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 #include "commit-slab.h"
  12 #include "prio-queue.h"
  13
  14 static struct commit_extra_header *read_commit_extra_header_lines(const char *buf, size_t len, const char **);
  15
  16 int save_commit_buffer = 1;
  17
  18 const char *commit_type = "commit";
  19 static int commit_count;
  20
  21 static struct commit *check_commit(struct object *obj,
  22                                    const unsigned char *sha1,
  23                                    int quiet)
  24 {
  25         if (obj->type != OBJ_COMMIT) {
  26                 if (!quiet)
  27                         error("Object %s is a %s, not a commit",
  28                               sha1_to_hex(sha1), typename(obj->type));
  29                 return NULL;
  30         }
  31         return (struct commit *) obj;
  32 }
  33
  34 struct commit *lookup_commit_reference_gently(const unsigned char *sha1,
  35                                               int quiet)
  36 {
  37         struct object *obj = deref_tag(parse_object(sha1), NULL, 0);
  38
  39         if (!obj)
  40                 return NULL;
  41         return check_commit(obj, sha1, quiet);
  42 }
  43
  44 struct commit *lookup_commit_reference(const unsigned char *sha1)
  45 {
  46         return lookup_commit_reference_gently(sha1, 0);
  47 }
  48
  49 struct commit *lookup_commit_or_die(const unsigned char *sha1, const char *ref_name)
  50 {
  51         struct commit *c = lookup_commit_reference(sha1);
  52         if (!c)
  53                 die(_("could not parse %s"), ref_name);
  54         if (hashcmp(sha1, c->object.sha1)) {
  55                 warning(_("%s %s is not a commit!"),
  56                         ref_name, sha1_to_hex(sha1));
  57         }
  58         return c;
  59 }
  60
  61 struct commit *lookup_commit(const unsigned char *sha1)
  62 {
  63         struct object *obj = lookup_object(sha1);
  64         if (!obj) {
  65                 struct commit *c = alloc_commit_node();
  66                 c->index = commit_count++;
  67                 return create_object(sha1, OBJ_COMMIT, c);
  68         }
  69         if (!obj->type)
  70                 obj->type = OBJ_COMMIT;
  71         return check_commit(obj, sha1, 0);
  72 }
  73
  74 struct commit *lookup_commit_reference_by_name(const char *name)
  75 {
  76         unsigned char sha1[20];
  77         struct commit *commit;
  78
  79         if (get_sha1_committish(name, sha1))
  80                 return NULL;
  81         commit = lookup_commit_reference(sha1);
  82         if (!commit || parse_commit(commit))
  83                 return NULL;
  84         return commit;
  85 }
  86
  87 static unsigned long parse_commit_date(const char *buf, const char *tail)
  88 {
  89         const char *dateptr;
  90
  91         if (buf + 6 >= tail)
  92                 return 0;
  93         if (memcmp(buf, "author", 6))
  94                 return 0;
  95         while (buf < tail && *buf++ != '\n')
  96                 /* nada */;
  97         if (buf + 9 >= tail)
  98                 return 0;
  99         if (memcmp(buf, "committer", 9))
 100                 return 0;
 101         while (buf < tail && *buf++ != '>')
 102                 /* nada */;
 103         if (buf >= tail)
 104                 return 0;
 105         dateptr = buf;
 106         while (buf < tail && *buf++ != '\n')
 107                 /* nada */;
 108         if (buf >= tail)
 109                 return 0;
 110         /* dateptr < buf && buf[-1] == '\n', so strtoul will stop at buf-1 */
 111         return strtoul(dateptr, NULL, 10);
 112 }
 113
 114 static struct commit_graft **commit_graft;
 115 static int commit_graft_alloc, commit_graft_nr;
 116
 117 static int commit_graft_pos(const unsigned char *sha1)
 118 {
 119         int lo, hi;
 120         lo = 0;
 121         hi = commit_graft_nr;
 122         while (lo < hi) {
 123                 int mi = (lo + hi) / 2;
 124                 struct commit_graft *graft = commit_graft[mi];
 125                 int cmp = hashcmp(sha1, graft->sha1);
 126                 if (!cmp)
 127                         return mi;
 128                 if (cmp < 0)
 129                         hi = mi;
 130                 else
 131                         lo = mi + 1;
 132         }
 133         return -lo - 1;
 134 }
 135
 136 int register_commit_graft(struct commit_graft *graft, int ignore_dups)
 137 {
 138         int pos = commit_graft_pos(graft->sha1);
 139
 140         if (0 <= pos) {
 141                 if (ignore_dups)
 142                         free(graft);
 143                 else {
 144                         free(commit_graft[pos]);
 145                         commit_graft[pos] = graft;
 146                 }
 147                 return 1;
 148         }
 149         pos = -pos - 1;
 150         if (commit_graft_alloc <= ++commit_graft_nr) {
 151                 commit_graft_alloc = alloc_nr(commit_graft_alloc);
 152                 commit_graft = xrealloc(commit_graft,
 153                                         sizeof(*commit_graft) *
 154                                         commit_graft_alloc);
 155         }
 156         if (pos < commit_graft_nr)
