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 static struct commit_extra_header *read_commit_extra_header_lines(const char *buf, size_t len, const char **);
  13
  14 int save_commit_buffer = 1;
  15
  16 const char *commit_type = "commit";
  17
  18 static struct commit *check_commit(struct object *obj,
  19                                    const unsigned char *sha1,
  20                                    int quiet)
  21 {
  22         if (obj->type != OBJ_COMMIT) {
  23                 if (!quiet)
  24                         error("Object %s is a %s, not a commit",
  25                               sha1_to_hex(sha1), typename(obj->type));
  26                 return NULL;
  27         }
  28         return (struct commit *) obj;
  29 }
  30
  31 struct commit *lookup_commit_reference_gently(const unsigned char *sha1,
  32                                               int quiet)
  33 {
  34         struct object *obj = deref_tag(parse_object(sha1), NULL, 0);
  35
  36         if (!obj)
  37                 return NULL;
  38         return check_commit(obj, sha1, quiet);
  39 }
  40
  41 struct commit *lookup_commit_reference(const unsigned char *sha1)
  42 {
  43         return lookup_commit_reference_gently(sha1, 0);
  44 }
  45
  46 struct commit *lookup_commit_or_die(const unsigned char *sha1, const char *ref_name)
  47 {
  48         struct commit *c = lookup_commit_reference(sha1);
  49         if (!c)
  50                 die(_("could not parse %s"), ref_name);
  51         if (hashcmp(sha1, c->object.sha1)) {
  52                 warning(_("%s %s is not a commit!"),
  53                         ref_name, sha1_to_hex(sha1));
  54         }
  55         return c;
  56 }
  57
  58 struct commit *lookup_commit(const unsigned char *sha1)
  59 {
  60         struct object *obj = lookup_object(sha1);
  61         if (!obj)
  62                 return create_object(sha1, OBJ_COMMIT, alloc_commit_node());
  63         if (!obj->type)
  64                 obj->type = OBJ_COMMIT;
  65         return check_commit(obj, sha1, 0);
  66 }
  67
  68 struct commit *lookup_commit_reference_by_name(const char *name)
  69 {
  70         unsigned char sha1[20];
  71         struct commit *commit;
  72
  73         if (get_sha1_committish(name, sha1))
  74                 return NULL;
  75         commit = lookup_commit_reference(sha1);
  76         if (!commit || parse_commit(commit))
  77                 return NULL;
  78         return commit;
  79 }
  80
  81 static unsigned long parse_commit_date(const char *buf, const char *tail)
  82 {
  83         const char *dateptr;
  84
  85         if (buf + 6 >= tail)
  86                 return 0;
  87         if (memcmp(buf, "author", 6))
  88                 return 0;
  89         while (buf < tail && *buf++ != '\n')
  90                 /* nada */;
  91         if (buf + 9 >= tail)
  92                 return 0;
  93         if (memcmp(buf, "committer", 9))
  94                 return 0;
  95         while (buf < tail && *buf++ != '>')
  96                 /* nada */;
  97         if (buf >= tail)
  98                 return 0;
  99         dateptr = buf;
 100         while (buf < tail && *buf++ != '\n')
 101                 /* nada */;
 102         if (buf >= tail)
 103                 return 0;
 104         /* dateptr < buf && buf[-1] == '\n', so strtoul will stop at buf-1 */
 105         return strtoul(dateptr, NULL, 10);
 106 }
 107
 108 static struct commit_graft **commit_graft;
 109 static int commit_graft_alloc, commit_graft_nr;
 110
 111 static int commit_graft_pos(const unsigned char *sha1)
 112 {
 113         int lo, hi;
 114         lo = 0;
 115         hi = commit_graft_nr;
 116         while (lo < hi) {
 117                 int mi = (lo + hi) / 2;
 118                 struct commit_graft *graft = commit_graft[mi];
 119                 int cmp = hashcmp(sha1, graft->sha1);
 120                 if (!cmp)
 121                         return mi;
 122                 if (cmp < 0)
 123                         hi = mi;
 124                 else
 125                         lo = mi + 1;
 126         }
 127         return -lo - 1;
 128 }
 129
 130 int register_commit_graft(struct commit_graft *graft, int ignore_dups)
 131 {
 132         int pos = commit_graft_pos(graft->sha1);
 133
 134         if (0 <= pos) {
 135                 if (ignore_dups)
 136                         free(graft);
 137                 else {
 138                         free(commit_graft[pos]);
 139                         commit_graft[pos] = graft;
 140                 }
 141                 return 1;
 142         }
 143         pos = -pos - 1;
 144         if (commit_graft_alloc <= ++commit_graft_nr) {
 145                 commit_graft_alloc = alloc_nr(commit_graft_alloc);
 146                 commit_graft = xrealloc(commit_graft,
 147                                         sizeof(*commit_graft) *
 148                                         commit_graft_alloc);
 149         }
 150         if (pos < commit_graft_nr)
 151                 memmove(commit_graft + pos + 1,
 152                         commit_graft + pos,
