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 #include "sha1-lookup.h"
  14
  15 static struct commit_extra_header *read_commit_extra_header_lines(const char *buf, size_t len, const char **);
  16
  17 int save_commit_buffer = 1;
  18
  19 const char *commit_type = "commit";
  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                 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 (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 const unsigned char *commit_graft_sha1_access(size_t index, void *table)
 117 {
 118         struct commit_graft **commit_graft_table = table;
 119         return commit_graft_table[index]->sha1;
 120 }
 121
 122 static int commit_graft_pos(const unsigned char *sha1)
 123 {
 124         return sha1_pos(sha1, commit_graft, commit_graft_nr,
 125                         commit_graft_sha1_access);
 126 }
 127
 128 int register_commit_graft(struct commit_graft *graft, int ignore_dups)
 129 {
 130         int pos = commit_graft_pos(graft->sha1);
 131
 132         if (0 <= pos) {
 133                 if (ignore_dups)
 134                         free(graft);
 135                 else {
 136                         free(commit_graft[pos]);
 137                         commit_graft[pos] = graft;
 138                 }
 139                 return 1;
 140         }
 141         pos = -pos - 1;
 142         ALLOC_GROW(commit_graft, commit_graft_nr + 1, commit_graft_alloc);
 143         commit_graft_nr++;
 144         if (pos < commit_graft_nr)
 145                 memmove(commit_graft + pos + 1,
 146                         commit_graft + pos,
 147                         (commit_graft_nr - pos - 1) *
 148                         sizeof(*commit_graft));
 149         commit_graft[pos] = graft;
 150         return 0;
 151 }
 152
 153 struct commit_graft *read_graft_line(char *buf, int len)
 154 {
 155         /* The format is just "Commit Parent1 Parent2 ...\n" */
 156         int i;
 157         struct commit_graft *graft = NULL;
 158
 159         while (len && isspace(buf[len-1]))
 160                 buf[--len] = '\0';
 161         if (buf[0] == '#' || buf[0] == '\0')
 162                 return NULL;
 163         if ((len + 1) % 41)
 164                 goto bad_graft_data;
 165         i = (len + 1) / 41 - 1;
 166         graft = xmalloc(sizeof(*graft) + 20 * i);
 167         graft->nr_parent = i;
 168         if (get_sha1_hex(buf, graft->sha1))
 169                 goto bad_graft_data;
 170         for (i = 40; i < len; i += 41) {
 171                 if (buf[i] != ' ')
 172                         goto bad_graft_data;
 173                 if (get_sha1_hex(buf + i + 1, graft->parent[i/41]))
 174                         goto bad_graft_data;
 175         }
 176         return graft;
 177
 178 bad_graft_data:
 179         error("bad graft data: %s", buf);
 180         free(graft);
 181         return NULL;
 182 }
 183
 184 static int read_graft_file(const char *graft_file)
 185 {
 186         FILE *fp = fopen(graft_file, "r");
 187         struct strbuf buf = STRBUF_INIT;
 188         if (!fp)
 189                 return -1;
 190         while (!strbuf_getwholeline(&buf, fp, '\n')) {
 191                 /* The format is just "Commit Parent1 Parent2 ...\n" */
 192                 struct commit_graft *graft = read_graft_line(buf.buf, buf.len);
 193                 if (!graft)
 194                         continue;
 195                 if (register_commit_graft(graft, 1))
 196                         error("duplicate graft data: %s", buf.buf);
 197         }
 198         fclose(fp);
 199         strbuf_release(&buf);
 200         return 0;
 201 }
 202
 203 static void prepare_commit_graft(void)
 204 {
 205         static int commit_graft_prepared;
 206         char *graft_file;
 207
 208         if (commit_graft_prepared)
 209                 return;
 210         graft_file = get_graft_file();
 211         read_graft_file(graft_file);
 212         /* make sure shallows are read */
 213         is_repository_shallow();
 214         commit_graft_prepared = 1;
 215 }
 216
 217 struct commit_graft *lookup_commit_graft(const unsigned char *sha1)
 218 {
 219         int pos;
 220         prepare_commit_graft();
 221         pos = commit_graft_pos(sha1);
 222         if (pos < 0)
 223                 return NULL;
 224         return commit_graft[pos];
 225 }
 226
 227 int for_each_commit_graft(each_commit_graft_fn fn, void *cb_data)
 228 {
 229         int i, ret;
 230         for (i = ret = 0; i < commit_graft_nr && !ret; i++)
 231                 ret = fn(commit_graft[i], cb_data);
 232         return ret;
 233 }
 234
 235 int unregister_shallow(const unsigned char *sha1)
 236 {
 237         int pos = commit_graft_pos(sha1);
 238         if (pos < 0)
 239                 return -1;
 240         if (pos + 1 < commit_graft_nr)
 241                 memmove(commit_graft + pos, commit_graft + pos + 1,
 242                                 sizeof(struct commit_graft *)
 243                                 * (commit_graft_nr - pos - 1));
 244         commit_graft_nr--;
 245         return 0;
 246 }
 247
 248 void set_commit_buffer(struct commit *commit, void *buffer)
 249 {
 250         commit->buffer = buffer;
 251 }
 252
 253 const void *get_cached_commit_buffer(const struct commit *commit)
 254 {
 255         return commit->buffer;
 256 }
 257
 258 const void *get_commit_buffer(const struct commit *commit)
 259 {
 260         const void *ret = get_cached_commit_buffer(commit);
 261         if (!ret) {
 262                 enum object_type type;
 263                 unsigned long size;
 264                 ret = read_sha1_file(commit->object.sha1, &type, &size);
 265                 if (!ret)
 266                         die("cannot read commit object %s",
 267                             sha1_to_hex(commit->object.sha1));
 268                 if (type != OBJ_COMMIT)
 269                         die("expected commit for %s, got %s",
 270                             sha1_to_hex(commit->object.sha1), typename(type));
 271         }
 272         return ret;
 273 }
 274
 275 void unuse_commit_buffer(const struct commit *commit, const void *buffer)
 276 {
 277         if (commit->buffer != buffer)
 278                 free((void *)buffer);
 279 }
 280
 281 void free_commit_buffer(struct commit *commit)
 282 {
 283         free(commit->buffer);
 284         commit->buffer = NULL;
 285 }
 286
 287 const void *detach_commit_buffer(struct commit *commit)
 288 {
 289         void *ret = commit->buffer;
 290         commit->buffer = NULL;
 291         return ret;
 292 }
 293
 294 int parse_commit_buffer(struct commit *item, const void *buffer, unsigned long size)
 295 {
 296         const char *tail = buffer;
 297         const char *bufptr = buffer;
 298         unsigned char parent[20];
 299         struct commit_list **pptr;
 300         struct commit_graft *graft;
 301
 302         if (item->object.parsed)
 303                 return 0;
 304         item->object.parsed = 1;
 305         tail += size;
 306         if (tail <= bufptr + 46 || memcmp(bufptr, "tree ", 5) || bufptr[45] != '\n')
 307                 return error("bogus commit object %s", sha1_to_hex(item->object.sha1));
 308         if (get_sha1_hex(bufptr + 5, parent) < 0)
 309                 return error("bad tree pointer in commit %s",
 310                              sha1_to_hex(item->object.sha1));
 311         item->tree = lookup_tree(parent);
 312         bufptr += 46; /* "tree " + "hex sha1" + "\n" */
 313         pptr = &item->parents;
 314
 315         graft = lookup_commit_graft(item->object.sha1);
 316         while (bufptr + 48 < tail && !memcmp(bufptr, "parent ", 7)) {
 317                 struct commit *new_parent;
 318
 319                 if (tail <= bufptr + 48 ||
 320                     get_sha1_hex(bufptr + 7, parent) ||
 321                     bufptr[47] != '\n')
 322                         return error("bad parents in commit %s", sha1_to_hex(item->object.sha1));
 323                 bufptr += 48;
 324                 /*
 325                  * The clone is shallow if nr_parent < 0, and we must
 326                  * not traverse its real parents even when we unhide them.