 157                 memmove(commit_graft + pos + 1,
 158                         commit_graft + pos,
 159                         (commit_graft_nr - pos - 1) *
 160                         sizeof(*commit_graft));
 161         commit_graft[pos] = graft;
 162         return 0;
 163 }
 164
 165 struct commit_graft *read_graft_line(char *buf, int len)
 166 {
 167         /* The format is just "Commit Parent1 Parent2 ...\n" */
 168         int i;
 169         struct commit_graft *graft = NULL;
 170
 171         while (len && isspace(buf[len-1]))
 172                 buf[--len] = '\0';
 173         if (buf[0] == '#' || buf[0] == '\0')
 174                 return NULL;
 175         if ((len + 1) % 41)
 176                 goto bad_graft_data;
 177         i = (len + 1) / 41 - 1;
 178         graft = xmalloc(sizeof(*graft) + 20 * i);
 179         graft->nr_parent = i;
 180         if (get_sha1_hex(buf, graft->sha1))
 181                 goto bad_graft_data;
 182         for (i = 40; i < len; i += 41) {
 183                 if (buf[i] != ' ')
 184                         goto bad_graft_data;
 185                 if (get_sha1_hex(buf + i + 1, graft->parent[i/41]))
 186                         goto bad_graft_data;
 187         }
 188         return graft;
 189
 190 bad_graft_data:
 191         error("bad graft data: %s", buf);
 192         free(graft);
 193         return NULL;
 194 }
 195
 196 static int read_graft_file(const char *graft_file)
 197 {
 198         FILE *fp = fopen(graft_file, "r");
 199         char buf[1024];
 200         if (!fp)
 201                 return -1;
 202         while (fgets(buf, sizeof(buf), fp)) {
 203                 /* The format is just "Commit Parent1 Parent2 ...\n" */
 204                 int len = strlen(buf);
 205                 struct commit_graft *graft = read_graft_line(buf, len);
 206                 if (!graft)
 207                         continue;
 208                 if (register_commit_graft(graft, 1))
 209                         error("duplicate graft data: %s", buf);
 210         }
 211         fclose(fp);
 212         return 0;
 213 }
 214
 215 static void prepare_commit_graft(void)
 216 {
 217         static int commit_graft_prepared;
 218         char *graft_file;
 219
 220         if (commit_graft_prepared)
 221                 return;
 222         graft_file = get_graft_file();
 223         read_graft_file(graft_file);
 224         /* make sure shallows are read */
 225         is_repository_shallow();
 226         commit_graft_prepared = 1;
 227 }
 228
 229 struct commit_graft *lookup_commit_graft(const unsigned char *sha1)
 230 {
 231         int pos;
 232         prepare_commit_graft();
 233         pos = commit_graft_pos(sha1);
 234         if (pos < 0)
 235                 return NULL;
 236         return commit_graft[pos];
 237 }
 238
 239 int for_each_commit_graft(each_commit_graft_fn fn, void *cb_data)
 240 {
 241         int i, ret;
 242         for (i = ret = 0; i < commit_graft_nr && !ret; i++)
 243                 ret = fn(commit_graft[i], cb_data);
 244         return ret;
 245 }
 246
 247 int unregister_shallow(const unsigned char *sha1)
 248 {
 249         int pos = commit_graft_pos(sha1);
 250         if (pos < 0)
 251                 return -1;
 252         if (pos + 1 < commit_graft_nr)
 253                 memmove(commit_graft + pos, commit_graft + pos + 1,
 254                                 sizeof(struct commit_graft *)
 255                                 * (commit_graft_nr - pos - 1));
 256         commit_graft_nr--;
 257         return 0;
 258 }
 259
 260 int parse_commit_buffer(struct commit *item, const void *buffer, unsigned long size)
 261 {
 262         const char *tail = buffer;
 263         const char *bufptr = buffer;
 264         unsigned char parent[20];
 265         struct commit_list **pptr;
 266         struct commit_graft *graft;
 267
 268         if (item->object.parsed)
 269                 return 0;
 270         item->object.parsed = 1;
 271         tail += size;
 272         if (tail <= bufptr + 46 || memcmp(bufptr, "tree ", 5) || bufptr[45] != '\n')
 273                 return error("bogus commit object %s", sha1_to_hex(item->object.sha1));
 274         if (get_sha1_hex(bufptr + 5, parent) < 0)
 275                 return error("bad tree pointer in commit %s",
 276                              sha1_to_hex(item->object.sha1));
 277         item->tree = lookup_tree(parent);
 278         bufptr += 46; /* "tree " + "hex sha1" + "\n" */
 279         pptr = &item->parents;
 280
 281         graft = lookup_commit_graft(item->object.sha1);
 282         while (bufptr + 48 < tail && !memcmp(bufptr, "parent ", 7)) {
 283                 struct commit *new_parent;
 284
 285                 if (tail <= bufptr + 48 ||
 286                     get_sha1_hex(bufptr + 7, parent) ||
 287                     bufptr[47] != '\n')
 288                         return error("bad parents in commit %s", sha1_to_hex(item->object.sha1));
 289                 bufptr += 48;
 290                 /*
 291                  * The clone is shallow if nr_parent < 0, and we must
 292                  * not traverse its real parents even when we unhide them.
