commit.c

   1 #include "cache.h"
   2 #include "tag.h"
   3 #include "commit.h"
   4 #include "commit-graph.h"
   5 #include "repository.h"
   6 #include "object-store.h"
   7 #include "pkt-line.h"
   8 #include "utf8.h"
   9 #include "diff.h"
  10 #include "revision.h"
  11 #include "notes.h"
  12 #include "alloc.h"
  13 #include "gpg-interface.h"
  14 #include "mergesort.h"
  15 #include "commit-slab.h"
  16 #include "prio-queue.h"
  17 #include "sha1-lookup.h"
  18 #include "wt-status.h"
  19 #include "advice.h"
  20 #include "refs.h"
  21 #include "commit-reach.h"
  22
  23 static struct commit_extra_header *read_commit_extra_header_lines(const char *buf, size_t len, const char **);
  24
  25 int save_commit_buffer = 1;
  26
  27 const char *commit_type = "commit";
  28
  29 struct commit *lookup_commit_reference_gently(struct repository *r,
  30                 const struct object_id *oid, int quiet)
  31 {
  32         struct object *obj = deref_tag(r,
  33                                        parse_object(r, oid),
  34                                        NULL, 0);
  35
  36         if (!obj)
  37                 return NULL;
  38         return object_as_type(r, obj, OBJ_COMMIT, quiet);
  39 }
  40
  41 struct commit *lookup_commit_reference(struct repository *r, const struct object_id *oid)
  42 {
  43         return lookup_commit_reference_gently(r, oid, 0);
  44 }
  45
  46 struct commit *lookup_commit_or_die(const struct object_id *oid, const char *ref_name)
  47 {
  48         struct commit *c = lookup_commit_reference(the_repository, oid);
  49         if (!c)
  50                 die(_("could not parse %s"), ref_name);
  51         if (!oideq(oid, &c->object.oid)) {
  52                 warning(_("%s %s is not a commit!"),
  53                         ref_name, oid_to_hex(oid));
  54         }
  55         return c;
  56 }
  57
  58 struct commit *lookup_commit(struct repository *r, const struct object_id *oid)
  59 {
  60         struct object *obj = lookup_object(r, oid);
  61         if (!obj)
  62                 return create_object(r, oid, alloc_commit_node(r));
  63         return object_as_type(r, obj, OBJ_COMMIT, 0);
  64 }
  65
  66 struct commit *lookup_commit_reference_by_name(const char *name)
  67 {
  68         struct object_id oid;
  69         struct commit *commit;
  70
  71         if (get_oid_committish(name, &oid))
  72                 return NULL;
  73         commit = lookup_commit_reference(the_repository, &oid);
  74         if (parse_commit(commit))
  75                 return NULL;
  76         return commit;
  77 }
  78
  79 static timestamp_t parse_commit_date(const char *buf, const char *tail)
  80 {
  81         const char *dateptr;
  82
  83         if (buf + 6 >= tail)
  84                 return 0;
  85         if (memcmp(buf, "author", 6))
  86                 return 0;
  87         while (buf < tail && *buf++ != '\n')
  88                 /* nada */;
  89         if (buf + 9 >= tail)
  90                 return 0;
  91         if (memcmp(buf, "committer", 9))
  92                 return 0;
  93         while (buf < tail && *buf++ != '>')
  94                 /* nada */;
  95         if (buf >= tail)
  96                 return 0;
  97         dateptr = buf;
  98         while (buf < tail && *buf++ != '\n')
  99                 /* nada */;
 100         if (buf >= tail)
 101                 return 0;
 102         /* dateptr < buf && buf[-1] == '\n', so parsing will stop at buf-1 */
 103         return parse_timestamp(dateptr, NULL, 10);
 104 }
 105
 106 static const unsigned char *commit_graft_sha1_access(size_t index, void *table)
 107 {
 108         struct commit_graft **commit_graft_table = table;
 109         return commit_graft_table[index]->oid.hash;
 110 }
 111
 112 static int commit_graft_pos(struct repository *r, const unsigned char *sha1)
 113 {
 114         return sha1_pos(sha1, r->parsed_objects->grafts,
 115                         r->parsed_objects->grafts_nr,
 116                         commit_graft_sha1_access);
 117 }
 118
 119 int register_commit_graft(struct repository *r, struct commit_graft *graft,
 120                           int ignore_dups)
 121 {
 122         int pos = commit_graft_pos(r, graft->oid.hash);
 123
 124         if (0 <= pos) {
 125                 if (ignore_dups)
 126                         free(graft);
 127                 else {
 128                         free(r->parsed_objects->grafts[pos]);
 129                         r->parsed_objects->grafts[pos] = graft;
 130                 }
 131                 return 1;
 132         }
 133         pos = -pos - 1;
 134         ALLOC_GROW(r->parsed_objects->grafts,
 135                    r->parsed_objects->grafts_nr + 1,
 136                    r->parsed_objects->grafts_alloc);
 137         r->parsed_objects->grafts_nr++;
 138         if (pos < r->parsed_objects->grafts_nr)
 139                 memmove(r->parsed_objects->grafts + pos + 1,
 140                         r->parsed_objects->grafts + pos,
 141                         (r->parsed_objects->grafts_nr - pos - 1) *
 142                         sizeof(*r->parsed_objects->grafts));
 143         r->parsed_objects->grafts[pos] = graft;
 144         return 0;
 145 }
 146
 147 struct commit_graft *read_graft_line(struct strbuf *line)
 148 {
 149         /* The format is just "Commit Parent1 Parent2 ...\n" */
 150         int i, phase;
 151         const char *tail = NULL;
 152         struct commit_graft *graft = NULL;
 153         struct object_id dummy_oid, *oid;
 154
 155         strbuf_rtrim(line);
 156         if (!line->len || line->buf[0] == '#')
 157                 return NULL;
 158         /*
 159          * phase 0 verifies line, counts hashes in line and allocates graft
 160          * phase 1 fills graft
 161          */
 162         for (phase = 0; phase < 2; phase++) {
 163                 oid = graft ? &graft->oid : &dummy_oid;
 164                 if (parse_oid_hex(line->buf, oid, &tail))
 165                         goto bad_graft_data;
 166                 for (i = 0; *tail != '\0'; i++) {
 167                         oid = graft ? &graft->parent[i] : &dummy_oid;
 168                         if (!isspace(*tail++) || parse_oid_hex(tail, oid, &tail))
 169                                 goto bad_graft_data;
 170                 }
 171                 if (!graft) {
 172                         graft = xmalloc(st_add(sizeof(*graft),
 173                                                st_mult(sizeof(struct object_id), i)));
 174                         graft->nr_parent = i;
 175                 }
 176         }
 177         return graft;
 178
 179 bad_graft_data:
 180         error("bad graft data: %s", line->buf);
 181         assert(!graft);
 182         return NULL;
 183 }
 184
 185 static int read_graft_file(struct repository *r, const char *graft_file)
 186 {
 187         FILE *fp = fopen_or_warn(graft_file, "r");
 188         struct strbuf buf = STRBUF_INIT;
 189         if (!fp)
 190                 return -1;
 191         if (advice_graft_file_deprecated)
 192                 advise(_("Support for <GIT_DIR>/info/grafts is deprecated\n"
 193                          "and will be removed in a future Git version.\n"
 194                          "\n"
 195                          "Please use \"git replace --convert-graft-file\"\n"
 196                          "to convert the grafts into replace refs.\n"
 197                          "\n"
 198                          "Turn this message off by running\n"
 199                          "\"git config advice.graftFileDeprecated false\""));
 200         while (!strbuf_getwholeline(&buf, fp, '\n')) {
 201                 /* The format is just "Commit Parent1 Parent2 ...\n" */