 327                  */
 328                 if (graft && (graft->nr_parent < 0 || grafts_replace_parents))
 329                         continue;
 330                 new_parent = lookup_commit(parent);
 331                 if (new_parent)
 332                         pptr = &commit_list_insert(new_parent, pptr)->next;
 333         }
 334         if (graft) {
 335                 int i;
 336                 struct commit *new_parent;
 337                 for (i = 0; i < graft->nr_parent; i++) {
 338                         new_parent = lookup_commit(graft->parent[i]);
 339                         if (!new_parent)
 340                                 continue;
 341                         pptr = &commit_list_insert(new_parent, pptr)->next;
 342                 }
 343         }
 344         item->date = parse_commit_date(bufptr, tail);
 345
 346         return 0;
 347 }
 348
 349 int parse_commit(struct commit *item)
 350 {
 351         enum object_type type;
 352         void *buffer;
 353         unsigned long size;
 354         int ret;
 355
 356         if (!item)
 357                 return -1;
 358         if (item->object.parsed)
 359                 return 0;
 360         buffer = read_sha1_file(item->object.sha1, &type, &size);
 361         if (!buffer)
 362                 return error("Could not read %s",
 363                              sha1_to_hex(item->object.sha1));
 364         if (type != OBJ_COMMIT) {
 365                 free(buffer);
 366                 return error("Object %s not a commit",
 367                              sha1_to_hex(item->object.sha1));
 368         }
 369         ret = parse_commit_buffer(item, buffer, size);
 370         if (save_commit_buffer && !ret) {
 371                 set_commit_buffer(item, buffer);
 372                 return 0;
 373         }
 374         free(buffer);
 375         return ret;
 376 }
 377
 378 void parse_commit_or_die(struct commit *item)
 379 {
 380         if (parse_commit(item))
 381                 die("unable to parse commit %s",
 382                     item ? sha1_to_hex(item->object.sha1) : "(null)");
 383 }
 384
 385 int find_commit_subject(const char *commit_buffer, const char **subject)
 386 {
 387         const char *eol;
 388         const char *p = commit_buffer;
 389
 390         while (*p && (*p != '\n' || p[1] != '\n'))
 391                 p++;
 392         if (*p) {
 393                 p += 2;
 394                 for (eol = p; *eol && *eol != '\n'; eol++)
 395                         ; /* do nothing */
 396         } else
 397                 eol = p;
 398
 399         *subject = p;
 400
 401         return eol - p;
 402 }
 403
 404 struct commit_list *commit_list_insert(struct commit *item, struct commit_list **list_p)
 405 {
 406         struct commit_list *new_list = xmalloc(sizeof(struct commit_list));
 407         new_list->item = item;
 408         new_list->next = *list_p;
 409         *list_p = new_list;
 410         return new_list;
 411 }
 412
 413 unsigned commit_list_count(const struct commit_list *l)
 414 {
 415         unsigned c = 0;
 416         for (; l; l = l->next )
 417                 c++;
 418         return c;
 419 }
 420
 421 struct commit_list *copy_commit_list(struct commit_list *list)
 422 {
 423         struct commit_list *head = NULL;
 424         struct commit_list **pp = &head;
 425         while (list) {
 426                 struct commit_list *new;
 427                 new = xmalloc(sizeof(struct commit_list));
 428                 new->item = list->item;
 429                 new->next = NULL;
 430                 *pp = new;
 431                 pp = &new->next;
 432                 list = list->next;
 433         }
 434         return head;
 435 }
 436
 437 void free_commit_list(struct commit_list *list)
 438 {
 439         while (list) {
 440                 struct commit_list *temp = list;
 441                 list = temp->next;
 442                 free(temp);
 443         }
 444 }
 445
 446 struct commit_list * commit_list_insert_by_date(struct commit *item, struct commit_list **list)
 447 {
 448         struct commit_list **pp = list;
 449         struct commit_list *p;
 450         while ((p = *pp) != NULL) {
 451                 if (p->item->date < item->date) {
 452                         break;
 453                 }
 454                 pp = &p->next;
 455         }
 456         return commit_list_insert(item, pp);
 457 }
 458
 459 static int commit_list_compare_by_date(const void *a, const void *b)
 460 {
 461         unsigned long a_date = ((const struct commit_list *)a)->item->date;
 462         unsigned long b_date = ((const struct commit_list *)b)->item->date;
 463         if (a_date < b_date)
 464                 return 1;
 465         if (a_date > b_date)
 466                 return -1;
 467         return 0;
 468 }
 469
 470 static void *commit_list_get_next(const void *a)
 471 {
 472         return ((const struct commit_list *)a)->next;
 473 }
 474
 475 static void commit_list_set_next(void *a, void *next)
 476 {
 477         ((struct commit_list *)a)->next = next;
 478 }
 479
 480 void commit_list_sort_by_date(struct commit_list **list)
 481 {
 482         *list = llist_mergesort(*list, commit_list_get_next, commit_list_set_next,
 483                                 commit_list_compare_by_date);
 484 }
 485
 486 struct commit *pop_most_recent_commit(struct commit_list **list,
 487                                       unsigned int mark)
 488 {
 489         struct commit *ret = (*list)->item;
 490         struct commit_list *parents = ret->parents;
 491         struct commit_list *old = *list;
 492
 493         *list = (*list)->next;
 494         free(old);
 495
 496         while (parents) {
 497                 struct commit *commit = parents->item;
 498                 if (!parse_commit(commit) && !(commit->object.flags & mark)) {
 499                         commit->object.flags |= mark;
 500                         commit_list_insert_by_date(commit, list);
 501                 }
 502                 parents = parents->next;
 503         }
 504         return ret;
 505 }
 506
 507 static void clear_commit_marks_1(struct commit_list **plist,
 508                                  struct commit *commit, unsigned int mark)
 509 {
 510         while (commit) {
 511                 struct commit_list *parents;
 512
 513                 if (!(mark & commit->object.flags))
 514                         return;
 515
 516                 commit->object.flags &= ~mark;
 517
 518                 parents = commit->parents;
 519                 if (!parents)
 520                         return;
 521
