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 #include "run-command.h"
  23 #include "shallow.h"
  24
  25 static struct commit_extra_header *read_commit_extra_header_lines(const char *buf, size_t len, const char **);
  26
  27 int save_commit_buffer = 1;
  28
  29 const char *commit_type = "commit";
  30
  31 struct commit *lookup_commit_reference_gently(struct repository *r,
  32                 const struct object_id *oid, int quiet)
  33 {
  34         struct object *obj = deref_tag(r,
  35                                        parse_object(r, oid),
  36                                        NULL, 0);
  37
  38         if (!obj)
  39                 return NULL;
  40         return object_as_type(r, obj, OBJ_COMMIT, quiet);
  41 }
  42
  43 struct commit *lookup_commit_reference(struct repository *r, const struct object_id *oid)
  44 {
  45         return lookup_commit_reference_gently(r, oid, 0);
  46 }
  47
  48 struct commit *lookup_commit_or_die(const struct object_id *oid, const char *ref_name)
  49 {
  50         struct commit *c = lookup_commit_reference(the_repository, oid);
  51         if (!c)
  52                 die(_("could not parse %s"), ref_name);
  53         if (!oideq(oid, &c->object.oid)) {
  54                 warning(_("%s %s is not a commit!"),
  55                         ref_name, oid_to_hex(oid));
  56         }
  57         return c;
  58 }
  59
  60 struct commit *lookup_commit(struct repository *r, const struct object_id *oid)
  61 {
  62         struct object *obj = lookup_object(r, oid);
  63         if (!obj)
  64                 return create_object(r, oid, alloc_commit_node(r));
  65         return object_as_type(r, obj, OBJ_COMMIT, 0);
  66 }
  67
  68 struct commit *lookup_commit_reference_by_name(const char *name)
  69 {
  70         struct object_id oid;
  71         struct commit *commit;
  72
  73         if (get_oid_committish(name, &oid))
  74                 return NULL;
  75         commit = lookup_commit_reference(the_repository, &oid);
  76         if (parse_commit(commit))
  77                 return NULL;
  78         return commit;
  79 }
  80
  81 static timestamp_t parse_commit_date(const char *buf, const char *tail)
  82 {
  83         const char *dateptr;
  84
  85         if (buf + 6 >= tail)
  86                 return 0;
  87         if (memcmp(buf, "author", 6))
  88                 return 0;
  89         while (buf < tail && *buf++ != '\n')
  90                 /* nada */;
  91         if (buf + 9 >= tail)
  92                 return 0;
  93         if (memcmp(buf, "committer", 9))
  94                 return 0;
  95         while (buf < tail && *buf++ != '>')
  96                 /* nada */;
  97         if (buf >= tail)
  98                 return 0;
  99         dateptr = buf;
 100         while (buf < tail && *buf++ != '\n')
 101                 /* nada */;
 102         if (buf >= tail)
 103                 return 0;
 104         /* dateptr < buf && buf[-1] == '\n', so parsing will stop at buf-1 */
 105         return parse_timestamp(dateptr, NULL, 10);
 106 }
 107
 108 static const unsigned char *commit_graft_sha1_access(size_t index, void *table)
 109 {
 110         struct commit_graft **commit_graft_table = table;
 111         return commit_graft_table[index]->oid.hash;
 112 }
 113
 114 int commit_graft_pos(struct repository *r, const unsigned char *sha1)
 115 {
 116         return sha1_pos(sha1, r->parsed_objects->grafts,
 117                         r->parsed_objects->grafts_nr,
 118                         commit_graft_sha1_access);
 119 }
 120
 121 int register_commit_graft(struct repository *r, struct commit_graft *graft,
 122                           int ignore_dups)
 123 {
 124         int pos = commit_graft_pos(r, graft->oid.hash);
 125
 126         if (0 <= pos) {
 127                 if (ignore_dups)
 128                         free(graft);
 129                 else {
 130                         free(r->parsed_objects->grafts[pos]);
 131                         r->parsed_objects->grafts[pos] = graft;
 132                 }
 133                 return 1;
 134         }
 135         pos = -pos - 1;
 136         ALLOC_GROW(r->parsed_objects->grafts,
 137                    r->parsed_objects->grafts_nr + 1,
 138                    r->parsed_objects->grafts_alloc);
 139         r->parsed_objects->grafts_nr++;
 140         if (pos < r->parsed_objects->grafts_nr)
 141                 memmove(r->parsed_objects->grafts + pos + 1,
 142                         r->parsed_objects->grafts + pos,
 143                         (r->parsed_objects->grafts_nr - pos - 1) *
 144                         sizeof(*r->parsed_objects->grafts));
 145         r->parsed_objects->grafts[pos] = graft;
 146         return 0;
 147 }
 148
 149 struct commit_graft *read_graft_line(struct strbuf *line)
 150 {
 151         /* The format is just "Commit Parent1 Parent2 ...\n" */
 152         int i, phase;
 153         const char *tail = NULL;
 154         struct commit_graft *graft = NULL;
 155         struct object_id dummy_oid, *oid;
 156
 157         strbuf_rtrim(line);
 158         if (!line->len || line->buf[0] == '#')
 159                 return NULL;
 160         /*
 161          * phase 0 verifies line, counts hashes in line and allocates graft
 162          * phase 1 fills graft
 163          */
 164         for (phase = 0; phase < 2; phase++) {
 165                 oid = graft ? &graft->oid : &dummy_oid;
 166                 if (parse_oid_hex(line->buf, oid, &tail))
 167                         goto bad_graft_data;
 168                 for (i = 0; *tail != '\0'; i++) {
 169                         oid = graft ? &graft->parent[i] : &dummy_oid;
 170                         if (!isspace(*tail++) || parse_oid_hex(tail, oid, &tail))
 171                                 goto bad_graft_data;
 172                 }
 173                 if (!graft) {
 174                         graft = xmalloc(st_add(sizeof(*graft),
 175                                                st_mult(sizeof(struct object_id), i)));
 176                         graft->nr_parent = i;
 177                 }
 178         }
 179         return graft;
 180
 181 bad_graft_data:
 182         error("bad graft data: %s", line->buf);
 183         assert(!graft);
 184         return NULL;
 185 }
 186
 187 static int read_graft_file(struct repository *r, const char *graft_file)
 188 {
 189         FILE *fp = fopen_or_warn(graft_file, "r");
 190         struct strbuf buf = STRBUF_INIT;
 191         if (!fp)
 192                 return -1;
 193         if (advice_graft_file_deprecated)
 194                 advise(_("Support for <GIT_DIR>/info/grafts is deprecated\n"
 195                          "and will be removed in a future Git version.\n"
 196                          "\n"
 197                          "Please use \"git replace --convert-graft-file\"\n"
 198                          "to convert the grafts into replace refs.\n"
 199                          "\n"
 200                          "Turn this message off by running\n"
 201                          "\"git config advice.graftFileDeprecated false\""));
 202         while (!strbuf_getwholeline(&buf, fp, '\n')) {