 202                 struct commit_graft *graft = read_graft_line(&buf);
 203                 if (!graft)
 204                         continue;
 205                 if (register_commit_graft(r, graft, 1))
 206                         error("duplicate graft data: %s", buf.buf);
 207         }
 208         fclose(fp);
 209         strbuf_release(&buf);
 210         return 0;
 211 }
 212
 213 void prepare_commit_graft(struct repository *r)
 214 {
 215         char *graft_file;
 216
 217         if (r->parsed_objects->commit_graft_prepared)
 218                 return;
 219         if (!startup_info->have_repository)
 220                 return;
 221
 222         graft_file = get_graft_file(r);
 223         read_graft_file(r, graft_file);
 224         /* make sure shallows are read */
 225         is_repository_shallow(r);
 226         r->parsed_objects->commit_graft_prepared = 1;
 227 }
 228
 229 struct commit_graft *lookup_commit_graft(struct repository *r, const struct object_id *oid)
 230 {
 231         int pos;
 232         prepare_commit_graft(r);
 233         pos = commit_graft_pos(r, oid->hash);
 234         if (pos < 0)
 235                 return NULL;
 236         return r->parsed_objects->grafts[pos];
 237 }
 238
 239 int for_each_commit_graft(each_commit_graft_fn fn, void *cb_data)
 240 {
 241         int i, ret;
 242         for (i = ret = 0; i < the_repository->parsed_objects->grafts_nr && !ret; i++)
 243                 ret = fn(the_repository->parsed_objects->grafts[i], cb_data);
 244         return ret;
 245 }
 246
 247 int unregister_shallow(const struct object_id *oid)
 248 {
 249         int pos = commit_graft_pos(the_repository, oid->hash);
 250         if (pos < 0)
 251                 return -1;
 252         if (pos + 1 < the_repository->parsed_objects->grafts_nr)
 253                 MOVE_ARRAY(the_repository->parsed_objects->grafts + pos,
 254                            the_repository->parsed_objects->grafts + pos + 1,
 255                            the_repository->parsed_objects->grafts_nr - pos - 1);
 256         the_repository->parsed_objects->grafts_nr--;
 257         return 0;
 258 }
 259
 260 struct commit_buffer {
 261         void *buffer;
 262         unsigned long size;
 263 };
 264 define_commit_slab(buffer_slab, struct commit_buffer);
 265
 266 struct buffer_slab *allocate_commit_buffer_slab(void)
 267 {
 268         struct buffer_slab *bs = xmalloc(sizeof(*bs));
 269         init_buffer_slab(bs);
 270         return bs;
 271 }
 272
 273 void free_commit_buffer_slab(struct buffer_slab *bs)
 274 {
 275         clear_buffer_slab(bs);
 276         free(bs);
 277 }
 278
 279 void set_commit_buffer(struct repository *r, struct commit *commit, void *buffer, unsigned long size)
 280 {
 281         struct commit_buffer *v = buffer_slab_at(
 282                 r->parsed_objects->buffer_slab, commit);
 283         v->buffer = buffer;
 284         v->size = size;
 285 }
 286
 287 const void *get_cached_commit_buffer(struct repository *r, const struct commit *commit, unsigned long *sizep)
 288 {
 289         struct commit_buffer *v = buffer_slab_peek(
 290                 r->parsed_objects->buffer_slab, commit);
 291         if (!v) {
 292                 if (sizep)
 293                         *sizep = 0;
 294                 return NULL;
 295         }
 296         if (sizep)
 297                 *sizep = v->size;
 298         return v->buffer;
 299 }
 300
 301 const void *repo_get_commit_buffer(struct repository *r,
 302                                    const struct commit *commit,
 303                                    unsigned long *sizep)
 304 {
 305         const void *ret = get_cached_commit_buffer(r, commit, sizep);
 306         if (!ret) {
 307                 enum object_type type;
 308                 unsigned long size;
 309                 ret = repo_read_object_file(r, &commit->object.oid, &type, &size);
 310                 if (!ret)
 311                         die("cannot read commit object %s",
 312                             oid_to_hex(&commit->object.oid));
 313                 if (type != OBJ_COMMIT)
 314                         die("expected commit for %s, got %s",
 315                             oid_to_hex(&commit->object.oid), type_name(type));
 316                 if (sizep)
 317                         *sizep = size;
 318         }
 319         return ret;
 320 }
 321
 322 void repo_unuse_commit_buffer(struct repository *r,
 323                               const struct commit *commit,
 324                               const void *buffer)
 325 {
 326         struct commit_buffer *v = buffer_slab_peek(
 327                 r->parsed_objects->buffer_slab, commit);
 328         if (!(v && v->buffer == buffer))
 329                 free((void *)buffer);
 330 }
 331
 332 void free_commit_buffer(struct parsed_object_pool *pool, struct commit *commit)
 333 {
 334         struct commit_buffer *v = buffer_slab_peek(
 335                 pool->buffer_slab, commit);
 336         if (v) {
 337                 FREE_AND_NULL(v->buffer);
 338                 v->size = 0;
 339         }
 340 }
 341
 342 static inline void set_commit_tree(struct commit *c, struct tree *t)
 343 {
 344         c->maybe_tree = t;
 345 }
 346
 347 struct tree *repo_get_commit_tree(struct repository *r,
 348                                   const struct commit *commit)
 349 {
 350         if (commit->maybe_tree || !commit->object.parsed)
 351                 return commit->maybe_tree;
 352
 353         if (commit->graph_pos != COMMIT_NOT_FROM_GRAPH)
 354                 return get_commit_tree_in_graph(r, commit);
 355
 356         return NULL;
 357 }
 358
 359 struct object_id *get_commit_tree_oid(const struct commit *commit)
 360 {
 361         struct tree *tree = get_commit_tree(commit);
 362         return tree ? &tree->object.oid : NULL;
 363 }
 364
 365 void release_commit_memory(struct parsed_object_pool *pool, struct commit *c)
 366 {
 367         set_commit_tree(c, NULL);
 368         free_commit_buffer(pool, c);
 369         c->index = 0;
 370         free_commit_list(c->parents);
 371
 372         c->object.parsed = 0;
 373 }
 374
 375 const void *detach_commit_buffer(struct commit *commit, unsigned long *sizep)
 376 {
 377         struct commit_buffer *v = buffer_slab_peek(
 378                 the_repository->parsed_objects->buffer_slab, commit);
 379         void *ret;
 380
 381         if (!v) {
 382                 if (sizep)
 383                         *sizep = 0;
 384                 return NULL;
 385         }
 386         ret = v->buffer;
 387         if (sizep)
 388                 *sizep = v->size;
 389
 390         v->buffer = NULL;
 391         v->size = 0;
 392         return ret;
 393 }
 394
 395 int parse_commit_buffer(struct repository *r, struct commit *item, const void *buffer, unsigned long size, int check_graph)
 396 {
 397         const char *tail = buffer;
 398         const char *bufptr = buffer;
 399         struct object_id parent;
 400         struct commit_list **pptr;
 401         struct commit_graft *graft;
 402         const int tree_entry_len = the_hash_algo->hexsz + 5;
 403         const int parent_entry_len = the_hash_algo->hexsz + 7;
 404         struct tree *tree;
 405
 406         if (item->object.parsed)
 407                 return 0;
 408
 409         if (item->parents) {
 410                 /*
 411                  * Presumably this is leftover from an earlier failed parse;
 412                  * clear it out in preparation for us re-parsing (we'll hit the
 413                  * same error, but that's good, since it lets our caller know
 414                  * the result cannot be trusted.