 522                 while ((parents = parents->next))
 523                         commit_list_insert(parents->item, plist);
 524
 525                 commit = commit->parents->item;
 526         }
 527 }
 528
 529 void clear_commit_marks_many(int nr, struct commit **commit, unsigned int mark)
 530 {
 531         struct commit_list *list = NULL;
 532
 533         while (nr--) {
 534                 commit_list_insert(*commit, &list);
 535                 commit++;
 536         }
 537         while (list)
 538                 clear_commit_marks_1(&list, pop_commit(&list), mark);
 539 }
 540
 541 void clear_commit_marks(struct commit *commit, unsigned int mark)
 542 {
 543         clear_commit_marks_many(1, &commit, mark);
 544 }
 545
 546 void clear_commit_marks_for_object_array(struct object_array *a, unsigned mark)
 547 {
 548         struct object *object;
 549         struct commit *commit;
 550         unsigned int i;
 551
 552         for (i = 0; i < a->nr; i++) {
 553                 object = a->objects[i].item;
 554                 commit = lookup_commit_reference_gently(object->sha1, 1);
 555                 if (commit)
 556                         clear_commit_marks(commit, mark);
 557         }
 558 }
 559
 560 struct commit *pop_commit(struct commit_list **stack)
 561 {
 562         struct commit_list *top = *stack;
 563         struct commit *item = top ? top->item : NULL;
 564
 565         if (top) {
 566                 *stack = top->next;
 567                 free(top);
 568         }
 569         return item;
 570 }
 571
 572 /*
 573  * Topological sort support
 574  */
 575
 576 /* count number of children that have not been emitted */
 577 define_commit_slab(indegree_slab, int);
 578
 579 /* record author-date for each commit object */
 580 define_commit_slab(author_date_slab, unsigned long);
 581
 582 static void record_author_date(struct author_date_slab *author_date,
 583                                struct commit *commit)
 584 {
 585         const char *buf, *line_end, *ident_line;
 586         const char *buffer = get_commit_buffer(commit);
 587         struct ident_split ident;
 588         char *date_end;
 589         unsigned long date;
 590
 591         for (buf = buffer; buf; buf = line_end + 1) {
 592                 line_end = strchrnul(buf, '\n');
 593                 ident_line = skip_prefix(buf, "author ");
 594                 if (!ident_line) {
 595                         if (!line_end[0] || line_end[1] == '\n')
 596                                 return; /* end of header */
 597                         continue;
 598                 }
 599                 if (split_ident_line(&ident,
 600                                      ident_line, line_end - ident_line) ||
 601                     !ident.date_begin || !ident.date_end)
 602                         goto fail_exit; /* malformed "author" line */
 603                 break;
 604         }
 605
 606         date = strtoul(ident.date_begin, &date_end, 10);
 607         if (date_end != ident.date_end)
 608                 goto fail_exit; /* malformed date */
 609         *(author_date_slab_at(author_date, commit)) = date;
 610
 611 fail_exit:
 612         unuse_commit_buffer(commit, buffer);
 613 }
 614
 615 static int compare_commits_by_author_date(const void *a_, const void *b_,
 616                                           void *cb_data)
 617 {
 618         const struct commit *a = a_, *b = b_;
 619         struct author_date_slab *author_date = cb_data;
 620         unsigned long a_date = *(author_date_slab_at(author_date, a));
 621         unsigned long b_date = *(author_date_slab_at(author_date, b));
 622
 623         /* newer commits with larger date first */
 624         if (a_date < b_date)
 625                 return 1;
 626         else if (a_date > b_date)
 627                 return -1;
 628         return 0;
 629 }
 630
 631 int compare_commits_by_commit_date(const void *a_, const void *b_, void *unused)
 632 {
 633         const struct commit *a = a_, *b = b_;
 634         /* newer commits with larger date first */
 635         if (a->date < b->date)
 636                 return 1;
 637         else if (a->date > b->date)
 638                 return -1;
 639         return 0;
 640 }
 641
 642 /*
 643  * Performs an in-place topological sort on the list supplied.
 644  */
 645 void sort_in_topological_order(struct commit_list **list, enum rev_sort_order sort_order)
 646 {
 647         struct commit_list *next, *orig = *list;
 648         struct commit_list **pptr;
 649         struct indegree_slab indegree;
 650         struct prio_queue queue;
 651         struct commit *commit;
 652         struct author_date_slab author_date;
 653
 654         if (!orig)
 655                 return;
 656         *list = NULL;
 657
 658         init_indegree_slab(&indegree);
 659         memset(&queue, '\0', sizeof(queue));
 660
 661         switch (sort_order) {
 662         default: /* REV_SORT_IN_GRAPH_ORDER */
 663                 queue.compare = NULL;
 664                 break;
 665         case REV_SORT_BY_COMMIT_DATE:
 666                 queue.compare = compare_commits_by_commit_date;
 667                 break;
 668         case REV_SORT_BY_AUTHOR_DATE:
 669                 init_author_date_slab(&author_date);
 670                 queue.compare = compare_commits_by_author_date;
 671                 queue.cb_data = &author_date;
 672                 break;
 673         }
 674
 675         /* Mark them and clear the indegree */
 676         for (next = orig; next; next = next->next) {
 677                 struct commit *commit = next->item;
 678                 *(indegree_slab_at(&indegree, commit)) = 1;
 679                 /* also record the author dates, if needed */
 680                 if (sort_order == REV_SORT_BY_AUTHOR_DATE)
 681                         record_author_date(&author_date, commit);
 682         }
 683
 684         /* update the indegree */
 685         for (next = orig; next; next = next->next) {
 686                 struct commit_list *parents = next->item->parents;
 687                 while (parents) {
 688                         struct commit *parent = parents->item;
 689                         int *pi = indegree_slab_at(&indegree, parent);
 690
 691                         if (*pi)
 692                                 (*pi)++;
 693                         parents = parents->next;
 694                 }
 695         }
 696
 697         /*
 698          * find the tips
 699          *
 700          * tips are nodes not reachable from any other node in the list
 701          *
 702          * the tips serve as a starting set for the work queue.
