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(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(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_position(commit) != 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         if (graft)
 427                 r->parsed_objects->substituted_parent = 1;
 428         while (bufptr + parent_entry_len < tail && !memcmp(bufptr, "parent ", 7)) {
 429                 struct commit *new_parent;
 430
 431                 if (tail <= bufptr + parent_entry_len + 1 ||
 432                     get_oid_hex(bufptr + 7, &parent) ||
 433                     bufptr[parent_entry_len] != '\n')
 434                         return error("bad parents in commit %s", oid_to_hex(&item->object.oid));
 435                 bufptr += parent_entry_len + 1;
 436                 /*
 437                  * The clone is shallow if nr_parent < 0, and we must
 438                  * not traverse its real parents even when we unhide them.
 439                  */
 440                 if (graft && (graft->nr_parent < 0 || grafts_replace_parents))
 441                         continue;
 442                 new_parent = lookup_commit(r, &parent);
 443                 if (!new_parent)
 444                         return error("bad parent %s in commit %s",
 445                                      oid_to_hex(&parent),
 446                                      oid_to_hex(&item->object.oid));
 447                 pptr = &commit_list_insert(new_parent, pptr)->next;
 448         }
 449         if (graft) {
 450                 int i;
 451                 struct commit *new_parent;
 452                 for (i = 0; i < graft->nr_parent; i++) {
 453                         new_parent = lookup_commit(r,
 454                                                    &graft->parent[i]);
 455                         if (!new_parent)
 456                                 return error("bad graft parent %s in commit %s",
 457                                              oid_to_hex(&graft->parent[i]),
 458                                              oid_to_hex(&item->object.oid));
 459                         pptr = &commit_list_insert(new_parent, pptr)->next;
 460                 }
 461         }
 462         item->date = parse_commit_date(bufptr, tail);
 463
 464         if (check_graph)
 465                 load_commit_graph_info(r, item);
 466
 467         item->object.parsed = 1;
 468         return 0;
 469 }
 470
 471 int repo_parse_commit_internal(struct repository *r,
 472                                struct commit *item,
 473                                int quiet_on_missing,
 474                                int use_commit_graph)
 475 {
 476         enum object_type type;
 477         void *buffer;
 478         unsigned long size;
 479         int ret;
 480
 481         if (!item)
 482                 return -1;
 483         if (item->object.parsed)
 484                 return 0;
 485         if (use_commit_graph && parse_commit_in_graph(r, item))
 486                 return 0;
 487         buffer = repo_read_object_file(r, &item->object.oid, &type, &size);
 488         if (!buffer)
 489                 return quiet_on_missing ? -1 :
 490                         error("Could not read %s",
 491                              oid_to_hex(&item->object.oid));
 492         if (type != OBJ_COMMIT) {
 493                 free(buffer);
 494                 return error("Object %s not a commit",
 495                              oid_to_hex(&item->object.oid));
 496         }
 497
 498         ret = parse_commit_buffer(r, item, buffer, size, 0);
 499         if (save_commit_buffer && !ret) {
 500                 set_commit_buffer(r, item, buffer, size);
 501                 return 0;
 502         }
 503         free(buffer);
 504         return ret;
 505 }
 506
 507 int repo_parse_commit_gently(struct repository *r,
 508                              struct commit *item, int quiet_on_missing)
 509 {
 510         return repo_parse_commit_internal(r, item, quiet_on_missing, 1);
 511 }
 512
 513 void parse_commit_or_die(struct commit *item)
 514 {
 515         if (parse_commit(item))
 516                 die("unable to parse commit %s",
 517                     item ? oid_to_hex(&item->object.oid) : "(null)");
 518 }
 519
 520 int find_commit_subject(const char *commit_buffer, const char **subject)
 521 {
 522         const char *eol;
