git.oblomov.eu Git - git/blob - 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
  24 static struct commit_extra_header *read_commit_extra_header_lines(const char *buf, size_t len, const char **);
  25
  26 int save_commit_buffer = 1;
  27
  28 const char *commit_type = "commit";
  29
  30 struct commit *lookup_commit_reference_gently(struct repository *r,
  31                 const struct object_id *oid, int quiet)
  32 {
  33         struct object *obj = deref_tag(r,
  34                                        parse_object(r, oid),
  35                                        NULL, 0);
  36
  37         if (!obj)
  38                 return NULL;
  39         return object_as_type(r, obj, OBJ_COMMIT, quiet);
  40 }
  41
  42 struct commit *lookup_commit_reference(struct repository *r, const struct object_id *oid)
  43 {
  44         return lookup_commit_reference_gently(r, oid, 0);
  45 }
  46
  47 struct commit *lookup_commit_or_die(const struct object_id *oid, const char *ref_name)
  48 {
  49         struct commit *c = lookup_commit_reference(the_repository, oid);
  50         if (!c)
  51                 die(_("could not parse %s"), ref_name);
  52         if (!oideq(oid, &c->object.oid)) {
  53                 warning(_("%s %s is not a commit!"),
  54                         ref_name, oid_to_hex(oid));
  55         }
  56         return c;
  57 }
  58
  59 struct commit *lookup_commit(struct repository *r, const struct object_id *oid)
  60 {
  61         struct object *obj = lookup_object(r, oid);
  62         if (!obj)
  63                 return create_object(r, oid, alloc_commit_node(r));
  64         return object_as_type(r, obj, OBJ_COMMIT, 0);
  65 }
  66
  67 struct commit *lookup_commit_reference_by_name(const char *name)
  68 {
  69         struct object_id oid;
  70         struct commit *commit;
  71
  72         if (get_oid_committish(name, &oid))
  73                 return NULL;
  74         commit = lookup_commit_reference(the_repository, &oid);
  75         if (parse_commit(commit))
  76                 return NULL;
  77         return commit;
  78 }
  79
  80 static timestamp_t parse_commit_date(const char *buf, const char *tail)
  81 {
  82         const char *dateptr;
  83
  84         if (buf + 6 >= tail)
  85                 return 0;
  86         if (memcmp(buf, "author", 6))
  87                 return 0;
  88         while (buf < tail && *buf++ != '\n')
  89                 /* nada */;
  90         if (buf + 9 >= tail)
  91                 return 0;
  92         if (memcmp(buf, "committer", 9))
  93                 return 0;
  94         while (buf < tail && *buf++ != '>')
  95                 /* nada */;
  96         if (buf >= tail)
  97                 return 0;
  98         dateptr = buf;
  99         while (buf < tail && *buf++ != '\n')
 100                 /* nada */;
 101         if (buf >= tail)
 102                 return 0;
 103         /* dateptr < buf && buf[-1] == '\n', so parsing will stop at buf-1 */
 104         return parse_timestamp(dateptr, NULL, 10);
 105 }
 106
 107 static const unsigned char *commit_graft_sha1_access(size_t index, void *table)
 108 {
 109         struct commit_graft **commit_graft_table = table;
 110         return commit_graft_table[index]->oid.hash;
 111 }
 112
 113 static int commit_graft_pos(struct repository *r, const unsigned char *sha1)
 114 {
 115         return sha1_pos(sha1, r->parsed_objects->grafts,
 116                         r->parsed_objects->grafts_nr,
 117                         commit_graft_sha1_access);
 118 }
 119
 120 int register_commit_graft(struct repository *r, struct commit_graft *graft,
 121                           int ignore_dups)
 122 {
 123         int pos = commit_graft_pos(r, graft->oid.hash);
 124
 125         if (0 <= pos) {
 126                 if (ignore_dups)
 127                         free(graft);
 128                 else {
 129                         free(r->parsed_objects->grafts[pos]);
 130                         r->parsed_objects->grafts[pos] = graft;
 131                 }
 132                 return 1;
 133         }
 134         pos = -pos - 1;
 135         ALLOC_GROW(r->parsed_objects->grafts,
 136                    r->parsed_objects->grafts_nr + 1,
 137                    r->parsed_objects->grafts_alloc);
 138         r->parsed_objects->grafts_nr++;
 139         if (pos < r->parsed_objects->grafts_nr)
 140                 memmove(r->parsed_objects->grafts + pos + 1,
 141                         r->parsed_objects->grafts + pos,
 142                         (r->parsed_objects->grafts_nr - pos - 1) *
 143                         sizeof(*r->parsed_objects->grafts));
 144         r->parsed_objects->grafts[pos] = graft;
 145         return 0;
 146 }
 147
 148 struct commit_graft *read_graft_line(struct strbuf *line)
 149 {
 150         /* The format is just "Commit Parent1 Parent2 ...\n" */
 151         int i, phase;
 152         const char *tail = NULL;
 153         struct commit_graft *graft = NULL;
 154         struct object_id dummy_oid, *oid;
 155
 156         strbuf_rtrim(line);
 157         if (!line->len || line->buf[0] == '#')
 158                 return NULL;
 159         /*
 160          * phase 0 verifies line, counts hashes in line and allocates graft
 161          * phase 1 fills graft
 162          */
 163         for (phase = 0; phase < 2; phase++) {
 164                 oid = graft ? &graft->oid : &dummy_oid;
 165                 if (parse_oid_hex(line->buf, oid, &tail))
 166                         goto bad_graft_data;
 167                 for (i = 0; *tail != '\0'; i++) {
 168                         oid = graft ? &graft->parent[i] : &dummy_oid;
 169                         if (!isspace(*tail++) || parse_oid_hex(tail, oid, &tail))
 170                                 goto bad_graft_data;
 171                 }
 172                 if (!graft) {
 173                         graft = xmalloc(st_add(sizeof(*graft),
 174                                                st_mult(sizeof(struct object_id), i)));
 175                         graft->nr_parent = i;
 176                 }
 177         }
 178         return graft;
 179
 180 bad_graft_data:
 181         error("bad graft data: %s", line->buf);
 182         assert(!graft);
 183         return NULL;
 184 }
 185
 186 static int read_graft_file(struct repository *r, const char *graft_file)
 187 {
 188         FILE *fp = fopen_or_warn(graft_file, "r");
 189         struct strbuf buf = STRBUF_INIT;
 190         if (!fp)
 191                 return -1;
 192         if (advice_graft_file_deprecated)
 193                 advise(_("Support for <GIT_DIR>/info/grafts is deprecated\n"
 194                          "and will be removed in a future Git version.\n"
 195                          "\n"
 196                          "Please use \"git replace --convert-graft-file\"\n"
 197                          "to convert the grafts into replace refs.\n"
 198                          "\n"
 199                          "Turn this message off by running\n"
 200                          "\"git config advice.graftFileDeprecated false\""));
 201         while (!strbuf_getwholeline(&buf, fp, '\n')) {
 202                 /* The format is just "Commit Parent1 Parent2 ...\n" */
 203                 struct commit_graft *graft = read_graft_line(&buf);
 204                 if (!graft)
 205                         continue;
 206                 if (register_commit_graft(r, graft, 1))