 415                  */
 416                 free_commit_list(item->parents);
 417                 item->parents = NULL;
 418         }
 419
 420         tail += size;
 421         if (tail <= bufptr + tree_entry_len + 1 || memcmp(bufptr, "tree ", 5) ||
 422                         bufptr[tree_entry_len] != '\n')
 423                 return error("bogus commit object %s", oid_to_hex(&item->object.oid));
 424         if (get_oid_hex(bufptr + 5, &parent) < 0)
 425                 return error("bad tree pointer in commit %s",
 426                              oid_to_hex(&item->object.oid));
 427         tree = lookup_tree(r, &parent);
 428         if (!tree)
 429                 return error("bad tree pointer %s in commit %s",
 430                              oid_to_hex(&parent),
 431                              oid_to_hex(&item->object.oid));
 432         set_commit_tree(item, tree);
 433         bufptr += tree_entry_len + 1; /* "tree " + "hex sha1" + "\n" */
 434         pptr = &item->parents;
 435
 436         graft = lookup_commit_graft(r, &item->object.oid);
 437         while (bufptr + parent_entry_len < tail && !memcmp(bufptr, "parent ", 7)) {
 438                 struct commit *new_parent;
 439
 440                 if (tail <= bufptr + parent_entry_len + 1 ||
 441                     get_oid_hex(bufptr + 7, &parent) ||
 442                     bufptr[parent_entry_len] != '\n')
 443                         return error("bad parents in commit %s", oid_to_hex(&item->object.oid));
 444                 bufptr += parent_entry_len + 1;
 445                 /*
 446                  * The clone is shallow if nr_parent < 0, and we must
 447                  * not traverse its real parents even when we unhide them.
 448                  */
 449                 if (graft && (graft->nr_parent < 0 || grafts_replace_parents))
 450                         continue;
 451                 new_parent = lookup_commit(r, &parent);
 452                 if (!new_parent)
 453                         return error("bad parent %s in commit %s",
 454                                      oid_to_hex(&parent),
 455                                      oid_to_hex(&item->object.oid));
 456                 pptr = &commit_list_insert(new_parent, pptr)->next;
 457         }
 458         if (graft) {
 459                 int i;
 460                 struct commit *new_parent;
 461                 for (i = 0; i < graft->nr_parent; i++) {
 462                         new_parent = lookup_commit(r,
 463                                                    &graft->parent[i]);
 464                         if (!new_parent)
 465                                 return error("bad graft parent %s in commit %s",
 466                                              oid_to_hex(&graft->parent[i]),
 467                                              oid_to_hex(&item->object.oid));
 468                         pptr = &commit_list_insert(new_parent, pptr)->next;
 469                 }
 470         }
 471         item->date = parse_commit_date(bufptr, tail);
 472
 473         if (check_graph)
 474                 load_commit_graph_info(r, item);
 475
 476         item->object.parsed = 1;
 477         return 0;
 478 }
 479
 480 int repo_parse_commit_internal(struct repository *r,
 481                                struct commit *item,
 482                                int quiet_on_missing,
 483                                int use_commit_graph)
 484 {
 485         enum object_type type;
 486         void *buffer;
 487         unsigned long size;
 488         int ret;
 489
 490         if (!item)
 491                 return -1;
 492         if (item->object.parsed)
 493                 return 0;
 494         if (use_commit_graph && parse_commit_in_graph(r, item))
 495                 return 0;
 496         buffer = repo_read_object_file(r, &item->object.oid, &type, &size);
 497         if (!buffer)
 498                 return quiet_on_missing ? -1 :
 499                         error("Could not read %s",
 500                              oid_to_hex(&item->object.oid));
 501         if (type != OBJ_COMMIT) {
 502                 free(buffer);
 503                 return error("Object %s not a commit",
 504                              oid_to_hex(&item->object.oid));
 505         }
 506
 507         ret = parse_commit_buffer(r, item, buffer, size, 0);
 508         if (save_commit_buffer && !ret) {
 509                 set_commit_buffer(r, item, buffer, size);
 510                 return 0;
 511         }
 512         free(buffer);
 513         return ret;
 514 }
 515
 516 int repo_parse_commit_gently(struct repository *r,
 517                              struct commit *item, int quiet_on_missing)
 518 {
 519         return repo_parse_commit_internal(r, item, quiet_on_missing, 1);
 520 }
 521
 522 void parse_commit_or_die(struct commit *item)
 523 {
 524         if (parse_commit(item))
 525                 die("unable to parse commit %s",
 526                     item ? oid_to_hex(&item->object.oid) : "(null)");
 527 }
 528
 529 int find_commit_subject(const char *commit_buffer, const char **subject)
 530 {
 531         const char *eol;
 532         const char *p = commit_buffer;
 533
 534         while (*p && (*p != '\n' || p[1] != '\n'))
 535                 p++;
 536         if (*p) {
 537                 p = skip_blank_lines(p + 2);
 538                 eol = strchrnul(p, '\n');
 539         } else
 540                 eol = p;
 541
 542         *subject = p;
 543
 544         return eol - p;
 545 }
 546
 547 struct commit_list *commit_list_insert(struct commit *item, struct commit_list **list_p)
 548 {
 549         struct commit_list *new_list = xmalloc(sizeof(struct commit_list));
 550         new_list->item = item;
 551         new_list->next = *list_p;
 552         *list_p = new_list;
 553         return new_list;
 554 }
 555
 556 unsigned commit_list_count(const struct commit_list *l)
 557 {
 558         unsigned c = 0;
 559         for (; l; l = l->next )
 560                 c++;
 561         return c;
 562 }
 563
 564 struct commit_list *copy_commit_list(struct commit_list *list)
 565 {
 566         struct commit_list *head = NULL;
 567         struct commit_list **pp = &head;
 568         while (list) {
 569                 pp = commit_list_append(list->item, pp);
 570                 list = list->next;
 571         }
 572         return head;
 573 }
 574
 575 void free_commit_list(struct commit_list *list)
 576 {
 577         while (list)
 578                 pop_commit(&list);
 579 }
 580
 581 struct commit_list * commit_list_insert_by_date(struct commit *item, struct commit_list **list)
 582 {
 583         struct commit_list **pp = list;
 584         struct commit_list *p;
 585         while ((p = *pp) != NULL) {
 586                 if (p->item->date < item->date) {
 587                         break;
 588                 }
 589                 pp = &p->next;
 590         }
 591         return commit_list_insert(item, pp);
 592 }
 593
 594 static int commit_list_compare_by_date(const void *a, const void *b)
 595 {
 596         timestamp_t a_date = ((const struct commit_list *)a)->item->date;
 597         timestamp_t b_date = ((const struct commit_list *)b)->item->date;
 598         if (a_date < b_date)
 599                 return 1;
 600         if (a_date > b_date)
 601                 return -1;
 602         return 0;
 603 }
 604
 605 static void *commit_list_get_next(const void *a)
 606 {
 607         return ((const struct commit_list *)a)->next;
 608 }
 609
 610 static void commit_list_set_next(void *a, void *next)
 611 {
 612         ((struct commit_list *)a)->next = next;
 613 }
 614
 615 void commit_list_sort_by_date(struct commit_list **list)
