)
3006 {
3007         struct commit *c = NULL;
3008         struct commit_list *l;
3009
3010         if (revs->boundary == 2) {
3011                 /*
3012                  * All of the normal commits have already been returned,
3013                  * and we are now returning boundary commits.
3014                  * create_boundary_commit_list() has populated
3015                  * revs->commits with the remaining commits to return.
3016                  */
3017                 c = pop_commit(&revs->commits);
3018                 if (c)
3019                         c->object.flags |= SHOWN;
3020                 return c;
3021         }
3022
3023         /*
3024          * If our max_count counter has reached zero, then we are done. We
3025          * don't simply return NULL because we still might need to show
3026          * boundary commits. But we want to avoid calling get_revision_1, which
3027          * might do a considerable amount of work finding the next commit only
3028          * for us to throw it away.
3029          *
3030          * If it is non-zero, then either we don't have a max_count at all
3031          * (-1), or it is still counting, in which case we decrement.
3032          */
3033         if (revs->max_count) {
3034                 c = get_revision_1(revs);
3035                 if (c) {
3036                         while (revs->skip_count > 0) {
3037                                 revs->skip_count--;
3038                                 c = get_revision_1(revs);
3039                                 if (!c)
3040                                         break;
3041                         }
3042                 }
3043
3044                 if (revs->max_count > 0)
3045                         revs->max_count--;
3046         }
3047
3048         if (c)
3049                 c->object.flags |= SHOWN;
3050
3051         if (!revs->boundary)
3052                 return c;
3053
3054         if (!c) {
3055                 /*
3056                  * get_revision_1() runs out the commits, and
3057                  * we are done computing the boundaries.
3058                  * switch to boundary commits output mode.
3059                  */
3060                 revs->boundary = 2;
3061
3062                 /*
3063                  * Update revs->commits to contain the list of
3064                  * boundary commits.
3065                  */
3066                 create_boundary_commit_list(revs);
3067
3068                 return get_revision_internal(revs);
3069         }
3070
3071         /*
3072          * boundary commits are the commits that are parents of the
3073          * ones we got from get_revision_1() but they themselves are
3074          * not returned from get_revision_1().  Before returning
3075          * 'c', we need to mark its parents that they could be boundaries.
3076          */
3077
3078         for (l = c->parents; l; l = l->next) {
3079                 struct object *p;
3080                 p = &(l->item->object);
3081                 if (p->flags & (CHILD_SHOWN | SHOWN))
3082                         continue;
3083                 p->flags |= CHILD_SHOWN;
3084                 gc_boundary(&revs->boundary_commits);
3085                 add_object_array(p, NULL, &revs->boundary_commits);
3086         }
3087
3088         return c;
3089 }
3090
3091 struct commit *get_revision(struct rev_info *revs)
3092 {
3093         struct commit *c;
3094         struct commit_list *reversed;
3095
3096         if (revs->reverse) {
3097                 reversed = NULL;
3098                 while ((c = get_revision_internal(revs)))
3099                         commit_list_insert(c, &reversed);
3100                 revs->commits = reversed;
3101                 revs->reverse = 0;
3102                 revs->reverse_output_stage = 1;
3103         }
3104
3105         if (revs->reverse_output_stage)
3106                 return pop_commit(&revs->commits);
3107
3108         c = get_revision_internal(revs);
3109         if (c && revs->graph)
3110                 graph_update(revs->graph, c);
3111         if (!c)
3112                 free_saved_parents(revs);
3113         return c;
3114 }
3115
3116 char *get_revision_mark(const struct rev_info *revs, const struct commit *commit)
3117 {
3118         if (commit->object.flags & BOUNDARY)
3119                 return "-";
3120         else if (commit->object.flags & UNINTERESTING)
3121                 return "^";
3122         else if (commit->object.flags & PATCHSAME)
3123                 return "=";
3124         else if (!revs || revs->left_right) {
3125                 if (commit->object.flags & SYMMETRIC_LEFT)
3126                         return "<";
3127                 else
3128                         return ">";
3129         } else if (revs->graph)
3130                 return "*";
3131         else if (revs->cherry_mark)
3132                 return "+";
3133         return "";
3134 }
3135
3136 void put_revision_mark(const struct rev_info *revs, const struct commit *commit)
3137 {
3138         char *mark = get_revision_mark(revs, commit);
3139         if (!strlen(mark))
3140                 return;
3141         fputs(mark, stdout);
3142         putchar(' ');
3143 }
3144
3145 define_commit_slab(saved_parents, struct commit_list *);
3146
3147 #define EMPTY_PARENT_LIST ((struct commit_list *)-1)
3148
3149 void save_parents(struct rev_info *revs, struct commit *commit)
3150 {
3151         struct commit_list **pp;
3152
3153         if (!revs->saved_parents_slab) {
3154                 revs->saved_parents_slab = xmalloc(sizeof(struct saved_parents));
3155                 init_saved_parents(revs->saved_parents_slab);
3156         }
3157
3158         pp = saved_parents_at(revs->saved_parents_slab, commit);
3159
3160         /*
3161          * When walking with reflogs, we may visit the same commit
3162          * several times: once for each appearance in the reflog.
3163          *
3164          * In this case, save_parents() will be called multiple times.
3165          * We want to keep only the first set of parents.  We need to
3166          * store a sentinel value for an empty (i.e., NULL) parent
3167          * list to distinguish it from a not-yet-saved list, however.
3168          */
3169         if (*pp)
3170                 return;
3171         if (commit->parents)
3172                 *pp = copy_commit_list(commit->parents);
3173         else
3174                 *pp = EMPTY_PARENT_LIST;
3175 }
3176
3177 struct commit_list *get_saved_parents(struct rev_info *revs, const struct commit *commit)
3178 {
3179         struct commit_list *parents;
3180
3181         if (!revs->saved_parents_slab)
3182                 return commit->parents;
3183
3184         parents = *saved_parents_at(revs->saved_parents_slab, commit);
3185         if (parents == EMPTY_PARENT_LIST)
3186                 return NULL;
3187         return parents;
3188 }
3189
3190 void free_saved_parents(struct rev_info *revs)
3191 {
3192         if (revs->saved_parents_slab)
3193                 clear_saved_parents(revs->saved_parents_slab);
3194 }