 207                         error("duplicate graft data: %s", buf.buf);
 208         }
 209         fclose(fp);
 210         strbuf_release(&buf);
 211         return 0;
 212 }
 213
 214 void prepare_commit_graft(struct repository *r)
 215 {
 216         char *graft_file;
 217
 218         if (r->parsed_objects->commit_graft_prepared)
 219                 return;
 220         if (!startup_info->have_repository)
 221                 return;
 222
 223         graft_file = get_graft_file(r);
 224         read_graft_file(r, graft_file);
 225         /* make sure shallows are read */
 226         is_repository_shallow(r);
 227         r->parsed_objects->commit_graft_prepared = 1;
 228 }
 229
 230 struct commit_graft *lookup_commit_graft(struct repository *r, const struct object_id *oid)
 231 {
 232         int pos;
 233         prepare_commit_graft(r);
 234         pos = commit_graft_pos(r, oid->hash);
 235         if (pos < 0)
 236                 return NULL;
 237         return r->parsed_objects->grafts[pos];
 238 }
 239
 240 int for_each_commit_graft(each_commit_graft_fn fn, void *cb_data)
 241 {
 242         int i, ret;
 243         for (i = ret = 0; i < the_repository->parsed_objects->grafts_nr && !ret; i++)
 244                 ret = fn(the_repository->parsed_objects->grafts[i], cb_data);
 245         return ret;
 246 }
 247
 248 int unregister_shallow(const struct object_id *oid)
 249 {
 250         int pos = commit_graft_pos(the_repository, oid->hash);
 251         if (pos < 0)
 252                 return -1;
 253         if (pos + 1 < the_repository->parsed_objects->grafts_nr)
 254                 MOVE_ARRAY(the_repository->parsed_objects->grafts + pos,
 255                            the_repository->parsed_objects->grafts + pos + 1,
 256                            the_repository->parsed_objects->grafts_nr - pos - 1);
 257         the_repository->parsed_objects->grafts_nr--;
 258         return 0;
 259 }
 260
 261 struct commit_buffer {
 262         void *buffer;
 263         unsigned long size;
 264 };
 265 define_commit_slab(buffer_slab, struct commit_buffer);
 266
 267 struct buffer_slab *allocate_commit_buffer_slab(void)
 268 {
 269         struct buffer_slab *bs = xmalloc(sizeof(*bs));
 270         init_buffer_slab(bs);
 271         return bs;
 272 }
 273
 274 void free_commit_buffer_slab(struct buffer_slab *bs)
 275 {
 276         clear_buffer_slab(bs);
 277         free(bs);
 278 }
 279
 280 void set_commit_buffer(struct repository *r, struct commit *commit, void *buffer, unsigned long size)
 281 {
 282         struct commit_buffer *v = buffer_slab_at(
 283                 r->parsed_objects->buffer_slab, commit);
 284         v->buffer = buffer;
 285         v->size = size;
 286 }
 287
 288 const void *get_cached_commit_buffer(struct repository *r, const struct commit *commit, unsigned long *sizep)
 289 {
 290         struct commit_buffer *v = buffer_slab_peek(
 291                 r->parsed_objects->buffer_slab, commit);
 292         if (!v) {
 293                 if (sizep)
 294                         *sizep = 0;
 295                 return NULL;
 296         }
 297         if (sizep)
 298                 *sizep = v->size;
 299         return v->buffer;
 300 }
 301
 302 const void *repo_get_commit_buffer(struct repository *r,
 303                                    const struct commit *commit,
 304                                    unsigned long *sizep)
 305 {
 306         const void *ret = get_cached_commit_buffer(r, commit, sizep);
 307         if (!ret) {
 308                 enum object_type type;
 309                 unsigned long size;
 310                 ret = repo_read_object_file(r, &commit->object.oid, &type, &size);
 311                 if (!ret)
 312                         die("cannot read commit object %s",
 313                             oid_to_hex(&commit->object.oid));
 314                 if (type != OBJ_COMMIT)
 315                         die("expected commit for %s, got %s",
 316                             oid_to_hex(&commit->object.oid), type_name(type));
 317                 if (sizep)
 318                         *sizep = size;
 319         }
 320         return ret;
 321 }
 322
 323 void repo_unuse_commit_buffer(struct repository *r,
 324                               const struct commit *commit,
 325                               const void *buffer)
 326 {
 327         struct commit_buffer *v = buffer_slab_peek(
 328                 r->parsed_objects->buffer_slab, commit);
 329         if (!(v && v->buffer == buffer))
 330                 free((void *)buffer);
 331 }
 332
 333 void free_commit_buffer(struct parsed_object_pool *pool, struct commit *commit)
 334 {
 335         struct commit_buffer *v = buffer_slab_peek(
 336                 pool->buffer_slab, commit);
 337         if (v) {
 338                 FREE_AND_NULL(v->buffer);
 339                 v->size = 0;
 340         }
 341 }
 342
 343 static inline void set_commit_tree(struct commit *c, struct tree *t)
 344 {
 345         c->maybe_tree = t;
 346 }
 347
 348 struct tree *repo_get_commit_tree(struct repository *r,
 349                                   const struct commit *commit)
 350 {
 351         if (commit->maybe_tree || !commit->object.parsed)
 352                 return commit->maybe_tree;
 353
 354         if (commit->graph_pos != COMMIT_NOT_FROM_GRAPH)
 355                 return get_commit_tree_in_graph(r, commit);
 356
 357         return NULL;
 358 }
 359
 360 struct object_id *get_commit_tree_oid(const struct commit *commit)
 361 {
 362         struct tree *tree = get_commit_tree(commit);
 363         return tree ? &tree->object.oid : NULL;
 364 }
 365
 366 void release_commit_memory(struct parsed_object_pool *pool, struct commit *c)
 367 {
 368         set_commit_tree(c, NULL);
 369         free_commit_buffer(pool, c);
 370         c->index = 0;
 371         free_commit_list(c->parents);
 372
 373         c->object.parsed = 0;
 374 }
 375
 376 const void *detach_commit_buffer(struct commit *commit, unsigned long *sizep)
 377 {
 378         struct commit_buffer *v = buffer_slab_peek(
 379                 the_repository->parsed_objects->buffer_slab, commit);
 380         void *ret;
 381
 382         if (!v) {
 383                 if (sizep)
 384                         *sizep = 0;
 385                 return NULL;
 386         }
 387         ret = v->buffer;
 388         if (sizep)
 389                 *sizep = v->size;
 390
 391         v->buffer = NULL;
 392         v->size = 0;
 393         return ret;
 394 }
 395
 396 int parse_commit_buffer(struct repository *r, struct commit *item, const void *buffer, unsigned long size, int check_graph)
 397 {
 398         const char *tail = buffer;
 399         const char *bufptr = buffer;
 400         struct object_id parent;
 401         struct commit_list **pptr;
 402         struct commit_graft *graft;
 403         const int tree_entry_len = the_hash_algo->hexsz + 5;
 404         const int parent_entry_len = the_hash_algo->hexsz + 7;
 405         struct tree *tree;
 406
 407         if (item->object.parsed)
 408                 return 0;
 409
 410         if (item->parents) {
 411                 /*
 412                  * Presumably this is leftover from an earlier failed parse;
 413                  * clear it out in preparation for us re-parsing (we'll hit the
 414                  * same error, but that's good, since it lets our caller know
 415                  * the result cannot be trusted.