 523         const char *p = commit_buffer;
 524
 525         while (*p && (*p != '\n' || p[1] != '\n'))
 526                 p++;
 527         if (*p) {
 528                 p = skip_blank_lines(p + 2);
 529                 eol = strchrnul(p, '\n');
 530         } else
 531                 eol = p;
 532
 533         *subject = p;
 534
 535         return eol - p;
 536 }
 537
 538 struct commit_list *commit_list_insert(struct commit *item, struct commit_list **list_p)
 539 {
 540         struct commit_list *new_list = xmalloc(sizeof(struct commit_list));
 541         new_list->item = item;
 542         new_list->next = *list_p;
 543         *list_p = new_list;
 544         return new_list;
 545 }
 546
 547 int commit_list_contains(struct commit *item, struct commit_list *list)
 548 {
 549         while (list) {
 550                 if (list->item == item)
 551                         return 1;
 552                 list = list->next;
 553         }
 554
 555         return 0;
 556 }
 557
 558 unsigned commit_list_count(const struct commit_list *l)
 559 {
 560         unsigned c = 0;
 561         for (; l; l = l->next )
 562                 c++;
 563         return c;
 564 }
 565
 566 struct commit_list *copy_commit_list(struct commit_list *list)
 567 {
 568         struct commit_list *head = NULL;
 569         struct commit_list **pp = &head;
 570         while (list) {
 571                 pp = commit_list_append(list->item, pp);
 572                 list = list->next;
 573         }
 574         return head;
 575 }
 576
 577 struct commit_list *reverse_commit_list(struct commit_list *list)
 578 {
 579         struct commit_list *next = NULL, *current, *backup;
 580         for (current = list; current; current = backup) {
 581                 backup = current->next;
 582                 current->next = next;
 583                 next = current;
 584         }
 585         return next;
 586 }
 587
 588 void free_commit_list(struct commit_list *list)
 589 {
 590         while (list)
 591                 pop_commit(&list);
 592 }
 593
 594 struct commit_list * commit_list_insert_by_date(struct commit *item, struct commit_list **list)
 595 {
 596         struct commit_list **pp = list;
 597         struct commit_list *p;
 598         while ((p = *pp) != NULL) {
 599                 if (p->item->date < item->date) {
 600                         break;
 601                 }
 602                 pp = &p->next;
 603         }
 604         return commit_list_insert(item, pp);
 605 }
 606
 607 static int commit_list_compare_by_date(const void *a, const void *b)
 608 {
 609         timestamp_t a_date = ((const struct commit_list *)a)->item->date;
 610         timestamp_t b_date = ((const struct commit_list *)b)->item->date;
 611         if (a_date < b_date)
 612                 return 1;
 613         if (a_date > b_date)
 614                 return -1;
 615         return 0;
 616 }
 617
 618 static void *commit_list_get_next(const void *a)
 619 {
 620         return ((const struct commit_list *)a)->next;
 621 }
 622
 623 static void commit_list_set_next(void *a, void *next)
 624 {
 625         ((struct commit_list *)a)->next = next;
 626 }
 627
 628 void commit_list_sort_by_date(struct commit_list **list)
 629 {
 630         *list = llist_mergesort(*list, commit_list_get_next, commit_list_set_next,
 631                                 commit_list_compare_by_date);
 632 }
 633
 634 struct commit *pop_most_recent_commit(struct commit_list **list,
 635                                       unsigned int mark)
 636 {
 637         struct commit *ret = pop_commit(list);
 638         struct commit_list *parents = ret->parents;
 639
 640         while (parents) {
 641                 struct commit *commit = parents->item;
 642                 if (!parse_commit(commit) && !(commit->object.flags & mark)) {
 643                         commit->object.flags |= mark;
 644                         commit_list_insert_by_date(commit, list);
 645                 }
 646                 parents = parents->next;
 647         }
 648         return ret;
 649 }
 650
 651 static void clear_commit_marks_1(struct commit_list **plist,
 652                                  struct commit *commit, unsigned int mark)
 653 {
 654         while (commit) {
 655                 struct commit_list *parents;
 656
 657                 if (!(mark & commit->object.flags))