 153                         (commit_graft_nr - pos - 1) *
 154                         sizeof(*commit_graft));
 155         commit_graft[pos] = graft;
 156         return 0;
 157 }
 158
 159 struct commit_graft *read_graft_line(char *buf, int len)
 160 {
 161         /* The format is just "Commit Parent1 Parent2 ...\n" */
 162         int i;
 163         struct commit_graft *graft = NULL;
 164
 165         while (len && isspace(buf[len-1]))
 166                 buf[--len] = '\0';
 167         if (buf[0] == '#' || buf[0] == '\0')
 168                 return NULL;
 169         if ((len + 1) % 41)
 170                 goto bad_graft_data;
 171         i = (len + 1) / 41 - 1;
 172         graft = xmalloc(sizeof(*graft) + 20 * i);
 173         graft->nr_parent = i;
 174         if (get_sha1_hex(buf, graft->sha1))
 175                 goto bad_graft_data;
 176         for (i = 40; i < len; i += 41) {
 177                 if (buf[i] != ' ')
 178                         goto bad_graft_data;
 179                 if (get_sha1_hex(buf + i + 1, graft->parent[i/41]))
 180                         goto bad_graft_data;
 181         }
 182         return graft;
 183
 184 bad_graft_data:
 185         error("bad graft data: %s", buf);
 186         free(graft);
 187         return NULL;
 188 }
 189
 190 static int read_graft_file(const char *graft_file)
 191 {
 192         FILE *fp = fopen(graft_file, "r");
 193         char buf[1024];
 194         if (!fp)
 195                 return -1;
 196         while (fgets(buf, sizeof(buf), fp)) {
 197                 /* The format is just "Commit Parent1 Parent2 ...\n" */
 198                 int len = strlen(buf);
 199                 struct commit_graft *graft = read_graft_line(buf, len);
 200                 if (!graft)
 201                         continue;
 202                 if (register_commit_graft(graft, 1))
 203                         error("duplicate graft data: %s", buf);
 204         }
 205         fclose(fp);
 206         return 0;
 207 }
 208
 209 static void prepare_commit_graft(void)
 210 {
 211         static int commit_graft_prepared;
 212         char *graft_file;
 213
 214         if (commit_graft_prepared)
 215                 return;
 216         graft_file = get_graft_file();
 217         read_graft_file(graft_file);
 218         /* make sure shallows are read */
 219         is_repository_shallow();
 220         commit_graft_prepared = 1;
 221 }
 222
 223 struct commit_graft *lookup_commit_graft(const unsigned char *sha1)
 224 {
 225         int pos;
 226         prepare_commit_graft();
 227         pos = commit_graft_pos(sha1);
 228         if (pos < 0)
 229                 return NULL;
 230         return commit_graft[pos];
 231 }
 232
 233 int for_each_commit_graft(each_commit_graft_fn fn, void *cb_data)
 234 {
 235         int i, ret;
 236         for (i = ret = 0; i < commit_graft_nr && !ret; i++)
 237                 ret = fn(commit_graft[i], cb_data);
 238         return ret;
 239 }
 240
 241 int unregister_shallow(const unsigned char *sha1)
 242 {
 243         int pos = commit_graft_pos(sha1);
 244         if (pos < 0)
 245                 return -1;
 246         if (pos + 1 < commit_graft_nr)
 247                 memmove(commit_graft + pos, commit_graft + pos + 1,
 248                                 sizeof(struct commit_graft *)
 249                                 * (commit_graft_nr - pos - 1));
 250         commit_graft_nr--;
 251         return 0;
 252 }
 253
 254 int parse_commit_buffer(struct commit *item, const void *buffer, unsigned long size)
 255 {
 256         const char *tail = buffer;
 257         const char *bufptr = buffer;
 258         unsigned char parent[20];
 259         struct commit_list **pptr;
 260         struct commit_graft *graft;
 261
 262         if (item->object.parsed)
 263                 return 0;
 264         item->object.parsed = 1;
 265         tail += size;
 266         if (tail <= bufptr + 46 || memcmp(bufptr, "tree ", 5) || bufptr[45] != '\n')
 267                 return error("bogus commit object %s", sha1_to_hex(item->object.sha1));
 268         if (get_sha1_hex(bufptr + 5, parent) < 0)
 269                 return error("bad tree pointer in commit %s",
 270                              sha1_to_hex(item->object.sha1));
 271         item->tree = lookup_tree(parent);
 272         bufptr += 46; /* "tree " + "hex sha1" + "\n" */
 273         pptr = &item->parents;
 274
 275         graft = lookup_commit_graft(item->object.sha1);
 276         while (bufptr + 48 < tail && !memcmp(bufptr, "parent ", 7)) {
 277                 struct commit *new_parent;
 278
 279                 if (tail <= bufptr + 48 ||
 280                     get_sha1_hex(bufptr + 7, parent) ||
 281                     bufptr[47] != '\n')
 282                         return error("bad parents in commit %s", sha1_to_hex(item->object.sha1));
 283                 bufptr += 48;
 284                 /*
 285                  * The clone is shallow if nr_parent < 0, and we must
 286                  * not traverse its real parents even when we unhide them.
