commit.c

   1 #include "cache.h"
   2 #include "tag.h"
   3 #include "commit.h"
   4 #include "pkt-line.h"
   5 #include "utf8.h"
   6 #include "diff.h"
   7 #include "revision.h"
   8 #include "notes.h"
   9 #include "gpg-interface.h"
  10 #include "mergesort.h"
  11
  12 int save_commit_buffer = 1;
  13
  14 const char *commit_type = "commit";
  15
  16 static struct commit *check_commit(struct object *obj,
  17                                    const unsigned char *sha1,
  18                                    int quiet)
  19 {
  20         if (obj->type != OBJ_COMMIT) {
  21                 if (!quiet)
  22                         error("Object %s is a %s, not a commit",
  23                               sha1_to_hex(sha1), typename(obj->type));
  24                 return NULL;
  25         }
  26         return (struct commit *) obj;
  27 }
  28
  29 struct commit *lookup_commit_reference_gently(const unsigned char *sha1,
  30                                               int quiet)
  31 {
  32         struct object *obj = deref_tag(parse_object(sha1), NULL, 0);
  33
  34         if (!obj)
  35                 return NULL;
  36         return check_commit(obj, sha1, quiet);
  37 }
  38
  39 struct commit *lookup_commit_reference(const unsigned char *sha1)
  40 {
  41         return lookup_commit_reference_gently(sha1, 0);
  42 }
  43
  44 struct commit *lookup_commit_or_die(const unsigned char *sha1, const char *ref_name)
  45 {
  46         struct commit *c = lookup_commit_reference(sha1);
  47         if (!c)
  48                 die(_("could not parse %s"), ref_name);
  49         if (hashcmp(sha1, c->object.sha1)) {
  50                 warning(_("%s %s is not a commit!"),
  51                         ref_name, sha1_to_hex(sha1));
  52         }
  53         return c;
  54 }
  55
  56 struct commit *lookup_commit(const unsigned char *sha1)
  57 {
  58         struct object *obj = lookup_object(sha1);
  59         if (!obj)
  60                 return create_object(sha1, OBJ_COMMIT, alloc_commit_node());
  61         if (!obj->type)
  62                 obj->type = OBJ_COMMIT;
  63         return check_commit(obj, sha1, 0);
  64 }
  65
  66 struct commit *lookup_commit_reference_by_name(const char *name)
  67 {
  68         unsigned char sha1[20];
  69         struct commit *commit;
  70
  71         if (get_sha1_committish(name, sha1))
  72                 return NULL;
  73         commit = lookup_commit_reference(sha1);
  74         if (!commit || parse_commit(commit))
  75                 return NULL;
  76         return commit;
  77 }
  78
  79 static unsigned long parse_commit_date(const char *buf, const char *tail)
  80 {
  81         const char *dateptr;
  82
  83         if (buf + 6 >= tail)
  84                 return 0;
  85         if (memcmp(buf, "author", 6))
  86                 return 0;
  87         while (buf < tail && *buf++ != '\n')
  88                 /* nada */;
  89         if (buf + 9 >= tail)
  90                 return 0;
  91         if (memcmp(buf, "committer", 9))
  92                 return 0;
  93         while (buf < tail && *buf++ != '>')
  94                 /* nada */;
  95         if (buf >= tail)
  96                 return 0;
  97         dateptr = buf;
  98         while (buf < tail && *buf++ != '\n')
  99                 /* nada */;
 100         if (buf >= tail)
 101                 return 0;
 102         /* dateptr < buf && buf[-1] == '\n', so strtoul will stop at buf-1 */
 103         return strtoul(dateptr, NULL, 10);
 104 }
 105
 106 static struct commit_graft **commit_graft;
 107 static int commit_graft_alloc, commit_graft_nr;
 108
 109 static int commit_graft_pos(const unsigned char *sha1)
 110 {
 111         int lo, hi;
 112         lo = 0;
 113         hi = commit_graft_nr;
 114         while (lo < hi) {
 115                 int mi = (lo + hi) / 2;
 116                 struct commit_graft *graft = commit_graft[mi];
 117                 int cmp = hashcmp(sha1, graft->sha1);
 118                 if (!cmp)
 119                         return mi;
 120                 if (cmp < 0)
 121                         hi = mi;
 122                 else
 123                         lo = mi + 1;
 124         }
 125         return -lo - 1;
 126 }
 127
 128 int register_commit_graft(struct commit_graft *graft, int ignore_dups)
 129 {
 130         int pos = commit_graft_pos(graft->sha1);
 131
 132         if (0 <= pos) {
 133                 if (ignore_dups)
 134                         free(graft);
 135                 else {
 136                         free(commit_graft[pos]);
 137                         commit_graft[pos] = graft;
 138                 }
 139                 return 1;
 140         }
 141         pos = -pos - 1;
 142         if (commit_graft_alloc <= ++commit_graft_nr) {
 143                 commit_graft_alloc = alloc_nr(commit_graft_alloc);
 144                 commit_graft = xrealloc(commit_graft,
 145                                         sizeof(*commit_graft) *
 146                                         commit_graft_alloc);
 147         }
 148         if (pos < commit_graft_nr)
 149                 memmove(commit_graft + pos + 1,