 658                         return;
 659
 660                 commit->object.flags &= ~mark;
 661
 662                 parents = commit->parents;
 663                 if (!parents)
 664                         return;
 665
 666                 while ((parents = parents->next))
 667                         commit_list_insert(parents->item, plist);
 668
 669                 commit = commit->parents->item;
 670         }
 671 }
 672
 673 void clear_commit_marks_many(int nr, struct commit **commit, unsigned int mark)
 674 {
 675         struct commit_list *list = NULL;
 676
 677         while (nr--) {
 678                 clear_commit_marks_1(&list, *commit, mark);
 679                 commit++;
 680         }
 681         while (list)
 682                 clear_commit_marks_1(&list, pop_commit(&list), mark);
 683 }
 684
 685 void clear_commit_marks(struct commit *commit, unsigned int mark)
 686 {
 687         clear_commit_marks_many(1, &commit, mark);
 688 }
 689
 690 struct commit *pop_commit(struct commit_list **stack)
 691 {
 692         struct commit_list *top = *stack;
 693         struct commit *item = top ? top->item : NULL;
 694
 695         if (top) {
 696                 *stack = top->next;
 697                 free(top);
 698         }
 699         return item;
 700 }
 701
 702 /*
 703  * Topological sort support
 704  */
 705
 706 /* count number of children that have not been emitted */
 707 define_commit_slab(indegree_slab, int);
 708
 709 define_commit_slab(author_date_slab, timestamp_t);
 710
 711 void record_author_date(struct author_date_slab *author_date,
 712                         struct commit *commit)
 713 {
 714         const char *buffer = get_commit_buffer(commit, NULL);
 715         struct ident_split ident;
 716         const char *ident_line;
 717         size_t ident_len;
 718         char *date_end;
 719         timestamp_t date;
 720
 721         ident_line = find_commit_header(buffer, "author", &ident_len);
 722         if (!ident_line)
 723                 goto fail_exit; /* no author line */
 724         if (split_ident_line(&ident, ident_line, ident_len) ||
 725             !ident.date_begin || !ident.date_end)
 726                 goto fail_exit; /* malformed "author" line */
 727
 728         date = parse_timestamp(ident.date_begin, &date_end, 10);
 729         if (date_end != ident.date_end)
 730                 goto fail_exit; /* malformed date */
 731         *(author_date_slab_at(author_date, commit)) = date;
 732
 733 fail_exit:
 734         unuse_commit_buffer(commit, buffer);
 735 }
 736
 737 int compare_commits_by_author_date(const void *a_, const void *b_,
 738                                    void *cb_data)
 739 {
 740         const struct commit *a = a_, *b = b_;
 741         struct author_date_slab *author_date = cb_data;
 742         timestamp_t a_date = *(author_date_slab_at(author_date, a));
 743         timestamp_t b_date = *(author_date_slab_at(author_date, b));
 744
 745         /* newer commits with larger date first */
 746         if (a_date < b_date)
 747                 return 1;
 748         else if (a_date > b_date)
 749                 return -1;
 750         return 0;
 751 }
 752
 753 int compare_commits_by_gen_then_commit_date(const void *a_, const void *b_, void *unused)
 754 {
 755         const struct commit *a = a_, *b = b_;
 756         const uint32_t generation_a = commit_graph_generation(a),
 757                        generation_b = commit_graph_generation(b);
 758
 759         /* newer commits first */
 760         if (generation_a < generation_b)
 761                 return 1;
 762         else if (generation_a > generation_b)
 763                 return -1;
 764
 765         /* use date as a heuristic when generations are equal */
 766         if (a->date < b->date)
 767                 return 1;
 768         else if (a->date > b->date)
 769                 return -1;
 770         return 0;
 771 }
 772
 773 int compare_commits_by_commit_date(const void *a_, const void *b_, void *unused)
 774 {
 775         const struct commit *a = a_, *b = b_;
 776         /* newer commits with larger date first */
 777         if (a->date < b->date)
 778                 return 1;
 779         else if (a->date > b->date)
 780                 return -1;
 781         return 0;
 782 }
 783
 784 /*
 785  * Performs an in-place topological sort on the list supplied.