 616 {
 617         *list = llist_mergesort(*list, commit_list_get_next, commit_list_set_next,
 618                                 commit_list_compare_by_date);
 619 }
 620
 621 struct commit *pop_most_recent_commit(struct commit_list **list,
 622                                       unsigned int mark)
 623 {
 624         struct commit *ret = pop_commit(list);
 625         struct commit_list *parents = ret->parents;
 626
 627         while (parents) {
 628                 struct commit *commit = parents->item;
 629                 if (!parse_commit(commit) && !(commit->object.flags & mark)) {
 630                         commit->object.flags |= mark;
 631                         commit_list_insert_by_date(commit, list);
 632                 }
 633                 parents = parents->next;
 634         }
 635         return ret;
 636 }
 637
 638 static void clear_commit_marks_1(struct commit_list **plist,
 639                                  struct commit *commit, unsigned int mark)
 640 {
 641         while (commit) {
 642                 struct commit_list *parents;
 643
 644                 if (!(mark & commit->object.flags))
 645                         return;
 646
 647                 commit->object.flags &= ~mark;
 648
 649                 parents = commit->parents;
 650                 if (!parents)
 651                         return;
 652
 653                 while ((parents = parents->next))
 654                         commit_list_insert(parents->item, plist);
 655
 656                 commit = commit->parents->item;
 657         }
 658 }
 659
 660 void clear_commit_marks_many(int nr, struct commit **commit, unsigned int mark)
 661 {
 662         struct commit_list *list = NULL;
 663
 664         while (nr--) {
 665                 clear_commit_marks_1(&list, *commit, mark);
 666                 commit++;
 667         }
 668         while (list)
 669                 clear_commit_marks_1(&list, pop_commit(&list), mark);
 670 }
 671
 672 void clear_commit_marks(struct commit *commit, unsigned int mark)
 673 {
 674         clear_commit_marks_many(1, &commit, mark);
 675 }
 676
 677 struct commit *pop_commit(struct commit_list **stack)
 678 {
 679         struct commit_list *top = *stack;
 680         struct commit *item = top ? top->item : NULL;
 681
 682         if (top) {
 683                 *stack = top->next;
 684                 free(top);
 685         }
 686         return item;
 687 }
 688
 689 /*
 690  * Topological sort support
 691  */
 692
 693 /* count number of children that have not been emitted */
 694 define_commit_slab(indegree_slab, int);
 695
 696 define_commit_slab(author_date_slab, timestamp_t);
 697
 698 void record_author_date(struct author_date_slab *author_date,
 699                         struct commit *commit)
 700 {
 701         const char *buffer = get_commit_buffer(commit, NULL);
 702         struct ident_split ident;
 703         const char *ident_line;
 704         size_t ident_len;
 705         char *date_end;
 706         timestamp_t date;
 707
 708         ident_line = find_commit_header(buffer, "author", &ident_len);
 709         if (!ident_line)
 710                 goto fail_exit; /* no author line */
 711         if (split_ident_line(&ident, ident_line, ident_len) ||
 712             !ident.date_begin || !ident.date_end)
 713                 goto fail_exit; /* malformed "author" line */
 714
 715         date = parse_timestamp(ident.date_begin, &date_end, 10);
 716         if (date_end != ident.date_end)
 717                 goto fail_exit; /* malformed date */
 718         *(author_date_slab_at(author_date, commit)) = date;
 719
 720 fail_exit:
 721         unuse_commit_buffer(commit, buffer);
 722 }
 723
 724 int compare_commits_by_author_date(const void *a_, const void *b_,
 725                                    void *cb_data)
 726 {
 727         const struct commit *a = a_, *b = b_;
 728         struct author_date_slab *author_date = cb_data;
 729         timestamp_t a_date = *(author_date_slab_at(author_date, a));
 730         timestamp_t b_date = *(author_date_slab_at(author_date, b));
 731
 732         /* newer commits with larger date first */
 733         if (a_date < b_date)
 734                 return 1;
 735         else if (a_date > b_date)
 736                 return -1;
 737         return 0;
 738 }
 739
 740 int compare_commits_by_gen_then_commit_date(const void *a_, const void *b_, void *unused)
 741 {
 742         const struct commit *a = a_, *b = b_;
 743
 744         /* newer commits first */
 745         if (a->generation < b->generation)
 746                 return 1;
 747         else if (a->generation > b->generation)
 748                 return -1;
 749
 750         /* use date as a heuristic when generations are equal */
 751         if (a->date < b->date)
 752                 return 1;
 753         else if (a->date > b->date)
 754                 return -1;
 755         return 0;
 756 }
 757
 758 int compare_commits_by_commit_date(const void *a_, const void *b_, void *unused)
 759 {
 760         const struct commit *a = a_, *b = b_;
 761         /* newer commits with larger date first */
 762         if (a->date < b->date)
 763                 return 1;
 764         else if (a->date > b->date)
 765                 return -1;
 766         return 0;
 767 }
 768
 769 /*
 770  * Performs an in-place topological sort on the list supplied.
 771  */
 772 void sort_in_topological_order(struct commit_list **list, enum rev_sort_order sort_order)
 773 {
 774         struct commit_list *next, *orig = *list;
 775         struct commit_list **pptr;
 776         struct indegree_slab indegree;
 777         struct prio_queue queue;
 778         struct commit *commit;
 779         struct author_date_slab author_date;
 780
 781         if (!orig)
 782                 return;
 783         *list = NULL;
 784
 785         init_indegree_slab(&indegree);
 786         memset(&queue, '\0', sizeof(queue));
 787
 788         switch (sort_order) {
 789         default: /* REV_SORT_IN_GRAPH_ORDER */
 790                 queue.compare = NULL;
 791                 break;
 792         case REV_SORT_BY_COMMIT_DATE:
 793                 queue.compare = compare_commits_by_commit_date;
 794                 break;
 795         case REV_SORT_BY_AUTHOR_DATE:
 796                 init_author_date_slab(&author_date);
 797                 queue.compare = compare_commits_by_author_date;
 798                 queue.cb_data = &author_date;
 799                 break;
 800         }
 801
 802         /* Mark them and clear the indegree */
 803         for (next = orig; next; next = next->next) {
 804                 struct commit *commit = next->item;
 805                 *(indegree_slab_at(&indegree, commit)) = 1;
 806                 /* also record the author dates, if needed */
 807                 if (sort_order == REV_SORT_BY_AUTHOR_DATE)
 808                         record_author_date(&author_date, commit);
 809         }
 810
 811         /* update the indegree */
 812         for (next = orig; next; next = next->next) {
 813                 struct commit_list *parents = next->item->parents;
 814                 while (parents) {
 815                         struct commit *parent = parents->item;
 816                         int *pi = indegree_slab_at(&indegree, parent);
 817
 818                         if (*pi)
 819                                 (*pi)++;
 820                         parents = parents->next;
 821                 }
 822         }
 823
 824         /*
 825          * find the tips
 826          *
 827          * tips are nodes not reachable from any other node in the list
 828          *
 829          * the tips serve as a starting set for the work queue.