 293                  */
 294                 if (graft && (graft->nr_parent < 0 || grafts_replace_parents))
 295                         continue;
 296                 new_parent = lookup_commit(parent);
 297                 if (new_parent)
 298                         pptr = &commit_list_insert(new_parent, pptr)->next;
 299         }
 300         if (graft) {
 301                 int i;
 302                 struct commit *new_parent;
 303                 for (i = 0; i < graft->nr_parent; i++) {
 304                         new_parent = lookup_commit(graft->parent[i]);
 305                         if (!new_parent)
 306                                 continue;
 307                         pptr = &commit_list_insert(new_parent, pptr)->next;
 308                 }
 309         }
 310         item->date = parse_commit_date(bufptr, tail);
 311
 312         return 0;
 313 }
 314
 315 int parse_commit(struct commit *item)
 316 {
 317         enum object_type type;
 318         void *buffer;
 319         unsigned long size;
 320         int ret;
 321
 322         if (!item)
 323                 return -1;
 324         if (item->object.parsed)
 325                 return 0;
 326         buffer = read_sha1_file(item->object.sha1, &type, &size);
 327         if (!buffer)
 328                 return error("Could not read %s",
 329                              sha1_to_hex(item->object.sha1));
 330         if (type != OBJ_COMMIT) {
 331                 free(buffer);
 332                 return error("Object %s not a commit",
 333                              sha1_to_hex(item->object.sha1));
 334         }
 335         ret = parse_commit_buffer(item, buffer, size);
 336         if (save_commit_buffer && !ret) {
 337                 item->buffer = buffer;
 338                 return 0;
 339         }
 340         free(buffer);
 341         return ret;
 342 }
 343
 344 int find_commit_subject(const char *commit_buffer, const char **subject)
 345 {
 346         const char *eol;
 347         const char *p = commit_buffer;
 348
 349         while (*p && (*p != '\n' || p[1] != '\n'))
 350                 p++;
 351         if (*p) {
 352                 p += 2;
 353                 for (eol = p; *eol && *eol != '\n'; eol++)
 354                         ; /* do nothing */
 355         } else
 356                 eol = p;
 357
 358         *subject = p;
 359
 360         return eol - p;
 361 }
 362
 363 struct commit_list *commit_list_insert(struct commit *item, struct commit_list **list_p)
 364 {
 365         struct commit_list *new_list = xmalloc(sizeof(struct commit_list));
 366         new_list->item = item;
 367         new_list->next = *list_p;
 368         *list_p = new_list;
 369         return new_list;
 370 }
 371
 372 unsigned commit_list_count(const struct commit_list *l)
 373 {
 374         unsigned c = 0;
 375         for (; l; l = l->next )
 376                 c++;
 377         return c;
 378 }
 379
 380 void free_commit_list(struct commit_list *list)
 381 {
 382         while (list) {
 383                 struct commit_list *temp = list;
 384                 list = temp->next;
 385                 free(temp);
 386         }
 387 }
 388
 389 struct commit_list * commit_list_insert_by_date(struct commit *item, struct commit_list **list)
 390 {
 391         struct commit_list **pp = list;
 392         struct commit_list *p;
 393         while ((p = *pp) != NULL) {
 394                 if (p->item->date < item->date) {
 395                         break;
 396                 }
 397                 pp = &p->next;
 398         }
 399         return commit_list_insert(item, pp);
 400 }
 401
 402 static int commit_list_compare_by_date(const void *a, const void *b)
 403 {
 404         unsigned long a_date = ((const struct commit_list *)a)->item->date;
 405         unsigned long b_date = ((const struct commit_list *)b)->item->date;
 406         if (a_date < b_date)
 407                 return 1;
 408         if (a_date > b_date)
 409                 return -1;
 410         return 0;
 411 }
 412
 413 static void *commit_list_get_next(const void *a)
 414 {
 415         return ((const struct commit_list *)a)->next;
 416 }
 417
 418 static void commit_list_set_next(void *a, void *next)
 419 {
 420         ((struct commit_list *)a)->next = next;
 421 }
 422
 423 void commit_list_sort_by_date(struct commit_list **list)
 424 {
 425         *list = llist_mergesort(*list, commit_list_get_next, commit_list_set_next,
 426                                 commit_list_compare_by_date);
 427 }
 428
 429 struct commit *pop_most_recent_commit(struct commit_list **list,
 430                                       unsigned int mark)
 431 {
 432         struct commit *ret = (*list)->item;
 433         struct commit_list *parents = ret->parents;
 434         struct commit_list *old = *list;
 435
 436         *list = (*list)->next;
 437         free(old);
 438
 439         while (parents) {
 440                 struct commit *commit = parents->item;
 441                 if (!parse_commit(commit) && !(commit->object.flags & mark)) {
 442                         commit->object.flags |= mark;
 443                         commit_list_insert_by_date(commit, list);
 444                 }
 445                 parents = parents->next;
 446         }
 447         return ret;
 448 }
 449
 450 static void clear_commit_marks_1(struct commit_list **plist,
 451                                  struct commit *commit, unsigned int mark)
 452 {
 453         while (commit) {
 454                 struct commit_list *parents;
 455
 456                 if (!(mark & commit->object.flags))
 457                         return;
 458
 459                 commit->object.flags &= ~mark;
 460
 461                 parents = commit->parents;
 462                 if (!parents)
 463                         return;
 464
 465                 while ((parents = parents->next))
 466                         commit_list_insert(parents->item, plist);
 467
 468                 commit = commit->parents->item;
 469         }
 470 }
 471
 472 void clear_commit_marks(struct commit *commit, unsigned int mark)
 473 {
 474         struct commit_list *list = NULL;
 475         commit_list_insert(commit, &list);
 476         while (list)
 477                 clear_commit_marks_1(&list, pop_commit(&list), mark);
 478 }
 479
 480 void clear_commit_marks_for_object_array(struct object_array *a, unsigned mark)
 481 {
 482         struct object *object;
 483         struct commit *commit;
 484         unsigned int i;
 485
 486         for (i = 0; i < a->nr; i++) {
 487                 object = a->objects[i].item;
 488                 commit = lookup_commit_reference_gently(object->sha1, 1);
 489                 if (commit)
 490                         clear_commit_marks(commit, mark);
 491         }
 492 }
 493
 494 struct commit *pop_commit(struct commit_list **stack)
 495 {
 496         struct commit_list *top = *stack;
 497         struct commit *item = top ? top->item : NULL;
 498
 499         if (top) {
 500                 *stack = top->next;
 501                 free(top);
 502         }
 503         return item;
 504 }
 505
 506 /*
 507  * Topological sort support
 508  */
 509
 510 /* count number of children that have not been emitted */
 511 define_commit_slab(indegree_slab, int);
 512
 513 static int compare_commits_by_commit_date(const void *a_, const void *b_, void *unused)
 514 {
 515         const struct commit *a = a_, *b = b_;
 516         /* newer commits with larger date first */
 517         if (a->date < b->date)
 518                 return 1;
 519         else if (a->date > b->date)
 520                 return -1;
 521         return 0;
 522 }
 523
 524 /*
 525  * Performs an in-place topological sort on the list supplied.