 786  */
 787 void sort_in_topological_order(struct commit_list **list, enum rev_sort_order sort_order)
 788 {
 789         struct commit_list *next, *orig = *list;
 790         struct commit_list **pptr;
 791         struct indegree_slab indegree;
 792         struct prio_queue queue;
 793         struct commit *commit;
 794         struct author_date_slab author_date;
 795
 796         if (!orig)
 797                 return;
 798         *list = NULL;
 799
 800         init_indegree_slab(&indegree);
 801         memset(&queue, '\0', sizeof(queue));
 802
 803         switch (sort_order) {
 804         default: /* REV_SORT_IN_GRAPH_ORDER */
 805                 queue.compare = NULL;
 806                 break;
 807         case REV_SORT_BY_COMMIT_DATE:
 808                 queue.compare = compare_commits_by_commit_date;
 809                 break;
 810         case REV_SORT_BY_AUTHOR_DATE:
 811                 init_author_date_slab(&author_date);
 812                 queue.compare = compare_commits_by_author_date;
 813                 queue.cb_data = &author_date;
 814                 break;
 815         }
 816
 817         /* Mark them and clear the indegree */
 818         for (next = orig; next; next = next->next) {
 819                 struct commit *commit = next->item;
 820                 *(indegree_slab_at(&indegree, commit)) = 1;
 821                 /* also record the author dates, if needed */
 822                 if (sort_order == REV_SORT_BY_AUTHOR_DATE)
 823                         record_author_date(&author_date, commit);
 824         }
 825
 826         /* update the indegree */
 827         for (next = orig; next; next = next->next) {
 828                 struct commit_list *parents = next->item->parents;
 829                 while (parents) {
 830                         struct commit *parent = parents->item;
 831                         int *pi = indegree_slab_at(&indegree, parent);
 832
 833                         if (*pi)
 834                                 (*pi)++;
 835                         parents = parents->next;
 836                 }
 837         }
 838
 839         /*
 840          * find the tips
 841          *
 842          * tips are nodes not reachable from any other node in the list
 843          *
 844          * the tips serve as a starting set for the work queue.
 845          */
 846         for (next = orig; next; next = next->next) {
 847                 struct commit *commit = next->item;
 848
 849                 if (*(indegree_slab_at(&indegree, commit)) == 1)
 850                         prio_queue_put(&queue, commit);
 851         }
 852
 853         /*
 854          * This is unfortunate; the initial tips need to be shown
 855          * in the order given from the revision traversal machinery.
 856          */
 857         if (sort_order == REV_SORT_IN_GRAPH_ORDER)
 858                 prio_queue_reverse(&queue);
 859
 860         /* We no longer need the commit list */
 861         free_commit_list(orig);
 862
 863         pptr = list;
 864         *list = NULL;
 865         while ((commit = prio_queue_get(&queue)) != NULL) {
 866                 struct commit_list *parents;
 867
 868                 for (parents = commit->parents; parents ; parents = parents->next) {
 869                         struct commit *parent = parents->item;
 870                         int *pi = indegree_slab_at(&indegree, parent);
 871
 872                         if (!*pi)
 873                                 continue;
 874
 875                         /*
 876                          * parents are only enqueued for emission
 877                          * when all their children have been emitted thereby
 878                          * guaranteeing topological order.
 879                          */
 880                         if (--(*pi) == 1)
 881                                 prio_queue_put(&queue, parent);
 882                 }
 883                 /*
 884                  * all children of commit have already been
 885                  * emitted. we can emit it now.