 416                  */
 417                 free_commit_list(item->parents);
 418                 item->parents = NULL;
 419         }
 420
 421         tail += size;
 422         if (tail <= bufptr + tree_entry_len + 1 || memcmp(bufptr, "tree ", 5) ||
 423                         bufptr[tree_entry_len] != '\n')
 424                 return error("bogus commit object %s", oid_to_hex(&item->object.oid));
 425         if (get_oid_hex(bufptr + 5, &parent) < 0)
 426                 return error("bad tree pointer in commit %s",
 427                              oid_to_hex(&item->object.oid));
 428         tree = lookup_tree(r, &parent);
 429         if (!tree)
 430                 return error("bad tree pointer %s in commit %s",
 431                              oid_to_hex(&parent),
 432                              oid_to_hex(&item->object.oid));
 433         set_commit_tree(item, tree);
 434         bufptr += tree_entry_len + 1; /* "tree " + "hex sha1" + "\n" */
 435         pptr = &item->parents;
 436
 437         graft = lookup_commit_graft(r, &item->object.oid);
 438         while (bufptr + parent_entry_len < tail && !memcmp(bufptr, "parent ", 7)) {
 439                 struct commit *new_parent;
 440
 441                 if (tail <= bufptr + parent_entry_len + 1 ||
 442                     get_oid_hex(bufptr + 7, &parent) ||
 443                     bufptr[parent_entry_len] != '\n')
 444                         return error("bad parents in commit %s", oid_to_hex(&item->object.oid));
 445                 bufptr += parent_entry_len + 1;
 446                 /*
 447                  * The clone is shallow if nr_parent < 0, and we must
 448                  * not traverse its real parents even when we unhide them.
 449                  */
 450                 if (graft && (graft->nr_parent < 0 || grafts_replace_parents))
 451                         continue;
 452                 new_parent = lookup_commit(r, &parent);
 453                 if (!new_parent)
 454                         return error("bad parent %s in commit %s",
 455                                      oid_to_hex(&parent),
 456                                      oid_to_hex(&item->object.oid));
 457                 pptr = &commit_list_insert(new_parent, pptr)->next;
 458         }
 459         if (graft) {
 460                 int i;
 461                 struct commit *new_parent;
 462                 for (i = 0; i < graft->nr_parent; i++) {
 463                         new_parent = lookup_commit(r,
 464                                                    &graft->parent[i]);
 465                         if (!new_parent)
 466                                 return error("bad graft parent %s in commit %s",
 467                                              oid_to_hex(&graft->parent[i]),
 468                                              oid_to_hex(&item->object.oid));
 469                         pptr = &commit_list_insert(new_parent, pptr)->next;
 470                 }
 471         }
 472         item->date = parse_commit_date(bufptr, tail);
 473
 474         if (check_graph)
 475                 load_commit_graph_info(r, item);
 476
 477         item->object.parsed = 1;
 478         return 0;
 479 }
 480
 481 int repo_parse_commit_internal(struct repository *r,
 482                                struct commit *item,
 483                                int quiet_on_missing,
 484                                int use_commit_graph)
 485 {
 486         enum object_type type;
 487         void *buffer;
 488         unsigned long size;
 489         int ret;
 490
 491         if (!item)
 492                 return -1;
 493         if (item->object.parsed)
 494                 return 0;
 495         if (use_commit_graph && parse_commit_in_graph(r, item))
 496                 return 0;
 497         buffer = repo_read_object_file(r, &item->object.oid, &type, &size);
 498         if (!buffer)
 499                 return quiet_on_missing ? -1 :
 500                         error("Could not read %s",
 501                              oid_to_hex(&item->object.oid));
 502         if (type != OBJ_COMMIT) {
 503                 free(buffer);
 504                 return error("Object %s not a commit",
 505                              oid_to_hex(&item->object.oid));
 506         }
 507
 508         ret = parse_commit_buffer(r, item, buffer, size, 0);
 509         if (save_commit_buffer && !ret) {
 510                 set_commit_buffer(r, item, buffer, size);
 511                 return 0;
 512         }
 513         free(buffer);
 514         return ret;
 515 }
 516
 517 int repo_parse_commit_gently(struct repository *r,
 518                              struct commit *item, int quiet_on_missing)
 519 {
 520         return repo_parse_commit_internal(r, item, quiet_on_missing, 1);
 521 }
 522
 523 void parse_commit_or_die(struct commit *item)
 524 {
 525         if (parse_commit(item))
 526                 die("unable to parse commit %s",
 527                     item ? oid_to_hex(&item->object.oid) : "(null)");
 528 }
 529
 530 int find_commit_subject(const char *commit_buffer, const char **subject)
 531 {
 532         const char *eol;
 533         const char *p = commit_buffer;
 534
 535         while (*p && (*p != '\n' || p[1] != '\n'))
 536                 p++;
 537         if (*p) {
 538                 p = skip_blank_lines(p + 2);
 539                 eol = strchrnul(p, '\n');
 540         } else
 541                 eol = p;
 542
 543         *subject = p;
 544
 545         return eol - p;
 546 }
 547
 548 struct commit_list *commit_list_insert(struct commit *item, struct commit_list **list_p)
 549 {
 550         struct commit_list *new_list = xmalloc(sizeof(struct commit_list));
 551         new_list->item = item;
 552         new_list->next = *list_p;
 553         *list_p = new_list;
 554         return new_list;
 555 }
 556
 557 unsigned commit_list_count(const struct commit_list *l)
 558 {
 559         unsigned c = 0;
 560         for (; l; l = l->next )
 561                 c++;
 562         return c;
 563 }
 564
 565 struct commit_list *copy_commit_list(struct commit_list *list)
 566 {
 567         struct commit_list *head = NULL;
 568         struct commit_list **pp = &head;
 569         while (list) {
 570                 pp = commit_list_append(list->item, pp);
 571                 list = list->next;
 572         }
 573         return head;
 574 }
 575
 576 void free_commit_list(struct commit_list *list)
 577 {
 578         while (list)
 579                 pop_commit(&list);
 580 }
 581
 582 struct commit_list * commit_list_insert_by_date(struct commit *item, struct commit_list **list)
 583 {
 584         struct commit_list **pp = list;
 585         struct commit_list *p;
 586         while ((p = *pp) != NULL) {
 587                 if (p->item->date < item->date) {
 588                         break;
 589                 }
 590                 pp = &p->next;
 591         }
 592         return commit_list_insert(item, pp);
 593 }
 594
 595 static int commit_list_compare_by_date(const void *a, const void *b)
 596 {
 597         timestamp_t a_date = ((const struct commit_list *)a)->item->date;
 598         timestamp_t b_date = ((const struct commit_list *)b)->item->date;
 599         if (a_date < b_date)
 600                 return 1;
 601         if (a_date > b_date)
 602                 return -1;
 603         return 0;
 604 }
 605
 606 static void *commit_list_get_next(const void *a)
 607 {
 608         return ((const struct commit_list *)a)->next;
 609 }
 610
 611 static void commit_list_set_next(void *a, void *next)
 612 {
 613         ((struct commit_list *)a)->next = next;
 614 }
 615
 616 void commit_list_sort_by_date(struct commit_list **list)
 617 {
 618         *list = llist_mergesort(*list, commit_list_get_next, commit_list_set_next,
 619                                 commit_list_compare_by_date);
 620 }
 621
 622 struct commit *pop_most_recent_commit(struct commit_list **list,