 526  */
 527 void sort_in_topological_order(struct commit_list **list, enum rev_sort_order sort_order)
 528 {
 529         struct commit_list *next, *orig = *list;
 530         struct commit_list **pptr;
 531         struct indegree_slab indegree;
 532         struct prio_queue queue;
 533         struct commit *commit;
 534
 535         if (!orig)
 536                 return;
 537         *list = NULL;
 538
 539         init_indegree_slab(&indegree);
 540         memset(&queue, '\0', sizeof(queue));
 541         switch (sort_order) {
 542         default: /* REV_SORT_IN_GRAPH_ORDER */
 543                 queue.compare = NULL;
 544                 break;
 545         case REV_SORT_BY_COMMIT_DATE:
 546                 queue.compare = compare_commits_by_commit_date;
 547                 break;
 548         }
 549
 550         /* Mark them and clear the indegree */
 551         for (next = orig; next; next = next->next) {
 552                 struct commit *commit = next->item;
 553                 *(indegree_slab_at(&indegree, commit)) = 1;
 554         }
 555
 556         /* update the indegree */
 557         for (next = orig; next; next = next->next) {
 558                 struct commit_list *parents = next->item->parents;
 559                 while (parents) {
 560                         struct commit *parent = parents->item;
 561                         int *pi = indegree_slab_at(&indegree, parent);
 562
 563                         if (*pi)
 564                                 (*pi)++;
 565                         parents = parents->next;
 566                 }
 567         }
 568
 569         /*
 570          * find the tips
 571          *
 572          * tips are nodes not reachable from any other node in the list
 573          *
 574          * the tips serve as a starting set for the work queue.
 575          */
 576         for (next = orig; next; next = next->next) {
 577                 struct commit *commit = next->item;
 578
 579                 if (*(indegree_slab_at(&indegree, commit)) == 1)
 580                         prio_queue_put(&queue, commit);
 581         }
 582
 583         /*
 584          * This is unfortunate; the initial tips need to be shown
 585          * in the order given from the revision traversal machinery.
 586          */
 587         if (sort_order == REV_SORT_IN_GRAPH_ORDER)
 588                 prio_queue_reverse(&queue);
 589
 590         /* We no longer need the commit list */
 591         free_commit_list(orig);
 592
 593         pptr = list;
 594         *list = NULL;
 595         while ((commit = prio_queue_get(&queue)) != NULL) {
 596                 struct commit_list *parents;
 597
 598                 for (parents = commit->parents; parents ; parents = parents->next) {
 599                         struct commit *parent = parents->item;
 600                         int *pi = indegree_slab_at(&indegree, parent);
 601
 602                         if (!*pi)
 603                                 continue;
 604
 605                         /*
 606                          * parents are only enqueued for emission
 607                          * when all their children have been emitted thereby
 608                          * guaranteeing topological order.
 609                          */
 610                         if (--(*pi) == 1)
 611                                 prio_queue_put(&queue, parent);
 612                 }
 613                 /*
 614                  * all children of commit have already been
 615                  * emitted. we can emit it now.
 616                  */
 617                 *(indegree_slab_at(&indegree, commit)) = 0;
 618
 619                 pptr = &commit_list_insert(commit, pptr)->next;
 620         }
 621
 622         clear_indegree_slab(&indegree);
 623         clear_prio_queue(&queue);
 624 }
 625
 626 /* merge-base stuff */
 627
 628 /* bits #0..15 in revision.h */
 629 #define PARENT1         (1u<<16)
 630 #define PARENT2         (1u<<17)
 631 #define STALE           (1u<<18)
 632 #define RESULT          (1u<<19)
 633
 634 static const unsigned all_flags = (PARENT1 | PARENT2 | STALE | RESULT);
 635
 636 static struct commit *interesting(struct commit_list *list)
 637 {
 638         while (list) {
 639                 struct commit *commit = list->item;
 640                 list = list->next;
 641                 if (commit->object.flags & STALE)
 642                         continue;
 643                 return commit;
 644         }
 645         return NULL;
 646 }
 647
 648 /* all input commits in one and twos[] must have been parsed! */
 649 static struct commit_list *paint_down_to_common(struct commit *one, int n, struct commit **twos)
 650 {
 651         struct commit_list *list = NULL;
 652         struct commit_list *result = NULL;
 653         int i;
 654
 655         one->object.flags |= PARENT1;
 656         commit_list_insert_by_date(one, &list);
 657         if (!n)
 658                 return list;
 659         for (i = 0; i < n; i++) {
 660                 twos[i]->object.flags |= PARENT2;
 661                 commit_list_insert_by_date(twos[i], &list);
 662         }
 663
 664         while (interesting(list)) {
 665                 struct commit *commit;
 666                 struct commit_list *parents;
 667                 struct commit_list *next;
 668                 int flags;
 669
 670                 commit = list->item;
 671                 next = list->next;
 672                 free(list);
 673                 list = next;
 674
 675                 flags = commit->object.flags & (PARENT1 | PARENT2 | STALE);
 676                 if (flags == (PARENT1 | PARENT2)) {
 677                         if (!(commit->object.flags & RESULT)) {
 678                                 commit->object.flags |= RESULT;
 679                                 commit_list_insert_by_date(commit, &result);
 680                         }
 681                         /* Mark parents of a found merge stale */
 682                         flags |= STALE;
 683                 }
 684                 parents = commit->parents;
 685                 while (parents) {
 686                         struct commit *p = parents->item;
 687                         parents = parents->next;
 688                         if ((p->object.flags & flags) == flags)
 689                                 continue;
 690                         if (parse_commit(p))
 691                                 return NULL;
 692                         p->object.flags |= flags;
 693                         commit_list_insert_by_date(p, &list);
 694                 }
 695         }
 696
 697         free_commit_list(list);
 698         return result;
 699 }
 700
 701 static struct commit_list *merge_bases_many(struct commit *one, int n, struct commit **twos)
 702 {
 703         struct commit_list *list = NULL;
 704         struct commit_list *result = NULL;
 705         int i;
 706
 707         for (i = 0; i < n; i++) {
 708                 if (one == twos[i])
 709                         /*
 710                          * We do not mark this even with RESULT so we do not
 711                          * have to clean it up.