 830          */
 831         for (next = orig; next; next = next->next) {
 832                 struct commit *commit = next->item;
 833
 834                 if (*(indegree_slab_at(&indegree, commit)) == 1)
 835                         prio_queue_put(&queue, commit);
 836         }
 837
 838         /*
 839          * This is unfortunate; the initial tips need to be shown
 840          * in the order given from the revision traversal machinery.
 841          */
 842         if (sort_order == REV_SORT_IN_GRAPH_ORDER)
 843                 prio_queue_reverse(&queue);
 844
 845         /* We no longer need the commit list */
 846         free_commit_list(orig);
 847
 848         pptr = list;
 849         *list = NULL;
 850         while ((commit = prio_queue_get(&queue)) != NULL) {
 851                 struct commit_list *parents;
 852
 853                 for (parents = commit->parents; parents ; parents = parents->next) {
 854                         struct commit *parent = parents->item;
 855                         int *pi = indegree_slab_at(&indegree, parent);
 856
 857                         if (!*pi)
 858                                 continue;
 859
 860                         /*
 861                          * parents are only enqueued for emission
 862                          * when all their children have been emitted thereby
 863                          * guaranteeing topological order.
 864                          */
 865                         if (--(*pi) == 1)
 866                                 prio_queue_put(&queue, parent);
 867                 }
 868                 /*
 869                  * all children of commit have already been
 870                  * emitted. we can emit it now.
 871                  */
 872                 *(indegree_slab_at(&indegree, commit)) = 0;
 873
 874                 pptr = &commit_list_insert(commit, pptr)->next;
 875         }
 876
 877         clear_indegree_slab(&indegree);
 878         clear_prio_queue(&queue);
 879         if (sort_order == REV_SORT_BY_AUTHOR_DATE)
 880                 clear_author_date_slab(&author_date);
 881 }
 882
 883 struct rev_collect {
 884         struct commit **commit;
 885         int nr;
 886         int alloc;
 887         unsigned int initial : 1;
 888 };
 889
 890 static void add_one_commit(struct object_id *oid, struct rev_collect *revs)
 891 {
 892         struct commit *commit;
 893
 894         if (is_null_oid(oid))
 895                 return;
 896
 897         commit = lookup_commit(the_repository, oid);
 898         if (!commit ||
 899             (commit->object.flags & TMP_MARK) ||
 900             parse_commit(commit))
 901                 return;
 902
 903         ALLOC_GROW(revs->commit, revs->nr + 1, revs->alloc);
 904         revs->commit[revs->nr++] = commit;
 905         commit->object.flags |= TMP_MARK;
 906 }
 907
 908 static int collect_one_reflog_ent(struct object_id *ooid, struct object_id *noid,
 909                                   const char *ident, timestamp_t timestamp,
 910                                   int tz, const char *message, void *cbdata)
 911 {
 912         struct rev_collect *revs = cbdata;
 913
 914         if (revs->initial) {
 915                 revs->initial = 0;
 916                 add_one_commit(ooid, revs);
 917         }
 918         add_one_commit(noid, revs);
 919         return 0;
 920 }
 921
 922 struct commit *get_fork_point(const char *refname, struct commit *commit)
 923 {
 924         struct object_id oid;
 925         struct rev_collect revs;
 926         struct commit_list *bases;
 927         int i;
 928         struct commit *ret = NULL;
 929
 930         memset(&revs, 0, sizeof(revs));
 931         revs.initial = 1;
 932         for_each_reflog_ent(refname, collect_one_reflog_ent, &revs);
 933
 934         if (!revs.nr && !get_oid(refname, &oid))
 935                 add_one_commit(&oid, &revs);
 936
 937         for (i = 0; i < revs.nr; i++)
 938                 revs.commit[i]->object.flags &= ~TMP_MARK;
 939
 940         bases = get_merge_bases_many(commit, revs.nr, revs.commit);
 941
 942         /*
 943          * There should be one and only one merge base, when we found
 944          * a common ancestor among reflog entries.
 945          */
 946         if (!bases || bases->next)
 947                 goto cleanup_return;
 948
 949         /* And the found one must be one of the reflog entries */
 950         for (i = 0; i < revs.nr; i++)
 951                 if (&bases->item->object == &revs.commit[i]->object)
 952                         break; /* found */
 953         if (revs.nr <= i)
 954                 goto cleanup_return;
 955
 956         ret = bases->item;
 957
 958 cleanup_return:
 959         free_commit_list(bases);
 960         return ret;
 961 }
 962
 963 static const char gpg_sig_header[] = "gpgsig";
 964 static const int gpg_sig_header_len = sizeof(gpg_sig_header) - 1;
 965
 966 static int do_sign_commit(struct strbuf *buf, const char *keyid)
 967 {
 968         struct strbuf sig = STRBUF_INIT;
 969         int inspos, copypos;
 970         const char *eoh;
 971
 972         /* find the end of the header */
 973         eoh = strstr(buf->buf, "\n\n");
 974         if (!eoh)
 975                 inspos = buf->len;
 976         else
 977                 inspos = eoh - buf->buf + 1;
 978
 979         if (!keyid || !*keyid)
 980                 keyid = get_signing_key();
 981         if (sign_buffer(buf, &sig, keyid)) {
 982                 strbuf_release(&sig);
 983                 return -1;
 984         }
 985
 986         for (copypos = 0; sig.buf[copypos]; ) {
 987                 const char *bol = sig.buf + copypos;
 988                 const char *eol = strchrnul(bol, '\n');
 989                 int len = (eol - bol) + !!*eol;
 990
 991                 if (!copypos) {
 992                         strbuf_insert(buf, inspos, gpg_sig_header, gpg_sig_header_len);
 993                         inspos += gpg_sig_header_len;
 994                 }
 995                 strbuf_insert(buf, inspos++, " ", 1);
 996                 strbuf_insert(buf, inspos, bol, len);
 997                 inspos += len;
 998                 copypos += len;
 999         }
1000         strbuf_release(&sig);
1001         return 0;
1002 }
1003
1004 int parse_signed_commit(const struct commit *commit,
1005                         struct strbuf *payload, struct strbuf *signature)
1006 {
1007
1008         unsigned long size;
1009         const char *buffer = get_commit_buffer(commit, &size);
1010         int in_signature, saw_signature = -1;
1011         const char *line, *tail;
1012
1013         line = buffer;
1014         tail = buffer + size;
1015         in_signature = 0;
1016         saw_signature = 0;
1017         while (line < tail) {
1018                 const char *sig = NULL;
1019                 const char *next = memchr(line, '\n', tail - line);
1020
1021                 next = next ? next + 1 : tail;
1022                 if (in_signature && line[0] == ' ')
1023                         sig = line + 1;
1024                 else if (starts_with(line, gpg_sig_header) &&
1025                          line[gpg_sig_header_len] == ' ')
1026                         sig = line + gpg_sig_header_len + 1;
1027                 if (sig) {
1028                         strbuf_add(signature, sig, next - sig);
1029                         saw_signature = 1;
1030                         in_signature = 1;
1031                 } else {
1032                         if (*line == '\n')
1033                                 /* dump the whole remainder of the buffer */
1034                                 next = tail;
1035                         strbuf_add(payload, line, next - line);
1036                         in_signature = 0;
1037                 }
1038                 line = next;
1039         }
1040         unuse_commit_buffer(commit, buffer);