 203                 /* The format is just "Commit Parent1 Parent2 ...\n" */
 204                 struct commit_graft *graft = read_graft_line(&buf);
 205                 if (!graft)
 206                         continue;
 207                 if (register_commit_graft(r, graft, 1))
 208                         error("duplicate graft data: %s", buf.buf);
 209         }
 210         fclose(fp);
 211         strbuf_release(&buf);
 212         return 0;
 213 }
 214
 215 void prepare_commit_graft(struct repository *r)
 216 {
 217         char *graft_file;
 218
 219         if (r->parsed_objects->commit_graft_prepared)
 220                 return;
 221         if (!startup_info->have_repository)
 222                 return;
 223
 224         graft_file = get_graft_file(r);
 225         read_graft_file(r, graft_file);
 226         /* make sure shallows are read */
 227         is_repository_shallow(r);
 228         r->parsed_objects->commit_graft_prepared = 1;
 229 }
 230
 231 struct commit_graft *lookup_commit_graft(struct repository *r, const struct object_id *oid)
 232 {
 233         int pos;
 234         prepare_commit_graft(r);
 235         pos = commit_graft_pos(r, oid->hash);
 236         if (pos < 0)
 237                 return NULL;
 238         return r->parsed_objects->grafts[pos];
 239 }
 240
 241 int for_each_commit_graft(each_commit_graft_fn fn, void *cb_data)
 242 {
 243         int i, ret;
 244         for (i = ret = 0; i < the_repository->parsed_objects->grafts_nr && !ret; i++)
 245                 ret = fn(the_repository->parsed_objects->grafts[i], cb_data);
 246         return ret;
 247 }
 248
 249 struct commit_buffer {
 250         void *buffer;
 251         unsigned long size;
 252 };
 253 define_commit_slab(buffer_slab, struct commit_buffer);
 254
 255 struct buffer_slab *allocate_commit_buffer_slab(void)
 256 {
 257         struct buffer_slab *bs = xmalloc(sizeof(*bs));
 258         init_buffer_slab(bs);
 259         return bs;
 260 }
 261
 262 void free_commit_buffer_slab(struct buffer_slab *bs)
 263 {
 264         clear_buffer_slab(bs);
 265         free(bs);
 266 }
 267
 268 void set_commit_buffer(struct repository *r, struct commit *commit, void *buffer, unsigned long size)
 269 {
 270         struct commit_buffer *v = buffer_slab_at(
 271                 r->parsed_objects->buffer_slab, commit);
 272         v->buffer = buffer;
 273         v->size = size;
 274 }
 275
 276 const void *get_cached_commit_buffer(struct repository *r, const struct commit *commit, unsigned long *sizep)
 277 {
 278         struct commit_buffer *v = buffer_slab_peek(
 279                 r->parsed_objects->buffer_slab, commit);
 280         if (!v) {
 281                 if (sizep)
 282                         *sizep = 0;
 283                 return NULL;
 284         }
 285         if (sizep)
 286                 *sizep = v->size;
 287         return v->buffer;
 288 }
 289
 290 const void *repo_get_commit_buffer(struct repository *r,
 291                                    const struct commit *commit,
 292                                    unsigned long *sizep)
 293 {
 294         const void *ret = get_cached_commit_buffer(r, commit, sizep);
 295         if (!ret) {
 296                 enum object_type type;
 297                 unsigned long size;
 298                 ret = repo_read_object_file(r, &commit->object.oid, &type, &size);
 299                 if (!ret)
 300                         die("cannot read commit object %s",
 301                             oid_to_hex(&commit->object.oid));
 302                 if (type != OBJ_COMMIT)
 303                         die("expected commit for %s, got %s",
 304                             oid_to_hex(&commit->object.oid), type_name(type));
 305                 if (sizep)
 306                         *sizep = size;
 307         }
 308         return ret;
 309 }
 310
 311 void repo_unuse_commit_buffer(struct repository *r,
 312                               const struct commit *commit,
 313                               const void *buffer)
 314 {
 315         struct commit_buffer *v = buffer_slab_peek(
 316                 r->parsed_objects->buffer_slab, commit);
 317         if (!(v && v->buffer == buffer))
 318                 free((void *)buffer);
 319 }
 320
 321 void free_commit_buffer(struct parsed_object_pool *pool, struct commit *commit)
 322 {
 323         struct commit_buffer *v = buffer_slab_peek(
 324                 pool->buffer_slab, commit);
 325         if (v) {
 326                 FREE_AND_NULL(v->buffer);
 327                 v->size = 0;
 328         }
 329 }
 330
 331 static inline void set_commit_tree(struct commit *c, struct tree *t)
 332 {
 333         c->maybe_tree = t;
 334 }
 335
 336 struct tree *repo_get_commit_tree(struct repository *r,
 337                                   const struct commit *commit)
 338 {
 339         if (commit->maybe_tree || !commit->object.parsed)
 340                 return commit->maybe_tree;
 341
 342         if (commit->graph_pos != COMMIT_NOT_FROM_GRAPH)
 343                 return get_commit_tree_in_graph(r, commit);
 344
 345         return NULL;
 346 }
 347
 348 struct object_id *get_commit_tree_oid(const struct commit *commit)
 349 {
 350         struct tree *tree = get_commit_tree(commit);
 351         return tree ? &tree->object.oid : NULL;
 352 }
 353
 354 void release_commit_memory(struct parsed_object_pool *pool, struct commit *c)
 355 {
 356         set_commit_tree(c, NULL);
 357         free_commit_buffer(pool, c);
 358         c->index = 0;
 359         free_commit_list(c->parents);
 360
 361         c->object.parsed = 0;
 362 }
 363
 364 const void *detach_commit_buffer(struct commit *commit, unsigned long *sizep)
 365 {
 366         struct commit_buffer *v = buffer_slab_peek(
 367                 the_repository->parsed_objects->buffer_slab, commit);
 368         void *ret;
 369
 370         if (!v) {
 371                 if (sizep)
 372                         *sizep = 0;
 373                 return NULL;
 374         }
 375         ret = v->buffer;
 376         if (sizep)
 377                 *sizep = v->size;
 378
 379         v->buffer = NULL;
 380         v->size = 0;
 381         return ret;
 382 }
 383
 384 int parse_commit_buffer(struct repository *r, struct commit *item, const void *buffer, unsigned long size, int check_graph)
 385 {
 386         const char *tail = buffer;
 387         const char *bufptr = buffer;
 388         struct object_id parent;
 389         struct commit_list **pptr;
 390         struct commit_graft *graft;
 391         const int tree_entry_len = the_hash_algo->hexsz + 5;
 392         const int parent_entry_len = the_hash_algo->hexsz + 7;
 393         struct tree *tree;
 394
 395         if (item->object.parsed)
 396                 return 0;
 397
 398         if (item->parents) {
 399                 /*
 400                  * Presumably this is leftover from an earlier failed parse;
 401                  * clear it out in preparation for us re-parsing (we'll hit the
 402                  * same error, but that's good, since it lets our caller know
 403                  * the result cannot be trusted.