 712                          */
 713                         return commit_list_insert(one, &result);
 714         }
 715
 716         if (parse_commit(one))
 717                 return NULL;
 718         for (i = 0; i < n; i++) {
 719                 if (parse_commit(twos[i]))
 720                         return NULL;
 721         }
 722
 723         list = paint_down_to_common(one, n, twos);
 724
 725         while (list) {
 726                 struct commit_list *next = list->next;
 727                 if (!(list->item->object.flags & STALE))
 728                         commit_list_insert_by_date(list->item, &result);
 729                 free(list);
 730                 list = next;
 731         }
 732         return result;
 733 }
 734
 735 struct commit_list *get_octopus_merge_bases(struct commit_list *in)
 736 {
 737         struct commit_list *i, *j, *k, *ret = NULL;
 738         struct commit_list **pptr = &ret;
 739
 740         for (i = in; i; i = i->next) {
 741                 if (!ret)
 742                         pptr = &commit_list_insert(i->item, pptr)->next;
 743                 else {
 744                         struct commit_list *new = NULL, *end = NULL;
 745
 746                         for (j = ret; j; j = j->next) {
 747                                 struct commit_list *bases;
 748                                 bases = get_merge_bases(i->item, j->item, 1);
 749                                 if (!new)
 750                                         new = bases;
 751                                 else
 752                                         end->next = bases;
 753                                 for (k = bases; k; k = k->next)
 754                                         end = k;
 755                         }
 756                         ret = new;
 757                 }
 758         }
 759         return ret;
 760 }
 761
 762 static int remove_redundant(struct commit **array, int cnt)
 763 {
 764         /*
 765          * Some commit in the array may be an ancestor of
 766          * another commit.  Move such commit to the end of
 767          * the array, and return the number of commits that
 768          * are independent from each other.
 769          */
 770         struct commit **work;
 771         unsigned char *redundant;
 772         int *filled_index;
 773         int i, j, filled;
 774
 775         work = xcalloc(cnt, sizeof(*work));
 776         redundant = xcalloc(cnt, 1);
 777         filled_index = xmalloc(sizeof(*filled_index) * (cnt - 1));
 778
 779         for (i = 0; i < cnt; i++)
 780                 parse_commit(array[i]);
 781         for (i = 0; i < cnt; i++) {
 782                 struct commit_list *common;
 783
 784                 if (redundant[i])
 785                         continue;
 786                 for (j = filled = 0; j < cnt; j++) {
 787                         if (i == j || redundant[j])
 788                                 continue;
 789                         filled_index[filled] = j;
 790                         work[filled++] = array[j];
 791                 }
 792                 common = paint_down_to_common(array[i], filled, work);
 793                 if (array[i]->object.flags & PARENT2)
 794                         redundant[i] = 1;
 795                 for (j = 0; j < filled; j++)
 796                         if (work[j]->object.flags & PARENT1)
 797                                 redundant[filled_index[j]] = 1;
 798                 clear_commit_marks(array[i], all_flags);
 799                 for (j = 0; j < filled; j++)
 800                         clear_commit_marks(work[j], all_flags);
 801                 free_commit_list(common);
 802         }
 803
 804         /* Now collect the result */
 805         memcpy(work, array, sizeof(*array) * cnt);
 806         for (i = filled = 0; i < cnt; i++)
 807                 if (!redundant[i])
 808                         array[filled++] = work[i];
 809         for (j = filled, i = 0; i < cnt; i++)
 810                 if (redundant[i])
 811                         array[j++] = work[i];
 812         free(work);
 813         free(redundant);
 814         free(filled_index);
 815         return filled;
 816 }
 817
 818 struct commit_list *get_merge_bases_many(struct commit *one,
 819                                          int n,
 820                                          struct commit **twos,
 821                                          int cleanup)
 822 {
 823         struct commit_list *list;
 824         struct commit **rslt;
 825         struct commit_list *result;
 826         int cnt, i;
 827
 828         result = merge_bases_many(one, n, twos);
 829         for (i = 0; i < n; i++) {
 830                 if (one == twos[i])
 831                         return result;
 832         }
 833         if (!result || !result->next) {
 834                 if (cleanup) {
 835                         clear_commit_marks(one, all_flags);
 836                         for (i = 0; i < n; i++)
 837                                 clear_commit_marks(twos[i], all_flags);
 838                 }
 839                 return result;
 840         }
 841
 842         /* There are more than one */
 843         cnt = 0;
 844         list = result;
 845         while (list) {
 846                 list = list->next;
 847                 cnt++;
 848         }
 849         rslt = xcalloc(cnt, sizeof(*rslt));
 850         for (list = result, i = 0; list; list = list->next)
 851                 rslt[i++] = list->item;
 852         free_commit_list(result);
 853
 854         clear_commit_marks(one, all_flags);
 855         for (i = 0; i < n; i++)
 856                 clear_commit_marks(twos[i], all_flags);
 857
 858         cnt = remove_redundant(rslt, cnt);
 859         result = NULL;
 860         for (i = 0; i < cnt; i++)
 861                 commit_list_insert_by_date(rslt[i], &result);
 862         free(rslt);
 863         return result;
 864 }
 865
 866 struct commit_list *get_merge_bases(struct commit *one, struct commit *two,
 867                                     int cleanup)
 868 {
 869         return get_merge_bases_many(one, 1, &two, cleanup);
 870 }
 871
 872 /*
 873  * Is "commit" a descendant of one of the elements on the "with_commit" list?
 874  */
 875 int is_descendant_of(struct commit *commit, struct commit_list *with_commit)
 876 {
 877         if (!with_commit)
 878                 return 1;
 879         while (with_commit) {
 880                 struct commit *other;
 881
 882                 other = with_commit->item;
 883                 with_commit = with_commit->next;
 884                 if (in_merge_bases(other, commit))
 885                         return 1;
 886         }
 887         return 0;
 888 }
 889
 890 /*
 891  * Is "commit" an ancestor of (i.e. reachable from) the "reference"?