 623                                       unsigned int mark)
 624 {
 625         struct commit *ret = pop_commit(list);
 626         struct commit_list *parents = ret->parents;
 627
 628         while (parents) {
 629                 struct commit *commit = parents->item;
 630                 if (!parse_commit(commit) && !(commit->object.flags & mark)) {
 631                         commit->object.flags |= mark;
 632                         commit_list_insert_by_date(commit, list);
 633                 }
 634                 parents = parents->next;
 635         }
 636         return ret;
 637 }
 638
 639 static void clear_commit_marks_1(struct commit_list **plist,
 640                                  struct commit *commit, unsigned int mark)
 641 {
 642         while (commit) {
 643                 struct commit_list *parents;
 644
 645                 if (!(mark & commit->object.flags))
 646                         return;
 647
 648                 commit->object.flags &= ~mark;
 649
 650                 parents = commit->parents;
 651                 if (!parents)
 652                         return;
 653
 654                 while ((parents = parents->next))
 655                         commit_list_insert(parents->item, plist);
 656
 657                 commit = commit->parents->item;
 658         }
 659 }
 660
 661 void clear_commit_marks_many(int nr, struct commit **commit, unsigned int mark)
 662 {
 663         struct commit_list *list = NULL;
 664
 665         while (nr--) {
 666                 clear_commit_marks_1(&list, *commit, mark);
 667                 commit++;
 668         }
 669         while (list)
 670                 clear_commit_marks_1(&list, pop_commit(&list), mark);
 671 }
 672
 673 void clear_commit_marks(struct commit *commit, unsigned int mark)
 674 {
 675         clear_commit_marks_many(1, &commit, mark);
 676 }
 677
 678 struct commit *pop_commit(struct commit_list **stack)
 679 {
 680         struct commit_list *top = *stack;
 681         struct commit *item = top ? top->item : NULL;
 682
 683         if (top) {
 684                 *stack = top->next;
 685                 free(top);
 686         }
 687         return item;
 688 }
 689
 690 /*
 691  * Topological sort support
 692  */
 693
 694 /* count number of children that have not been emitted */
 695 define_commit_slab(indegree_slab, int);
 696
 697 define_commit_slab(author_date_slab, timestamp_t);
 698
 699 void record_author_date(struct author_date_slab *author_date,
 700                         struct commit *commit)
 701 {
 702         const char *buffer = get_commit_buffer(commit, NULL);
 703         struct ident_split ident;
 704         const char *ident_line;
 705         size_t ident_len;
 706         char *date_end;
 707         timestamp_t date;
 708
 709         ident_line = find_commit_header(buffer, "author", &ident_len);
 710         if (!ident_line)
 711                 goto fail_exit; /* no author line */
 712         if (split_ident_line(&ident, ident_line, ident_len) ||
 713             !ident.date_begin || !ident.date_end)
 714                 goto fail_exit; /* malformed "author" line */
 715
 716         date = parse_timestamp(ident.date_begin, &date_end, 10);
 717         if (date_end != ident.date_end)
 718                 goto fail_exit; /* malformed date */
 719         *(author_date_slab_at(author_date, commit)) = date;
 720
 721 fail_exit:
 722         unuse_commit_buffer(commit, buffer);
 723 }
 724
 725 int compare_commits_by_author_date(const void *a_, const void *b_,
 726                                    void *cb_data)
 727 {
 728         const struct commit *a = a_, *b = b_;
 729         struct author_date_slab *author_date = cb_data;
 730         timestamp_t a_date = *(author_date_slab_at(author_date, a));
 731         timestamp_t b_date = *(author_date_slab_at(author_date, b));
 732
 733         /* newer commits with larger date first */
 734         if (a_date < b_date)
 735                 return 1;
 736         else if (a_date > b_date)
 737                 return -1;
 738         return 0;
 739 }
 740
 741 int compare_commits_by_gen_then_commit_date(const void *a_, const void *b_, void *unused)
 742 {
 743         const struct commit *a = a_, *b = b_;
 744
 745         /* newer commits first */
 746         if (a->generation < b->generation)
 747                 return 1;
 748         else if (a->generation > b->generation)
 749                 return -1;
 750
 751         /* use date as a heuristic when generations are equal */
 752         if (a->date < b->date)
 753                 return 1;
 754         else if (a->date > b->date)
 755                 return -1;
 756         return 0;
 757 }
 758
 759 int compare_commits_by_commit_date(const void *a_, const void *b_, void *unused)
 760 {
 761         const struct commit *a = a_, *b = b_;
 762         /* newer commits with larger date first */
 763         if (a->date < b->date)
 764                 return 1;
 765         else if (a->date > b->date)
 766                 return -1;
 767         return 0;
 768 }
 769
 770 /*
 771  * Performs an in-place topological sort on the list supplied.
 772  */
 773 void sort_in_topological_order(struct commit_list **list, enum rev_sort_order sort_order)
 774 {
 775         struct commit_list *next, *orig = *list;
 776         struct commit_list **pptr;
 777         struct indegree_slab indegree;
 778         struct prio_queue queue;
 779         struct commit *commit;
 780         struct author_date_slab author_date;
 781
 782         if (!orig)
 783                 return;
 784         *list = NULL;
 785
 786         init_indegree_slab(&indegree);
 787         memset(&queue, '\0', sizeof(queue));
 788
 789         switch (sort_order) {
 790         default: /* REV_SORT_IN_GRAPH_ORDER */
 791                 queue.compare = NULL;
 792                 break;
 793         case REV_SORT_BY_COMMIT_DATE:
 794                 queue.compare = compare_commits_by_commit_date;
 795                 break;
 796         case REV_SORT_BY_AUTHOR_DATE:
 797                 init_author_date_slab(&author_date);
 798                 queue.compare = compare_commits_by_author_date;
 799                 queue.cb_data = &author_date;
 800                 break;
 801         }
 802
 803         /* Mark them and clear the indegree */
 804         for (next = orig; next; next = next->next) {
 805                 struct commit *commit = next->item;
 806                 *(indegree_slab_at(&indegree, commit)) = 1;
 807                 /* also record the author dates, if needed */
 808                 if (sort_order == REV_SORT_BY_AUTHOR_DATE)
 809                         record_author_date(&author_date, commit);
 810         }
 811
 812         /* update the indegree */
 813         for (next = orig; next; next = next->next) {
 814                 struct commit_list *parents = next->item->parents;
 815                 while (parents) {
 816                         struct commit *parent = parents->item;
 817                         int *pi = indegree_slab_at(&indegree, parent);
 818
 819                         if (*pi)
 820                                 (*pi)++;
 821                         parents = parents->next;
 822                 }
 823         }
 824
 825         /*
 826          * find the tips
 827          *
 828          * tips are nodes not reachable from any other node in the list
 829          *
 830          * the tips serve as a starting set for the work queue.
 831          */
 832         for (next = orig; next; next = next->next) {
 833                 struct commit *commit = next->item;
 834
 835                 if (*(indegree_slab_at(&indegree, commit)) == 1)
 836                         prio_queue_put(&queue, commit);
 837         }
 838
 839         /*
 840          * This is unfortunate; the initial tips need to be shown
 841          * in the order given from the revision traversal machinery.