 287                  */
 288                 if (graft && (graft->nr_parent < 0 || grafts_replace_parents))
 289                         continue;
 290                 new_parent = lookup_commit(parent);
 291                 if (new_parent)
 292                         pptr = &commit_list_insert(new_parent, pptr)->next;
 293         }
 294         if (graft) {
 295                 int i;
 296                 struct commit *new_parent;
 297                 for (i = 0; i < graft->nr_parent; i++) {
 298                         new_parent = lookup_commit(graft->parent[i]);
 299                         if (!new_parent)
 300                                 continue;
 301                         pptr = &commit_list_insert(new_parent, pptr)->next;
 302                 }
 303         }
 304         item->date = parse_commit_date(bufptr, tail);
 305
 306         return 0;
 307 }
 308
 309 int parse_commit(struct commit *item)
 310 {
 311         enum object_type type;
 312         void *buffer;
 313         unsigned long size;
 314         int ret;
 315
 316         if (!item)
 317                 return -1;
 318         if (item->object.parsed)
 319                 return 0;
 320         buffer = read_sha1_file(item->object.sha1, &type, &size);
 321         if (!buffer)
 322                 return error("Could not read %s",
 323                              sha1_to_hex(item->object.sha1));
 324         if (type != OBJ_COMMIT) {
 325                 free(buffer);
 326                 return error("Object %s not a commit",
 327                              sha1_to_hex(item->object.sha1));
 328         }
 329         ret = parse_commit_buffer(item, buffer, size);
 330         if (save_commit_buffer && !ret) {
 331                 item->buffer = buffer;
 332                 return 0;
 333         }
 334         free(buffer);
 335         return ret;
 336 }
 337
 338 int find_commit_subject(const char *commit_buffer, const char **subject)
 339 {
 340         const char *eol;
 341         const char *p = commit_buffer;
 342
 343         while (*p && (*p != '\n' || p[1] != '\n'))
 344                 p++;
 345         if (*p) {
 346                 p += 2;
 347                 for (eol = p; *eol && *eol != '\n'; eol++)
 348                         ; /* do nothing */
 349         } else
 350                 eol = p;
 351
 352         *subject = p;
 353
 354         return eol - p;
 355 }
 356
 357 struct commit_list *commit_list_insert(struct commit *item, struct commit_list **list_p)
 358 {
 359         struct commit_list *new_list = xmalloc(sizeof(struct commit_list));
 360         new_list->item = item;
 361         new_list->next = *list_p;
 362         *list_p = new_list;
 363         return new_list;
 364 }
 365
 366 unsigned commit_list_count(const struct commit_list *l)
 367 {
 368         unsigned c = 0;
 369         for (; l; l = l->next )
 370                 c++;
 371         return c;
 372 }
 373
 374 void free_commit_list(struct commit_list *list)
 375 {
 376         while (list) {
 377                 struct commit_list *temp = list;
 378                 list = temp->next;
 379                 free(temp);
 380         }
 381 }
 382
 383 struct commit_list * commit_list_insert_by_date(struct commit *item, struct commit_list **list)
 384 {
 385         struct commit_list **pp = list;
 386         struct commit_list *p;
 387         while ((p = *pp) != NULL) {
 388                 if (p->item->date < item->date) {
 389                         break;
 390                 }
 391                 pp = &p->next;
 392         }
 393         return commit_list_insert(item, pp);
 394 }
 395
 396 static int commit_list_compare_by_date(const void *a, const void *b)
 397 {
 398         unsigned long a_date = ((const struct commit_list *)a)->item->date;
 399         unsigned long b_date = ((const struct commit_list *)b)->item->date;
 400         if (a_date < b_date)
 401                 return 1;
 402         if (a_date > b_date)
 403                 return -1;
 404         return 0;
 405 }
 406
 407 static void *commit_list_get_next(const void *a)
 408 {
 409         return ((const struct commit_list *)a)->next;
 410 }
 411
 412 static void commit_list_set_next(void *a, void *next)
 413 {
 414         ((struct commit_list *)a)->next = next;
 415 }
 416
 417 void commit_list_sort_by_date(struct commit_list **list)
 418 {
 419         *list = llist_mergesort(*list, commit_list_get_next, commit_list_set_next,
 420                                 commit_list_compare_by_date);
 421 }
 422
 423 struct commit *pop_most_recent_commit(struct commit_list **list,
 424                                       unsigned int mark)
 425 {
 426         struct commit *ret = (*list)->item;
 427         struct commit_list *parents = ret->parents;
 428         struct commit_list *old = *list;
 429
 430         *list = (*list)->next;
 431         free(old);
 432
 433         while (parents) {
 434                 struct commit *commit = parents->item;
 435                 if (!parse_commit(commit) && !(commit->object.flags & mark)) {
 436                         commit->object.flags |= mark;
 437                         commit_list_insert_by_date(commit, list);
 438                 }
 439                 parents = parents->next;
 440         }
 441         return ret;
 442 }
 443
 444 static void clear_commit_marks_1(struct commit_list **plist,
 445                                  struct commit *commit, unsigned int mark)
 446 {
 447         while (commit) {
 448                 struct commit_list *parents;
 449
 450                 if (!(mark & commit->object.flags))
 451                         return;
 452
 453                 commit->object.flags &= ~mark;
 454
 455                 parents = commit->parents;
 456                 if (!parents)
 457                         return;
 458
 459                 while ((parents = parents->next))
 460                         commit_list_insert(parents->item, plist);
 461
 462                 commit = commit->parents->item;
 463         }
 464 }
 465
 466 void clear_commit_marks_many(int nr, struct commit **commit, unsigned int mark)
 467 {
 468         struct commit_list *list = NULL;
 469
 470         while (nr--) {
 471                 commit_list_insert(*commit, &list);
 472                 commit++;
 473         }
 474         while (list)
 475                 clear_commit_marks_1(&list, pop_commit(&list), mark);
 476 }
 477
 478 void clear_commit_marks(struct commit *commit, unsigned int mark)
 479 {
 480         clear_commit_marks_many(1, &commit, mark);
 481 }
 482
 483 void clear_commit_marks_for_object_array(struct object_array *a, unsigned mark)
 484 {
 485         struct object *object;
 486         struct commit *commit;
 487         unsigned int i;
 488
 489         for (i = 0; i < a->nr; i++) {
 490                 object = a->objects[i].item;
 491                 commit = lookup_commit_reference_gently(object->sha1, 1);
 492                 if (commit)
 493                         clear_commit_marks(commit, mark);
 494         }
 495 }
 496
 497 struct commit *pop_commit(struct commit_list **stack)
 498 {
 499         struct commit_list *top = *stack;
 500         struct commit *item = top ? top->item : NULL;
 501
 502         if (top) {
 503                 *stack = top->next;
 504                 free(top);
 505         }
 506         return item;
 507 }
 508
 509 /*
 510  * Performs an in-place topological sort on the list supplied.