 150                         commit_graft + pos,
 151                         (commit_graft_nr - pos - 1) *
 152                         sizeof(*commit_graft));
 153         commit_graft[pos] = graft;
 154         return 0;
 155 }
 156
 157 struct commit_graft *read_graft_line(char *buf, int len)
 158 {
 159         /* The format is just "Commit Parent1 Parent2 ...\n" */
 160         int i;
 161         struct commit_graft *graft = NULL;
 162
 163         while (len && isspace(buf[len-1]))
 164                 buf[--len] = '\0';
 165         if (buf[0] == '#' || buf[0] == '\0')
 166                 return NULL;
 167         if ((len + 1) % 41)
 168                 goto bad_graft_data;
 169         i = (len + 1) / 41 - 1;
 170         graft = xmalloc(sizeof(*graft) + 20 * i);
 171         graft->nr_parent = i;
 172         if (get_sha1_hex(buf, graft->sha1))
 173                 goto bad_graft_data;
 174         for (i = 40; i < len; i += 41) {
 175                 if (buf[i] != ' ')
 176                         goto bad_graft_data;
 177                 if (get_sha1_hex(buf + i + 1, graft->parent[i/41]))
 178                         goto bad_graft_data;
 179         }
 180         return graft;
 181
 182 bad_graft_data:
 183         error("bad graft data: %s", buf);
 184         free(graft);
 185         return NULL;
 186 }
 187
 188 static int read_graft_file(const char *graft_file)
 189 {
 190         FILE *fp = fopen(graft_file, "r");
 191         char buf[1024];
 192         if (!fp)
 193                 return -1;
 194         while (fgets(buf, sizeof(buf), fp)) {
 195                 /* The format is just "Commit Parent1 Parent2 ...\n" */
 196                 int len = strlen(buf);
 197                 struct commit_graft *graft = read_graft_line(buf, len);
 198                 if (!graft)
 199                         continue;
 200                 if (register_commit_graft(graft, 1))
 201                         error("duplicate graft data: %s", buf);
 202         }
 203         fclose(fp);
 204         return 0;
 205 }
 206
 207 static void prepare_commit_graft(void)
 208 {
 209         static int commit_graft_prepared;
 210         char *graft_file;
 211
 212         if (commit_graft_prepared)
 213                 return;
 214         graft_file = get_graft_file();
 215         read_graft_file(graft_file);
 216         /* make sure shallows are read */
 217         is_repository_shallow();
 218         commit_graft_prepared = 1;
 219 }
 220
 221 struct commit_graft *lookup_commit_graft(const unsigned char *sha1)
 222 {
 223         int pos;
 224         prepare_commit_graft();
 225         pos = commit_graft_pos(sha1);
 226         if (pos < 0)
 227                 return NULL;
 228         return commit_graft[pos];
 229 }
 230
 231 int for_each_commit_graft(each_commit_graft_fn fn, void *cb_data)
 232 {
 233         int i, ret;
 234         for (i = ret = 0; i < commit_graft_nr && !ret; i++)
 235                 ret = fn(commit_graft[i], cb_data);
 236         return ret;
 237 }
 238
 239 int unregister_shallow(const unsigned char *sha1)
 240 {
 241         int pos = commit_graft_pos(sha1);
 242         if (pos < 0)
 243                 return -1;
 244         if (pos + 1 < commit_graft_nr)
 245                 memmove(commit_graft + pos, commit_graft + pos + 1,
 246                                 sizeof(struct commit_graft *)
 247                                 * (commit_graft_nr - pos - 1));
 248         commit_graft_nr--;
 249         return 0;
 250 }
 251
 252 int parse_commit_buffer(struct commit *item, const void *buffer, unsigned long size)
 253 {
 254         const char *tail = buffer;
 255         const char *bufptr = buffer;
 256         unsigned char parent[20];
 257         struct commit_list **pptr;
 258         struct commit_graft *graft;
 259
 260         if (item->object.parsed)
 261                 return 0;
 262         item->object.parsed = 1;
 263         tail += size;
 264         if (tail <= bufptr + 46 || memcmp(bufptr, "tree ", 5) || bufptr[45] != '\n')
 265                 return error("bogus commit object %s", sha1_to_hex(item->object.sha1));
 266         if (get_sha1_hex(bufptr + 5, parent) < 0)
 267                 return error("bad tree pointer in commit %s",
 268                              sha1_to_hex(item->object.sha1));
 269         item->tree = lookup_tree(parent);
 270         bufptr += 46; /* "tree " + "hex sha1" + "\n" */
 271         pptr = &item->parents;
 272
 273         graft = lookup_commit_graft(item->object.sha1);
 274         while (bufptr + 48 < tail && !memcmp(bufptr, "parent ", 7)) {
 275                 struct commit *new_parent;
 276
 277                 if (tail <= bufptr + 48 ||
 278                     get_sha1_hex(bufptr + 7, parent) ||
 279                     bufptr[47] != '\n')
 280                         return error("bad parents in commit %s", sha1_to_hex(item->object.sha1));
 281                 bufptr += 48;
 282                 /*
 283                  * The clone is shallow if nr_parent < 0, and we must
 284                  * not traverse its real parents even when we unhide them.