 703          */
 704         for (next = orig; next; next = next->next) {
 705                 struct commit *commit = next->item;
 706
 707                 if (*(indegree_slab_at(&indegree, commit)) == 1)
 708                         prio_queue_put(&queue, commit);
 709         }
 710
 711         /*
 712          * This is unfortunate; the initial tips need to be shown
 713          * in the order given from the revision traversal machinery.
 714          */
 715         if (sort_order == REV_SORT_IN_GRAPH_ORDER)
 716                 prio_queue_reverse(&queue);
 717
 718         /* We no longer need the commit list */
 719         free_commit_list(orig);
 720
 721         pptr = list;
 722         *list = NULL;
 723         while ((commit = prio_queue_get(&queue)) != NULL) {
 724                 struct commit_list *parents;
 725
 726                 for (parents = commit->parents; parents ; parents = parents->next) {
 727                         struct commit *parent = parents->item;
 728                         int *pi = indegree_slab_at(&indegree, parent);
 729
 730                         if (!*pi)
 731                                 continue;
 732
 733                         /*
 734                          * parents are only enqueued for emission
 735                          * when all their children have been emitted thereby
 736                          * guaranteeing topological order.
 737                          */
 738                         if (--(*pi) == 1)
 739                                 prio_queue_put(&queue, parent);
 740                 }
 741                 /*
 742                  * all children of commit have already been
 743                  * emitted. we can emit it now.
 744                  */
 745                 *(indegree_slab_at(&indegree, commit)) = 0;
 746
 747                 pptr = &commit_list_insert(commit, pptr)->next;
 748         }
 749
 750         clear_indegree_slab(&indegree);
 751         clear_prio_queue(&queue);
 752         if (sort_order == REV_SORT_BY_AUTHOR_DATE)
 753                 clear_author_date_slab(&author_date);
 754 }
 755
 756 /* merge-base stuff */
 757
 758 /* Remember to update object flag allocation in object.h */
 759 #define PARENT1         (1u<<16)
 760 #define PARENT2         (1u<<17)
 761 #define STALE           (1u<<18)
 762 #define RESULT          (1u<<19)
 763
 764 static const unsigned all_flags = (PARENT1 | PARENT2 | STALE | RESULT);
 765
 766 static struct commit *interesting(struct commit_list *list)
 767 {
 768         while (list) {
 769                 struct commit *commit = list->item;
 770                 list = list->next;
 771                 if (commit->object.flags & STALE)
 772                         continue;
 773                 return commit;
 774         }
 775         return NULL;
 776 }
 777
 778 /* all input commits in one and twos[] must have been parsed! */
 779 static struct commit_list *paint_down_to_common(struct commit *one, int n, struct commit **twos)
 780 {
 781         struct commit_list *list = NULL;
 782         struct commit_list *result = NULL;
 783         int i;
 784
 785         one->object.flags |= PARENT1;
 786         commit_list_insert_by_date(one, &list);
 787         if (!n)
 788                 return list;
 789         for (i = 0; i < n; i++) {
 790                 twos[i]->object.flags |= PARENT2;
 791                 commit_list_insert_by_date(twos[i], &list);
 792         }
 793
 794         while (interesting(list)) {
 795                 struct commit *commit;
 796                 struct commit_list *parents;
 797                 struct commit_list *next;
 798                 int flags;
 799
 800                 commit = list->item;
 801                 next = list->next;
 802                 free(list);
 803                 list = next;
 804
 805                 flags = commit->object.flags & (PARENT1 | PARENT2 | STALE);
 806                 if (flags == (PARENT1 | PARENT2)) {
 807                         if (!(commit->object.flags & RESULT)) {
 808                                 commit->object.flags |= RESULT;
 809                                 commit_list_insert_by_date(commit, &result);
 810                         }
 811                         /* Mark parents of a found merge stale */
 812                         flags |= STALE;
 813                 }
 814                 parents = commit->parents;
 815                 while (parents) {
 816                         struct commit *p = parents->item;
 817                         parents = parents->next;
 818                         if ((p->object.flags & flags) == flags)
 819                                 continue;
 820                         if (parse_commit(p))
 821                                 return NULL;
 822                         p->object.flags |= flags;
 823                         commit_list_insert_by_date(p, &list);
 824                 }
 825         }
 826
 827         free_commit_list(list);
 828         return result;
 829 }
 830
 831 static struct commit_list *merge_bases_many(struct commit *one, int n, struct commit **twos)
 832 {
 833         struct commit_list *list = NULL;
 834         struct commit_list *result = NULL;
 835         int i;
 836
 837         for (i = 0; i < n; i++) {
 838                 if (one == twos[i])
 839                         /*
 840                          * We do not mark this even with RESULT so we do not
 841                          * have to clean it up.
 842                          */
 843                         return commit_list_insert(one, &result);
 844         }
 845
 846         if (parse_commit(one))
 847                 return NULL;
 848         for (i = 0; i < n; i++) {
 849                 if (parse_commit(twos[i]))
 850                         return NULL;
 851         }
 852
 853         list = paint_down_to_common(one, n, twos);
 854
 855         while (list) {
 856                 struct commit_list *next = list->next;
 857                 if (!(list->item->object.flags & STALE))
 858                         commit_list_insert_by_date(list->item, &result);
 859                 free(list);
 860                 list = next;
 861         }
 862         return result;
 863 }
 864
 865 struct commit_list *get_octopus_merge_bases(struct commit_list *in)
 866 {
 867         struct commit_list *i, *j, *k, *ret = NULL;
 868
 869         if (!in)
 870                 return ret;
 871
 872         commit_list_insert(in->item, &ret);
 873
 874         for (i = in->next; i; i = i->next) {
 875                 struct commit_list *new = NULL, *end = NULL;
 876
 877                 for (j = ret; j; j = j->next) {
 878                         struct commit_list *bases;
 879                         bases = get_merge_bases(i->item, j->item, 1);
 880                         if (!new)
 881                                 new = bases;
 882                         else
 883                                 end->next = bases;
 884                         for (k = bases; k; k = k->next)
 885                                 end = k;
 886                 }
 887                 ret = new;
 888         }
 889         return ret;
 890 }
 891
 892 static int remove_redundant(struct commit **array, int cnt)
 893 {
 894         /*
 895          * Some commit in the array may be an ancestor of
 896          * another commit.  Move such commit to the end of
 897          * the array, and return the number of commits that
 898          * are independent from each other.