 511  */
 512 void sort_in_topological_order(struct commit_list ** list, int lifo)
 513 {
 514         struct commit_list *next, *orig = *list;
 515         struct commit_list *work, **insert;
 516         struct commit_list **pptr;
 517
 518         if (!orig)
 519                 return;
 520         *list = NULL;
 521
 522         /* Mark them and clear the indegree */
 523         for (next = orig; next; next = next->next) {
 524                 struct commit *commit = next->item;
 525                 commit->indegree = 1;
 526         }
 527
 528         /* update the indegree */
 529         for (next = orig; next; next = next->next) {
 530                 struct commit_list * parents = next->item->parents;
 531                 while (parents) {
 532                         struct commit *parent = parents->item;
 533
 534                         if (parent->indegree)
 535                                 parent->indegree++;
 536                         parents = parents->next;
 537                 }
 538         }
 539
 540         /*
 541          * find the tips
 542          *
 543          * tips are nodes not reachable from any other node in the list
 544          *
 545          * the tips serve as a starting set for the work queue.
 546          */
 547         work = NULL;
 548         insert = &work;
 549         for (next = orig; next; next = next->next) {
 550                 struct commit *commit = next->item;
 551
 552                 if (commit->indegree == 1)
 553                         insert = &commit_list_insert(commit, insert)->next;
 554         }
 555
 556         /* process the list in topological order */
 557         if (!lifo)
 558                 commit_list_sort_by_date(&work);
 559
 560         pptr = list;
 561         *list = NULL;
 562         while (work) {
 563                 struct commit *commit;
 564                 struct commit_list *parents, *work_item;
 565
 566                 work_item = work;
 567                 work = work_item->next;
 568                 work_item->next = NULL;
 569
 570                 commit = work_item->item;
 571                 for (parents = commit->parents; parents ; parents = parents->next) {
 572                         struct commit *parent = parents->item;
 573
 574                         if (!parent->indegree)
 575                                 continue;
 576
 577                         /*
 578                          * parents are only enqueued for emission
 579                          * when all their children have been emitted thereby
 580                          * guaranteeing topological order.
 581                          */
 582                         if (--parent->indegree == 1) {
 583                                 if (!lifo)
 584                                         commit_list_insert_by_date(parent, &work);
 585                                 else
 586                                         commit_list_insert(parent, &work);
 587                         }
 588                 }
 589                 /*
 590                  * work_item is a commit all of whose children
 591                  * have already been emitted. we can emit it now.
 592                  */
 593                 commit->indegree = 0;
 594                 *pptr = work_item;
 595                 pptr = &work_item->next;
 596         }
 597 }
 598
 599 /* merge-base stuff */
 600
 601 /* bits #0..15 in revision.h */
 602 #define PARENT1         (1u<<16)
 603 #define PARENT2         (1u<<17)
 604 #define STALE           (1u<<18)
 605 #define RESULT          (1u<<19)
 606
 607 static const unsigned all_flags = (PARENT1 | PARENT2 | STALE | RESULT);
 608
 609 static struct commit *interesting(struct commit_list *list)
 610 {
 611         while (list) {
 612                 struct commit *commit = list->item;
 613                 list = list->next;
 614                 if (commit->object.flags & STALE)
 615                         continue;
 616                 return commit;
 617         }
 618         return NULL;
 619 }
 620
 621 /* all input commits in one and twos[] must have been parsed! */
 622 static struct commit_list *paint_down_to_common(struct commit *one, int n, struct commit **twos)
 623 {
 624         struct commit_list *list = NULL;
 625         struct commit_list *result = NULL;
 626         int i;
 627
 628         one->object.flags |= PARENT1;
 629         commit_list_insert_by_date(one, &list);
 630         if (!n)
 631                 return list;
 632         for (i = 0; i < n; i++) {
 633                 twos[i]->object.flags |= PARENT2;
 634                 commit_list_insert_by_date(twos[i], &list);
 635         }
 636
 637         while (interesting(list)) {
 638                 struct commit *commit;
 639                 struct commit_list *parents;
 640                 struct commit_list *next;
 641                 int flags;
 642
 643                 commit = list->item;
 644                 next = list->next;
 645                 free(list);
 646                 list = next;
 647
 648                 flags = commit->object.flags & (PARENT1 | PARENT2 | STALE);
 649                 if (flags == (PARENT1 | PARENT2)) {
 650                         if (!(commit->object.flags & RESULT)) {
 651                                 commit->object.flags |= RESULT;
 652                                 commit_list_insert_by_date(commit, &result);
 653                         }
 654                         /* Mark parents of a found merge stale */
 655                         flags |= STALE;
 656                 }
 657                 parents = commit->parents;
 658                 while (parents) {
 659                         struct commit *p = parents->item;
 660                         parents = parents->next;
 661                         if ((p->object.flags & flags) == flags)
 662                                 continue;
 663                         if (parse_commit(p))
 664                                 return NULL;
 665                         p->object.flags |= flags;
 666                         commit_list_insert_by_date(p, &list);
 667                 }
 668         }
 669
 670         free_commit_list(list);
 671         return result;
 672 }
 673
 674 static struct commit_list *merge_bases_many(struct commit *one, int n, struct commit **twos)
 675 {
 676         struct commit_list *list = NULL;
 677         struct commit_list *result = NULL;
 678         int i;
 679
 680         for (i = 0; i < n; i++) {
 681                 if (one == twos[i])
 682                         /*
 683                          * We do not mark this even with RESULT so we do not
 684                          * have to clean it up.