 886                  */
 887                 *(indegree_slab_at(&indegree, commit)) = 0;
 888
 889                 pptr = &commit_list_insert(commit, pptr)->next;
 890         }
 891
 892         clear_indegree_slab(&indegree);
 893         clear_prio_queue(&queue);
 894         if (sort_order == REV_SORT_BY_AUTHOR_DATE)
 895                 clear_author_date_slab(&author_date);
 896 }
 897
 898 struct rev_collect {
 899         struct commit **commit;
 900         int nr;
 901         int alloc;
 902         unsigned int initial : 1;
 903 };
 904
 905 static void add_one_commit(struct object_id *oid, struct rev_collect *revs)
 906 {
 907         struct commit *commit;
 908
 909         if (is_null_oid(oid))
 910                 return;
 911
 912         commit = lookup_commit(the_repository, oid);
 913         if (!commit ||
 914             (commit->object.flags & TMP_MARK) ||
 915             parse_commit(commit))
 916                 return;
 917
 918         ALLOC_GROW(revs->commit, revs->nr + 1, revs->alloc);
 919         revs->commit[revs->nr++] = commit;
 920         commit->object.flags |= TMP_MARK;
 921 }
 922
 923 static int collect_one_reflog_ent(struct object_id *ooid, struct object_id *noid,
 924                                   const char *ident, timestamp_t timestamp,
 925                                   int tz, const char *message, void *cbdata)
 926 {
 927         struct rev_collect *revs = cbdata;
 928
 929         if (revs->initial) {
 930                 revs->initial = 0;
 931                 add_one_commit(ooid, revs);
 932         }
 933         add_one_commit(noid, revs);
 934         return 0;
 935 }
 936
 937 struct commit *get_fork_point(const char *refname, struct commit *commit)
 938 {
 939         struct object_id oid;
 940         struct rev_collect revs;
 941         struct commit_list *bases;
 942         int i;
 943         struct commit *ret = NULL;
 944         char *full_refname;
 945
 946         switch (dwim_ref(refname, strlen(refname), &oid, &full_refname, 0)) {
 947         case 0:
 948                 die("No such ref: '%s'", refname);
 949         case 1:
 950                 break; /* good */
 951         default:
 952                 die("Ambiguous refname: '%s'", refname);
 953         }
 954
 955         memset(&revs, 0, sizeof(revs));
 956         revs.initial = 1;
 957         for_each_reflog_ent(full_refname, collect_one_reflog_ent, &revs);
 958
 959         if (!revs.nr)
 960                 add_one_commit(&oid, &revs);
 961
 962         for (i = 0; i < revs.nr; i++)
 963                 revs.commit[i]->object.flags &= ~TMP_MARK;
 964
 965         bases = get_merge_bases_many(commit, revs.nr, revs.commit);
 966
 967         /*
 968          * There should be one and only one merge base, when we found
 969          * a common ancestor among reflog entries.
 970          */
 971         if (!bases || bases->next)
 972                 goto cleanup_return;
 973
 974         /* And the found one must be one of the reflog entries */
 975         for (i = 0; i < revs.nr; i++)
 976                 if (&bases->item->object == &revs.commit[i]->object)
 977                         break; /* found */
 978         if (revs.nr <= i)
 979                 goto cleanup_return;
 980
 981         ret = bases->item;
 982
 983 cleanup_return:
 984         free_commit_list(bases);
 985         free(full_refname);
 986         return ret;
 987 }
 988
 989 /*
 990  * Indexed by hash algorithm identifier.
 991  */
 992 static const char *gpg_sig_headers[] = {
 993         NULL,
 994         "gpgsig",
 995         "gpgsig-sha256",
 996 };
 997
 998 static int do_sign_commit(struct strbuf *buf, const char *keyid)
 999 {
1000         struct strbuf sig = STRBUF_INIT;
1001         int inspos, copypos;
1002         const char *eoh;
1003         const char *gpg_sig_header = gpg_sig_headers[hash_algo_by_ptr(the_hash_algo)];
1004         int gpg_sig_header_len = strlen(gpg_sig_header);
1005
1006         /* find the end of the header */
1007         eoh = strstr(buf->buf, "\n\n");
1008         if (!eoh)
1009                 inspos = buf->len;
1010         else
1011                 inspos = eoh - buf->buf + 1;
1012
1013         if (!keyid || !*keyid)
1014                 keyid = get_signing_key();
1015         if (sign_buffer(buf, &sig, keyid)) {
1016                 strbuf_release(&sig);
1017                 return -1;
1018         }
1019
1020         for (copypos = 0; sig.buf[copypos]; ) {
1021                 const char *bol = sig.buf + copypos;
1022                 const char *eol = strchrnul(bol, '\n');
1023                 int len = (eol - bol) + !!*eol;
1024
1025                 if (!copypos) {
1026                         strbuf_insert(buf, inspos, gpg_sig_header, gpg_sig_header_len);
1027                         inspos += gpg_sig_header_len;
1028                 }
1029                 strbuf_insertstr(buf, inspos++, " ");
1030                 strbuf_insert(buf, inspos, bol, len);
1031                 inspos += len;
1032                 copypos += len;
1033         }
1034         strbuf_release(&sig);
1035         return 0;
1036 }
1037
1038 int parse_signed_commit(const struct commit *commit,