 285                  */
 286                 if (graft && (graft->nr_parent < 0 || grafts_replace_parents))
 287                         continue;
 288                 new_parent = lookup_commit(parent);
 289                 if (new_parent)
 290                         pptr = &commit_list_insert(new_parent, pptr)->next;
 291         }
 292         if (graft) {
 293                 int i;
 294                 struct commit *new_parent;
 295                 for (i = 0; i < graft->nr_parent; i++) {
 296                         new_parent = lookup_commit(graft->parent[i]);
 297                         if (!new_parent)
 298                                 continue;
 299                         pptr = &commit_list_insert(new_parent, pptr)->next;
 300                 }
 301         }
 302         item->date = parse_commit_date(bufptr, tail);
 303
 304         return 0;
 305 }
 306
 307 int parse_commit(struct commit *item)
 308 {
 309         enum object_type type;
 310         void *buffer;
 311         unsigned long size;
 312         int ret;
 313
 314         if (!item)
 315                 return -1;
 316         if (item->object.parsed)
 317                 return 0;
 318         buffer = read_sha1_file(item->object.sha1, &type, &size);
 319         if (!buffer)
 320                 return error("Could not read %s",
 321                              sha1_to_hex(item->object.sha1));
 322         if (type != OBJ_COMMIT) {
 323                 free(buffer);
 324                 return error("Object %s not a commit",
 325                              sha1_to_hex(item->object.sha1));
 326         }
 327         ret = parse_commit_buffer(item, buffer, size);
 328         if (save_commit_buffer && !ret) {
 329                 item->buffer = buffer;
 330                 return 0;
 331         }
 332         free(buffer);
 333         return ret;
 334 }
 335
 336 int find_commit_subject(const char *commit_buffer, const char **subject)
 337 {
 338         const char *eol;
 339         const char *p = commit_buffer;
 340
 341         while (*p && (*p != '\n' || p[1] != '\n'))
 342                 p++;
 343         if (*p) {
 344                 p += 2;
 345                 for (eol = p; *eol && *eol != '\n'; eol++)
 346                         ; /* do nothing */
 347         } else
 348                 eol = p;
 349
 350         *subject = p;
 351
 352         return eol - p;
 353 }
 354
 355 struct commit_list *commit_list_insert(struct commit *item, struct commit_list **list_p)
 356 {
 357         struct commit_list *new_list = xmalloc(sizeof(struct commit_list));
 358         new_list->item = item;
 359         new_list->next = *list_p;
 360         *list_p = new_list;
 361         return new_list;
 362 }
 363
 364 unsigned commit_list_count(const struct commit_list *l)
 365 {
 366         unsigned c = 0;
 367         for (; l; l = l->next )
 368                 c++;
 369         return c;
 370 }
 371
 372 void free_commit_list(struct commit_list *list)
 373 {
 374         while (list) {
 375                 struct commit_list *temp = list;
 376                 list = temp->next;
 377                 free(temp);
 378         }
 379 }
 380
 381 struct commit_list * commit_list_insert_by_date(struct commit *item, struct commit_list **list)
 382 {
 383         struct commit_list **pp = list;
 384         struct commit_list *p;
 385         while ((p = *pp) != NULL) {
 386                 if (p->item->date < item->date) {
 387                         break;
 388                 }
 389                 pp = &p->next;
 390         }
 391         return commit_list_insert(item, pp);
 392 }
 393
 394 static int commit_list_compare_by_date(const void *a, const void *b)
 395 {
 396         unsigned long a_date = ((const struct commit_list *)a)->item->date;
 397         unsigned long b_date = ((const struct commit_list *)b)->item->date;
 398         if (a_date < b_date)
 399                 return 1;
 400         if (a_date > b_date)
 401                 return -1;
 402         return 0;
 403 }
 404
 405 static void *commit_list_get_next(const void *a)
 406 {
 407         return ((const struct commit_list *)a)->next;
 408 }
 409
 410 static void commit_list_set_next(void *a, void *next)
 411 {
 412         ((struct commit_list *)a)->next = next;
 413 }
 414
 415 void commit_list_sort_by_date(struct commit_list **list)
 416 {
 417         *list = llist_mergesort(*list, commit_list_get_next, commit_list_set_next,
 418                                 commit_list_compare_by_date);
 419 }
 420
 421 struct commit *pop_most_recent_commit(struct commit_list **list,
 422                                       unsigned int mark)
 423 {
 424         struct commit *ret = (*list)->item;
 425         struct commit_list *parents = ret->parents;
 426         struct commit_list *old = *list;
 427
 428         *list = (*list)->next;
 429         free(old);
 430
 431         while (parents) {
 432                 struct commit *commit = parents->item;
 433                 if (!parse_commit(commit) && !(commit->object.flags & mark)) {
 434                         commit->object.flags |= mark;
 435                         commit_list_insert_by_date(commit, list);
 436                 }
 437                 parents = parents->next;
 438         }
 439         return ret;
 440 }
 441
 442 static void clear_commit_marks_1(struct commit_list **plist,
 443                                  struct commit *commit, unsigned int mark)
 444 {
 445         while (commit) {
 446                 struct commit_list *parents;
 447
 448                 if (!(mark & commit->object.flags))
 449                         return;
 450
 451                 commit->object.flags &= ~mark;
 452
 453                 parents = commit->parents;
 454                 if (!parents)
 455                         return;
 456
 457                 while ((parents = parents->next))
 458                         commit_list_insert(parents->item, plist);
 459
 460                 commit = commit->parents->item;
 461         }
 462 }
 463
 464 void clear_commit_marks(struct commit *commit, unsigned int mark)
 465 {
 466         struct commit_list *list = NULL;
 467         commit_list_insert(commit, &list);
 468         while (list)
 469                 clear_commit_marks_1(&list, pop_commit(&list), mark);
 470 }
 471
 472 void clear_commit_marks_for_object_array(struct object_array *a, unsigned mark)
 473 {
 474         struct object *object;
 475         struct commit *commit;
 476         unsigned int i;
 477
 478         for (i = 0; i < a->nr; i++) {
 479                 object = a->objects[i].item;
 480                 commit = lookup_commit_reference_gently(object->sha1, 1);
 481                 if (commit)
 482                         clear_commit_marks(commit, mark);
 483         }
 484 }
 485
 486 struct commit *pop_commit(struct commit_list **stack)
 487 {
 488         struct commit_list *top = *stack;
 489         struct commit *item = top ? top->item : NULL;
 490
 491         if (top) {
 492                 *stack = top->next;
 493                 free(top);
 494         }
 495         return item;
 496 }
 497
 498 /*
 499  * Performs an in-place topological sort on the list supplied.