 404                  */
 405                 free_commit_list(item->parents);
 406                 item->parents = NULL;
 407         }
 408
 409         tail += size;
 410         if (tail <= bufptr + tree_entry_len + 1 || memcmp(bufptr, "tree ", 5) ||
 411                         bufptr[tree_entry_len] != '\n')
 412                 return error("bogus commit object %s", oid_to_hex(&item->object.oid));
 413         if (get_oid_hex(bufptr + 5, &parent) < 0)
 414                 return error("bad tree pointer in commit %s",
 415                              oid_to_hex(&item->object.oid));
 416         tree = lookup_tree(r, &parent);
 417         if (!tree)
 418                 return error("bad tree pointer %s in commit %s",
 419                              oid_to_hex(&parent),
 420                              oid_to_hex(&item->object.oid));
 421         set_commit_tree(item, tree);
 422         bufptr += tree_entry_len + 1; /* "tree " + "hex sha1" + "\n" */
 423         pptr = &item->parents;
 424
 425         graft = lookup_commit_graft(r, &item->object.oid);
 426         while (bufptr + parent_entry_len < tail && !memcmp(bufptr, "parent ", 7)) {
 427                 struct commit *new_parent;
 428
 429                 if (tail <= bufptr + parent_entry_len + 1 ||
 430                     get_oid_hex(bufptr + 7, &parent) ||
 431                     bufptr[parent_entry_len] != '\n')
 432                         return error("bad parents in commit %s", oid_to_hex(&item->object.oid));
 433                 bufptr += parent_entry_len + 1;
 434                 /*
 435                  * The clone is shallow if nr_parent < 0, and we must
 436                  * not traverse its real parents even when we unhide them.
 437                  */
 438                 if (graft && (graft->nr_parent < 0 || grafts_replace_parents))
 439                         continue;
 440                 new_parent = lookup_commit(r, &parent);
 441                 if (!new_parent)
 442                         return error("bad parent %s in commit %s",
 443                                      oid_to_hex(&parent),
 444                                      oid_to_hex(&item->object.oid));
 445                 pptr = &commit_list_insert(new_parent, pptr)->next;
 446         }
 447         if (graft) {
 448                 int i;
 449                 struct commit *new_parent;
 450                 for (i = 0; i < graft->nr_parent; i++) {
 451                         new_parent = lookup_commit(r,
 452                                                    &graft->parent[i]);
 453                         if (!new_parent)
 454                                 return error("bad graft parent %s in commit %s",
 455                                              oid_to_hex(&graft->parent[i]),
 456                                              oid_to_hex(&item->object.oid));
 457                         pptr = &commit_list_insert(new_parent, pptr)->next;
 458                 }
 459         }
 460         item->date = parse_commit_date(bufptr, tail);
 461
 462         if (check_graph)
 463                 load_commit_graph_info(r, item);
 464
 465         item->object.parsed = 1;
 466         return 0;
 467 }
 468
 469 int repo_parse_commit_internal(struct repository *r,
 470                                struct commit *item,
 471                                int quiet_on_missing,
 472                                int use_commit_graph)
 473 {
 474         enum object_type type;
 475         void *buffer;
 476         unsigned long size;
 477         int ret;
 478
 479         if (!item)
 480                 return -1;
 481         if (item->object.parsed)
 482                 return 0;
 483         if (use_commit_graph && parse_commit_in_graph(r, item))
 484                 return 0;
 485         buffer = repo_read_object_file(r, &item->object.oid, &type, &size);
 486         if (!buffer)
 487                 return quiet_on_missing ? -1 :
 488                         error("Could not read %s",
 489                              oid_to_hex(&item->object.oid));
 490         if (type != OBJ_COMMIT) {
 491                 free(buffer);
 492                 return error("Object %s not a commit",
 493                              oid_to_hex(&item->object.oid));
 494         }
 495
 496         ret = parse_commit_buffer(r, item, buffer, size, 0);
 497         if (save_commit_buffer && !ret) {
 498                 set_commit_buffer(r, item, buffer, size);
 499                 return 0;
 500         }
 501         free(buffer);
 502         return ret;
 503 }
 504
 505 int repo_parse_commit_gently(struct repository *r,
 506                              struct commit *item, int quiet_on_missing)
 507 {
 508         return repo_parse_commit_internal(r, item, quiet_on_missing, 1);
 509 }
 510
 511 void parse_commit_or_die(struct commit *item)
 512 {
 513         if (parse_commit(item))
 514                 die("unable to parse commit %s",
 515                     item ? oid_to_hex(&item->object.oid) : "(null)");
 516 }
 517
 518 int find_commit_subject(const char *commit_buffer, const char **subject)
 519 {
 520         const char *eol;
 521         const char *p = commit_buffer;
 522
 523         while (*p && (*p != '\n' || p[1] != '\n'))
 524                 p++;
 525         if (*p) {
 526                 p = skip_blank_lines(p + 2);
 527                 eol = strchrnul(p, '\n');
 528         } else
 529                 eol = p;
 530
 531         *subject = p;
 532
 533         return eol - p;
 534 }
 535
 536 struct commit_list *commit_list_insert(struct commit *item, struct commit_list **list_p)
 537 {
 538         struct commit_list *new_list = xmalloc(sizeof(struct commit_list));
 539         new_list->item = item;
 540         new_list->next = *list_p;
 541         *list_p = new_list;
 542         return new_list;
 543 }
 544
 545 unsigned commit_list_count(const struct commit_list *l)
 546 {
 547         unsigned c = 0;
 548         for (; l; l = l->next )
 549                 c++;
 550         return c;
 551 }
 552
 553 struct commit_list *copy_commit_list(struct commit_list *list)
 554 {
 555         struct commit_list *head = NULL;
 556         struct commit_list **pp = &head;
 557         while (list) {
 558                 pp = commit_list_append(list->item, pp);
 559                 list = list->next;
 560         }
 561         return head;
 562 }
 563
 564 void free_commit_list(struct commit_list *list)
 565 {
 566         while (list)
 567                 pop_commit(&list);
 568 }
 569
 570 struct commit_list * commit_list_insert_by_date(struct commit *item, struct commit_list **list)
 571 {
 572         struct commit_list **pp = list;
 573         struct commit_list *p;
 574         while ((p = *pp) != NULL) {
 575                 if (p->item->date < item->date) {
 576                         break;
 577                 }
 578                 pp = &p->next;
 579         }
 580         return commit_list_insert(item, pp);
 581 }
 582
 583 static int commit_list_compare_by_date(const void *a, const void *b)
 584 {
 585         timestamp_t a_date = ((const struct commit_list *)a)->item->date;
 586         timestamp_t b_date = ((const struct commit_list *)b)->item->date;
 587         if (a_date < b_date)
 588                 return 1;
 589         if (a_date > b_date)
 590                 return -1;
 591         return 0;
 592 }
 593
 594 static void *commit_list_get_next(const void *a)
 595 {
 596         return ((const struct commit_list *)a)->next;
 597 }
 598
 599 static void commit_list_set_next(void *a, void *next)
 600 {
 601         ((struct commit_list *)a)->next = next;
 602 }
 603
 604 void commit_list_sort_by_date(struct commit_list **list)
 605 {
 606         *list = llist_mergesort(*list, commit_list_get_next, commit_list_set_next,
 607                                 commit_list_compare_by_date);
 608 }
 609
 610 struct commit *pop_most_recent_commit(struct commit_list **list,
 611                                       unsigned int mark)
 612 {
 613         struct commit *ret = pop_commit(list);
 614         struct commit_list *parents = ret->parents;
 615
 616         while (parents) {
 617                 struct commit *commit = parents->item;
 618                 if (!parse_commit(commit) && !(commit->object.flags & mark)) {
 619                         commit->object.flags |= mark;