 685                          */
 686                         return commit_list_insert(one, &result);
 687         }
 688
 689         if (parse_commit(one))
 690                 return NULL;
 691         for (i = 0; i < n; i++) {
 692                 if (parse_commit(twos[i]))
 693                         return NULL;
 694         }
 695
 696         list = paint_down_to_common(one, n, twos);
 697
 698         while (list) {
 699                 struct commit_list *next = list->next;
 700                 if (!(list->item->object.flags & STALE))
 701                         commit_list_insert_by_date(list->item, &result);
 702                 free(list);
 703                 list = next;
 704         }
 705         return result;
 706 }
 707
 708 struct commit_list *get_octopus_merge_bases(struct commit_list *in)
 709 {
 710         struct commit_list *i, *j, *k, *ret = NULL;
 711         struct commit_list **pptr = &ret;
 712
 713         for (i = in; i; i = i->next) {
 714                 if (!ret)
 715                         pptr = &commit_list_insert(i->item, pptr)->next;
 716                 else {
 717                         struct commit_list *new = NULL, *end = NULL;
 718
 719                         for (j = ret; j; j = j->next) {
 720                                 struct commit_list *bases;
 721                                 bases = get_merge_bases(i->item, j->item, 1);
 722                                 if (!new)
 723                                         new = bases;
 724                                 else
 725                                         end->next = bases;
 726                                 for (k = bases; k; k = k->next)
 727                                         end = k;
 728                         }
 729                         ret = new;
 730                 }
 731         }
 732         return ret;
 733 }
 734
 735 static int remove_redundant(struct commit **array, int cnt)
 736 {
 737         /*
 738          * Some commit in the array may be an ancestor of
 739          * another commit.  Move such commit to the end of
 740          * the array, and return the number of commits that
 741          * are independent from each other.
 742          */
 743         struct commit **work;
 744         unsigned char *redundant;
 745         int *filled_index;
 746         int i, j, filled;
 747
 748         work = xcalloc(cnt, sizeof(*work));
 749         redundant = xcalloc(cnt, 1);
 750         filled_index = xmalloc(sizeof(*filled_index) * (cnt - 1));
 751
 752         for (i = 0; i < cnt; i++)
 753                 parse_commit(array[i]);
 754         for (i = 0; i < cnt; i++) {
 755                 struct commit_list *common;
 756
 757                 if (redundant[i])
 758                         continue;
 759                 for (j = filled = 0; j < cnt; j++) {
 760                         if (i == j || redundant[j])
 761                                 continue;
 762                         filled_index[filled] = j;
 763                         work[filled++] = array[j];
 764                 }
 765                 common = paint_down_to_common(array[i], filled, work);
 766                 if (array[i]->object.flags & PARENT2)
 767                         redundant[i] = 1;
 768                 for (j = 0; j < filled; j++)
 769                         if (work[j]->object.flags & PARENT1)
 770                                 redundant[filled_index[j]] = 1;
 771                 clear_commit_marks(array[i], all_flags);
 772                 for (j = 0; j < filled; j++)
 773                         clear_commit_marks(work[j], all_flags);
 774                 free_commit_list(common);
 775         }
 776
 777         /* Now collect the result */
 778         memcpy(work, array, sizeof(*array) * cnt);
 779         for (i = filled = 0; i < cnt; i++)
 780                 if (!redundant[i])
 781                         array[filled++] = work[i];
 782         for (j = filled, i = 0; i < cnt; i++)
 783                 if (redundant[i])
 784                         array[j++] = work[i];
 785         free(work);
 786         free(redundant);
 787         free(filled_index);
 788         return filled;
 789 }
 790
 791 struct commit_list *get_merge_bases_many(struct commit *one,
 792                                          int n,
 793                                          struct commit **twos,
 794                                          int cleanup)
 795 {
 796         struct commit_list *list;
 797         struct commit **rslt;
 798         struct commit_list *result;
 799         int cnt, i;
 800
 801         result = merge_bases_many(one, n, twos);
 802         for (i = 0; i < n; i++) {
 803                 if (one == twos[i])
 804                         return result;
 805         }
 806         if (!result || !result->next) {
 807                 if (cleanup) {
 808                         clear_commit_marks(one, all_flags);
 809                         clear_commit_marks_many(n, twos, all_flags);
 810                 }
 811                 return result;
 812         }
 813
 814         /* There are more than one */
 815         cnt = 0;
 816         list = result;
 817         while (list) {
 818                 list = list->next;
 819                 cnt++;
 820         }
 821         rslt = xcalloc(cnt, sizeof(*rslt));
 822         for (list = result, i = 0; list; list = list->next)
 823                 rslt[i++] = list->item;
 824         free_commit_list(result);
 825
 826         clear_commit_marks(one, all_flags);
 827         clear_commit_marks_many(n, twos, all_flags);
 828
 829         cnt = remove_redundant(rslt, cnt);
 830         result = NULL;
 831         for (i = 0; i < cnt; i++)
 832                 commit_list_insert_by_date(rslt[i], &result);
 833         free(rslt);
 834         return result;
 835 }
 836
 837 struct commit_list *get_merge_bases(struct commit *one, struct commit *two,
 838                                     int cleanup)
 839 {
 840         return get_merge_bases_many(one, 1, &two, cleanup);
 841 }
 842
 843 /*
 844  * Is "commit" a decendant of one of the elements on the "with_commit" list?
 845  */
 846 int is_descendant_of(struct commit *commit, struct commit_list *with_commit)
 847 {
 848         if (!with_commit)
 849                 return 1;
 850         while (with_commit) {
 851                 struct commit *other;
 852
 853                 other = with_commit->item;
 854                 with_commit = with_commit->next;
 855                 if (in_merge_bases(other, commit))
 856                         return 1;
 857         }
 858         return 0;
 859 }
 860
 861 /*
 862  * Is "commit" an ancestor of one of the "references"?
 863  */
 864 int in_merge_bases_many(struct commit *commit, int nr_reference, struct commit **reference)
 865 {
 866         struct commit_list *bases;
 867         int ret = 0, i;
 868
 869         if (parse_commit(commit))
 870                 return ret;
 871         for (i = 0; i < nr_reference; i++)
 872                 if (parse_commit(reference[i]))
 873                         return ret;
 874
 875         bases = paint_down_to_common(commit, nr_reference, reference);
 876         if (commit->object.flags & PARENT2)
 877                 ret = 1;
 878         clear_commit_marks(commit, all_flags);
 879         clear_commit_marks_many(nr_reference, reference, all_flags);
 880         free_commit_list(bases);
 881         return ret;
 882 }
 883
 884 /*
 885  * Is "commit" an ancestor of (i.e. reachable from) the "reference"?