 500  */
 501 void sort_in_topological_order(struct commit_list ** list, int lifo)
 502 {
 503         struct commit_list *next, *orig = *list;
 504         struct commit_list *work, **insert;
 505         struct commit_list **pptr;
 506
 507         if (!orig)
 508                 return;
 509         *list = NULL;
 510
 511         /* Mark them and clear the indegree */
 512         for (next = orig; next; next = next->next) {
 513                 struct commit *commit = next->item;
 514                 commit->indegree = 1;
 515         }
 516
 517         /* update the indegree */
 518         for (next = orig; next; next = next->next) {
 519                 struct commit_list * parents = next->item->parents;
 520                 while (parents) {
 521                         struct commit *parent = parents->item;
 522
 523                         if (parent->indegree)
 524                                 parent->indegree++;
 525                         parents = parents->next;
 526                 }
 527         }
 528
 529         /*
 530          * find the tips
 531          *
 532          * tips are nodes not reachable from any other node in the list
 533          *
 534          * the tips serve as a starting set for the work queue.
 535          */
 536         work = NULL;
 537         insert = &work;
 538         for (next = orig; next; next = next->next) {
 539                 struct commit *commit = next->item;
 540
 541                 if (commit->indegree == 1)
 542                         insert = &commit_list_insert(commit, insert)->next;
 543         }
 544
 545         /* process the list in topological order */
 546         if (!lifo)
 547                 commit_list_sort_by_date(&work);
 548
 549         pptr = list;
 550         *list = NULL;
 551         while (work) {
 552                 struct commit *commit;
 553                 struct commit_list *parents, *work_item;
 554
 555                 work_item = work;
 556                 work = work_item->next;
 557                 work_item->next = NULL;
 558
 559                 commit = work_item->item;
 560                 for (parents = commit->parents; parents ; parents = parents->next) {
 561                         struct commit *parent = parents->item;
 562
 563                         if (!parent->indegree)
 564                                 continue;
 565
 566                         /*
 567                          * parents are only enqueued for emission
 568                          * when all their children have been emitted thereby
 569                          * guaranteeing topological order.
 570                          */
 571                         if (--parent->indegree == 1) {
 572                                 if (!lifo)
 573                                         commit_list_insert_by_date(parent, &work);
 574                                 else
 575                                         commit_list_insert(parent, &work);
 576                         }
 577                 }
 578                 /*
 579                  * work_item is a commit all of whose children
 580                  * have already been emitted. we can emit it now.
 581                  */
 582                 commit->indegree = 0;
 583                 *pptr = work_item;
 584                 pptr = &work_item->next;
 585         }
 586 }
 587
 588 /* merge-base stuff */
 589
 590 /* bits #0..15 in revision.h */
 591 #define PARENT1         (1u<<16)
 592 #define PARENT2         (1u<<17)
 593 #define STALE           (1u<<18)
 594 #define RESULT          (1u<<19)
 595
 596 static const unsigned all_flags = (PARENT1 | PARENT2 | STALE | RESULT);
 597
 598 static struct commit *interesting(struct commit_list *list)
 599 {
 600         while (list) {
 601                 struct commit *commit = list->item;
 602                 list = list->next;
 603                 if (commit->object.flags & STALE)
 604                         continue;
 605                 return commit;
 606         }
 607         return NULL;
 608 }
 609
 610 static struct commit_list *paint_down_to_common(struct commit *one, int n, struct commit **twos)
 611 {
 612         struct commit_list *list = NULL;
 613         struct commit_list *result = NULL;
 614         int i;
 615
 616         one->object.flags |= PARENT1;
 617         commit_list_insert_by_date(one, &list);
 618         for (i = 0; i < n; i++) {
 619                 twos[i]->object.flags |= PARENT2;
 620                 commit_list_insert_by_date(twos[i], &list);
 621         }
 622
 623         while (interesting(list)) {
 624                 struct commit *commit;
 625                 struct commit_list *parents;
 626                 struct commit_list *next;
 627                 int flags;
 628
 629                 commit = list->item;
 630                 next = list->next;
 631                 free(list);
 632                 list = next;
 633
 634                 flags = commit->object.flags & (PARENT1 | PARENT2 | STALE);
 635                 if (flags == (PARENT1 | PARENT2)) {
 636                         if (!(commit->object.flags & RESULT)) {
 637                                 commit->object.flags |= RESULT;
 638                                 commit_list_insert_by_date(commit, &result);
 639                         }
 640                         /* Mark parents of a found merge stale */
 641                         flags |= STALE;
 642                 }
 643                 parents = commit->parents;
 644                 while (parents) {
 645                         struct commit *p = parents->item;
 646                         parents = parents->next;
 647                         if ((p->object.flags & flags) == flags)
 648                                 continue;
 649                         if (parse_commit(p))
 650                                 return NULL;
 651                         p->object.flags |= flags;
 652                         commit_list_insert_by_date(p, &list);
 653                 }
 654         }
 655
 656         free_commit_list(list);
 657         return result;
 658 }
 659
 660 static struct commit_list *merge_bases_many(struct commit *one, int n, struct commit **twos)
 661 {
 662         struct commit_list *list = NULL;
 663         struct commit_list *result = NULL;
 664         int i;
 665
 666         for (i = 0; i < n; i++) {
 667                 if (one == twos[i])
 668                         /*
 669                          * We do not mark this even with RESULT so we do not
 670                          * have to clean it up.