 620                         commit_list_insert_by_date(commit, list);
 621                 }
 622                 parents = parents->next;
 623         }
 624         return ret;
 625 }
 626
 627 static void clear_commit_marks_1(struct commit_list **plist,
 628                                  struct commit *commit, unsigned int mark)
 629 {
 630         while (commit) {
 631                 struct commit_list *parents;
 632
 633                 if (!(mark & commit->object.flags))
 634                         return;
 635
 636                 commit->object.flags &= ~mark;
 637
 638                 parents = commit->parents;
 639                 if (!parents)
 640                         return;
 641
 642                 while ((parents = parents->next))
 643                         commit_list_insert(parents->item, plist);
 644
 645                 commit = commit->parents->item;
 646         }
 647 }
 648
 649 void clear_commit_marks_many(int nr, struct commit **commit, unsigned int mark)
 650 {
 651         struct commit_list *list = NULL;
 652
 653         while (nr--) {
 654                 clear_commit_marks_1(&list, *commit, mark);
 655                 commit++;
 656         }
 657         while (list)
 658                 clear_commit_marks_1(&list, pop_commit(&list), mark);
 659 }
 660
 661 void clear_commit_marks(struct commit *commit, unsigned int mark)
 662 {
 663         clear_commit_marks_many(1, &commit, mark);
 664 }
 665
 666 struct commit *pop_commit(struct commit_list **stack)
 667 {
 668         struct commit_list *top = *stack;
 669         struct commit *item = top ? top->item : NULL;
 670
 671         if (top) {
 672                 *stack = top->next;
 673                 free(top);
 674         }
 675         return item;
 676 }
 677
 678 /*
 679  * Topological sort support
 680  */
 681
 682 /* count number of children that have not been emitted */
 683 define_commit_slab(indegree_slab, int);
 684
 685 define_commit_slab(author_date_slab, timestamp_t);
 686
 687 void record_author_date(struct author_date_slab *author_date,
 688                         struct commit *commit)
 689 {
 690         const char *buffer = get_commit_buffer(commit, NULL);
 691         struct ident_split ident;
 692         const char *ident_line;
 693         size_t ident_len;
 694         char *date_end;
 695         timestamp_t date;
 696
 697         ident_line = find_commit_header(buffer, "author", &ident_len);
 698         if (!ident_line)
 699                 goto fail_exit; /* no author line */
 700         if (split_ident_line(&ident, ident_line, ident_len) ||
 701             !ident.date_begin || !ident.date_end)
 702                 goto fail_exit; /* malformed "author" line */
 703
 704         date = parse_timestamp(ident.date_begin, &date_end, 10);
 705         if (date_end != ident.date_end)
 706                 goto fail_exit; /* malformed date */
 707         *(author_date_slab_at(author_date, commit)) = date;
 708
 709 fail_exit:
 710         unuse_commit_buffer(commit, buffer);
 711 }
 712
 713 int compare_commits_by_author_date(const void *a_, const void *b_,
 714                                    void *cb_data)
 715 {
 716         const struct commit *a = a_, *b = b_;
 717         struct author_date_slab *author_date = cb_data;
 718         timestamp_t a_date = *(author_date_slab_at(author_date, a));
 719         timestamp_t b_date = *(author_date_slab_at(author_date, b));
 720
 721         /* newer commits with larger date first */
 722         if (a_date < b_date)
 723                 return 1;
 724         else if (a_date > b_date)
 725                 return -1;
 726         return 0;
 727 }
 728
 729 int compare_commits_by_gen_then_commit_date(const void *a_, const void *b_, void *unused)
 730 {
 731         const struct commit *a = a_, *b = b_;
 732
 733         /* newer commits first */
 734         if (a->generation < b->generation)
 735                 return 1;
 736         else if (a->generation > b->generation)
 737                 return -1;
 738
 739         /* use date as a heuristic when generations are equal */
 740         if (a->date < b->date)
 741                 return 1;
 742         else if (a->date > b->date)
 743                 return -1;
 744         return 0;
 745 }
 746
 747 int compare_commits_by_commit_date(const void *a_, const void *b_, void *unused)
 748 {
 749         const struct commit *a = a_, *b = b_;
 750         /* newer commits with larger date first */
 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 /*
 759  * Performs an in-place topological sort on the list supplied.
 760  */
 761 void sort_in_topological_order(struct commit_list **list, enum rev_sort_order sort_order)
 762 {
 763         struct commit_list *next, *orig = *list;
 764         struct commit_list **pptr;
 765         struct indegree_slab indegree;
 766         struct prio_queue queue;
 767         struct commit *commit;
 768         struct author_date_slab author_date;
 769
 770         if (!orig)
 771                 return;
 772         *list = NULL;
 773
 774         init_indegree_slab(&indegree);
 775         memset(&queue, '\0', sizeof(queue));
 776
 777         switch (sort_order) {
 778         default: /* REV_SORT_IN_GRAPH_ORDER */
 779                 queue.compare = NULL;
 780                 break;
 781         case REV_SORT_BY_COMMIT_DATE:
 782                 queue.compare = compare_commits_by_commit_date;
 783                 break;
 784         case REV_SORT_BY_AUTHOR_DATE:
 785                 init_author_date_slab(&author_date);
 786                 queue.compare = compare_commits_by_author_date;
 787                 queue.cb_data = &author_date;
 788                 break;
 789         }
 790
 791         /* Mark them and clear the indegree */
 792         for (next = orig; next; next = next->next) {
 793                 struct commit *commit = next->item;
 794                 *(indegree_slab_at(&indegree, commit)) = 1;
 795                 /* also record the author dates, if needed */
 796                 if (sort_order == REV_SORT_BY_AUTHOR_DATE)
 797                         record_author_date(&author_date, commit);
 798         }
 799
 800         /* update the indegree */
 801         for (next = orig; next; next = next->next) {
 802                 struct commit_list *parents = next->item->parents;
 803                 while (parents) {
 804                         struct commit *parent = parents->item;
 805                         int *pi = indegree_slab_at(&indegree, parent);
 806
 807                         if (*pi)
 808                                 (*pi)++;
 809                         parents = parents->next;
 810                 }
 811         }
 812
 813         /*
 814          * find the tips
 815          *
 816          * tips are nodes not reachable from any other node in the list
 817          *
 818          * the tips serve as a starting set for the work queue.
 819          */
 820         for (next = orig; next; next = next->next) {
 821                 struct commit *commit = next->item;
 822
 823                 if (*(indegree_slab_at(&indegree, commit)) == 1)
 824                         prio_queue_put(&queue, commit);
 825         }
 826
 827         /*
 828          * This is unfortunate; the initial tips need to be shown
 829          * in the order given from the revision traversal machinery.
 830          */
 831         if (sort_order == REV_SORT_IN_GRAPH_ORDER)
 832                 prio_queue_reverse(&queue);
 833
 834         /* We no longer need the commit list */
 835         free_commit_list(orig);
 836
 837         pptr = list;
 838         *list = NULL;
 839         while ((commit = prio_queue_get(&queue)) != NULL) {
 840                 struct commit_list *parents;
 841
 842                 for (parents = commit->parents; parents ; parents = parents->next) {
 843                         struct commit *parent = parents->item;
 844                         int *pi = indegree_slab_at(&indegree, parent);
 845
 846                         if (!*pi)
 847                                 continue;
 848
 849                         /*
 850                          * parents are only enqueued for emission
 851                          * when all their children have been emitted thereby
 852                          * guaranteeing topological order.