 899          */
 900         struct commit **work;
 901         unsigned char *redundant;
 902         int *filled_index;
 903         int i, j, filled;
 904
 905         work = xcalloc(cnt, sizeof(*work));
 906         redundant = xcalloc(cnt, 1);
 907         filled_index = xmalloc(sizeof(*filled_index) * (cnt - 1));
 908
 909         for (i = 0; i < cnt; i++)
 910                 parse_commit(array[i]);
 911         for (i = 0; i < cnt; i++) {
 912                 struct commit_list *common;
 913
 914                 if (redundant[i])
 915                         continue;
 916                 for (j = filled = 0; j < cnt; j++) {
 917                         if (i == j || redundant[j])
 918                                 continue;
 919                         filled_index[filled] = j;
 920                         work[filled++] = array[j];
 921                 }
 922                 common = paint_down_to_common(array[i], filled, work);
 923                 if (array[i]->object.flags & PARENT2)
 924                         redundant[i] = 1;
 925                 for (j = 0; j < filled; j++)
 926                         if (work[j]->object.flags & PARENT1)
 927                                 redundant[filled_index[j]] = 1;
 928                 clear_commit_marks(array[i], all_flags);
 929                 for (j = 0; j < filled; j++)
 930                         clear_commit_marks(work[j], all_flags);
 931                 free_commit_list(common);
 932         }
 933
 934         /* Now collect the result */
 935         memcpy(work, array, sizeof(*array) * cnt);
 936         for (i = filled = 0; i < cnt; i++)
 937                 if (!redundant[i])
 938                         array[filled++] = work[i];
 939         for (j = filled, i = 0; i < cnt; i++)
 940                 if (redundant[i])
 941                         array[j++] = work[i];
 942         free(work);
 943         free(redundant);
 944         free(filled_index);
 945         return filled;
 946 }
 947
 948 struct commit_list *get_merge_bases_many(struct commit *one,
 949                                          int n,
 950                                          struct commit **twos,
 951                                          int cleanup)
 952 {
 953         struct commit_list *list;
 954         struct commit **rslt;
 955         struct commit_list *result;
 956         int cnt, i;
 957
 958         result = merge_bases_many(one, n, twos);
 959         for (i = 0; i < n; i++) {
 960                 if (one == twos[i])
 961                         return result;
 962         }
 963         if (!result || !result->next) {
 964                 if (cleanup) {
 965                         clear_commit_marks(one, all_flags);
 966                         clear_commit_marks_many(n, twos, all_flags);
 967                 }
 968                 return result;
 969         }
 970
 971         /* There are more than one */
 972         cnt = 0;
 973         list = result;
 974         while (list) {
 975                 list = list->next;
 976                 cnt++;
 977         }
 978         rslt = xcalloc(cnt, sizeof(*rslt));
 979         for (list = result, i = 0; list; list = list->next)
 980                 rslt[i++] = list->item;
 981         free_commit_list(result);
 982
 983         clear_commit_marks(one, all_flags);
 984         clear_commit_marks_many(n, twos, all_flags);
 985
 986         cnt = remove_redundant(rslt, cnt);
 987         result = NULL;
 988         for (i = 0; i < cnt; i++)
 989                 commit_list_insert_by_date(rslt[i], &result);
 990         free(rslt);
 991         return result;
 992 }
 993
 994 struct commit_list *get_merge_bases(struct commit *one, struct commit *two,
 995                                     int cleanup)
 996 {
 997         return get_merge_bases_many(one, 1, &two, cleanup);
 998 }
 999
1000 /*
1001  * Is "commit" a descendant of one of the elements on the "with_commit" list?
1002  */
1003 int is_descendant_of(struct commit *commit, struct commit_list *with_commit)
1004 {
1005         if (!with_commit)
1006                 return 1;
1007         while (with_commit) {
1008                 struct commit *other;
1009
1010                 other = with_commit->item;
1011                 with_commit = with_commit->next;
1012                 if (in_merge_bases(other, commit))
1013                         return 1;
1014         }
1015         return 0;
1016 }
1017
1018 /*
1019  * Is "commit" an ancestor of one of the "references"?
1020  */
1021 int in_merge_bases_many(struct commit *commit, int nr_reference, struct commit **reference)
1022 {
1023         struct commit_list *bases;
1024         int ret = 0, i;
1025
1026         if (parse_commit(commit))
1027                 return ret;
1028         for (i = 0; i < nr_reference; i++)
1029                 if (parse_commit(reference[i]))
1030                         return ret;
1031
1032         bases = paint_down_to_common(commit, nr_reference, reference);
1033         if (commit->object.flags & PARENT2)
1034                 ret = 1;
1035         clear_commit_marks(commit, all_flags);
1036         clear_commit_marks_many(nr_reference, reference, all_flags);
1037         free_commit_list(bases);
1038         return ret;
1039 }
1040
1041 /*
1042  * Is "commit" an ancestor of (i.e. reachable from) the "reference"?