 671                          */
 672                         return commit_list_insert(one, &result);
 673         }
 674
 675         if (parse_commit(one))
 676                 return NULL;
 677         for (i = 0; i < n; i++) {
 678                 if (parse_commit(twos[i]))
 679                         return NULL;
 680         }
 681
 682         list = paint_down_to_common(one, n, twos);
 683
 684         while (list) {
 685                 struct commit_list *next = list->next;
 686                 if (!(list->item->object.flags & STALE))
 687                         commit_list_insert_by_date(list->item, &result);
 688                 free(list);
 689                 list = next;
 690         }
 691         return result;
 692 }
 693
 694 struct commit_list *get_octopus_merge_bases(struct commit_list *in)
 695 {
 696         struct commit_list *i, *j, *k, *ret = NULL;
 697         struct commit_list **pptr = &ret;
 698
 699         for (i = in; i; i = i->next) {
 700                 if (!ret)
 701                         pptr = &commit_list_insert(i->item, pptr)->next;
 702                 else {
 703                         struct commit_list *new = NULL, *end = NULL;
 704
 705                         for (j = ret; j; j = j->next) {
 706                                 struct commit_list *bases;
 707                                 bases = get_merge_bases(i->item, j->item, 1);
 708                                 if (!new)
 709                                         new = bases;
 710                                 else
 711                                         end->next = bases;
 712                                 for (k = bases; k; k = k->next)
 713                                         end = k;
 714                         }
 715                         ret = new;
 716                 }
 717         }
 718         return ret;
 719 }
 720
 721 static int remove_redundant(struct commit **array, int cnt)
 722 {
 723         /*
 724          * Some commit in the array may be an ancestor of
 725          * another commit.  Move such commit to the end of
 726          * the array, and return the number of commits that
 727          * are independent from each other.
 728          */
 729         struct commit **work;
 730         unsigned char *redundant;
 731         int *filled_index;
 732         int i, j, filled;
 733
 734         work = xcalloc(cnt, sizeof(*work));
 735         redundant = xcalloc(cnt, 1);
 736         filled_index = xmalloc(sizeof(*filled_index) * (cnt - 1));
 737
 738         for (i = 0; i < cnt; i++) {
 739                 struct commit_list *common;
 740
 741                 if (redundant[i])
 742                         continue;
 743                 for (j = filled = 0; j < cnt; j++) {
 744                         if (i == j || redundant[j])
 745                                 continue;
 746                         filled_index[filled] = j;
 747                         work[filled++] = array[j];
 748                 }
 749                 common = paint_down_to_common(array[i], filled, work);
 750                 if (array[i]->object.flags & PARENT2)
 751                         redundant[i] = 1;
 752                 for (j = 0; j < filled; j++)
 753                         if (work[j]->object.flags & PARENT1)
 754                                 redundant[filled_index[j]] = 1;
 755                 clear_commit_marks(array[i], all_flags);
 756                 for (j = 0; j < filled; j++)
 757                         clear_commit_marks(work[j], all_flags);
 758                 free_commit_list(common);
 759         }
 760
 761         /* Now collect the result */
 762         memcpy(work, array, sizeof(*array) * cnt);
 763         for (i = filled = 0; i < cnt; i++)
 764                 if (!redundant[i])
 765                         array[filled++] = work[i];
 766         for (j = filled, i = 0; i < cnt; i++)
 767                 if (redundant[i])
 768                         array[j++] = work[i];
 769         free(work);
 770         free(redundant);
 771         free(filled_index);
 772         return filled;
 773 }
 774
 775 struct commit_list *get_merge_bases_many(struct commit *one,
 776                                          int n,
 777                                          struct commit **twos,
 778                                          int cleanup)
 779 {
 780         struct commit_list *list;
 781         struct commit **rslt;
 782         struct commit_list *result;
 783         int cnt, i;
 784
 785         result = merge_bases_many(one, n, twos);
 786         for (i = 0; i < n; i++) {
 787                 if (one == twos[i])
 788                         return result;
 789         }
 790         if (!result || !result->next) {
 791                 if (cleanup) {
 792                         clear_commit_marks(one, all_flags);
 793                         for (i = 0; i < n; i++)
 794                                 clear_commit_marks(twos[i], all_flags);
 795                 }
 796                 return result;
 797         }
 798
 799         /* There are more than one */
 800         cnt = 0;
 801         list = result;
 802         while (list) {
 803                 list = list->next;
 804                 cnt++;
 805         }
 806         rslt = xcalloc(cnt, sizeof(*rslt));
 807         for (list = result, i = 0; list; list = list->next)
 808                 rslt[i++] = list->item;
 809         free_commit_list(result);
 810
 811         clear_commit_marks(one, all_flags);
 812         for (i = 0; i < n; i++)
 813                 clear_commit_marks(twos[i], all_flags);
 814
 815         cnt = remove_redundant(rslt, cnt);
 816         result = NULL;
 817         for (i = 0; i < cnt; i++)
 818                 commit_list_insert_by_date(rslt[i], &result);
 819         free(rslt);
 820         return result;
 821 }
 822
 823 struct commit_list *get_merge_bases(struct commit *one, struct commit *two,
 824                                     int cleanup)
 825 {
 826         return get_merge_bases_many(one, 1, &two, cleanup);
 827 }
 828
 829 /*
 830  * Is "commit" a decendant of one of the elements on the "with_commit" list?
 831  */
 832 int is_descendant_of(struct commit *commit, struct commit_list *with_commit)
 833 {
 834         if (!with_commit)
 835                 return 1;
 836         while (with_commit) {
 837                 struct commit *other;
 838
 839                 other = with_commit->item;
 840                 with_commit = with_commit->next;
 841                 if (in_merge_bases(other, commit))
 842                         return 1;
 843         }
 844         return 0;
 845 }
 846
 847 /*
 848  * Is "commit" an ancestor of (i.e. reachable from) the "reference"?