 853                          */
 854                         if (--(*pi) == 1)
 855                                 prio_queue_put(&queue, parent);
 856                 }
 857                 /*
 858                  * all children of commit have already been
 859                  * emitted. we can emit it now.
 860                  */
 861                 *(indegree_slab_at(&indegree, commit)) = 0;
 862
 863                 pptr = &commit_list_insert(commit, pptr)->next;
 864         }
 865
 866         clear_indegree_slab(&indegree);
 867         clear_prio_queue(&queue);
 868         if (sort_order == REV_SORT_BY_AUTHOR_DATE)
 869                 clear_author_date_slab(&author_date);
 870 }
 871
 872 struct rev_collect {
 873         struct commit **commit;
 874         int nr;
 875         int alloc;
 876         unsigned int initial : 1;
 877 };
 878
 879 static void add_one_commit(struct object_id *oid, struct rev_collect *revs)
 880 {
 881         struct commit *commit;
 882
 883         if (is_null_oid(oid))
 884                 return;
 885
 886         commit = lookup_commit(the_repository, oid);
 887         if (!commit ||
 888             (commit->object.flags & TMP_MARK) ||
 889             parse_commit(commit))
 890                 return;
 891
 892         ALLOC_GROW(revs->commit, revs->nr + 1, revs->alloc);
 893         revs->commit[revs->nr++] = commit;
 894         commit->object.flags |= TMP_MARK;
 895 }
 896
 897 static int collect_one_reflog_ent(struct object_id *ooid, struct object_id *noid,
 898                                   const char *ident, timestamp_t timestamp,
 899                                   int tz, const char *message, void *cbdata)
 900 {
 901         struct rev_collect *revs = cbdata;
 902
 903         if (revs->initial) {
 904                 revs->initial = 0;
 905                 add_one_commit(ooid, revs);
 906         }
 907         add_one_commit(noid, revs);
 908         return 0;
 909 }
 910
 911 struct commit *get_fork_point(const char *refname, struct commit *commit)
 912 {
 913         struct object_id oid;
 914         struct rev_collect revs;
 915         struct commit_list *bases;
 916         int i;
 917         struct commit *ret = NULL;
 918         char *full_refname;
 919
 920         switch (dwim_ref(refname, strlen(refname), &oid, &full_refname)) {
 921         case 0:
 922                 die("No such ref: '%s'", refname);
 923         case 1:
 924                 break; /* good */
 925         default:
 926                 die("Ambiguous refname: '%s'", refname);
 927         }
 928
 929         memset(&revs, 0, sizeof(revs));
 930         revs.initial = 1;
 931         for_each_reflog_ent(full_refname, collect_one_reflog_ent, &revs);
 932
 933         if (!revs.nr)
 934                 add_one_commit(&oid, &revs);
 935
 936         for (i = 0; i < revs.nr; i++)
 937                 revs.commit[i]->object.flags &= ~TMP_MARK;
 938
 939         bases = get_merge_bases_many(commit, revs.nr, revs.commit);
 940
 941         /*
 942          * There should be one and only one merge base, when we found
 943          * a common ancestor among reflog entries.
 944          */
 945         if (!bases || bases->next)
 946                 goto cleanup_return;
 947
 948         /* And the found one must be one of the reflog entries */
 949         for (i = 0; i < revs.nr; i++)
 950                 if (&bases->item->object == &revs.commit[i]->object)
 951                         break; /* found */
 952         if (revs.nr <= i)
 953                 goto cleanup_return;
 954
 955         ret = bases->item;
 956
 957 cleanup_return:
 958         free_commit_list(bases);
 959         free(full_refname);
 960         return ret;
 961 }
 962
 963 /*
 964  * Indexed by hash algorithm identifier.
 965  */
 966 static const char *gpg_sig_headers[] = {
 967         NULL,
 968         "gpgsig",
 969         "gpgsig-sha256",
 970 };
 971
 972 static int do_sign_commit(struct strbuf *buf, const char *keyid)
 973 {
 974         struct strbuf sig = STRBUF_INIT;
 975         int inspos, copypos;
 976         const char *eoh;
 977         const char *gpg_sig_header = gpg_sig_headers[hash_algo_by_ptr(the_hash_algo)];
 978         int gpg_sig_header_len = strlen(gpg_sig_header);
 979
 980         /* find the end of the header */
 981         eoh = strstr(buf->buf, "\n\n");
 982         if (!eoh)
 983                 inspos = buf->len;
 984         else
 985                 inspos = eoh - buf->buf + 1;
 986
 987         if (!keyid || !*keyid)
 988                 keyid = get_signing_key();
 989         if (sign_buffer(buf, &sig, keyid)) {
 990                 strbuf_release(&sig);
 991                 return -1;
 992         }
 993
 994         for (copypos = 0; sig.buf[copypos]; ) {
 995                 const char *bol = sig.buf + copypos;
 996                 const char *eol = strchrnul(bol, '\n');
 997                 int len = (eol - bol) + !!*eol;
 998
 999                 if (!copypos) {
1000                         strbuf_insert(buf, inspos, gpg_sig_header, gpg_sig_header_len);
1001                         inspos += gpg_sig_header_len;
1002                 }
1003                 strbuf_insertstr(buf, inspos++, " ");
1004                 strbuf_insert(buf, inspos, bol, len);
1005                 inspos += len;
1006                 copypos += len;
1007         }
1008         strbuf_release(&sig);
1009         return 0;
1010 }
1011
1012 int parse_signed_commit(const struct commit *commit,
1013                         struct strbuf *payload, struct strbuf *signature)
1014 {
1015
1016         unsigned long size;
1017         const char *buffer = get_commit_buffer(commit, &size);
1018         int in_signature, saw_signature = -1;
1019         const char *line, *tail;
1020         const char *gpg_sig_header = gpg_sig_headers[hash_algo_by_ptr(the_hash_algo)];
1021         int gpg_sig_header_len = strlen(gpg_sig_header);
1022
1023         line = buffer;
1024         tail = buffer + size;
1025         in_signature = 0;
1026         saw_signature = 0;
1027         while (line < tail) {
1028                 const char *sig = NULL;
1029                 const char *next = memchr(line, '\n', tail - line);
1030
1031                 next = next ? next + 1 : tail;
1032                 if (in_signature && line[0] == ' ')
1033                         sig = line + 1;
1034                 else if (starts_with(line, gpg_sig_header) &&
1035                          line[gpg_sig_header_len] == ' ')
1036                         sig = line + gpg_sig_header_len + 1;
1037                 if (sig) {
1038                         strbuf_add(signature, sig, next - sig);
1039                         saw_signature = 1;
1040                         in_signature = 1;
1041                 } else {
1042                         if (*line == '\n')
1043                                 /* dump the whole remainder of the buffer */
1044                                 next = tail;
1045                         strbuf_add(payload, line, next - line);
1046                         in_signature = 0;
1047                 }
1048                 line = next;
1049         }
1050         unuse_commit_buffer(commit, buffer);
1051         return saw_signature;
1052 }
1053
1054 int remove_signature(struct strbuf *buf)
1055 {
1056         const char *line = buf->buf;
1057         const char *tail = buf->buf + buf->len;
1058         int in_signature = 0;
1059         const char *sig_start = NULL;
1060         const char *sig_end = NULL;
1061
1062         while (line < tail) {
1063                 const char *next = memchr(line, '\n', tail - line);
1064                 next = next ? next + 1 : tail;
1065
1066                 if (in_signature && line[0] == ' ')
1067                         sig_end = next;
1068                 else if (starts_with(line, "gpgsig")) {