1039                         struct strbuf *payload, struct strbuf *signature,
1040                         const struct git_hash_algo *algop)
1041 {
1042
1043         unsigned long size;
1044         const char *buffer = get_commit_buffer(commit, &size);
1045         int in_signature = 0, saw_signature = 0, other_signature = 0;
1046         const char *line, *tail, *p;
1047         const char *gpg_sig_header = gpg_sig_headers[hash_algo_by_ptr(algop)];
1048
1049         line = buffer;
1050         tail = buffer + size;
1051         while (line < tail) {
1052                 const char *sig = NULL;
1053                 const char *next = memchr(line, '\n', tail - line);
1054
1055                 next = next ? next + 1 : tail;
1056                 if (in_signature && line[0] == ' ')
1057                         sig = line + 1;
1058                 else if (skip_prefix(line, gpg_sig_header, &p) &&
1059                          *p == ' ') {
1060                         sig = line + strlen(gpg_sig_header) + 1;
1061                         other_signature = 0;
1062                 }
1063                 else if (starts_with(line, "gpgsig"))
1064                         other_signature = 1;
1065                 else if (other_signature && line[0] != ' ')
1066                         other_signature = 0;
1067                 if (sig) {
1068                         strbuf_add(signature, sig, next - sig);
1069                         saw_signature = 1;
1070                         in_signature = 1;
1071                 } else {
1072                         if (*line == '\n')
1073                                 /* dump the whole remainder of the buffer */
1074                                 next = tail;
1075                         if (!other_signature)
1076                                 strbuf_add(payload, line, next - line);
1077                         in_signature = 0;
1078                 }
1079                 line = next;
1080         }
1081         unuse_commit_buffer(commit, buffer);
1082         return saw_signature;
1083 }
1084
1085 int remove_signature(struct strbuf *buf)
1086 {
1087         const char *line = buf->buf;
1088         const char *tail = buf->buf + buf->len;
1089         int in_signature = 0;
1090         struct sigbuf {
1091                 const char *start;
1092                 const char *end;
1093         } sigs[2], *sigp = &sigs[0];
1094         int i;
1095         const char *orig_buf = buf->buf;
1096
1097         memset(sigs, 0, sizeof(sigs));
1098
1099         while (line < tail) {
1100                 const char *next = memchr(line, '\n', tail - line);
1101                 next = next ? next + 1 : tail;
1102
1103                 if (in_signature && line[0] == ' ')
1104                         sigp->end = next;
1105                 else if (starts_with(line, "gpgsig")) {
1106                         int i;
1107                         for (i = 1; i < GIT_HASH_NALGOS; i++) {
1108                                 const char *p;
1109                                 if (skip_prefix(line, gpg_sig_headers[i], &p) &&
1110                                     *p == ' ') {
1111                                         sigp->start = line;
1112                                         sigp->end = next;
1113                                         in_signature = 1;
1114                                 }
1115                         }
1116                 } else {
1117                         if (*line == '\n')
1118                                 /* dump the whole remainder of the buffer */
1119                                 next = tail;
1120                         if (in_signature && sigp - sigs != ARRAY_SIZE(sigs))
1121                                 sigp++;
1122                         in_signature = 0;
1123                 }
1124                 line = next;
1125         }
1126
1127         for (i = ARRAY_SIZE(sigs) - 1; i >= 0; i--)
1128                 if (sigs[i].start)
1129                         strbuf_remove(buf, sigs[i].start - orig_buf, sigs[i].end - sigs[i].start);
1130
1131         return sigs[0].start != NULL;
1132 }
1133
1134 static void handle_signed_tag(struct commit *parent, struct commit_extra_header ***tail)
1135 {
1136         struct merge_remote_desc *desc;
1137         struct commit_extra_header *mergetag;
1138         char *buf;
1139         unsigned long size;
1140         enum object_type type;
1141         struct strbuf payload = STRBUF_INIT;
1142         struct strbuf signature = STRBUF_INIT;
1143
1144         desc = merge_remote_util(parent);
1145         if (!desc || !desc->obj)
1146                 return;
1147         buf = read_object_file(&desc->obj->oid, &type, &size);
1148         if (!buf || type != OBJ_TAG)
1149                 goto free_return;
1150         if (!parse_signature(buf, size, &payload, &signature))
1151                 goto free_return;
1152         /*
1153          * We could verify this signature and either omit the tag when
1154          * it does not validate, but the integrator may not have the
1155          * public key of the signer of the tag he is merging, while a
1156          * later auditor may have it while auditing, so let's not run
1157          * verify-signed-buffer here for now...