 886  */
 887 int in_merge_bases(struct commit *commit, struct commit *reference)
 888 {
 889         return in_merge_bases_many(commit, 1, &reference);
 890 }
 891
 892 struct commit_list *reduce_heads(struct commit_list *heads)
 893 {
 894         struct commit_list *p;
 895         struct commit_list *result = NULL, **tail = &result;
 896         struct commit **array;
 897         int num_head, i;
 898
 899         if (!heads)
 900                 return NULL;
 901
 902         /* Uniquify */
 903         for (p = heads; p; p = p->next)
 904                 p->item->object.flags &= ~STALE;
 905         for (p = heads, num_head = 0; p; p = p->next) {
 906                 if (p->item->object.flags & STALE)
 907                         continue;
 908                 p->item->object.flags |= STALE;
 909                 num_head++;
 910         }
 911         array = xcalloc(sizeof(*array), num_head);
 912         for (p = heads, i = 0; p; p = p->next) {
 913                 if (p->item->object.flags & STALE) {
 914                         array[i++] = p->item;
 915                         p->item->object.flags &= ~STALE;
 916                 }
 917         }
 918         num_head = remove_redundant(array, num_head);
 919         for (i = 0; i < num_head; i++)
 920                 tail = &commit_list_insert(array[i], tail)->next;
 921         return result;
 922 }
 923
 924 static const char gpg_sig_header[] = "gpgsig";
 925 static const int gpg_sig_header_len = sizeof(gpg_sig_header) - 1;
 926
 927 static int do_sign_commit(struct strbuf *buf, const char *keyid)
 928 {
 929         struct strbuf sig = STRBUF_INIT;
 930         int inspos, copypos;
 931
 932         /* find the end of the header */
 933         inspos = strstr(buf->buf, "\n\n") - buf->buf + 1;
 934
 935         if (!keyid || !*keyid)
 936                 keyid = get_signing_key();
 937         if (sign_buffer(buf, &sig, keyid)) {
 938                 strbuf_release(&sig);
 939                 return -1;
 940         }
 941
 942         for (copypos = 0; sig.buf[copypos]; ) {
 943                 const char *bol = sig.buf + copypos;
 944                 const char *eol = strchrnul(bol, '\n');
 945                 int len = (eol - bol) + !!*eol;
 946
 947                 if (!copypos) {
 948                         strbuf_insert(buf, inspos, gpg_sig_header, gpg_sig_header_len);
 949                         inspos += gpg_sig_header_len;
 950                 }
 951                 strbuf_insert(buf, inspos++, " ", 1);
 952                 strbuf_insert(buf, inspos, bol, len);
 953                 inspos += len;
 954                 copypos += len;
 955         }
 956         strbuf_release(&sig);
 957         return 0;
 958 }
 959
 960 int parse_signed_commit(const unsigned char *sha1,
 961                         struct strbuf *payload, struct strbuf *signature)
 962 {
 963         unsigned long size;
 964         enum object_type type;
 965         char *buffer = read_sha1_file(sha1, &type, &size);
 966         int in_signature, saw_signature = -1;
 967         char *line, *tail;
 968
 969         if (!buffer || type != OBJ_COMMIT)
 970                 goto cleanup;
 971
 972         line = buffer;
 973         tail = buffer + size;
 974         in_signature = 0;
 975         saw_signature = 0;
 976         while (line < tail) {
 977                 const char *sig = NULL;
 978                 char *next = memchr(line, '\n', tail - line);
 979
 980                 next = next ? next + 1 : tail;
 981                 if (in_signature && line[0] == ' ')
 982                         sig = line + 1;
 983                 else if (!prefixcmp(line, gpg_sig_header) &&
 984                          line[gpg_sig_header_len] == ' ')
 985                         sig = line + gpg_sig_header_len + 1;
 986                 if (sig) {
 987                         strbuf_add(signature, sig, next - sig);
 988                         saw_signature = 1;
 989                         in_signature = 1;
 990                 } else {
 991                         if (*line == '\n')
 992                                 /* dump the whole remainder of the buffer */
 993                                 next = tail;
 994                         strbuf_add(payload, line, next - line);
 995                         in_signature = 0;
 996                 }
 997                 line = next;
 998         }
 999  cleanup:
1000         free(buffer);
1001         return saw_signature;
1002 }
1003
1004 static void handle_signed_tag(struct commit *parent, struct commit_extra_header ***tail)
1005 {
1006         struct merge_remote_desc *desc;
1007         struct commit_extra_header *mergetag;
1008         char *buf;
1009         unsigned long size, len;
1010         enum object_type type;
1011
1012         desc = merge_remote_util(parent);
1013         if (!desc || !desc->obj)
1014                 return;
1015         buf = read_sha1_file(desc->obj->sha1, &type, &size);
1016         if (!buf || type != OBJ_TAG)
1017                 goto free_return;
1018         len = parse_signature(buf, size);
1019         if (size == len)
1020                 goto free_return;
1021         /*
1022          * We could verify this signature and either omit the tag when
1023          * it does not validate, but the integrator may not have the
1024          * public key of the signer of the tag he is merging, while a
1025          * later auditor may have it while auditing, so let's not run
1026          * verify-signed-buffer here for now...
1027          *
1028          * if (verify_signed_buffer(buf, len, buf + len, size - len, ...))