 842          */
 843         if (sort_order == REV_SORT_IN_GRAPH_ORDER)
 844                 prio_queue_reverse(&queue);
 845
 846         /* We no longer need the commit list */
 847         free_commit_list(orig);
 848
 849         pptr = list;
 850         *list = NULL;
 851         while ((commit = prio_queue_get(&queue)) != NULL) {
 852                 struct commit_list *parents;
 853
 854                 for (parents = commit->parents; parents ; parents = parents->next) {
 855                         struct commit *parent = parents->item;
 856                         int *pi = indegree_slab_at(&indegree, parent);
 857
 858                         if (!*pi)
 859                                 continue;
 860
 861                         /*
 862                          * parents are only enqueued for emission
 863                          * when all their children have been emitted thereby
 864                          * guaranteeing topological order.
 865                          */
 866                         if (--(*pi) == 1)
 867                                 prio_queue_put(&queue, parent);
 868                 }
 869                 /*
 870                  * all children of commit have already been
 871                  * emitted. we can emit it now.
 872                  */
 873                 *(indegree_slab_at(&indegree, commit)) = 0;
 874
 875                 pptr = &commit_list_insert(commit, pptr)->next;
 876         }
 877
 878         clear_indegree_slab(&indegree);
 879         clear_prio_queue(&queue);
 880         if (sort_order == REV_SORT_BY_AUTHOR_DATE)
 881                 clear_author_date_slab(&author_date);
 882 }
 883
 884 struct rev_collect {
 885         struct commit **commit;
 886         int nr;
 887         int alloc;
 888         unsigned int initial : 1;
 889 };
 890
 891 static void add_one_commit(struct object_id *oid, struct rev_collect *revs)
 892 {
 893         struct commit *commit;
 894
 895         if (is_null_oid(oid))
 896                 return;
 897
 898         commit = lookup_commit(the_repository, oid);
 899         if (!commit ||
 900             (commit->object.flags & TMP_MARK) ||
 901             parse_commit(commit))
 902                 return;
 903
 904         ALLOC_GROW(revs->commit, revs->nr + 1, revs->alloc);
 905         revs->commit[revs->nr++] = commit;
 906         commit->object.flags |= TMP_MARK;
 907 }
 908
 909 static int collect_one_reflog_ent(struct object_id *ooid, struct object_id *noid,
 910                                   const char *ident, timestamp_t timestamp,
 911                                   int tz, const char *message, void *cbdata)
 912 {
 913         struct rev_collect *revs = cbdata;
 914
 915         if (revs->initial) {
 916                 revs->initial = 0;
 917                 add_one_commit(ooid, revs);
 918         }
 919         add_one_commit(noid, revs);
 920         return 0;
 921 }
 922
 923 struct commit *get_fork_point(const char *refname, struct commit *commit)
 924 {
 925         struct object_id oid;
 926         struct rev_collect revs;
 927         struct commit_list *bases;
 928         int i;
 929         struct commit *ret = NULL;
 930         char *full_refname;
 931
 932         switch (dwim_ref(refname, strlen(refname), &oid, &full_refname)) {
 933         case 0:
 934                 die("No such ref: '%s'", refname);
 935         case 1:
 936                 break; /* good */
 937         default:
 938                 die("Ambiguous refname: '%s'", refname);
 939         }
 940
 941         memset(&revs, 0, sizeof(revs));
 942         revs.initial = 1;
 943         for_each_reflog_ent(full_refname, collect_one_reflog_ent, &revs);
 944
 945         if (!revs.nr)
 946                 add_one_commit(&oid, &revs);
 947
 948         for (i = 0; i < revs.nr; i++)
 949                 revs.commit[i]->object.flags &= ~TMP_MARK;
 950
 951         bases = get_merge_bases_many(commit, revs.nr, revs.commit);
 952
 953         /*
 954          * There should be one and only one merge base, when we found
 955          * a common ancestor among reflog entries.
 956          */
 957         if (!bases || bases->next)
 958                 goto cleanup_return;
 959
 960         /* And the found one must be one of the reflog entries */
 961         for (i = 0; i < revs.nr; i++)
 962                 if (&bases->item->object == &revs.commit[i]->object)
 963                         break; /* found */
 964         if (revs.nr <= i)
 965                 goto cleanup_return;
 966
 967         ret = bases->item;
 968
 969 cleanup_return:
 970         free_commit_list(bases);
 971         free(full_refname);
 972         return ret;
 973 }
 974
 975 /*
 976  * Indexed by hash algorithm identifier.
 977  */
 978 static const char *gpg_sig_headers[] = {
 979         NULL,
 980         "gpgsig",
 981         "gpgsig-sha256",
 982 };
 983
 984 static int do_sign_commit(struct strbuf *buf, const char *keyid)
 985 {
 986         struct strbuf sig = STRBUF_INIT;
 987         int inspos, copypos;
 988         const char *eoh;
 989         const char *gpg_sig_header = gpg_sig_headers[hash_algo_by_ptr(the_hash_algo)];
 990         int gpg_sig_header_len = strlen(gpg_sig_header);
 991
 992         /* find the end of the header */
 993         eoh = strstr(buf->buf, "\n\n");
 994         if (!eoh)
 995                 inspos = buf->len;
 996         else
 997                 inspos = eoh - buf->buf + 1;
 998
 999         if (!keyid || !*keyid)
1000                 keyid = get_signing_key();
1001         if (sign_buffer(buf, &sig, keyid)) {
1002                 strbuf_release(&sig);
1003                 return -1;
1004         }
1005
1006         for (copypos = 0; sig.buf[copypos]; ) {
1007                 const char *bol = sig.buf + copypos;
1008                 const char *eol = strchrnul(bol, '\n');
1009                 int len = (eol - bol) + !!*eol;
1010
1011                 if (!copypos) {
1012                         strbuf_insert(buf, inspos, gpg_sig_header, gpg_sig_header_len);
1013                         inspos += gpg_sig_header_len;
1014                 }
1015                 strbuf_insertstr(buf, inspos++, " ");
1016                 strbuf_insert(buf, inspos, bol, len);
1017                 inspos += len;
1018                 copypos += len;
1019         }
1020         strbuf_release(&sig);
1021         return 0;
1022 }
1023
1024 int parse_signed_commit(const struct commit *commit,
1025                         struct strbuf *payload, struct strbuf *signature)
1026 {
1027
1028         unsigned long size;
1029         const char *buffer = get_commit_buffer(commit, &size);
1030         int in_signature, saw_signature = -1;
1031         const char *line, *tail;
1032         const char *gpg_sig_header = gpg_sig_headers[hash_algo_by_ptr(the_hash_algo)];
1033         int gpg_sig_header_len = strlen(gpg_sig_header);
1034
1035         line = buffer;
1036         tail = buffer + size;
1037         in_signature = 0;
1038         saw_signature = 0;
1039         while (line < tail) {
1040                 const char *sig = NULL;
1041                 const char *next = memchr(line, '\n', tail - line);
1042
1043                 next = next ? next + 1 : tail;
1044                 if (in_signature && line[0] == ' ')
1045                         sig = line + 1;
1046                 else if (starts_with(line, gpg_sig_header) &&
1047                          line[gpg_sig_header_len] == ' ')
1048                         sig = line + gpg_sig_header_len + 1;
1049                 if (sig) {
1050                         strbuf_add(signature, sig, next - sig);
1051                         saw_signature = 1;
1052                         in_signature = 1;
1053                 } else {
1054                         if (*line == '\n')
1055                                 /* dump the whole remainder of the buffer */
1056                                 next = tail;
1057                         strbuf_add(payload, line, next - line);
1058                         in_signature = 0;
1059                 }
1060                 line = next;
1061         }
1062         unuse_commit_buffer(commit, buffer);