 892  */
 893 int in_merge_bases(struct commit *commit, struct commit *reference)
 894 {
 895         struct commit_list *bases;
 896         int ret = 0;
 897
 898         if (parse_commit(commit) || parse_commit(reference))
 899                 return ret;
 900
 901         bases = paint_down_to_common(commit, 1, &reference);
 902         if (commit->object.flags & PARENT2)
 903                 ret = 1;
 904         clear_commit_marks(commit, all_flags);
 905         clear_commit_marks(reference, all_flags);
 906         free_commit_list(bases);
 907         return ret;
 908 }
 909
 910 struct commit_list *reduce_heads(struct commit_list *heads)
 911 {
 912         struct commit_list *p;
 913         struct commit_list *result = NULL, **tail = &result;
 914         struct commit **array;
 915         int num_head, i;
 916
 917         if (!heads)
 918                 return NULL;
 919
 920         /* Uniquify */
 921         for (p = heads; p; p = p->next)
 922                 p->item->object.flags &= ~STALE;
 923         for (p = heads, num_head = 0; p; p = p->next) {
 924                 if (p->item->object.flags & STALE)
 925                         continue;
 926                 p->item->object.flags |= STALE;
 927                 num_head++;
 928         }
 929         array = xcalloc(sizeof(*array), num_head);
 930         for (p = heads, i = 0; p; p = p->next) {
 931                 if (p->item->object.flags & STALE) {
 932                         array[i++] = p->item;
 933                         p->item->object.flags &= ~STALE;
 934                 }
 935         }
 936         num_head = remove_redundant(array, num_head);
 937         for (i = 0; i < num_head; i++)
 938                 tail = &commit_list_insert(array[i], tail)->next;
 939         return result;
 940 }
 941
 942 static const char gpg_sig_header[] = "gpgsig";
 943 static const int gpg_sig_header_len = sizeof(gpg_sig_header) - 1;
 944
 945 static int do_sign_commit(struct strbuf *buf, const char *keyid)
 946 {
 947         struct strbuf sig = STRBUF_INIT;
 948         int inspos, copypos;
 949
 950         /* find the end of the header */
 951         inspos = strstr(buf->buf, "\n\n") - buf->buf + 1;
 952
 953         if (!keyid || !*keyid)
 954                 keyid = get_signing_key();
 955         if (sign_buffer(buf, &sig, keyid)) {
 956                 strbuf_release(&sig);
 957                 return -1;
 958         }
 959
 960         for (copypos = 0; sig.buf[copypos]; ) {
 961                 const char *bol = sig.buf + copypos;
 962                 const char *eol = strchrnul(bol, '\n');
 963                 int len = (eol - bol) + !!*eol;
 964
 965                 if (!copypos) {
 966                         strbuf_insert(buf, inspos, gpg_sig_header, gpg_sig_header_len);
 967                         inspos += gpg_sig_header_len;
 968                 }
 969                 strbuf_insert(buf, inspos++, " ", 1);
 970                 strbuf_insert(buf, inspos, bol, len);
 971                 inspos += len;
 972                 copypos += len;
 973         }
 974         strbuf_release(&sig);
 975         return 0;
 976 }
 977
 978 int parse_signed_commit(const unsigned char *sha1,
 979                         struct strbuf *payload, struct strbuf *signature)
 980 {
 981         unsigned long size;
 982         enum object_type type;
 983         char *buffer = read_sha1_file(sha1, &type, &size);
 984         int in_signature, saw_signature = -1;
 985         char *line, *tail;
 986
 987         if (!buffer || type != OBJ_COMMIT)
 988                 goto cleanup;
 989
 990         line = buffer;
 991         tail = buffer + size;
 992         in_signature = 0;
 993         saw_signature = 0;
 994         while (line < tail) {
 995                 const char *sig = NULL;
 996                 char *next = memchr(line, '\n', tail - line);
 997
 998                 next = next ? next + 1 : tail;
 999                 if (in_signature && line[0] == ' ')
1000                         sig = line + 1;
1001                 else if (!prefixcmp(line, gpg_sig_header) &&
1002                          line[gpg_sig_header_len] == ' ')
1003                         sig = line + gpg_sig_header_len + 1;
1004                 if (sig) {
1005                         strbuf_add(signature, sig, next - sig);
1006                         saw_signature = 1;
1007                         in_signature = 1;
1008                 } else {
1009                         if (*line == '\n')
1010                                 /* dump the whole remainder of the buffer */
1011                                 next = tail;
1012                         strbuf_add(payload, line, next - line);
1013                         in_signature = 0;
1014                 }
1015                 line = next;
1016         }
1017  cleanup:
1018         free(buffer);
1019         return saw_signature;
1020 }
1021
1022 static void handle_signed_tag(struct commit *parent, struct commit_extra_header ***tail)
1023 {
1024         struct merge_remote_desc *desc;
1025         struct commit_extra_header *mergetag;
1026         char *buf;
1027         unsigned long size, len;
1028         enum object_type type;
1029
1030         desc = merge_remote_util(parent);
1031         if (!desc || !desc->obj)
1032                 return;
1033         buf = read_sha1_file(desc->obj->sha1, &type, &size);
1034         if (!buf || type != OBJ_TAG)
1035                 goto free_return;
1036         len = parse_signature(buf, size);
1037         if (size == len)
1038                 goto free_return;
1039         /*
1040          * We could verify this signature and either omit the tag when
1041          * it does not validate, but the integrator may not have the
1042          * public key of the signer of the tag he is merging, while a
1043          * later auditor may have it while auditing, so let's not run
1044          * verify-signed-buffer here for now...
1045          *
1046          * if (verify_signed_buffer(buf, len, buf + len, size - len, ...))