1041         return saw_signature;
1042 }
1043
1044 int remove_signature(struct strbuf *buf)
1045 {
1046         const char *line = buf->buf;
1047         const char *tail = buf->buf + buf->len;
1048         int in_signature = 0;
1049         const char *sig_start = NULL;
1050         const char *sig_end = NULL;
1051
1052         while (line < tail) {
1053                 const char *next = memchr(line, '\n', tail - line);
1054                 next = next ? next + 1 : tail;
1055
1056                 if (in_signature && line[0] == ' ')
1057                         sig_end = next;
1058                 else if (starts_with(line, gpg_sig_header) &&
1059                          line[gpg_sig_header_len] == ' ') {
1060                         sig_start = line;
1061                         sig_end = next;
1062                         in_signature = 1;
1063                 } else {
1064                         if (*line == '\n')
1065                                 /* dump the whole remainder of the buffer */
1066                                 next = tail;
1067                         in_signature = 0;
1068                 }
1069                 line = next;
1070         }
1071
1072         if (sig_start)
1073                 strbuf_remove(buf, sig_start - buf->buf, sig_end - sig_start);
1074
1075         return sig_start != NULL;
1076 }
1077
1078 static void handle_signed_tag(struct commit *parent, struct commit_extra_header ***tail)
1079 {
1080         struct merge_remote_desc *desc;
1081         struct commit_extra_header *mergetag;
1082         char *buf;
1083         unsigned long size, len;
1084         enum object_type type;
1085
1086         desc = merge_remote_util(parent);
1087         if (!desc || !desc->obj)
1088                 return;
1089         buf = read_object_file(&desc->obj->oid, &type, &size);
1090         if (!buf || type != OBJ_TAG)
1091                 goto free_return;
1092         len = parse_signature(buf, size);
1093         if (size == len)
1094                 goto free_return;
1095         /*
1096          * We could verify this signature and either omit the tag when
1097          * it does not validate, but the integrator may not have the
1098          * public key of the signer of the tag he is merging, while a
1099          * later auditor may have it while auditing, so let's not run
1100          * verify-signed-buffer here for now...
1101          *
1102          * if (verify_signed_buffer(buf, len, buf + len, size - len, ...))
1103          *      warn("warning: signed tag unverified.");
1104          */
1105         mergetag = xcalloc(1, sizeof(*mergetag));
1106         mergetag->key = xstrdup("mergetag");
1107         mergetag->value = buf;
1108         mergetag->len = size;
1109
1110         **tail = mergetag;
1111         *tail = &mergetag->next;
1112         return;
1113
1114 free_return:
1115         free(buf);
1116 }
1117
1118 int check_commit_signature(const struct commit *commit, struct signature_check *sigc)
1119 {
1120         struct strbuf payload = STRBUF_INIT;
1121         struct strbuf signature = STRBUF_INIT;
1122         int ret = 1;
1123
1124         sigc->result = 'N';
1125
1126         if (parse_signed_commit(commit, &payload, &signature) <= 0)
1127                 goto out;
1128         ret = check_signature(payload.buf, payload.len, signature.buf,
1129                 signature.len, sigc);
1130
1131  out:
1132         strbuf_release(&payload);
1133         strbuf_release(&signature);
1134
1135         return ret;
1136 }
1137
1138 void verify_merge_signature(struct commit *commit, int verbosity)
1139 {
1140         char hex[GIT_MAX_HEXSZ + 1];
1141         struct signature_check signature_check;
1142         memset(&signature_check, 0, sizeof(signature_check));
1143
1144         check_commit_signature(commit, &signature_check);
1145
1146         find_unique_abbrev_r(hex, &commit->object.oid, DEFAULT_ABBREV);
1147         switch (signature_check.result) {
1148         case 'G':
1149                 break;
1150         case 'U':
1151                 die(_("Commit %s has an untrusted GPG signature, "
1152                       "allegedly by %s."), hex, signature_check.signer);
1153         case 'B':
1154                 die(_("Commit %s has a bad GPG signature "
1155                       "allegedly by %s."), hex, signature_check.signer);
1156         default: /* 'N' */
1157                 die(_("Commit %s does not have a GPG signature."), hex);
1158         }
1159         if (verbosity >= 0 && signature_check.result == 'G')
1160                 printf(_("Commit %s has a good GPG signature by %s\n"),
1161                        hex, signature_check.signer);
1162
1163         signature_check_clear(&signature_check);
1164 }
1165
1166 void append_merge_tag_headers(struct commit_list *parents,
1167                               struct commit_extra_header ***tail)
1168 {
1169         while (parents) {
1170                 struct commit *parent = parents->item;
1171                 handle_signed_tag(parent, tail);
1172                 parents = parents->next;
1173         }
1174 }
1175
1176 static void add_extra_header(struct strbuf *buffer,
1177                              struct commit_extra_header *extra)
1178 {
1179         strbuf_addstr(buffer, extra->key);
1180         if (extra->len)
1181                 strbuf_add_lines(buffer, " ", extra->value, extra->len);
1182         else
1183                 strbuf_addch(buffer, '\n');
1184 }
1185
1186 struct commit_extra_header *read_commit_extra_headers(struct commit *commit,
1187                                                       const char **exclude)
1188 {
1189         struct commit_extra_header *extra = NULL;
1190         unsigned long size;
1191         const char *buffer = get_commit_buffer(commit, &size);
1192         extra = read_commit_extra_header_lines(buffer, size, exclude);
1193         unuse_commit_buffer(commit, buffer);
1194         return extra;
1195 }
1196
1197 int for_each_mergetag(each_mergetag_fn fn, struct commit *commit, void *data)
1198 {
1199         struct commit_extra_header *extra, *to_free;
1200         int res = 0;
1201
1202         to_free = read_commit_extra_headers(commit, NULL);
1203         for (extra = to_free; !res && extra; extra = extra->next) {
1204                 if (strcmp(extra->key, "mergetag"))
1205                         continue; /* not a merge tag */
1206                 res = fn(commit, extra, data);
1207         }
1208         free_commit_extra_headers(to_free);
1209         return res;
1210 }
1211
1212 static inline int standard_header_field(const char *field, size_t len)
1213 {
1214         return ((len == 4 && !memcmp(field, "tree", 4)) ||
1215                 (len == 6 && !memcmp(field, "parent", 6)) ||
1216                 (len == 6 && !memcmp(field, "author", 6)) ||
1217                 (len == 9 && !memcmp(field, "committer", 9)) ||
1218                 (len == 8 && !memcmp(field, "encoding", 8)));
1219 }
1220
1221 static int excluded_header_field(const char *field, size_t len, const char **exclude)
1222 {
1223         if (!exclude)
1224                 return 0;
1225
1226         while (*exclude) {
1227                 size_t xlen = strlen(*exclude);
1228                 if (len == xlen && !memcmp(field, *exclude, xlen))
1229                         return 1;
1230                 exclude++;
1231         }
1232         return 0;
1233 }
1234
1235 static struct commit_extra_header *read_commit_extra_header_lines(
1236         const char *buffer, size_t size,
1237         const char **exclude)
1238 {
1239         struct commit_extra_header *extra = NULL, **tail = &extra, *it = NULL;
1240         const char *line, *next, *eof, *eob;
1241         struct strbuf buf = STRBUF_INIT;
1242
1243         for (line = buffer, eob = line + size;
1244              line < eob && *line != '\n';
1245              line = next) {