 849  */
 850 int in_merge_bases(struct commit *commit, struct commit *reference)
 851 {
 852         struct commit_list *bases;
 853         int ret = 0;
 854
 855         if (parse_commit(commit) || parse_commit(reference))
 856                 return ret;
 857
 858         bases = paint_down_to_common(commit, 1, &reference);
 859         if (commit->object.flags & PARENT2)
 860                 ret = 1;
 861         clear_commit_marks(commit, all_flags);
 862         clear_commit_marks(reference, all_flags);
 863         free_commit_list(bases);
 864         return ret;
 865 }
 866
 867 struct commit_list *reduce_heads(struct commit_list *heads)
 868 {
 869         struct commit_list *p;
 870         struct commit_list *result = NULL, **tail = &result;
 871         struct commit **array;
 872         int num_head, i;
 873
 874         if (!heads)
 875                 return NULL;
 876
 877         /* Uniquify */
 878         for (p = heads; p; p = p->next)
 879                 p->item->object.flags &= ~STALE;
 880         for (p = heads, num_head = 0; p; p = p->next) {
 881                 if (p->item->object.flags & STALE)
 882                         continue;
 883                 p->item->object.flags |= STALE;
 884                 num_head++;
 885         }
 886         array = xcalloc(sizeof(*array), num_head);
 887         for (p = heads, i = 0; p; p = p->next) {
 888                 if (p->item->object.flags & STALE) {
 889                         array[i++] = p->item;
 890                         p->item->object.flags &= ~STALE;
 891                 }
 892         }
 893         num_head = remove_redundant(array, num_head);
 894         for (i = 0; i < num_head; i++)
 895                 tail = &commit_list_insert(array[i], tail)->next;
 896         return result;
 897 }
 898
 899 static const char gpg_sig_header[] = "gpgsig";
 900 static const int gpg_sig_header_len = sizeof(gpg_sig_header) - 1;
 901
 902 static int do_sign_commit(struct strbuf *buf, const char *keyid)
 903 {
 904         struct strbuf sig = STRBUF_INIT;
 905         int inspos, copypos;
 906
 907         /* find the end of the header */
 908         inspos = strstr(buf->buf, "\n\n") - buf->buf + 1;
 909
 910         if (!keyid || !*keyid)
 911                 keyid = get_signing_key();
 912         if (sign_buffer(buf, &sig, keyid)) {
 913                 strbuf_release(&sig);
 914                 return -1;
 915         }
 916
 917         for (copypos = 0; sig.buf[copypos]; ) {
 918                 const char *bol = sig.buf + copypos;
 919                 const char *eol = strchrnul(bol, '\n');
 920                 int len = (eol - bol) + !!*eol;
 921
 922                 if (!copypos) {
 923                         strbuf_insert(buf, inspos, gpg_sig_header, gpg_sig_header_len);
 924                         inspos += gpg_sig_header_len;
 925                 }
 926                 strbuf_insert(buf, inspos++, " ", 1);
 927                 strbuf_insert(buf, inspos, bol, len);
 928                 inspos += len;
 929                 copypos += len;
 930         }
 931         strbuf_release(&sig);
 932         return 0;
 933 }
 934
 935 int parse_signed_commit(const unsigned char *sha1,
 936                         struct strbuf *payload, struct strbuf *signature)
 937 {
 938         unsigned long size;
 939         enum object_type type;
 940         char *buffer = read_sha1_file(sha1, &type, &size);
 941         int in_signature, saw_signature = -1;
 942         char *line, *tail;
 943
 944         if (!buffer || type != OBJ_COMMIT)
 945                 goto cleanup;
 946
 947         line = buffer;
 948         tail = buffer + size;
 949         in_signature = 0;
 950         saw_signature = 0;
 951         while (line < tail) {
 952                 const char *sig = NULL;
 953                 char *next = memchr(line, '\n', tail - line);
 954
 955                 next = next ? next + 1 : tail;
 956                 if (in_signature && line[0] == ' ')
 957                         sig = line + 1;
 958                 else if (!prefixcmp(line, gpg_sig_header) &&
 959                          line[gpg_sig_header_len] == ' ')
 960                         sig = line + gpg_sig_header_len + 1;
 961                 if (sig) {
 962                         strbuf_add(signature, sig, next - sig);
 963                         saw_signature = 1;
 964                         in_signature = 1;
 965                 } else {
 966                         if (*line == '\n')
 967                                 /* dump the whole remainder of the buffer */
 968                                 next = tail;
 969                         strbuf_add(payload, line, next - line);
 970                         in_signature = 0;
 971                 }
 972                 line = next;
 973         }
 974  cleanup:
 975         free(buffer);
 976         return saw_signature;
 977 }
 978
 979 static void handle_signed_tag(struct commit *parent, struct commit_extra_header ***tail)
 980 {
 981         struct merge_remote_desc *desc;
 982         struct commit_extra_header *mergetag;
 983         char *buf;
 984         unsigned long size, len;
 985         enum object_type type;
 986
 987         desc = merge_remote_util(parent);
 988         if (!desc || !desc->obj)
 989                 return;
 990         buf = read_sha1_file(desc->obj->sha1, &type, &size);
 991         if (!buf || type != OBJ_TAG)
 992                 goto free_return;
 993         len = parse_signature(buf, size);
 994         if (size == len)
 995                 goto free_return;
 996         /*
 997          * We could verify this signature and either omit the tag when
 998          * it does not validate, but the integrator may not have the
 999          * public key of the signer of the tag he is merging, while a
1000          * later auditor may have it while auditing, so let's not run
1001          * verify-signed-buffer here for now...
1002          *
1003          * if (verify_signed_buffer(buf, len, buf + len, size - len, ...))