1069                         int i;
1070                         for (i = 1; i < GIT_HASH_NALGOS; i++) {
1071                                 const char *p;
1072                                 if (skip_prefix(line, gpg_sig_headers[i], &p) &&
1073                                     *p == ' ') {
1074                                         sig_start = line;
1075                                         sig_end = next;
1076                                         in_signature = 1;
1077                                 }
1078                         }
1079                 } else {
1080                         if (*line == '\n')
1081                                 /* dump the whole remainder of the buffer */
1082                                 next = tail;
1083                         in_signature = 0;
1084                 }
1085                 line = next;
1086         }
1087
1088         if (sig_start)
1089                 strbuf_remove(buf, sig_start - buf->buf, sig_end - sig_start);
1090
1091         return sig_start != NULL;
1092 }
1093
1094 static void handle_signed_tag(struct commit *parent, struct commit_extra_header ***tail)
1095 {
1096         struct merge_remote_desc *desc;
1097         struct commit_extra_header *mergetag;
1098         char *buf;
1099         unsigned long size, len;
1100         enum object_type type;
1101
1102         desc = merge_remote_util(parent);
1103         if (!desc || !desc->obj)
1104                 return;
1105         buf = read_object_file(&desc->obj->oid, &type, &size);
1106         if (!buf || type != OBJ_TAG)
1107                 goto free_return;
1108         len = parse_signature(buf, size);
1109         if (size == len)
1110                 goto free_return;
1111         /*
1112          * We could verify this signature and either omit the tag when
1113          * it does not validate, but the integrator may not have the
1114          * public key of the signer of the tag he is merging, while a
1115          * later auditor may have it while auditing, so let's not run
1116          * verify-signed-buffer here for now...
1117          *
1118          * if (verify_signed_buffer(buf, len, buf + len, size - len, ...))
1119          *      warn("warning: signed tag unverified.");
1120          */
1121         mergetag = xcalloc(1, sizeof(*mergetag));
1122         mergetag->key = xstrdup("mergetag");
1123         mergetag->value = buf;
1124         mergetag->len = size;
1125
1126         **tail = mergetag;
1127         *tail = &mergetag->next;
1128         return;
1129
1130 free_return:
1131         free(buf);
1132 }
1133
1134 int check_commit_signature(const struct commit *commit, struct signature_check *sigc)
1135 {
1136         struct strbuf payload = STRBUF_INIT;
1137         struct strbuf signature = STRBUF_INIT;
1138         int ret = 1;
1139
1140         sigc->result = 'N';
1141
1142         if (parse_signed_commit(commit, &payload, &signature) <= 0)
1143                 goto out;
1144         ret = check_signature(payload.buf, payload.len, signature.buf,
1145                 signature.len, sigc);
1146
1147  out:
1148         strbuf_release(&payload);
1149         strbuf_release(&signature);
1150
1151         return ret;
1152 }
1153
1154 void verify_merge_signature(struct commit *commit, int verbosity,
1155                             int check_trust)
1156 {
1157         char hex[GIT_MAX_HEXSZ + 1];
1158         struct signature_check signature_check;
1159         int ret;
1160         memset(&signature_check, 0, sizeof(signature_check));
1161
1162         ret = check_commit_signature(commit, &signature_check);
1163
1164         find_unique_abbrev_r(hex, &commit->object.oid, DEFAULT_ABBREV);
1165         switch (signature_check.result) {
1166         case 'G':
1167                 if (ret || (check_trust && signature_check.trust_level < TRUST_MARGINAL))
1168                         die(_("Commit %s has an untrusted GPG signature, "
1169                               "allegedly by %s."), hex, signature_check.signer);
1170                 break;
1171         case 'B':
1172                 die(_("Commit %s has a bad GPG signature "
1173                       "allegedly by %s."), hex, signature_check.signer);
1174         default: /* 'N' */
1175                 die(_("Commit %s does not have a GPG signature."), hex);
1176         }
1177         if (verbosity >= 0 && signature_check.result == 'G')
1178                 printf(_("Commit %s has a good GPG signature by %s\n"),
1179                        hex, signature_check.signer);
1180
1181         signature_check_clear(&signature_check);
1182 }
1183
1184 void append_merge_tag_headers(struct commit_list *parents,
1185                               struct commit_extra_header ***tail)
1186 {
1187         while (parents) {
1188                 struct commit *parent = parents->item;
1189                 handle_signed_tag(parent, tail);
1190                 parents = parents->next;
1191         }
1192 }
1193
1194 static void add_extra_header(struct strbuf *buffer,
1195                              struct commit_extra_header *extra)
1196 {
1197         strbuf_addstr(buffer, extra->key);
1198         if (extra->len)
1199                 strbuf_add_lines(buffer, " ", extra->value, extra->len);
1200         else
1201                 strbuf_addch(buffer, '\n');
1202 }
1203
1204 struct commit_extra_header *read_commit_extra_headers(struct commit *commit,
1205                                                       const char **exclude)
1206 {
1207         struct commit_extra_header *extra = NULL;
1208         unsigned long size;
1209         const char *buffer = get_commit_buffer(commit, &size);
1210         extra = read_commit_extra_header_lines(buffer, size, exclude);
1211         unuse_commit_buffer(commit, buffer);
1212         return extra;
1213 }
1214
1215 int for_each_mergetag(each_mergetag_fn fn, struct commit *commit, void *data)
1216 {
1217         struct commit_extra_header *extra, *to_free;
1218         int res = 0;
1219
1220         to_free = read_commit_extra_headers(commit, NULL);
1221         for (extra = to_free; !res && extra; extra = extra->next) {
1222                 if (strcmp(extra->key, "mergetag"))
1223                         continue; /* not a merge tag */
1224                 res = fn(commit, extra, data);
1225         }
1226         free_commit_extra_headers(to_free);
1227         return res;
1228 }
1229
1230 static inline int standard_header_field(const char *field, size_t len)
1231 {
1232         return ((len == 4 && !memcmp(field, "tree", 4)) ||
1233                 (len == 6 && !memcmp(field, "parent", 6)) ||
1234                 (len == 6 && !memcmp(field, "author", 6)) ||
1235                 (len == 9 && !memcmp(field, "committer", 9)) ||
1236                 (len == 8 && !memcmp(field, "encoding", 8)));
1237 }
1238
1239 static int excluded_header_field(const char *field, size_t len, const char **exclude)
1240 {
1241         if (!exclude)
1242                 return 0;
1243
1244         while (*exclude) {
1245                 size_t xlen = strlen(*exclude);
1246                 if (len == xlen && !memcmp(field, *exclude, xlen))
1247                         return 1;
1248                 exclude++;
1249         }
1250         return 0;
1251 }
1252
1253 static struct commit_extra_header *read_commit_extra_header_lines(
1254         const char *buffer, size_t size,
1255         const char **exclude)
1256 {
1257         struct commit_extra_header *extra = NULL, **tail = &extra, *it = NULL;
1258         const char *line, *next, *eof, *eob;
1259         struct strbuf buf = STRBUF_INIT;
1260
1261         for (line = buffer, eob = line + size;
1262              line < eob && *line != '\n';
1263              line = next) {
1264                 next = memchr(line, '\n', eob - line);