1246                 next = memchr(line, '\n', eob - line);
1247                 next = next ? next + 1 : eob;
1248                 if (*line == ' ') {
1249                         /* continuation */
1250                         if (it)
1251                                 strbuf_add(&buf, line + 1, next - (line + 1));
1252                         continue;
1253                 }
1254                 if (it)
1255                         it->value = strbuf_detach(&buf, &it->len);
1256                 strbuf_reset(&buf);
1257                 it = NULL;
1258
1259                 eof = memchr(line, ' ', next - line);
1260                 if (!eof)
1261                         eof = next;
1262                 else if (standard_header_field(line, eof - line) ||
1263                          excluded_header_field(line, eof - line, exclude))
1264                         continue;
1265
1266                 it = xcalloc(1, sizeof(*it));
1267                 it->key = xmemdupz(line, eof-line);
1268                 *tail = it;
1269                 tail = &it->next;
1270                 if (eof + 1 < next)
1271                         strbuf_add(&buf, eof + 1, next - (eof + 1));
1272         }
1273         if (it)
1274                 it->value = strbuf_detach(&buf, &it->len);
1275         return extra;
1276 }
1277
1278 void free_commit_extra_headers(struct commit_extra_header *extra)
1279 {
1280         while (extra) {
1281                 struct commit_extra_header *next = extra->next;
1282                 free(extra->key);
1283                 free(extra->value);
1284                 free(extra);
1285                 extra = next;
1286         }
1287 }
1288
1289 int commit_tree(const char *msg, size_t msg_len, const struct object_id *tree,
1290                 struct commit_list *parents, struct object_id *ret,
1291                 const char *author, const char *sign_commit)
1292 {
1293         struct commit_extra_header *extra = NULL, **tail = &extra;
1294         int result;
1295
1296         append_merge_tag_headers(parents, &tail);
1297         result = commit_tree_extended(msg, msg_len, tree, parents, ret,
1298                                       author, sign_commit, extra);
1299         free_commit_extra_headers(extra);
1300         return result;
1301 }
1302
1303 static int find_invalid_utf8(const char *buf, int len)
1304 {
1305         int offset = 0;
1306         static const unsigned int max_codepoint[] = {
1307                 0x7f, 0x7ff, 0xffff, 0x10ffff
1308         };
1309
1310         while (len) {
1311                 unsigned char c = *buf++;
1312                 int bytes, bad_offset;
1313                 unsigned int codepoint;
1314                 unsigned int min_val, max_val;
1315
1316                 len--;
1317                 offset++;
1318
1319                 /* Simple US-ASCII? No worries. */
1320                 if (c < 0x80)
1321                         continue;
1322
1323                 bad_offset = offset-1;
1324
1325                 /*
1326                  * Count how many more high bits set: that's how
1327                  * many more bytes this sequence should have.
1328                  */
1329                 bytes = 0;
1330                 while (c & 0x40) {
1331                         c <<= 1;
1332                         bytes++;
1333                 }
1334
1335                 /*
1336                  * Must be between 1 and 3 more bytes.  Longer sequences result in
1337                  * codepoints beyond U+10FFFF, which are guaranteed never to exist.
1338                  */
1339                 if (bytes < 1 || 3 < bytes)
1340                         return bad_offset;
1341
1342                 /* Do we *have* that many bytes? */
1343                 if (len < bytes)
1344                         return bad_offset;
1345
1346                 /*
1347                  * Place the encoded bits at the bottom of the value and compute the
1348                  * valid range.
1349                  */
1350                 codepoint = (c & 0x7f) >> bytes;
1351                 min_val = max_codepoint[bytes-1] + 1;
1352                 max_val = max_codepoint[bytes];
1353
1354                 offset += bytes;
1355                 len -= bytes;
1356
1357                 /* And verify that they are good continuation bytes */
1358                 do {
1359                         codepoint <<= 6;
1360                         codepoint |= *buf & 0x3f;
1361                         if ((*buf++ & 0xc0) != 0x80)
1362                                 return bad_offset;
1363                 } while (--bytes);
1364
1365                 /* Reject codepoints that are out of range for the sequence length. */
1366                 if (codepoint < min_val || codepoint > max_val)
1367                         return bad_offset;
1368                 /* Surrogates are only for UTF-16 and cannot be encoded in UTF-8. */
1369                 if ((codepoint & 0x1ff800) == 0xd800)
1370                         return bad_offset;
1371                 /* U+xxFFFE and U+xxFFFF are guaranteed non-characters. */
1372                 if ((codepoint & 0xfffe) == 0xfffe)
1373                         return bad_offset;
1374                 /* So are anything in the range U+FDD0..U+FDEF. */
1375                 if (codepoint >= 0xfdd0 && codepoint <= 0xfdef)
1376                         return bad_offset;
1377         }
1378         return -1;
1379 }
1380
1381 /*
1382  * This verifies that the buffer is in proper utf8 format.
1383  *
1384  * If it isn't, it assumes any non-utf8 characters are Latin1,
1385  * and does the conversion.
1386  */
1387 static int verify_utf8(struct strbuf *buf)
1388 {
1389         int ok = 1;
1390         long pos = 0;
1391
1392         for (;;) {
1393                 int bad;
1394                 unsigned char c;
1395                 unsigned char replace[2];
1396
1397                 bad = find_invalid_utf8(buf->buf + pos, buf->len - pos);
1398                 if (bad < 0)
1399                         return ok;
1400                 pos += bad;
1401                 ok = 0;
1402                 c = buf->buf[pos];
1403                 strbuf_remove(buf, pos, 1);
1404
1405                 /* We know 'c' must be in the range 128-255 */
1406                 replace[0] = 0xc0 + (c >> 6);
1407                 replace[1] = 0x80 + (c & 0x3f);
1408                 strbuf_insert(buf, pos, replace, 2);
1409                 pos += 2;
1410         }
1411 }
1412
1413 static const char commit_utf8_warn[] =
1414 N_("Warning: commit message did not conform to UTF-8.\n"
1415    "You may want to amend it after fixing the message, or set the config\n"
1416    "variable i18n.commitencoding to the encoding your project uses.\n");
1417
1418 int commit_tree_extended(const char *msg, size_t msg_len,
1419                          const struct object_id *tree,
1420                          struct commit_list *parents, struct object_id *ret,
1421                          const char *author, const char *sign_commit,
1422                          struct commit_extra_header *extra)
1423 {
1424         int result;
1425         int encoding_is_utf8;
1426         struct strbuf buffer;
1427
1428         assert_oid_type(tree, OBJ_TREE);
1429
1430         if (memchr(msg, '\0', msg_len))
1431                 return error("a NUL byte in commit log message not allowed.");
1432
1433         /* Not having i18n.commitencoding is the same as having utf-8 */
1434         encoding_is_utf8 = is_encoding_utf8(git_commit_encoding);
1435
1436         strbuf_init(&buffer, 8192); /* should avoid reallocs for the headers */
1437         strbuf_addf(&buffer, "tree %s\n", oid_to_hex(tree));
1438
1439         /*
1440          * NOTE! This ordering means that the same exact tree merged with a
1441          * different order of parents will be a _different_ changeset even
1442          * if everything else stays the same.