1004          *      warn("warning: signed tag unverified.");
1005          */
1006         mergetag = xcalloc(1, sizeof(*mergetag));
1007         mergetag->key = xstrdup("mergetag");
1008         mergetag->value = buf;
1009         mergetag->len = size;
1010
1011         **tail = mergetag;
1012         *tail = &mergetag->next;
1013         return;
1014
1015 free_return:
1016         free(buf);
1017 }
1018
1019 void append_merge_tag_headers(struct commit_list *parents,
1020                               struct commit_extra_header ***tail)
1021 {
1022         while (parents) {
1023                 struct commit *parent = parents->item;
1024                 handle_signed_tag(parent, tail);
1025                 parents = parents->next;
1026         }
1027 }
1028
1029 static void add_extra_header(struct strbuf *buffer,
1030                              struct commit_extra_header *extra)
1031 {
1032         strbuf_addstr(buffer, extra->key);
1033         if (extra->len)
1034                 strbuf_add_lines(buffer, " ", extra->value, extra->len);
1035         else
1036                 strbuf_addch(buffer, '\n');
1037 }
1038
1039 struct commit_extra_header *read_commit_extra_headers(struct commit *commit,
1040                                                       const char **exclude)
1041 {
1042         struct commit_extra_header *extra = NULL;
1043         unsigned long size;
1044         enum object_type type;
1045         char *buffer = read_sha1_file(commit->object.sha1, &type, &size);
1046         if (buffer && type == OBJ_COMMIT)
1047                 extra = read_commit_extra_header_lines(buffer, size, exclude);
1048         free(buffer);
1049         return extra;
1050 }
1051
1052 static inline int standard_header_field(const char *field, size_t len)
1053 {
1054         return ((len == 4 && !memcmp(field, "tree ", 5)) ||
1055                 (len == 6 && !memcmp(field, "parent ", 7)) ||
1056                 (len == 6 && !memcmp(field, "author ", 7)) ||
1057                 (len == 9 && !memcmp(field, "committer ", 10)) ||
1058                 (len == 8 && !memcmp(field, "encoding ", 9)));
1059 }
1060
1061 static int excluded_header_field(const char *field, size_t len, const char **exclude)
1062 {
1063         if (!exclude)
1064                 return 0;
1065
1066         while (*exclude) {
1067                 size_t xlen = strlen(*exclude);
1068                 if (len == xlen &&
1069                     !memcmp(field, *exclude, xlen) && field[xlen] == ' ')
1070                         return 1;
1071                 exclude++;
1072         }
1073         return 0;
1074 }
1075
1076 struct commit_extra_header *read_commit_extra_header_lines(const char *buffer, size_t size,
1077                                                            const char **exclude)
1078 {
1079         struct commit_extra_header *extra = NULL, **tail = &extra, *it = NULL;
1080         const char *line, *next, *eof, *eob;
1081         struct strbuf buf = STRBUF_INIT;
1082
1083         for (line = buffer, eob = line + size;
1084              line < eob && *line != '\n';
1085              line = next) {
1086                 next = memchr(line, '\n', eob - line);
1087                 next = next ? next + 1 : eob;
1088                 if (*line == ' ') {
1089                         /* continuation */
1090                         if (it)
1091                                 strbuf_add(&buf, line + 1, next - (line + 1));
1092                         continue;
1093                 }
1094                 if (it)
1095                         it->value = strbuf_detach(&buf, &it->len);
1096                 strbuf_reset(&buf);
1097                 it = NULL;
1098
1099                 eof = strchr(line, ' ');
1100                 if (next <= eof)
1101                         eof = next;
1102
1103                 if (standard_header_field(line, eof - line) ||
1104                     excluded_header_field(line, eof - line, exclude))
1105                         continue;
1106
1107                 it = xcalloc(1, sizeof(*it));
1108                 it->key = xmemdupz(line, eof-line);
1109                 *tail = it;
1110                 tail = &it->next;
1111                 if (eof + 1 < next)
1112                         strbuf_add(&buf, eof + 1, next - (eof + 1));
1113         }
1114         if (it)
1115                 it->value = strbuf_detach(&buf, &it->len);
1116         return extra;
1117 }
1118
1119 void free_commit_extra_headers(struct commit_extra_header *extra)
1120 {
1121         while (extra) {
1122                 struct commit_extra_header *next = extra->next;
1123                 free(extra->key);
1124                 free(extra->value);
1125                 free(extra);
1126                 extra = next;
1127         }
1128 }
1129
1130 int commit_tree(const struct strbuf *msg, unsigned char *tree,
1131                 struct commit_list *parents, unsigned char *ret,
1132                 const char *author, const char *sign_commit)
1133 {
1134         struct commit_extra_header *extra = NULL, **tail = &extra;
1135         int result;
1136
1137         append_merge_tag_headers(parents, &tail);
1138         result = commit_tree_extended(msg, tree, parents, ret,
1139                                       author, sign_commit, extra);
1140         free_commit_extra_headers(extra);
1141         return result;
1142 }
1143
1144 static int find_invalid_utf8(const char *buf, int len)
1145 {
1146         int offset = 0;
1147
1148         while (len) {
1149                 unsigned char c = *buf++;
1150                 int bytes, bad_offset;
1151
1152                 len--;
1153                 offset++;
1154
1155                 /* Simple US-ASCII? No worries. */
1156                 if (c < 0x80)
1157                         continue;
1158
1159                 bad_offset = offset-1;
1160
1161                 /*
1162                  * Count how many more high bits set: that's how
1163                  * many more bytes this sequence should have.
1164                  */
1165                 bytes = 0;
1166                 while (c & 0x40) {
1167                         c <<= 1;
1168                         bytes++;
1169                 }
1170
1171                 /* Must be between 1 and 5 more bytes */
1172                 if (bytes < 1 || bytes > 5)
1173                         return bad_offset;
1174
1175                 /* Do we *have* that many bytes? */
1176                 if (len < bytes)
1177                         return bad_offset;
1178
1179                 offset += bytes;
1180                 len -= bytes;
1181
1182                 /* And verify that they are good continuation bytes */
1183                 do {
1184                         if ((*buf++ & 0xc0) != 0x80)
1185                                 return bad_offset;
1186                 } while (--bytes);
1187
1188                 /* We could/should check the value and length here too */
1189         }
1190         return -1;
1191 }
1192
1193 /*
1194  * This verifies that the buffer is in proper utf8 format.