1265                 next = next ? next + 1 : eob;
1266                 if (*line == ' ') {
1267                         /* continuation */
1268                         if (it)
1269                                 strbuf_add(&buf, line + 1, next - (line + 1));
1270                         continue;
1271                 }
1272                 if (it)
1273                         it->value = strbuf_detach(&buf, &it->len);
1274                 strbuf_reset(&buf);
1275                 it = NULL;
1276
1277                 eof = memchr(line, ' ', next - line);
1278                 if (!eof)
1279                         eof = next;
1280                 else if (standard_header_field(line, eof - line) ||
1281                          excluded_header_field(line, eof - line, exclude))
1282                         continue;
1283
1284                 it = xcalloc(1, sizeof(*it));
1285                 it->key = xmemdupz(line, eof-line);
1286                 *tail = it;
1287                 tail = &it->next;
1288                 if (eof + 1 < next)
1289                         strbuf_add(&buf, eof + 1, next - (eof + 1));
1290         }
1291         if (it)
1292                 it->value = strbuf_detach(&buf, &it->len);
1293         return extra;
1294 }
1295
1296 void free_commit_extra_headers(struct commit_extra_header *extra)
1297 {
1298         while (extra) {
1299                 struct commit_extra_header *next = extra->next;
1300                 free(extra->key);
1301                 free(extra->value);
1302                 free(extra);
1303                 extra = next;
1304         }
1305 }
1306
1307 int commit_tree(const char *msg, size_t msg_len, const struct object_id *tree,
1308                 struct commit_list *parents, struct object_id *ret,
1309                 const char *author, const char *sign_commit)
1310 {
1311         struct commit_extra_header *extra = NULL, **tail = &extra;
1312         int result;
1313
1314         append_merge_tag_headers(parents, &tail);
1315         result = commit_tree_extended(msg, msg_len, tree, parents, ret,
1316                                       author, sign_commit, extra);
1317         free_commit_extra_headers(extra);
1318         return result;
1319 }
1320
1321 static int find_invalid_utf8(const char *buf, int len)
1322 {
1323         int offset = 0;
1324         static const unsigned int max_codepoint[] = {
1325                 0x7f, 0x7ff, 0xffff, 0x10ffff
1326         };
1327
1328         while (len) {
1329                 unsigned char c = *buf++;
1330                 int bytes, bad_offset;
1331                 unsigned int codepoint;
1332                 unsigned int min_val, max_val;
1333
1334                 len--;
1335                 offset++;
1336
1337                 /* Simple US-ASCII? No worries. */
1338                 if (c < 0x80)
1339                         continue;
1340
1341                 bad_offset = offset-1;
1342
1343                 /*
1344                  * Count how many more high bits set: that's how
1345                  * many more bytes this sequence should have.
1346                  */
1347                 bytes = 0;
1348                 while (c & 0x40) {
1349                         c <<= 1;
1350                         bytes++;
1351                 }
1352
1353                 /*
1354                  * Must be between 1 and 3 more bytes.  Longer sequences result in
1355                  * codepoints beyond U+10FFFF, which are guaranteed never to exist.
1356                  */
1357                 if (bytes < 1 || 3 < bytes)
1358                         return bad_offset;
1359
1360                 /* Do we *have* that many bytes? */
1361                 if (len < bytes)
1362                         return bad_offset;
1363
1364                 /*
1365                  * Place the encoded bits at the bottom of the value and compute the
1366                  * valid range.
1367                  */
1368                 codepoint = (c & 0x7f) >> bytes;
1369                 min_val = max_codepoint[bytes-1] + 1;
1370                 max_val = max_codepoint[bytes];
1371
1372                 offset += bytes;
1373                 len -= bytes;
1374
1375                 /* And verify that they are good continuation bytes */
1376                 do {
1377                         codepoint <<= 6;
1378                         codepoint |= *buf & 0x3f;
1379                         if ((*buf++ & 0xc0) != 0x80)
1380                                 return bad_offset;
1381                 } while (--bytes);
1382
1383                 /* Reject codepoints that are out of range for the sequence length. */
1384                 if (codepoint < min_val || codepoint > max_val)
1385                         return bad_offset;
1386                 /* Surrogates are only for UTF-16 and cannot be encoded in UTF-8. */
1387                 if ((codepoint & 0x1ff800) == 0xd800)
1388                         return bad_offset;
1389                 /* U+xxFFFE and U+xxFFFF are guaranteed non-characters. */
1390                 if ((codepoint & 0xfffe) == 0xfffe)
1391                         return bad_offset;
1392                 /* So are anything in the range U+FDD0..U+FDEF. */
1393                 if (codepoint >= 0xfdd0 && codepoint <= 0xfdef)
1394                         return bad_offset;
1395         }
1396         return -1;
1397 }
1398
1399 /*
1400  * This verifies that the buffer is in proper utf8 format.
1401  *
1402  * If it isn't, it assumes any non-utf8 characters are Latin1,
1403  * and does the conversion.
1404  */
1405 static int verify_utf8(struct strbuf *buf)
1406 {
1407         int ok = 1;
1408         long pos = 0;
1409
1410         for (;;) {
1411                 int bad;
1412                 unsigned char c;
1413                 unsigned char replace[2];
1414
1415                 bad = find_invalid_utf8(buf->buf + pos, buf->len - pos);
1416                 if (bad < 0)
1417                         return ok;
1418                 pos += bad;
1419                 ok = 0;
1420                 c = buf->buf[pos];
1421                 strbuf_remove(buf, pos, 1);
1422
1423                 /* We know 'c' must be in the range 128-255 */
1424                 replace[0] = 0xc0 + (c >> 6);
1425                 replace[1] = 0x80 + (c & 0x3f);
1426                 strbuf_insert(buf, pos, replace, 2);
1427                 pos += 2;
1428         }
1429 }
1430
1431 static const char commit_utf8_warn[] =
1432 N_("Warning: commit message did not conform to UTF-8.\n"
1433    "You may want to amend it after fixing the message, or set the config\n"
1434    "variable i18n.commitencoding to the encoding your project uses.\n");
1435
1436 int commit_tree_extended(const char *msg, size_t msg_len,
1437                          const struct object_id *tree,
1438                          struct commit_list *parents, struct object_id *ret,
1439                          const char *author, const char *sign_commit,
1440                          struct commit_extra_header *extra)
1441 {
1442         int result;
1443         int encoding_is_utf8;
1444         struct strbuf buffer;
1445
1446         assert_oid_type(tree, OBJ_TREE);
1447
1448         if (memchr(msg, '\0', msg_len))
1449                 return error("a NUL byte in commit log message not allowed.");
1450
1451         /* Not having i18n.commitencoding is the same as having utf-8 */
1452         encoding_is_utf8 = is_encoding_utf8(git_commit_encoding);
1453
1454         strbuf_init(&buffer, 8192); /* should avoid reallocs for the headers */
1455         strbuf_addf(&buffer, "tree %s\n", oid_to_hex(tree));
1456
1457         /*
1458          * NOTE! This ordering means that the same exact tree merged with a
1459          * different order of parents will be a _different_ changeset even
1460          * if everything else stays the same.