1029          *      warn("warning: signed tag unverified.");
1030          */
1031         mergetag = xcalloc(1, sizeof(*mergetag));
1032         mergetag->key = xstrdup("mergetag");
1033         mergetag->value = buf;
1034         mergetag->len = size;
1035
1036         **tail = mergetag;
1037         *tail = &mergetag->next;
1038         return;
1039
1040 free_return:
1041         free(buf);
1042 }
1043
1044 void append_merge_tag_headers(struct commit_list *parents,
1045                               struct commit_extra_header ***tail)
1046 {
1047         while (parents) {
1048                 struct commit *parent = parents->item;
1049                 handle_signed_tag(parent, tail);
1050                 parents = parents->next;
1051         }
1052 }
1053
1054 static void add_extra_header(struct strbuf *buffer,
1055                              struct commit_extra_header *extra)
1056 {
1057         strbuf_addstr(buffer, extra->key);
1058         if (extra->len)
1059                 strbuf_add_lines(buffer, " ", extra->value, extra->len);
1060         else
1061                 strbuf_addch(buffer, '\n');
1062 }
1063
1064 struct commit_extra_header *read_commit_extra_headers(struct commit *commit,
1065                                                       const char **exclude)
1066 {
1067         struct commit_extra_header *extra = NULL;
1068         unsigned long size;
1069         enum object_type type;
1070         char *buffer = read_sha1_file(commit->object.sha1, &type, &size);
1071         if (buffer && type == OBJ_COMMIT)
1072                 extra = read_commit_extra_header_lines(buffer, size, exclude);
1073         free(buffer);
1074         return extra;
1075 }
1076
1077 static inline int standard_header_field(const char *field, size_t len)
1078 {
1079         return ((len == 4 && !memcmp(field, "tree ", 5)) ||
1080                 (len == 6 && !memcmp(field, "parent ", 7)) ||
1081                 (len == 6 && !memcmp(field, "author ", 7)) ||
1082                 (len == 9 && !memcmp(field, "committer ", 10)) ||
1083                 (len == 8 && !memcmp(field, "encoding ", 9)));
1084 }
1085
1086 static int excluded_header_field(const char *field, size_t len, const char **exclude)
1087 {
1088         if (!exclude)
1089                 return 0;
1090
1091         while (*exclude) {
1092                 size_t xlen = strlen(*exclude);
1093                 if (len == xlen &&
1094                     !memcmp(field, *exclude, xlen) && field[xlen] == ' ')
1095                         return 1;
1096                 exclude++;
1097         }
1098         return 0;
1099 }
1100
1101 static struct commit_extra_header *read_commit_extra_header_lines(
1102         const char *buffer, size_t size,
1103         const char **exclude)
1104 {
1105         struct commit_extra_header *extra = NULL, **tail = &extra, *it = NULL;
1106         const char *line, *next, *eof, *eob;
1107         struct strbuf buf = STRBUF_INIT;
1108
1109         for (line = buffer, eob = line + size;
1110              line < eob && *line != '\n';
1111              line = next) {
1112                 next = memchr(line, '\n', eob - line);
1113                 next = next ? next + 1 : eob;
1114                 if (*line == ' ') {
1115                         /* continuation */
1116                         if (it)
1117                                 strbuf_add(&buf, line + 1, next - (line + 1));
1118                         continue;
1119                 }
1120                 if (it)
1121                         it->value = strbuf_detach(&buf, &it->len);
1122                 strbuf_reset(&buf);
1123                 it = NULL;
1124
1125                 eof = strchr(line, ' ');
1126                 if (next <= eof)
1127                         eof = next;
1128
1129                 if (standard_header_field(line, eof - line) ||
1130                     excluded_header_field(line, eof - line, exclude))
1131                         continue;
1132
1133                 it = xcalloc(1, sizeof(*it));
1134                 it->key = xmemdupz(line, eof-line);
1135                 *tail = it;
1136                 tail = &it->next;
1137                 if (eof + 1 < next)
1138                         strbuf_add(&buf, eof + 1, next - (eof + 1));
1139         }
1140         if (it)
1141                 it->value = strbuf_detach(&buf, &it->len);
1142         return extra;
1143 }
1144
1145 void free_commit_extra_headers(struct commit_extra_header *extra)
1146 {
1147         while (extra) {
1148                 struct commit_extra_header *next = extra->next;
1149                 free(extra->key);
1150                 free(extra->value);
1151                 free(extra);
1152                 extra = next;
1153         }
1154 }
1155
1156 int commit_tree(const struct strbuf *msg, unsigned char *tree,
1157                 struct commit_list *parents, unsigned char *ret,
1158                 const char *author, const char *sign_commit)
1159 {
1160         struct commit_extra_header *extra = NULL, **tail = &extra;
1161         int result;
1162
1163         append_merge_tag_headers(parents, &tail);
1164         result = commit_tree_extended(msg, tree, parents, ret,
1165                                       author, sign_commit, extra);
1166         free_commit_extra_headers(extra);
1167         return result;
1168 }
1169
1170 static int find_invalid_utf8(const char *buf, int len)
1171 {
1172         int offset = 0;
1173
1174         while (len) {
1175                 unsigned char c = *buf++;
1176                 int bytes, bad_offset;
1177
1178                 len--;
1179                 offset++;
1180
1181                 /* Simple US-ASCII? No worries. */
1182                 if (c < 0x80)
1183                         continue;
1184
1185                 bad_offset = offset-1;
1186
1187                 /*
1188                  * Count how many more high bits set: that's how
1189                  * many more bytes this sequence should have.
1190                  */
1191                 bytes = 0;
1192                 while (c & 0x40) {
1193                         c <<= 1;
1194                         bytes++;
1195                 }
1196
1197                 /* Must be between 1 and 5 more bytes */
1198                 if (bytes < 1 || bytes > 5)
1199                         return bad_offset;
1200
1201                 /* Do we *have* that many bytes? */
1202                 if (len < bytes)
1203                         return bad_offset;
1204
1205                 offset += bytes;
1206                 len -= bytes;
1207
1208                 /* And verify that they are good continuation bytes */
1209                 do {
1210                         if ((*buf++ & 0xc0) != 0x80)
1211                                 return bad_offset;
1212                 } while (--bytes);
1213
1214                 /* We could/should check the value and length here too */
1215         }
1216         return -1;
1217 }
1218
1219 /*
1220  * This verifies that the buffer is in proper utf8 format.