1063         return saw_signature;
1064 }
1065
1066 int remove_signature(struct strbuf *buf)
1067 {
1068         const char *line = buf->buf;
1069         const char *tail = buf->buf + buf->len;
1070         int in_signature = 0;
1071         const char *sig_start = NULL;
1072         const char *sig_end = NULL;
1073
1074         while (line < tail) {
1075                 const char *next = memchr(line, '\n', tail - line);
1076                 next = next ? next + 1 : tail;
1077
1078                 if (in_signature && line[0] == ' ')
1079                         sig_end = next;
1080                 else if (starts_with(line, "gpgsig")) {
1081                         int i;
1082                         for (i = 1; i < GIT_HASH_NALGOS; i++) {
1083                                 const char *p;
1084                                 if (skip_prefix(line, gpg_sig_headers[i], &p) &&
1085                                     *p == ' ') {
1086                                         sig_start = line;
1087                                         sig_end = next;
1088                                         in_signature = 1;
1089                                 }
1090                         }
1091                 } else {
1092                         if (*line == '\n')
1093                                 /* dump the whole remainder of the buffer */
1094                                 next = tail;
1095                         in_signature = 0;
1096                 }
1097                 line = next;
1098         }
1099
1100         if (sig_start)
1101                 strbuf_remove(buf, sig_start - buf->buf, sig_end - sig_start);
1102
1103         return sig_start != NULL;
1104 }
1105
1106 static void handle_signed_tag(struct commit *parent, struct commit_extra_header ***tail)
1107 {
1108         struct merge_remote_desc *desc;
1109         struct commit_extra_header *mergetag;
1110         char *buf;
1111         unsigned long size, len;
1112         enum object_type type;
1113
1114         desc = merge_remote_util(parent);
1115         if (!desc || !desc->obj)
1116                 return;
1117         buf = read_object_file(&desc->obj->oid, &type, &size);
1118         if (!buf || type != OBJ_TAG)
1119                 goto free_return;
1120         len = parse_signature(buf, size);
1121         if (size == len)
1122                 goto free_return;
1123         /*
1124          * We could verify this signature and either omit the tag when
1125          * it does not validate, but the integrator may not have the
1126          * public key of the signer of the tag he is merging, while a
1127          * later auditor may have it while auditing, so let's not run
1128          * verify-signed-buffer here for now...
1129          *
1130          * if (verify_signed_buffer(buf, len, buf + len, size - len, ...))
1131          *      warn("warning: signed tag unverified.");
1132          */
1133         mergetag = xcalloc(1, sizeof(*mergetag));
1134         mergetag->key = xstrdup("mergetag");
1135         mergetag->value = buf;
1136         mergetag->len = size;
1137
1138         **tail = mergetag;
1139         *tail = &mergetag->next;
1140         return;
1141
1142 free_return:
1143         free(buf);
1144 }
1145
1146 int check_commit_signature(const struct commit *commit, struct signature_check *sigc)
1147 {
1148         struct strbuf payload = STRBUF_INIT;
1149         struct strbuf signature = STRBUF_INIT;
1150         int ret = 1;
1151
1152         sigc->result = 'N';
1153
1154         if (parse_signed_commit(commit, &payload, &signature) <= 0)
1155                 goto out;
1156         ret = check_signature(payload.buf, payload.len, signature.buf,
1157                 signature.len, sigc);
1158
1159  out:
1160         strbuf_release(&payload);
1161         strbuf_release(&signature);
1162
1163         return ret;
1164 }
1165
1166 void verify_merge_signature(struct commit *commit, int verbosity,
1167                             int check_trust)
1168 {
1169         char hex[GIT_MAX_HEXSZ + 1];
1170         struct signature_check signature_check;
1171         int ret;
1172         memset(&signature_check, 0, sizeof(signature_check));
1173
1174         ret = check_commit_signature(commit, &signature_check);
1175
1176         find_unique_abbrev_r(hex, &commit->object.oid, DEFAULT_ABBREV);
1177         switch (signature_check.result) {
1178         case 'G':
1179                 if (ret || (check_trust && signature_check.trust_level < TRUST_MARGINAL))
1180                         die(_("Commit %s has an untrusted GPG signature, "
1181                               "allegedly by %s."), hex, signature_check.signer);
1182                 break;
1183         case 'B':
1184                 die(_("Commit %s has a bad GPG signature "
1185                       "allegedly by %s."), hex, signature_check.signer);
1186         default: /* 'N' */
1187                 die(_("Commit %s does not have a GPG signature."), hex);
1188         }
1189         if (verbosity >= 0 && signature_check.result == 'G')
1190                 printf(_("Commit %s has a good GPG signature by %s\n"),
1191                        hex, signature_check.signer);
1192
1193         signature_check_clear(&signature_check);
1194 }
1195
1196 void append_merge_tag_headers(struct commit_list *parents,
1197                               struct commit_extra_header ***tail)
1198 {
1199         while (parents) {
1200                 struct commit *parent = parents->item;
1201                 handle_signed_tag(parent, tail);
1202                 parents = parents->next;
1203         }
1204 }
1205
1206 static void add_extra_header(struct strbuf *buffer,
1207                              struct commit_extra_header *extra)
1208 {
1209         strbuf_addstr(buffer, extra->key);
1210         if (extra->len)
1211                 strbuf_add_lines(buffer, " ", extra->value, extra->len);
1212         else
1213                 strbuf_addch(buffer, '\n');
1214 }
1215
1216 struct commit_extra_header *read_commit_extra_headers(struct commit *commit,
1217                                                       const char **exclude)
1218 {
1219         struct commit_extra_header *extra = NULL;
1220         unsigned long size;
1221         const char *buffer = get_commit_buffer(commit, &size);
1222         extra = read_commit_extra_header_lines(buffer, size, exclude);
1223         unuse_commit_buffer(commit, buffer);
1224         return extra;
1225 }
1226
1227 int for_each_mergetag(each_mergetag_fn fn, struct commit *commit, void *data)
1228 {
1229         struct commit_extra_header *extra, *to_free;
1230         int res = 0;
1231
1232         to_free = read_commit_extra_headers(commit, NULL);
1233         for (extra = to_free; !res && extra; extra = extra->next) {
1234                 if (strcmp(extra->key, "mergetag"))
1235                         continue; /* not a merge tag */
1236                 res = fn(commit, extra, data);
1237         }
1238         free_commit_extra_headers(to_free);
1239         return res;
1240 }
1241
1242 static inline int standard_header_field(const char *field, size_t len)
1243 {
1244         return ((len == 4 && !memcmp(field, "tree", 4)) ||
1245                 (len == 6 && !memcmp(field, "parent", 6)) ||
1246                 (len == 6 && !memcmp(field, "author", 6)) ||
1247                 (len == 9 && !memcmp(field, "committer", 9)) ||
1248                 (len == 8 && !memcmp(field, "encoding", 8)));
1249 }
1250
1251 static int excluded_header_field(const char *field, size_t len, const char **exclude)
1252 {
1253         if (!exclude)
1254                 return 0;
1255
1256         while (*exclude) {
1257                 size_t xlen = strlen(*exclude);
1258                 if (len == xlen && !memcmp(field, *exclude, xlen))
1259                         return 1;
1260                 exclude++;
1261         }
1262         return 0;
1263 }
1264
1265 static struct commit_extra_header *read_commit_extra_header_lines(
1266         const char *buffer, size_t size,
1267         const char **exclude)
1268 {
1269         struct commit_extra_header *extra = NULL, **tail = &extra, *it = NULL;