1461          */
1462         while (parents) {
1463                 struct commit *parent = pop_commit(&parents);
1464                 strbuf_addf(&buffer, "parent %s\n",
1465                             oid_to_hex(&parent->object.oid));
1466         }
1467
1468         /* Person/date information */
1469         if (!author)
1470                 author = git_author_info(IDENT_STRICT);
1471         strbuf_addf(&buffer, "author %s\n", author);
1472         strbuf_addf(&buffer, "committer %s\n", git_committer_info(IDENT_STRICT));
1473         if (!encoding_is_utf8)
1474                 strbuf_addf(&buffer, "encoding %s\n", git_commit_encoding);
1475
1476         while (extra) {
1477                 add_extra_header(&buffer, extra);
1478                 extra = extra->next;
1479         }
1480         strbuf_addch(&buffer, '\n');
1481
1482         /* And add the comment */
1483         strbuf_add(&buffer, msg, msg_len);
1484
1485         /* And check the encoding */
1486         if (encoding_is_utf8 && !verify_utf8(&buffer))
1487                 fprintf(stderr, _(commit_utf8_warn));
1488
1489         if (sign_commit && do_sign_commit(&buffer, sign_commit)) {
1490                 result = -1;
1491                 goto out;
1492         }
1493
1494         result = write_object_file(buffer.buf, buffer.len, commit_type, ret);
1495 out:
1496         strbuf_release(&buffer);
1497         return result;
1498 }
1499
1500 define_commit_slab(merge_desc_slab, struct merge_remote_desc *);
1501 static struct merge_desc_slab merge_desc_slab = COMMIT_SLAB_INIT(1, merge_desc_slab);
1502
1503 struct merge_remote_desc *merge_remote_util(struct commit *commit)
1504 {
1505         return *merge_desc_slab_at(&merge_desc_slab, commit);
1506 }
1507
1508 void set_merge_remote_desc(struct commit *commit,
1509                            const char *name, struct object *obj)
1510 {
1511         struct merge_remote_desc *desc;
1512         FLEX_ALLOC_STR(desc, name, name);
1513         desc->obj = obj;
1514         *merge_desc_slab_at(&merge_desc_slab, commit) = desc;
1515 }
1516
1517 struct commit *get_merge_parent(const char *name)
1518 {
1519         struct object *obj;
1520         struct commit *commit;
1521         struct object_id oid;
1522         if (get_oid(name, &oid))
1523                 return NULL;
1524         obj = parse_object(the_repository, &oid);
1525         commit = (struct commit *)peel_to_type(name, 0, obj, OBJ_COMMIT);
1526         if (commit && !merge_remote_util(commit))
1527                 set_merge_remote_desc(commit, name, obj);
1528         return commit;
1529 }
1530
1531 /*
1532  * Append a commit to the end of the commit_list.
1533  *
1534  * next starts by pointing to the variable that holds the head of an
1535  * empty commit_list, and is updated to point to the "next" field of
1536  * the last item on the list as new commits are appended.
1537  *
1538  * Usage example:
1539  *
1540  *     struct commit_list *list;
1541  *     struct commit_list **next = &list;
1542  *
1543  *     next = commit_list_append(c1, next);
1544  *     next = commit_list_append(c2, next);
1545  *     assert(commit_list_count(list) == 2);
1546  *     return list;
1547  */
1548 struct commit_list **commit_list_append(struct commit *commit,
1549                                         struct commit_list **next)
1550 {
1551         struct commit_list *new_commit = xmalloc(sizeof(struct commit_list));
1552         new_commit->item = commit;
1553         *next = new_commit;
1554         new_commit->next = NULL;
1555         return &new_commit->next;
1556 }
1557
1558 const char *find_commit_header(const char *msg, const char *key, size_t *out_len)
1559 {
1560         int key_len = strlen(key);
1561         const char *line = msg;
1562
1563         while (line) {
1564                 const char *eol = strchrnul(line, '\n');
1565
1566                 if (line == eol)
1567                         return NULL;
1568
1569                 if (eol - line > key_len &&
1570                     !strncmp(line, key, key_len) &&
1571                     line[key_len] == ' ') {
1572                         *out_len = eol - line - key_len - 1;
1573                         return line + key_len + 1;
1574                 }
1575                 line = *eol ? eol + 1 : NULL;
1576         }
1577         return NULL;
1578 }
1579
1580 /*
1581  * Inspect the given string and determine the true "end" of the log message, in
1582  * order to find where to put a new Signed-off-by: line.  Ignored are
1583  * trailing comment lines and blank lines.  To support "git commit -s
1584  * --amend" on an existing commit, we also ignore "Conflicts:".  To
1585  * support "git commit -v", we truncate at cut lines.
1586  *
1587  * Returns the number of bytes from the tail to ignore, to be fed as
1588  * the second parameter to append_signoff().
1589  */
1590 size_t ignore_non_trailer(const char *buf, size_t len)
1591 {
1592         size_t boc = 0;
1593         size_t bol = 0;
1594         int in_old_conflicts_block = 0;
1595         size_t cutoff = wt_status_locate_end(buf, len);
1596
1597         while (bol < cutoff) {
1598                 const char *next_line = memchr(buf + bol, '\n', len - bol);
1599
1600                 if (!next_line)
1601                         next_line = buf + len;
1602                 else
1603                         next_line++;
1604
1605                 if (buf[bol] == comment_line_char || buf[bol] == '\n') {
1606                         /* is this the first of the run of comments? */
1607                         if (!boc)
1608                                 boc = bol;
1609                         /* otherwise, it is just continuing */
1610                 } else if (starts_with(buf + bol, "Conflicts:\n")) {
1611                         in_old_conflicts_block = 1;
1612                         if (!boc)
1613                                 boc = bol;
1614                 } else if (in_old_conflicts_block && buf[bol] == '\t') {
1615                         ; /* a pathname in the conflicts block */
1616                 } else if (boc) {
1617                         /* the previous was not trailing comment */
1618                         boc = 0;
1619                         in_old_conflicts_block = 0;
1620                 }
1621                 bol = next_line - buf;
1622         }
1623         return boc ? len - boc : len - cutoff;
1624 }
1625
1626 int run_commit_hook(int editor_is_used, const char *index_file,
1627                     const char *name, ...)
1628 {
1629         struct argv_array hook_env = ARGV_ARRAY_INIT;
1630         va_list args;
1631         int ret;
1632
1633         argv_array_pushf(&hook_env, "GIT_INDEX_FILE=%s", index_file);
1634
1635         /*
1636          * Let the hook know that no editor will be launched.
1637          */
1638         if (!editor_is_used)
1639                 argv_array_push(&hook_env, "GIT_EDITOR=:");
1640
1641         va_start(args, name);
1642         ret = run_hook_ve(hook_env.argv,name, args);
1643         va_end(args);
1644         argv_array_clear(&hook_env);
1645
1646         return ret;
1647 }