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