1195  *
1196  * If it isn't, it assumes any non-utf8 characters are Latin1,
1197  * and does the conversion.
1198  *
1199  * Fixme: we should probably also disallow overlong forms and
1200  * invalid characters. But we don't do that currently.
1201  */
1202 static int verify_utf8(struct strbuf *buf)
1203 {
1204         int ok = 1;
1205         long pos = 0;
1206
1207         for (;;) {
1208                 int bad;
1209                 unsigned char c;
1210                 unsigned char replace[2];
1211
1212                 bad = find_invalid_utf8(buf->buf + pos, buf->len - pos);
1213                 if (bad < 0)
1214                         return ok;
1215                 pos += bad;
1216                 ok = 0;
1217                 c = buf->buf[pos];
1218                 strbuf_remove(buf, pos, 1);
1219
1220                 /* We know 'c' must be in the range 128-255 */
1221                 replace[0] = 0xc0 + (c >> 6);
1222                 replace[1] = 0x80 + (c & 0x3f);
1223                 strbuf_insert(buf, pos, replace, 2);
1224                 pos += 2;
1225         }
1226 }
1227
1228 static const char commit_utf8_warn[] =
1229 "Warning: commit message did not conform to UTF-8.\n"
1230 "You may want to amend it after fixing the message, or set the config\n"
1231 "variable i18n.commitencoding to the encoding your project uses.\n";
1232
1233 int commit_tree_extended(const struct strbuf *msg, unsigned char *tree,
1234                          struct commit_list *parents, unsigned char *ret,
1235                          const char *author, const char *sign_commit,
1236                          struct commit_extra_header *extra)
1237 {
1238         int result;
1239         int encoding_is_utf8;
1240         struct strbuf buffer;
1241
1242         assert_sha1_type(tree, OBJ_TREE);
1243
1244         if (memchr(msg->buf, '\0', msg->len))
1245                 return error("a NUL byte in commit log message not allowed.");
1246
1247         /* Not having i18n.commitencoding is the same as having utf-8 */
1248         encoding_is_utf8 = is_encoding_utf8(git_commit_encoding);
1249
1250         strbuf_init(&buffer, 8192); /* should avoid reallocs for the headers */
1251         strbuf_addf(&buffer, "tree %s\n", sha1_to_hex(tree));
1252
1253         /*
1254          * NOTE! This ordering means that the same exact tree merged with a
1255          * different order of parents will be a _different_ changeset even
1256          * if everything else stays the same.
1257          */
1258         while (parents) {
1259                 struct commit_list *next = parents->next;
1260                 struct commit *parent = parents->item;
1261
1262                 strbuf_addf(&buffer, "parent %s\n",
1263                             sha1_to_hex(parent->object.sha1));
1264                 free(parents);
1265                 parents = next;
1266         }
1267
1268         /* Person/date information */
1269         if (!author)
1270                 author = git_author_info(IDENT_STRICT);
1271         strbuf_addf(&buffer, "author %s\n", author);
1272         strbuf_addf(&buffer, "committer %s\n", git_committer_info(IDENT_STRICT));
1273         if (!encoding_is_utf8)
1274                 strbuf_addf(&buffer, "encoding %s\n", git_commit_encoding);
1275
1276         while (extra) {
1277                 add_extra_header(&buffer, extra);
1278                 extra = extra->next;
1279         }
1280         strbuf_addch(&buffer, '\n');
1281
1282         /* And add the comment */
1283         strbuf_addbuf(&buffer, msg);
1284
1285         /* And check the encoding */
1286         if (encoding_is_utf8 && !verify_utf8(&buffer))
1287                 fprintf(stderr, commit_utf8_warn);
1288
1289         if (sign_commit && do_sign_commit(&buffer, sign_commit))
1290                 return -1;
1291
1292         result = write_sha1_file(buffer.buf, buffer.len, commit_type, ret);
1293         strbuf_release(&buffer);
1294         return result;
1295 }
1296
1297 struct commit *get_merge_parent(const char *name)
1298 {
1299         struct object *obj;
1300         struct commit *commit;
1301         unsigned char sha1[20];
1302         if (get_sha1(name, sha1))
1303                 return NULL;
1304         obj = parse_object(sha1);
1305         commit = (struct commit *)peel_to_type(name, 0, obj, OBJ_COMMIT);
1306         if (commit && !commit->util) {
1307                 struct merge_remote_desc *desc;
1308                 desc = xmalloc(sizeof(*desc));
1309                 desc->obj = obj;
1310                 desc->name = strdup(name);
1311                 commit->util = desc;
1312         }
1313         return commit;
1314 }
1315
1316 /*
1317  * Append a commit to the end of the commit_list.
1318  *
1319  * next starts by pointing to the variable that holds the head of an
1320  * empty commit_list, and is updated to point to the "next" field of
1321  * the last item on the list as new commits are appended.
1322  *
1323  * Usage example:
1324  *
1325  *     struct commit_list *list;
1326  *     struct commit_list **next = &list;
1327  *
1328  *     next = commit_list_append(c1, next);
1329  *     next = commit_list_append(c2, next);
1330  *     assert(commit_list_count(list) == 2);
1331  *     return list;
1332  */
1333 struct commit_list **commit_list_append(struct commit *commit,
1334                                         struct commit_list **next)
1335 {
1336         struct commit_list *new = xmalloc(sizeof(struct commit_list));
1337         new->item = commit;
1338         *next = new;
1339         new->next = NULL;
1340         return &new->next;
1341 }