1047          *      warn("warning: signed tag unverified.");
1048          */
1049         mergetag = xcalloc(1, sizeof(*mergetag));
1050         mergetag->key = xstrdup("mergetag");
1051         mergetag->value = buf;
1052         mergetag->len = size;
1053
1054         **tail = mergetag;
1055         *tail = &mergetag->next;
1056         return;
1057
1058 free_return:
1059         free(buf);
1060 }
1061
1062 void append_merge_tag_headers(struct commit_list *parents,
1063                               struct commit_extra_header ***tail)
1064 {
1065         while (parents) {
1066                 struct commit *parent = parents->item;
1067                 handle_signed_tag(parent, tail);
1068                 parents = parents->next;
1069         }
1070 }
1071
1072 static void add_extra_header(struct strbuf *buffer,
1073                              struct commit_extra_header *extra)
1074 {
1075         strbuf_addstr(buffer, extra->key);
1076         if (extra->len)
1077                 strbuf_add_lines(buffer, " ", extra->value, extra->len);
1078         else
1079                 strbuf_addch(buffer, '\n');
1080 }
1081
1082 struct commit_extra_header *read_commit_extra_headers(struct commit *commit,
1083                                                       const char **exclude)
1084 {
1085         struct commit_extra_header *extra = NULL;
1086         unsigned long size;
1087         enum object_type type;
1088         char *buffer = read_sha1_file(commit->object.sha1, &type, &size);
1089         if (buffer && type == OBJ_COMMIT)
1090                 extra = read_commit_extra_header_lines(buffer, size, exclude);
1091         free(buffer);
1092         return extra;
1093 }
1094
1095 static inline int standard_header_field(const char *field, size_t len)
1096 {
1097         return ((len == 4 && !memcmp(field, "tree ", 5)) ||
1098                 (len == 6 && !memcmp(field, "parent ", 7)) ||
1099                 (len == 6 && !memcmp(field, "author ", 7)) ||
1100                 (len == 9 && !memcmp(field, "committer ", 10)) ||
1101                 (len == 8 && !memcmp(field, "encoding ", 9)));
1102 }
1103
1104 static int excluded_header_field(const char *field, size_t len, const char **exclude)
1105 {
1106         if (!exclude)
1107                 return 0;
1108
1109         while (*exclude) {
1110                 size_t xlen = strlen(*exclude);
1111                 if (len == xlen &&
1112                     !memcmp(field, *exclude, xlen) && field[xlen] == ' ')
1113                         return 1;
1114                 exclude++;
1115         }
1116         return 0;
1117 }
1118
1119 static struct commit_extra_header *read_commit_extra_header_lines(
1120         const char *buffer, size_t size,
1121         const char **exclude)
1122 {
1123         struct commit_extra_header *extra = NULL, **tail = &extra, *it = NULL;
1124         const char *line, *next, *eof, *eob;
1125         struct strbuf buf = STRBUF_INIT;
1126
1127         for (line = buffer, eob = line + size;
1128              line < eob && *line != '\n';
1129              line = next) {
1130                 next = memchr(line, '\n', eob - line);
1131                 next = next ? next + 1 : eob;
1132                 if (*line == ' ') {
1133                         /* continuation */
1134                         if (it)
1135                                 strbuf_add(&buf, line + 1, next - (line + 1));
1136                         continue;
1137                 }
1138                 if (it)
1139                         it->value = strbuf_detach(&buf, &it->len);
1140                 strbuf_reset(&buf);
1141                 it = NULL;
1142
1143                 eof = strchr(line, ' ');
1144                 if (next <= eof)
1145                         eof = next;
1146
1147                 if (standard_header_field(line, eof - line) ||
1148                     excluded_header_field(line, eof - line, exclude))
1149                         continue;
1150
1151                 it = xcalloc(1, sizeof(*it));
1152                 it->key = xmemdupz(line, eof-line);
1153                 *tail = it;
1154                 tail = &it->next;
1155                 if (eof + 1 < next)
1156                         strbuf_add(&buf, eof + 1, next - (eof + 1));
1157         }
1158         if (it)
1159                 it->value = strbuf_detach(&buf, &it->len);
1160         return extra;
1161 }
1162
1163 void free_commit_extra_headers(struct commit_extra_header *extra)
1164 {
1165         while (extra) {
1166                 struct commit_extra_header *next = extra->next;
1167                 free(extra->key);
1168                 free(extra->value);
1169                 free(extra);
1170                 extra = next;
1171         }
1172 }
1173
1174 int commit_tree(const struct strbuf *msg, unsigned char *tree,
1175                 struct commit_list *parents, unsigned char *ret,
1176                 const char *author, const char *sign_commit)
1177 {
1178         struct commit_extra_header *extra = NULL, **tail = &extra;
1179         int result;
1180
1181         append_merge_tag_headers(parents, &tail);
1182         result = commit_tree_extended(msg, tree, parents, ret,
1183                                       author, sign_commit, extra);
1184         free_commit_extra_headers(extra);
1185         return result;
1186 }
1187
1188 static int find_invalid_utf8(const char *buf, int len)
1189 {
1190         int offset = 0;
1191
1192         while (len) {
1193                 unsigned char c = *buf++;
1194                 int bytes, bad_offset;
1195
1196                 len--;
1197                 offset++;
1198
1199                 /* Simple US-ASCII? No worries. */
1200                 if (c < 0x80)
1201                         continue;
1202
1203                 bad_offset = offset-1;
1204
1205                 /*
1206                  * Count how many more high bits set: that's how
1207                  * many more bytes this sequence should have.
1208                  */
1209                 bytes = 0;
1210                 while (c & 0x40) {
1211                         c <<= 1;
1212                         bytes++;
1213                 }
1214
1215                 /* Must be between 1 and 5 more bytes */
1216                 if (bytes < 1 || bytes > 5)
1217                         return bad_offset;
1218
1219                 /* Do we *have* that many bytes? */
1220                 if (len < bytes)
1221                         return bad_offset;
1222
1223                 offset += bytes;
1224                 len -= bytes;
1225
1226                 /* And verify that they are good continuation bytes */
1227                 do {
1228                         if ((*buf++ & 0xc0) != 0x80)
1229                                 return bad_offset;
1230                 } while (--bytes);
1231
1232                 /* We could/should check the value and length here too */
1233         }
1234         return -1;
1235 }
1236
1237 /*
1238  * This verifies that the buffer is in proper utf8 format.