1270         const char *line, *next, *eof, *eob;
1271         struct strbuf buf = STRBUF_INIT;
1272
1273         for (line = buffer, eob = line + size;
1274              line < eob && *line != '\n';
1275              line = next) {
1276                 next = memchr(line, '\n', eob - line);
1277                 next = next ? next + 1 : eob;
1278                 if (*line == ' ') {
1279                         /* continuation */
1280                         if (it)
1281                                 strbuf_add(&buf, line + 1, next - (line + 1));
1282                         continue;
1283                 }
1284                 if (it)
1285                         it->value = strbuf_detach(&buf, &it->len);
1286                 strbuf_reset(&buf);
1287                 it = NULL;
1288
1289                 eof = memchr(line, ' ', next - line);
1290                 if (!eof)
1291                         eof = next;
1292                 else if (standard_header_field(line, eof - line) ||
1293                          excluded_header_field(line, eof - line, exclude))
1294                         continue;
1295
1296                 it = xcalloc(1, sizeof(*it));
1297                 it->key = xmemdupz(line, eof-line);
1298                 *tail = it;
1299                 tail = &it->next;
1300                 if (eof + 1 < next)
1301                         strbuf_add(&buf, eof + 1, next - (eof + 1));
1302         }
1303         if (it)
1304                 it->value = strbuf_detach(&buf, &it->len);
1305         return extra;
1306 }
1307
1308 void free_commit_extra_headers(struct commit_extra_header *extra)
1309 {
1310         while (extra) {
1311                 struct commit_extra_header *next = extra->next;
1312                 free(extra->key);
1313                 free(extra->value);
1314                 free(extra);
1315                 extra = next;
1316         }
1317 }
1318
1319 int commit_tree(const char *msg, size_t msg_len, const struct object_id *tree,
1320                 struct commit_list *parents, struct object_id *ret,
1321                 const char *author, const char *sign_commit)
1322 {
1323         struct commit_extra_header *extra = NULL, **tail = &extra;
1324         int result;
1325
1326         append_merge_tag_headers(parents, &tail);
1327         result = commit_tree_extended(msg, msg_len, tree, parents, ret, author,
1328                                       NULL, sign_commit, extra);
1329         free_commit_extra_headers(extra);
1330         return result;
1331 }
1332
1333 static int find_invalid_utf8(const char *buf, int len)
1334 {
1335         int offset = 0;
1336         static const unsigned int max_codepoint[] = {
1337                 0x7f, 0x7ff, 0xffff, 0x10ffff
1338         };
1339
1340         while (len) {
1341                 unsigned char c = *buf++;
1342                 int bytes, bad_offset;
1343                 unsigned int codepoint;
1344                 unsigned int min_val, max_val;
1345
1346                 len--;
1347                 offset++;
1348
1349                 /* Simple US-ASCII? No worries. */
1350                 if (c < 0x80)
1351                         continue;
1352
1353                 bad_offset = offset-1;
1354
1355                 /*
1356                  * Count how many more high bits set: that's how
1357                  * many more bytes this sequence should have.
1358                  */
1359                 bytes = 0;
1360                 while (c & 0x40) {
1361                         c <<= 1;
1362                         bytes++;
1363                 }
1364
1365                 /*
1366                  * Must be between 1 and 3 more bytes.  Longer sequences result in
1367                  * codepoints beyond U+10FFFF, which are guaranteed never to exist.
1368                  */
1369                 if (bytes < 1 || 3 < bytes)
1370                         return bad_offset;
1371
1372                 /* Do we *have* that many bytes? */
1373                 if (len < bytes)
1374                         return bad_offset;
1375
1376                 /*
1377                  * Place the encoded bits at the bottom of the value and compute the
1378                  * valid range.
1379                  */
1380                 codepoint = (c & 0x7f) >> bytes;
1381                 min_val = max_codepoint[bytes-1] + 1;
1382                 max_val = max_codepoint[bytes];
1383
1384                 offset += bytes;
1385                 len -= bytes;
1386
1387                 /* And verify that they are good continuation bytes */
1388                 do {
1389                         codepoint <<= 6;
1390                         codepoint |= *buf & 0x3f;
1391                         if ((*buf++ & 0xc0) != 0x80)
1392                                 return bad_offset;
1393                 } while (--bytes);
1394
1395                 /* Reject codepoints that are out of range for the sequence length. */
1396                 if (codepoint < min_val || codepoint > max_val)
1397                         return bad_offset;
1398                 /* Surrogates are only for UTF-16 and cannot be encoded in UTF-8. */
1399                 if ((codepoint & 0x1ff800) == 0xd800)
1400                         return bad_offset;
1401                 /* U+xxFFFE and U+xxFFFF are guaranteed non-characters. */
1402                 if ((codepoint & 0xfffe) == 0xfffe)
1403                         return bad_offset;
1404                 /* So are anything in the range U+FDD0..U+FDEF. */
1405                 if (codepoint >= 0xfdd0 && codepoint <= 0xfdef)
1406                         return bad_offset;
1407         }
1408         return -1;
1409 }
1410
1411 /*
1412  * This verifies that the buffer is in proper utf8 format.
1413  *
1414  * If it isn't, it assumes any non-utf8 characters are Latin1,
1415  * and does the conversion.
1416  */
1417 static int verify_utf8(struct strbuf *buf)
1418 {
1419         int ok = 1;
1420         long pos = 0;
1421
1422         for (;;) {
1423                 int bad;
1424                 unsigned char c;
1425                 unsigned char replace[2];
1426
1427                 bad = find_invalid_utf8(buf->buf + pos, buf->len - pos);
1428                 if (bad < 0)
1429                         return ok;
1430                 pos += bad;
1431                 ok = 0;
1432                 c = buf->buf[pos];
1433                 strbuf_remove(buf, pos, 1);
1434
1435                 /* We know 'c' must be in the range 128-255 */
1436                 replace[0] = 0xc0 + (c >> 6);
1437                 replace[1] = 0x80 + (c & 0x3f);
1438                 strbuf_insert(buf, pos, replace, 2);
1439                 pos += 2;
1440         }
1441 }
1442
1443 static const char commit_utf8_warn[] =
1444 N_("Warning: commit message did not conform to UTF-8.\n"
1445    "You may want to amend it after fixing the message, or set the config\n"
1446    "variable i18n.commitencoding to the encoding your project uses.\n");
1447
1448 int commit_tree_extended(const char *msg, size_t msg_len,
1449                          const struct object_id *tree,
1450                          struct commit_list *parents, struct object_id *ret,
1451                          const char *author, const char *committer,
1452                          const char *sign_commit,
1453                          struct commit_extra_header *extra)
1454 {
1455         int result;
1456         int encoding_is_utf8;
1457         struct strbuf buffer;
1458
1459         assert_oid_type(tree, OBJ_TREE);
1460
1461         if (memchr(msg, '\0', msg_len))
1462                 return error("a NUL byte in commit log message not allowed.");
1463
1464         /* Not having i18n.commitencoding is the same as having utf-8 */
1465         encoding_is_utf8 = is_encoding_utf8(git_commit_encoding);
1466
1467         strbuf_init(&buffer, 8192); /* should avoid reallocs for the headers */
1468         strbuf_addf(&buffer, "tree %s\n", oid_to_hex(tree));
1469
1470         /*
1471          * NOTE! This ordering means that the same exact tree merged with a
1472          * different order of parents will be a _different_ changeset even
1473          * if everything else stays the same.