1221  *
1222  * If it isn't, it assumes any non-utf8 characters are Latin1,
1223  * and does the conversion.
1224  *
1225  * Fixme: we should probably also disallow overlong forms and
1226  * invalid characters. But we don't do that currently.
1227  */
1228 static int verify_utf8(struct strbuf *buf)
1229 {
1230         int ok = 1;
1231         long pos = 0;
1232
1233         for (;;) {
1234                 int bad;
1235                 unsigned char c;
1236                 unsigned char replace[2];
1237
1238                 bad = find_invalid_utf8(buf->buf + pos, buf->len - pos);
1239                 if (bad < 0)
1240                         return ok;
1241                 pos += bad;
1242                 ok = 0;
1243                 c = buf->buf[pos];
1244                 strbuf_remove(buf, pos, 1);
1245
1246                 /* We know 'c' must be in the range 128-255 */
1247                 replace[0] = 0xc0 + (c >> 6);
1248                 replace[1] = 0x80 + (c & 0x3f);
1249                 strbuf_insert(buf, pos, replace, 2);
1250                 pos += 2;
1251         }
1252 }
1253
1254 static const char commit_utf8_warn[] =
1255 "Warning: commit message did not conform to UTF-8.\n"
1256 "You may want to amend it after fixing the message, or set the config\n"
1257 "variable i18n.commitencoding to the encoding your project uses.\n";
1258
1259 int commit_tree_extended(const struct strbuf *msg, unsigned char *tree,
1260                          struct commit_list *parents, unsigned char *ret,
1261                          const char *author, const char *sign_commit,
1262                          struct commit_extra_header *extra)
1263 {
1264         int result;
1265         int encoding_is_utf8;
1266         struct strbuf buffer;
1267
1268         assert_sha1_type(tree, OBJ_TREE);
1269
1270         if (memchr(msg->buf, '\0', msg->len))
1271                 return error("a NUL byte in commit log message not allowed.");
1272
1273         /* Not having i18n.commitencoding is the same as having utf-8 */
1274         encoding_is_utf8 = is_encoding_utf8(git_commit_encoding);
1275
1276         strbuf_init(&buffer, 8192); /* should avoid reallocs for the headers */
1277         strbuf_addf(&buffer, "tree %s\n", sha1_to_hex(tree));
1278
1279         /*
1280          * NOTE! This ordering means that the same exact tree merged with a
1281          * different order of parents will be a _different_ changeset even
1282          * if everything else stays the same.
1283          */
1284         while (parents) {
1285                 struct commit_list *next = parents->next;
1286                 struct commit *parent = parents->item;
1287
1288                 strbuf_addf(&buffer, "parent %s\n",
1289                             sha1_to_hex(parent->object.sha1));
1290                 free(parents);
1291                 parents = next;
1292         }
1293
1294         /* Person/date information */
1295         if (!author)
1296                 author = git_author_info(IDENT_STRICT);
1297         strbuf_addf(&buffer, "author %s\n", author);
1298         strbuf_addf(&buffer, "committer %s\n", git_committer_info(IDENT_STRICT));
1299         if (!encoding_is_utf8)
1300                 strbuf_addf(&buffer, "encoding %s\n", git_commit_encoding);
1301
1302         while (extra) {
1303                 add_extra_header(&buffer, extra);
1304                 extra = extra->next;
1305         }
1306         strbuf_addch(&buffer, '\n');
1307
1308         /* And add the comment */
1309         strbuf_addbuf(&buffer, msg);
1310
1311         /* And check the encoding */
1312         if (encoding_is_utf8 && !verify_utf8(&buffer))
1313                 fprintf(stderr, commit_utf8_warn);
1314
1315         if (sign_commit && do_sign_commit(&buffer, sign_commit))
1316                 return -1;
1317
1318         result = write_sha1_file(buffer.buf, buffer.len, commit_type, ret);
1319         strbuf_release(&buffer);
1320         return result;
1321 }
1322
1323 struct commit *get_merge_parent(const char *name)
1324 {
1325         struct object *obj;
1326         struct commit *commit;
1327         unsigned char sha1[20];
1328         if (get_sha1(name, sha1))
1329                 return NULL;
1330         obj = parse_object(sha1);
1331         commit = (struct commit *)peel_to_type(name, 0, obj, OBJ_COMMIT);
1332         if (commit && !commit->util) {
1333                 struct merge_remote_desc *desc;
1334                 desc = xmalloc(sizeof(*desc));
1335                 desc->obj = obj;
1336                 desc->name = strdup(name);
1337                 commit->util = desc;
1338         }
1339         return commit;
1340 }
1341
1342 /*
1343  * Append a commit to the end of the commit_list.
1344  *
1345  * next starts by pointing to the variable that holds the head of an
1346  * empty commit_list, and is updated to point to the "next" field of
1347  * the last item on the list as new commits are appended.
1348  *
1349  * Usage example:
1350  *
1351  *     struct commit_list *list;
1352  *     struct commit_list **next = &list;
1353  *
1354  *     next = commit_list_append(c1, next);
1355  *     next = commit_list_append(c2, next);
1356  *     assert(commit_list_count(list) == 2);
1357  *     return list;
1358  */
1359 struct commit_list **commit_list_append(struct commit *commit,
1360                                         struct commit_list **next)
1361 {
1362         struct commit_list *new = xmalloc(sizeof(struct commit_list));
1363         new->item = commit;
1364         *next = new;
1365         new->next = NULL;
1366         return &new->next;
1367 }
1368
1369 void print_commit_list(struct commit_list *list,
1370                        const char *format_cur,
1371                        const char *format_last)
1372 {
1373         for ( ; list; list = list->next) {
1374                 const char *format = list->next ? format_cur : format_last;
1375                 printf(format, sha1_to_hex(list->item->object.sha1));
1376         }
1377 }