1443          */
1444         while (parents) {
1445                 struct commit *parent = pop_commit(&parents);
1446                 strbuf_addf(&buffer, "parent %s\n",
1447                             oid_to_hex(&parent->object.oid));
1448         }
1449
1450         /* Person/date information */
1451         if (!author)
1452                 author = git_author_info(IDENT_STRICT);
1453         strbuf_addf(&buffer, "author %s\n", author);
1454         strbuf_addf(&buffer, "committer %s\n", git_committer_info(IDENT_STRICT));
1455         if (!encoding_is_utf8)
1456                 strbuf_addf(&buffer, "encoding %s\n", git_commit_encoding);
1457
1458         while (extra) {
1459                 add_extra_header(&buffer, extra);
1460                 extra = extra->next;
1461         }
1462         strbuf_addch(&buffer, '\n');
1463
1464         /* And add the comment */
1465         strbuf_add(&buffer, msg, msg_len);
1466
1467         /* And check the encoding */
1468         if (encoding_is_utf8 && !verify_utf8(&buffer))
1469                 fprintf(stderr, _(commit_utf8_warn));
1470
1471         if (sign_commit && do_sign_commit(&buffer, sign_commit)) {
1472                 result = -1;
1473                 goto out;
1474         }
1475
1476         result = write_object_file(buffer.buf, buffer.len, commit_type, ret);
1477 out:
1478         strbuf_release(&buffer);
1479         return result;
1480 }
1481
1482 define_commit_slab(merge_desc_slab, struct merge_remote_desc *);
1483 static struct merge_desc_slab merge_desc_slab = COMMIT_SLAB_INIT(1, merge_desc_slab);
1484
1485 struct merge_remote_desc *merge_remote_util(struct commit *commit)
1486 {
1487         return *merge_desc_slab_at(&merge_desc_slab, commit);
1488 }
1489
1490 void set_merge_remote_desc(struct commit *commit,
1491                            const char *name, struct object *obj)
1492 {
1493         struct merge_remote_desc *desc;
1494         FLEX_ALLOC_STR(desc, name, name);
1495         desc->obj = obj;
1496         *merge_desc_slab_at(&merge_desc_slab, commit) = desc;
1497 }
1498
1499 struct commit *get_merge_parent(const char *name)
1500 {
1501         struct object *obj;
1502         struct commit *commit;
1503         struct object_id oid;
1504         if (get_oid(name, &oid))
1505                 return NULL;
1506         obj = parse_object(the_repository, &oid);
1507         commit = (struct commit *)peel_to_type(name, 0, obj, OBJ_COMMIT);
1508         if (commit && !merge_remote_util(commit))
1509                 set_merge_remote_desc(commit, name, obj);
1510         return commit;
1511 }
1512
1513 /*
1514  * Append a commit to the end of the commit_list.
1515  *
1516  * next starts by pointing to the variable that holds the head of an
1517  * empty commit_list, and is updated to point to the "next" field of
1518  * the last item on the list as new commits are appended.
1519  *
1520  * Usage example:
1521  *
1522  *     struct commit_list *list;
1523  *     struct commit_list **next = &list;
1524  *
1525  *     next = commit_list_append(c1, next);
1526  *     next = commit_list_append(c2, next);
1527  *     assert(commit_list_count(list) == 2);
1528  *     return list;
1529  */
1530 struct commit_list **commit_list_append(struct commit *commit,
1531                                         struct commit_list **next)
1532 {
1533         struct commit_list *new_commit = xmalloc(sizeof(struct commit_list));
1534         new_commit->item = commit;
1535         *next = new_commit;
1536         new_commit->next = NULL;
1537         return &new_commit->next;
1538 }
1539
1540 const char *find_commit_header(const char *msg, const char *key, size_t *out_len)
1541 {
1542         int key_len = strlen(key);
1543         const char *line = msg;
1544
1545         while (line) {
1546                 const char *eol = strchrnul(line, '\n');
1547
1548                 if (line == eol)
1549                         return NULL;
1550
1551                 if (eol - line > key_len &&
1552                     !strncmp(line, key, key_len) &&
1553                     line[key_len] == ' ') {
1554                         *out_len = eol - line - key_len - 1;
1555                         return line + key_len + 1;
1556                 }
1557                 line = *eol ? eol + 1 : NULL;
1558         }
1559         return NULL;
1560 }
1561
1562 /*
1563  * Inspect the given string and determine the true "end" of the log message, in
1564  * order to find where to put a new Signed-off-by: line.  Ignored are
1565  * trailing comment lines and blank lines.  To support "git commit -s
1566  * --amend" on an existing commit, we also ignore "Conflicts:".  To
1567  * support "git commit -v", we truncate at cut lines.
1568  *
1569  * Returns the number of bytes from the tail to ignore, to be fed as
1570  * the second parameter to append_signoff().
1571  */
1572 size_t ignore_non_trailer(const char *buf, size_t len)
1573 {
1574         size_t boc = 0;
1575         size_t bol = 0;
1576         int in_old_conflicts_block = 0;
1577         size_t cutoff = wt_status_locate_end(buf, len);
1578
1579         while (bol < cutoff) {
1580                 const char *next_line = memchr(buf + bol, '\n', len - bol);
1581
1582                 if (!next_line)
1583                         next_line = buf + len;
1584                 else
1585                         next_line++;
1586
1587                 if (buf[bol] == comment_line_char || buf[bol] == '\n') {
1588                         /* is this the first of the run of comments? */
1589                         if (!boc)
1590                                 boc = bol;
1591                         /* otherwise, it is just continuing */
1592                 } else if (starts_with(buf + bol, "Conflicts:\n")) {
1593                         in_old_conflicts_block = 1;
1594                         if (!boc)
1595                                 boc = bol;
1596                 } else if (in_old_conflicts_block && buf[bol] == '\t') {
1597                         ; /* a pathname in the conflicts block */
1598                 } else if (boc) {
1599                         /* the previous was not trailing comment */
1600                         boc = 0;
1601                         in_old_conflicts_block = 0;
1602                 }
1603                 bol = next_line - buf;
1604         }
1605         return boc ? len - boc : len - cutoff;
1606 }