1474          */
1475         while (parents) {
1476                 struct commit *parent = pop_commit(&parents);
1477                 strbuf_addf(&buffer, "parent %s\n",
1478                             oid_to_hex(&parent->object.oid));
1479         }
1480
1481         /* Person/date information */
1482         if (!author)
1483                 author = git_author_info(IDENT_STRICT);
1484         strbuf_addf(&buffer, "author %s\n", author);
1485         if (!committer)
1486                 committer = git_committer_info(IDENT_STRICT);
1487         strbuf_addf(&buffer, "committer %s\n", committer);
1488         if (!encoding_is_utf8)
1489                 strbuf_addf(&buffer, "encoding %s\n", git_commit_encoding);
1490
1491         while (extra) {
1492                 add_extra_header(&buffer, extra);
1493                 extra = extra->next;
1494         }
1495         strbuf_addch(&buffer, '\n');
1496
1497         /* And add the comment */
1498         strbuf_add(&buffer, msg, msg_len);
1499
1500         /* And check the encoding */
1501         if (encoding_is_utf8 && !verify_utf8(&buffer))
1502                 fprintf(stderr, _(commit_utf8_warn));
1503
1504         if (sign_commit && do_sign_commit(&buffer, sign_commit)) {
1505                 result = -1;
1506                 goto out;
1507         }
1508
1509         result = write_object_file(buffer.buf, buffer.len, commit_type, ret);
1510 out:
1511         strbuf_release(&buffer);
1512         return result;
1513 }
1514
1515 define_commit_slab(merge_desc_slab, struct merge_remote_desc *);
1516 static struct merge_desc_slab merge_desc_slab = COMMIT_SLAB_INIT(1, merge_desc_slab);
1517
1518 struct merge_remote_desc *merge_remote_util(struct commit *commit)
1519 {
1520         return *merge_desc_slab_at(&merge_desc_slab, commit);
1521 }
1522
1523 void set_merge_remote_desc(struct commit *commit,
1524                            const char *name, struct object *obj)
1525 {
1526         struct merge_remote_desc *desc;
1527         FLEX_ALLOC_STR(desc, name, name);
1528         desc->obj = obj;
1529         *merge_desc_slab_at(&merge_desc_slab, commit) = desc;
1530 }
1531
1532 struct commit *get_merge_parent(const char *name)
1533 {
1534         struct object *obj;
1535         struct commit *commit;
1536         struct object_id oid;
1537         if (get_oid(name, &oid))
1538                 return NULL;
1539         obj = parse_object(the_repository, &oid);
1540         commit = (struct commit *)peel_to_type(name, 0, obj, OBJ_COMMIT);
1541         if (commit && !merge_remote_util(commit))
1542                 set_merge_remote_desc(commit, name, obj);
1543         return commit;
1544 }
1545
1546 /*
1547  * Append a commit to the end of the commit_list.
1548  *
1549  * next starts by pointing to the variable that holds the head of an
1550  * empty commit_list, and is updated to point to the "next" field of
1551  * the last item on the list as new commits are appended.
1552  *
1553  * Usage example:
1554  *
1555  *     struct commit_list *list;
1556  *     struct commit_list **next = &list;
1557  *
1558  *     next = commit_list_append(c1, next);
1559  *     next = commit_list_append(c2, next);
1560  *     assert(commit_list_count(list) == 2);
1561  *     return list;
1562  */
1563 struct commit_list **commit_list_append(struct commit *commit,
1564                                         struct commit_list **next)
1565 {
1566         struct commit_list *new_commit = xmalloc(sizeof(struct commit_list));
1567         new_commit->item = commit;
1568         *next = new_commit;
1569         new_commit->next = NULL;
1570         return &new_commit->next;
1571 }
1572
1573 const char *find_commit_header(const char *msg, const char *key, size_t *out_len)
1574 {
1575         int key_len = strlen(key);
1576         const char *line = msg;
1577
1578         while (line) {
1579                 const char *eol = strchrnul(line, '\n');
1580
1581                 if (line == eol)
1582                         return NULL;
1583
1584                 if (eol - line > key_len &&
1585                     !strncmp(line, key, key_len) &&
1586                     line[key_len] == ' ') {
1587                         *out_len = eol - line - key_len - 1;
1588                         return line + key_len + 1;
1589                 }
1590                 line = *eol ? eol + 1 : NULL;
1591         }
1592         return NULL;
1593 }
1594
1595 /*
1596  * Inspect the given string and determine the true "end" of the log message, in
1597  * order to find where to put a new Signed-off-by: line.  Ignored are
1598  * trailing comment lines and blank lines.  To support "git commit -s
1599  * --amend" on an existing commit, we also ignore "Conflicts:".  To
1600  * support "git commit -v", we truncate at cut lines.
1601  *
1602  * Returns the number of bytes from the tail to ignore, to be fed as
1603  * the second parameter to append_signoff().
1604  */
1605 size_t ignore_non_trailer(const char *buf, size_t len)
1606 {
1607         size_t boc = 0;
1608         size_t bol = 0;
1609         int in_old_conflicts_block = 0;
1610         size_t cutoff = wt_status_locate_end(buf, len);
1611
1612         while (bol < cutoff) {
1613                 const char *next_line = memchr(buf + bol, '\n', len - bol);
1614
1615                 if (!next_line)
1616                         next_line = buf + len;
1617                 else
1618                         next_line++;
1619
1620                 if (buf[bol] == comment_line_char || buf[bol] == '\n') {
1621                         /* is this the first of the run of comments? */
1622                         if (!boc)
1623                                 boc = bol;
1624                         /* otherwise, it is just continuing */
1625                 } else if (starts_with(buf + bol, "Conflicts:\n")) {
1626                         in_old_conflicts_block = 1;
1627                         if (!boc)
1628                                 boc = bol;
1629                 } else if (in_old_conflicts_block && buf[bol] == '\t') {
1630                         ; /* a pathname in the conflicts block */
1631                 } else if (boc) {
1632                         /* the previous was not trailing comment */
1633                         boc = 0;
1634                         in_old_conflicts_block = 0;
1635                 }
1636                 bol = next_line - buf;
1637         }
1638         return boc ? len - boc : len - cutoff;
1639 }
1640
1641 int run_commit_hook(int editor_is_used, const char *index_file,
1642                     const char *name, ...)
1643 {
1644         struct argv_array hook_env = ARGV_ARRAY_INIT;
1645         va_list args;
1646         int ret;
1647
1648         argv_array_pushf(&hook_env, "GIT_INDEX_FILE=%s", index_file);
1649
1650         /*
1651          * Let the hook know that no editor will be launched.
1652          */
1653         if (!editor_is_used)
1654                 argv_array_push(&hook_env, "GIT_EDITOR=:");
1655
1656         va_start(args, name);
1657         ret = run_hook_ve(hook_env.argv,name, args);
1658         va_end(args);
1659         argv_array_clear(&hook_env);
1660
1661         return ret;
1662 }