1043  */
1044 int in_merge_bases(struct commit *commit, struct commit *reference)
1045 {
1046         return in_merge_bases_many(commit, 1, &reference);
1047 }
1048
1049 struct commit_list *reduce_heads(struct commit_list *heads)
1050 {
1051         struct commit_list *p;
1052         struct commit_list *result = NULL, **tail = &result;
1053         struct commit **array;
1054         int num_head, i;
1055
1056         if (!heads)
1057                 return NULL;
1058
1059         /* Uniquify */
1060         for (p = heads; p; p = p->next)
1061                 p->item->object.flags &= ~STALE;
1062         for (p = heads, num_head = 0; p; p = p->next) {
1063                 if (p->item->object.flags & STALE)
1064                         continue;
1065                 p->item->object.flags |= STALE;
1066                 num_head++;
1067         }
1068         array = xcalloc(sizeof(*array), num_head);
1069         for (p = heads, i = 0; p; p = p->next) {
1070                 if (p->item->object.flags & STALE) {
1071                         array[i++] = p->item;
1072                         p->item->object.flags &= ~STALE;
1073                 }
1074         }
1075         num_head = remove_redundant(array, num_head);
1076         for (i = 0; i < num_head; i++)
1077                 tail = &commit_list_insert(array[i], tail)->next;
1078         return result;
1079 }
1080
1081 static const char gpg_sig_header[] = "gpgsig";
1082 static const int gpg_sig_header_len = sizeof(gpg_sig_header) - 1;
1083
1084 static int do_sign_commit(struct strbuf *buf, const char *keyid)
1085 {
1086         struct strbuf sig = STRBUF_INIT;
1087         int inspos, copypos;
1088
1089         /* find the end of the header */
1090         inspos = strstr(buf->buf, "\n\n") - buf->buf + 1;
1091
1092         if (!keyid || !*keyid)
1093                 keyid = get_signing_key();
1094         if (sign_buffer(buf, &sig, keyid)) {
1095                 strbuf_release(&sig);
1096                 return -1;
1097         }
1098
1099         for (copypos = 0; sig.buf[copypos]; ) {
1100                 const char *bol = sig.buf + copypos;
1101                 const char *eol = strchrnul(bol, '\n');
1102                 int len = (eol - bol) + !!*eol;
1103
1104                 if (!copypos) {
1105                         strbuf_insert(buf, inspos, gpg_sig_header, gpg_sig_header_len);
1106                         inspos += gpg_sig_header_len;
1107                 }
1108                 strbuf_insert(buf, inspos++, " ", 1);
1109                 strbuf_insert(buf, inspos, bol, len);
1110                 inspos += len;
1111                 copypos += len;
1112         }
1113         strbuf_release(&sig);
1114         return 0;
1115 }
1116
1117 int parse_signed_commit(const unsigned char *sha1,
1118                         struct strbuf *payload, struct strbuf *signature)
1119 {
1120         unsigned long size;
1121         enum object_type type;
1122         char *buffer = read_sha1_file(sha1, &type, &size);
1123         int in_signature, saw_signature = -1;
1124         char *line, *tail;
1125
1126         if (!buffer || type != OBJ_COMMIT)
1127                 goto cleanup;
1128
1129         line = buffer;
1130         tail = buffer + size;
1131         in_signature = 0;
1132         saw_signature = 0;
1133         while (line < tail) {
1134                 const char *sig = NULL;
1135                 char *next = memchr(line, '\n', tail - line);
1136
1137                 next = next ? next + 1 : tail;
1138                 if (in_signature && line[0] == ' ')
1139                         sig = line + 1;
1140                 else if (starts_with(line, gpg_sig_header) &&
1141                          line[gpg_sig_header_len] == ' ')
1142                         sig = line + gpg_sig_header_len + 1;
1143                 if (sig) {
1144                         strbuf_add(signature, sig, next - sig);
1145                         saw_signature = 1;
1146                         in_signature = 1;
1147                 } else {
1148                         if (*line == '\n')
1149                                 /* dump the whole remainder of the buffer */
1150                                 next = tail;
1151                         strbuf_add(payload, line, next - line);
1152                         in_signature = 0;
1153                 }
1154                 line = next;
1155         }
1156  cleanup:
1157         free(buffer);
1158         return saw_signature;
1159 }
1160
1161 static void handle_signed_tag(struct commit *parent, struct commit_extra_header ***tail)
1162 {
1163         struct merge_remote_desc *desc;
1164         struct commit_extra_header *mergetag;
1165         char *buf;
1166         unsigned long size, len;
1167         enum object_type type;
1168
1169         desc = merge_remote_util(parent);
1170         if (!desc || !desc->obj)
1171                 return;
1172         buf = read_sha1_file(desc->obj->sha1, &type, &size);
1173         if (!buf || type != OBJ_TAG)
1174                 goto free_return;
1175         len = parse_signature(buf, size);
1176         if (size == len)
1177                 goto free_return;
1178         /*
1179          * We could verify this signature and either omit the tag when
1180          * it does not validate, but the integrator may not have the
1181          * public key of the signer of the tag he is merging, while a
1182          * later auditor may have it while auditing, so let's not run
1183          * verify-signed-buffer here for now...
1184          *
1185          * if (verify_signed_buffer(buf, len, buf + len, size - len, ...))
1186          *      warn("warning: signed tag unverified.");
1187          */
1188         mergetag = xcalloc(1, sizeof(*mergetag));
1189         mergetag->key = xstrdup("mergetag");
1190         mergetag->value = buf;
1191         mergetag->len = size;
1192
1193         **tail = mergetag;
1194         *tail = &mergetag->next;
1195         return;
1196
1197 free_return:
1198         free(buf);
1199 }
1200
1201 static struct {
1202         char result;
1203         const char *check;
1204 } sigcheck_gpg_status[] = {
1205         { 'G', "\n[GNUPG:] GOODSIG " },
1206         { 'B', "\n[GNUPG:] BADSIG " },
1207         { 'U', "\n[GNUPG:] TRUST_NEVER" },
1208         { 'U', "\n[GNUPG:] TRUST_UNDEFINED" },
1209 };
1210
1211 static void parse_gpg_output(struct signature_check *sigc)
1212 {
1213         const char *buf = sigc->gpg_status;
1214         int i;
1215
1216         /* Iterate over all search strings */
1217         for (i = 0; i < ARRAY_SIZE(sigcheck_gpg_status); i++) {
1218                 const char *found, *next;
1219
1220                 found = skip_prefix(buf, sigcheck_gpg_status[i].check + 1);
1221                 if (!found) {
1222                         found = strstr(buf, sigcheck_gpg_status[i].check);
1223                         if (!found)
1224                                 continue;
1225                         found += strlen(sigcheck_gpg_status[i].check);
1226                 }
1227                 sigc->result = sigcheck_gpg_status[i].result;
1228                 /* The trust messages are not followed by key/signer information */
1229                 if (sigc->result != 'U') {
1230                         sigc->key = xmemdupz(found, 16);
1231                         found += 17;
1232                         next = strchrnul(found, '\n');
1233                         sigc->signer = xmemdupz(found, next - found);
1234                 }
1235         }
1236 }
1237
1238 void check_commit_signature(const struct commit* commit, struct signature_check *sigc)
1239 {
1240         struct strbuf payload = STRBUF_INIT;