1158          *
1159          * if (verify_signed_buffer(buf, len, buf + len, size - len, ...))
1160          *      warn("warning: signed tag unverified.");
1161          */
1162         mergetag = xcalloc(1, sizeof(*mergetag));
1163         mergetag->key = xstrdup("mergetag");
1164         mergetag->value = buf;
1165         mergetag->len = size;
1166
1167         **tail = mergetag;
1168         *tail = &mergetag->next;
1169         strbuf_release(&payload);
1170         strbuf_release(&signature);
1171         return;
1172
1173 free_return:
1174         free(buf);
1175 }
1176
1177 int check_commit_signature(const struct commit *commit, struct signature_check *sigc)
1178 {
1179         struct strbuf payload = STRBUF_INIT;
1180         struct strbuf signature = STRBUF_INIT;
1181         int ret = 1;
1182
1183         sigc->result = 'N';
1184
1185         if (parse_signed_commit(commit, &payload, &signature, the_hash_algo) <= 0)
1186                 goto out;
1187         ret = check_signature(payload.buf, payload.len, signature.buf,
1188                 signature.len, sigc);
1189
1190  out:
1191         strbuf_release(&payload);
1192         strbuf_release(&signature);
1193
1194         return ret;
1195 }
1196
1197 void verify_merge_signature(struct commit *commit, int verbosity,
1198                             int check_trust)
1199 {
1200         char hex[GIT_MAX_HEXSZ + 1];
1201         struct signature_check signature_check;
1202         int ret;
1203         memset(&signature_check, 0, sizeof(signature_check));
1204
1205         ret = check_commit_signature(commit, &signature_check);
1206
1207         find_unique_abbrev_r(hex, &commit->object.oid, DEFAULT_ABBREV);
1208         switch (signature_check.result) {
1209         case 'G':
1210                 if (ret || (check_trust && signature_check.trust_level < TRUST_MARGINAL))
1211                         die(_("Commit %s has an untrusted GPG signature, "
1212                               "allegedly by %s."), hex, signature_check.signer);
1213                 break;
1214         case 'B':
1215                 die(_("Commit %s has a bad GPG signature "
1216                       "allegedly by %s."), hex, signature_check.signer);
1217         default: /* 'N' */
1218                 die(_("Commit %s does not have a GPG signature."), hex);
1219         }
1220         if (verbosity >= 0 && signature_check.result == 'G')
1221                 printf(_("Commit %s has a good GPG signature by %s\n"),
1222                        hex, signature_check.signer);
1223
1224         signature_check_clear(&signature_check);
1225 }
1226
1227 void append_merge_tag_headers(struct commit_list *parents,
1228                               struct commit_extra_header ***tail)
1229 {
1230         while (parents) {
1231                 struct commit *parent = parents->item;
1232                 handle_signed_tag(parent, tail);
1233                 parents = parents->next;
1234         }
1235 }
1236
1237 static void add_extra_header(struct strbuf *buffer,
1238                              struct commit_extra_header *extra)
1239 {
1240         strbuf_addstr(buffer, extra->key);
1241         if (extra->len)
1242                 strbuf_add_lines(buffer, " ", extra->value, extra->len);
1243         else
1244                 strbuf_addch(buffer, '\n');
1245 }
1246
1247 struct commit_extra_header *read_commit_extra_headers(struct commit *commit,
1248                                                       const char **exclude)
1249 {
1250         struct commit_extra_header *extra = NULL;
1251         unsigned long size;
1252         const char *buffer = get_commit_buffer(commit, &size);
1253         extra = read_commit_extra_header_lines(buffer, size, exclude);
1254         unuse_commit_buffer(commit, buffer);
1255         return extra;
1256 }
1257
1258 int for_each_mergetag(each_mergetag_fn fn, struct commit *commit, void *data)
1259 {
1260         struct commit_extra_header *extra, *to_free;
1261         int res = 0;
1262
1263         to_free = read_commit_extra_headers(commit, NULL);
1264         for (extra = to_free; !res && extra; extra = extra->next) {