1239  *
1240  * If it isn't, it assumes any non-utf8 characters are Latin1,
1241  * and does the conversion.
1242  *
1243  * Fixme: we should probably also disallow overlong forms and
1244  * invalid characters. But we don't do that currently.
1245  */
1246 static int verify_utf8(struct strbuf *buf)
1247 {
1248         int ok = 1;
1249         long pos = 0;
1250
1251         for (;;) {
1252                 int bad;
1253                 unsigned char c;
1254                 unsigned char replace[2];
1255
1256                 bad = find_invalid_utf8(buf->buf + pos, buf->len - pos);
1257                 if (bad < 0)
1258                         return ok;
1259                 pos += bad;
1260                 ok = 0;
1261                 c = buf->buf[pos];
1262                 strbuf_remove(buf, pos, 1);
1263
1264                 /* We know 'c' must be in the range 128-255 */
1265                 replace[0] = 0xc0 + (c >> 6);
1266                 replace[1] = 0x80 + (c & 0x3f);
1267                 strbuf_insert(buf, pos, replace, 2);
1268                 pos += 2;
1269         }
1270 }
1271
1272 static const char commit_utf8_warn[] =
1273 "Warning: commit message did not conform to UTF-8.\n"
1274 "You may want to amend it after fixing the message, or set the config\n"
1275 "variable i18n.commitencoding to the encoding your project uses.\n";
1276
1277 int commit_tree_extended(const struct strbuf *msg, unsigned char *tree,
1278                          struct commit_list *parents, unsigned char *ret,
1279                          const char *author, const char *sign_commit,
1280                          struct commit_extra_header *extra)
1281 {
1282         int result;
1283         int encoding_is_utf8;
1284         struct strbuf buffer;
1285
1286         assert_sha1_type(tree, OBJ_TREE);
1287
1288         if (memchr(msg->buf, '\0', msg->len))
1289                 return error("a NUL byte in commit log message not allowed.");
1290
1291         /* Not having i18n.commitencoding is the same as having utf-8 */
1292         encoding_is_utf8 = is_encoding_utf8(git_commit_encoding);
1293
1294         strbuf_init(&buffer, 8192); /* should avoid reallocs for the headers */
1295         strbuf_addf(&buffer, "tree %s\n", sha1_to_hex(tree));
1296
1297         /*
1298          * NOTE! This ordering means that the same exact tree merged with a
1299          * different order of parents will be a _different_ changeset even
1300          * if everything else stays the same.
1301          */
1302         while (parents) {
1303                 struct commit_list *next = parents->next;
1304                 struct commit *parent = parents->item;
1305
1306                 strbuf_addf(&buffer, "parent %s\n",
1307                             sha1_to_hex(parent->object.sha1));
1308                 free(parents);
1309                 parents = next;
1310         }
1311
1312         /* Person/date information */
1313         if (!author)
1314                 author = git_author_info(IDENT_STRICT);
1315         strbuf_addf(&buffer, "author %s\n", author);
1316         strbuf_addf(&buffer, "committer %s\n", git_committer_info(IDENT_STRICT));
1317         if (!encoding_is_utf8)
1318                 strbuf_addf(&buffer, "encoding %s\n", git_commit_encoding);
1319
1320         while (extra) {
1321                 add_extra_header(&buffer, extra);
1322                 extra = extra->next;
1323         }
1324         strbuf_addch(&buffer, '\n');
1325
1326         /* And add the comment */
1327         strbuf_addbuf(&buffer, msg);
1328
1329         /* And check the encoding */
1330         if (encoding_is_utf8 && !verify_utf8(&buffer))
1331                 fprintf(stderr, commit_utf8_warn);
1332
1333         if (sign_commit && do_sign_commit(&buffer, sign_commit))
1334                 return -1;
1335
1336         result = write_sha1_file(buffer.buf, buffer.len, commit_type, ret);
1337         strbuf_release(&buffer);
1338         return result;
1339 }
1340
1341 struct commit *get_merge_parent(const char *name)
1342 {
1343         struct object *obj;
1344         struct commit *commit;
1345         unsigned char sha1[20];
1346         if (get_sha1(name, sha1))
1347                 return NULL;
1348         obj = parse_object(sha1);
1349         commit = (struct commit *)peel_to_type(name, 0, obj, OBJ_COMMIT);
1350         if (commit && !commit->util) {
1351                 struct merge_remote_desc *desc;
1352                 desc = xmalloc(sizeof(*desc));
1353                 desc->obj = obj;
1354                 desc->name = strdup(name);
1355                 commit->util = desc;
1356         }
1357         return commit;
1358 }
1359
1360 /*
1361  * Append a commit to the end of the commit_list.
1362  *
1363  * next starts by pointing to the variable that holds the head of an
1364  * empty commit_list, and is updated to point to the "next" field of
1365  * the last item on the list as new commits are appended.
1366  *
1367  * Usage example:
1368  *
1369  *     struct commit_list *list;
1370  *     struct commit_list **next = &list;
1371  *
1372  *     next = commit_list_append(c1, next);
1373  *     next = commit_list_append(c2, next);
1374  *     assert(commit_list_count(list) == 2);
1375  *     return list;
1376  */
1377 struct commit_list **commit_list_append(struct commit *commit,
1378                                         struct commit_list **next)
1379 {
1380         struct commit_list *new = xmalloc(sizeof(struct commit_list));
1381         new->item = commit;
1382         *next = new;
1383         new->next = NULL;
1384         return &new->next;
1385 }
1386
1387 void print_commit_list(struct commit_list *list,
1388                        const char *format_cur,
1389                        const char *format_last)
1390 {
1391         for ( ; list; list = list->next) {
1392                 const char *format = list->next ? format_cur : format_last;
1393                 printf(format, sha1_to_hex(list->item->object.sha1));
1394         }
1395 }