1241         struct strbuf signature = STRBUF_INIT;
1242         struct strbuf gpg_output = STRBUF_INIT;
1243         struct strbuf gpg_status = STRBUF_INIT;
1244         int status;
1245
1246         sigc->result = 'N';
1247
1248         if (parse_signed_commit(commit->object.sha1,
1249                                 &payload, &signature) <= 0)
1250                 goto out;
1251         status = verify_signed_buffer(payload.buf, payload.len,
1252                                       signature.buf, signature.len,
1253                                       &gpg_output, &gpg_status);
1254         if (status && !gpg_output.len)
1255                 goto out;
1256         sigc->gpg_output = strbuf_detach(&gpg_output, NULL);
1257         sigc->gpg_status = strbuf_detach(&gpg_status, NULL);
1258         parse_gpg_output(sigc);
1259
1260  out:
1261         strbuf_release(&gpg_status);
1262         strbuf_release(&gpg_output);
1263         strbuf_release(&payload);
1264         strbuf_release(&signature);
1265 }
1266
1267
1268
1269 void append_merge_tag_headers(struct commit_list *parents,
1270                               struct commit_extra_header ***tail)
1271 {
1272         while (parents) {
1273                 struct commit *parent = parents->item;
1274                 handle_signed_tag(parent, tail);
1275                 parents = parents->next;
1276         }
1277 }
1278
1279 static void add_extra_header(struct strbuf *buffer,
1280                              struct commit_extra_header *extra)
1281 {
1282         strbuf_addstr(buffer, extra->key);
1283         if (extra->len)
1284                 strbuf_add_lines(buffer, " ", extra->value, extra->len);
1285         else
1286                 strbuf_addch(buffer, '\n');
1287 }
1288
1289 struct commit_extra_header *read_commit_extra_headers(struct commit *commit,
1290                                                       const char **exclude)
1291 {
1292         struct commit_extra_header *extra = NULL;
1293         unsigned long size;
1294         enum object_type type;
1295         char *buffer = read_sha1_file(commit->object.sha1, &type, &size);
1296         if (buffer && type == OBJ_COMMIT)
1297                 extra = read_commit_extra_header_lines(buffer, size, exclude);
1298         free(buffer);
1299         return extra;
1300 }
1301
1302 static inline int standard_header_field(const char *field, size_t len)
1303 {
1304         return ((len == 4 && !memcmp(field, "tree ", 5)) ||
1305                 (len == 6 && !memcmp(field, "parent ", 7)) ||
1306                 (len == 6 && !memcmp(field, "author ", 7)) ||
1307                 (len == 9 && !memcmp(field, "committer ", 10)) ||
1308                 (len == 8 && !memcmp(field, "encoding ", 9)));
1309 }
1310
1311 static int excluded_header_field(const char *field, size_t len, const char **exclude)
1312 {
1313         if (!exclude)
1314                 return 0;
1315
1316         while (*exclude) {
1317                 size_t xlen = strlen(*exclude);
1318                 if (len == xlen &&
1319                     !memcmp(field, *exclude, xlen) && field[xlen] == ' ')
1320                         return 1;
1321                 exclude++;
1322         }
1323         return 0;
1324 }
1325
1326 static struct commit_extra_header *read_commit_extra_header_lines(
1327         const char *buffer, size_t size,
1328         const char **exclude)
1329 {
1330         struct commit_extra_header *extra = NULL, **tail = &extra, *it = NULL;
1331         const char *line, *next, *eof, *eob;
1332         struct strbuf buf = STRBUF_INIT;
1333
1334         for (line = buffer, eob = line + size;
1335              line < eob && *line != '\n';
1336              line = next) {
1337                 next = memchr(line, '\n', eob - line);
1338                 next = next ? next + 1 : eob;
1339                 if (*line == ' ') {
1340                         /* continuation */
1341                         if (it)
1342                                 strbuf_add(&buf, line + 1, next - (line + 1));
1343                         continue;
1344                 }
1345                 if (it)
1346                         it->value = strbuf_detach(&buf, &it->len);
1347                 strbuf_reset(&buf);
1348                 it = NULL;
1349
1350                 eof = strchr(line, ' ');
1351                 if (next <= eof)
1352                         eof = next;
1353
1354                 if (standard_header_field(line, eof - line) ||
1355                     excluded_header_field(line, eof - line, exclude))
1356                         continue;
1357
1358                 it = xcalloc(1, sizeof(*it));
1359                 it->key = xmemdupz(line, eof-line);
1360                 *tail = it;
1361                 tail = &it->next;
1362                 if (eof + 1 < next)
1363                         strbuf_add(&buf, eof + 1, next - (eof + 1));
1364         }
1365         if (it)
1366                 it->value = strbuf_detach(&buf, &it->len);
1367         return extra;
1368 }
1369
1370 void free_commit_extra_headers(struct commit_extra_header *extra)
1371 {
1372         while (extra) {
1373                 struct commit_extra_header *next = extra->next;
1374                 free(extra->key);
1375                 free(extra->value);
1376                 free(extra);
1377                 extra = next;
1378         }
1379 }
1380
1381 int commit_tree(const char *msg, size_t msg_len,
1382                 const unsigned char *tree,
1383                 struct commit_list *parents, unsigned char *ret,
1384                 const char *author, const char *sign_commit)
1385 {
1386         struct commit_extra_header *extra = NULL, **tail = &extra;
1387         int result;
1388
1389         append_merge_tag_headers(parents, &tail);
1390         result = commit_tree_extended(msg, msg_len, tree, parents, ret,
1391                                       author, sign_commit, extra);
1392         free_commit_extra_headers(extra);
1393         return result;
1394 }
1395
1396 static int find_invalid_utf8(const char *buf, int len)
1397 {
1398         int offset = 0;
1399         static const unsigned int max_codepoint[] = {
1400                 0x7f, 0x7ff, 0xffff, 0x10ffff
1401         };
1402
1403         while (len) {
1404                 unsigned char c = *buf++;
1405                 int bytes, bad_offset;
1406                 unsigned int codepoint;
1407                 unsigned int min_val, max_val;
1408
1409                 len--;
1410                 offset++;
1411
1412                 /* Simple US-ASCII? No worries. */
1413                 if (c < 0x80)
1414                         continue;
1415
1416                 bad_offset = offset-1;
1417
1418                 /*
1419                  * Count how many more high bits set: that's how
1420                  * many more bytes this sequence should have.
1421                  */
1422                 bytes = 0;
1423                 while (c & 0x40) {
1424                         c <<= 1;
1425                         bytes++;
1426                 }
1427
1428                 /*
1429                  * Must be between 1 and 3 more bytes.  Longer sequences result in
1430                  * codepoints beyond U+10FFFF, which are guaranteed never to exist.
1431                  */
1432                 if (bytes < 1 || 3 < bytes)
1433                         return bad_offset;
1434
1435                 /* Do we *have* that many bytes? */
1436                 if (len < bytes)
1437                         return bad_offset;
1438
1439                 /*
1440                  * Place the encoded bits at the bottom of the value and compute the
1441                  * valid range.
1442                  */
1443                 codepoint = (c & 0x7f) >> bytes;
1444                 min_val = max_codepoint[bytes-1] + 1;
1445                 max_val = max_codepoint[bytes];
1446
1447                 offset += bytes;
1448                 len -= bytes;
1449
1450                 /* And verify that they are good continuation bytes */
1451                 do {
1452                         codepoint <<= 6;
1453                         codepoint |= *buf & 0x3f;
1454                         if ((*buf++ & 0xc0) != 0x80)
1455                                 return bad_offset;
1456                 } while (--bytes);
1457
1458                 /* Reject codepoints that are out of range for the sequence length. */
1459                 if (codepoint < min_val || codepoint > max_val)
1460                         return bad_offset;
1461                 /* Surrogates are only for UTF-16 and cannot be encoded in UTF-8. */
1462                 if ((codepoint & 0x1ff800) == 0xd800)
1463                         return bad_offset;
1464                 /* U+xxFFFE and U+xxFFFF are guaranteed non-characters. */
1465                 if ((codepoint & 0xfffe) == 0xfffe)
1466                         return bad_offset;
1467                 /* So are anything in the range U+FDD0..U+FDEF. */
1468                 if (codepoint >= 0xfdd0 && codepoint <= 0xfdef)
1469                         return bad_offset;
1470         }
1471         return -1;
1472 }
1473
1474 /*
1475  * This verifies that the buffer is in proper utf8 format.
1476  *
1477  * If it isn't, it assumes any non-utf8 characters are Latin1,
1478  * and does the conversion.
1479  */
1480 static int verify_utf8(struct strbuf *buf)
1481 {
1482         int ok = 1;
1483         long pos = 0;
1484
1485         for (;;) {
1486                 int bad;
1487                 unsigned char c;
1488                 unsigned char replace[2];
1489
1490                 bad = find_invalid_utf8(buf->buf + pos, buf->len - pos);
1491                 if (bad < 0)
1492                         return ok;
1493                 pos += bad;
1494                 ok = 0;
1495                 c = buf->buf[pos];
1496                 strbuf_remove(buf, pos, 1);
1497
1498                 /* We know 'c' must be in the range 128-255 */
1499                 replace[0] = 0xc0 + (c >> 6);
1500                 replace[1] = 0x80 + (c & 0x3f);
1501                 strbuf_insert(buf, pos, replace, 2);
1502                 pos += 2;
1503         }
1504 }
1505
1506 static const char commit_utf8_warn[] =
1507 "Warning: commit message did not conform to UTF-8.\n"
1508 "You may want to amend it after fixing the message, or set the config\n"
1509 "variable i18n.commitencoding to the encoding your project uses.\n";
1510
1511 int commit_tree_extended(const char *msg, size_t msg_len,
1512                          const unsigned char *tree,
1513                          struct commit_list *parents, unsigned char *ret,
1514                          const char *author, const char *sign_commit,
1515                          struct commit_extra_header *extra)
1516 {
1517         int result;
1518         int encoding_is_utf8;
1519         struct strbuf buffer;
1520
1521         assert_sha1_type(tree, OBJ_TREE);
1522
1523         if (memchr(msg, '\0', msg_len))
1524                 return error("a NUL byte in commit log message not allowed.");
1525
1526         /* Not having i18n.commitencoding is the same as having utf-8 */
1527         encoding_is_utf8 = is_encoding_utf8(git_commit_encoding);
1528
1529         strbuf_init(&buffer, 8192); /* should avoid reallocs for the headers */
1530         strbuf_addf(&buffer, "tree %s\n", sha1_to_hex(tree));
1531
1532         /*
1533          * NOTE! This ordering means that the same exact tree merged with a
1534          * different order of parents will be a _different_ changeset even
1535          * if everything else stays the same.
1536          */
1537         while (parents) {
1538                 struct commit_list *next = parents->next;
1539                 struct commit *parent = parents->item;
1540
1541                 strbuf_addf(&buffer, "parent %s\n",
1542                             sha1_to_hex(parent->object.sha1));
1543                 free(parents);
1544                 parents = next;
1545         }
1546
1547         /* Person/date information */
1548         if (!author)
1549                 author = git_author_info(IDENT_STRICT);
1550         strbuf_addf(&buffer, "author %s\n", author);
1551         strbuf_addf(&buffer, "committer %s\n", git_committer_info(IDENT_STRICT));
1552         if (!encoding_is_utf8)
1553                 strbuf_addf(&buffer, "encoding %s\n", git_commit_encoding);
1554
1555         while (extra) {
1556                 add_extra_header(&buffer, extra);
1557                 extra = extra->next;
1558         }
1559         strbuf_addch(&buffer, '\n');
1560
1561         /* And add the comment */
1562         strbuf_add(&buffer, msg, msg_len);
1563
1564         /* And check the encoding */
1565         if (encoding_is_utf8 && !verify_utf8(&buffer))
1566                 fprintf(stderr, commit_utf8_warn);
1567
1568         if (sign_commit && do_sign_commit(&buffer, sign_commit))
1569                 return -1;
1570
1571         result = write_sha1_file(buffer.buf, buffer.len, commit_type, ret);
1572         strbuf_release(&buffer);
1573         return result;
1574 }
1575
1576 struct commit *get_merge_parent(const char *name)
1577 {
1578         struct object *obj;
1579         struct commit *commit;
1580         unsigned char sha1[20];
1581         if (get_sha1(name, sha1))
1582                 return NULL;
1583         obj = parse_object(sha1);
1584         commit = (struct commit *)peel_to_type(name, 0, obj, OBJ_COMMIT);
1585         if (commit && !commit->util) {
1586                 struct merge_remote_desc *desc;
1587                 desc = xmalloc(sizeof(*desc));
1588                 desc->obj = obj;
1589                 desc->name = strdup(name);
1590                 commit->util = desc;
1591         }
1592         return commit;
1593 }
1594
1595 /*
1596  * Append a commit to the end of the commit_list.
1597  *
1598  * next starts by pointing to the variable that holds the head of an
1599  * empty commit_list, and is updated to point to the "next" field of
1600  * the last item on the list as new commits are appended.
1601  *
1602  * Usage example:
1603  *
1604  *     struct commit_list *list;
1605  *     struct commit_list **next = &list;
1606  *
1607  *     next = commit_list_append(c1, next);
1608  *     next = commit_list_append(c2, next);
1609  *     assert(commit_list_count(list) == 2);
1610  *     return list;
1611  */
1612 struct commit_list **commit_list_append(struct commit *commit,
1613                                         struct commit_list **next)
1614 {
1615         struct commit_list *new = xmalloc(sizeof(struct commit_list));
1616         new->item = commit;
1617         *next = new;
1618         new->next = NULL;
1619         return &new->next;
1620 }
1621
1622 void print_commit_list(struct commit_list *list,
1623                        const char *format_cur,
1624                        const char *format_last)
1625 {
1626         for ( ; list; list = list->next) {
1627                 const char *format = list->next ? format_cur : format_last;
1628                 printf(format, sha1_to_hex(list->item->object.sha1));
1629         }
1630 }