1265                 if (strcmp(extra->key, "mergetag"))
1266                         continue; /* not a merge tag */
1267                 res = fn(commit, extra, data);
1268         }
1269         free_commit_extra_headers(to_free);
1270         return res;
1271 }
1272
1273 static inline int standard_header_field(const char *field, size_t len)
1274 {
1275         return ((len == 4 && !memcmp(field, "tree", 4)) ||
1276                 (len == 6 && !memcmp(field, "parent", 6)) ||
1277                 (len == 6 && !memcmp(field, "author", 6)) ||
1278                 (len == 9 && !memcmp(field, "committer", 9)) ||
1279                 (len == 8 && !memcmp(field, "encoding", 8)));
1280 }
1281
1282 static int excluded_header_field(const char *field, size_t len, const char **exclude)
1283 {
1284         if (!exclude)
1285                 return 0;
1286
1287         while (*exclude) {
1288                 size_t xlen = strlen(*exclude);
1289                 if (len == xlen && !memcmp(field, *exclude, xlen))
1290                         return 1;
1291                 exclude++;
1292         }
1293         return 0;
1294 }
1295
1296 static struct commit_extra_header *read_commit_extra_header_lines(
1297         const char *buffer, size_t size,
1298         const char **exclude)
1299 {
1300         struct commit_extra_header *extra = NULL, **tail = &extra, *it = NULL;
1301         const char *line, *next, *eof, *eob;
1302         struct strbuf buf = STRBUF_INIT;
1303
1304         for (line = buffer, eob = line + size;
1305              line < eob && *line != '\n';
1306              line = next) {
1307                 next = memchr(line, '\n', eob - line);
1308                 next = next ? next + 1 : eob;
1309                 if (*line == ' ') {
1310                         /* continuation */
1311                         if (it)
1312                                 strbuf_add(&buf, line + 1, next - (line + 1));
1313                         continue;
1314                 }
1315                 if (it)
1316                         it->value = strbuf_detach(&buf, &it->len);
1317                 strbuf_reset(&buf);
1318                 it = NULL;
1319
1320                 eof = memchr(line, ' ', next - line);
1321                 if (!eof)
1322                         eof = next;
1323                 else if (standard_header_field(line, eof - line) ||
1324                          excluded_header_field(line, eof - line, exclude))
1325                         continue;
1326
1327                 it = xcalloc(1, sizeof(*it));
1328                 it->key = xmemdupz(line, eof-line);
1329                 *tail = it;
1330                 tail = &it->next;
1331                 if (eof + 1 < next)
1332                         strbuf_add(&buf, eof + 1, next - (eof + 1));
1333         }
1334         if (it)
1335                 it->value = strbuf_detach(&buf, &it->len);
1336         return extra;
1337 }
1338
1339 void free_commit_extra_headers(struct commit_extra_header *extra)
1340 {
1341         while (extra) {
1342                 struct commit_extra_header *next = extra->next;
1343                 free(extra->key);
1344                 free(extra->value);
1345                 free(extra);
1346                 extra = next;
1347         }
1348 }
1349
1350 int commit_tree(const char *msg, size_t msg_len, const struct object_id *tree,
1351                 struct commit_list *parents, struct object_id *ret,
1352                 const char *author, const char *sign_commit)
1353 {
1354         struct commit_extra_header *extra = NULL, **tail = &extra;
1355         int result;
1356
1357         append_merge_tag_headers(parents, &tail);
1358         result = commit_tree_extended(msg, msg_len, tree, parents, ret, author,
1359                                       NULL, sign_commit, extra);
1360         free_commit_extra_headers(extra);
1361         return result;
1362 }
1363
1364 static int find_invalid_utf8(const char *buf, int len)
1365 {
1366         int offset = 0;
1367         static const unsigned int max_codepoint[] = {
1368                 0x7f, 0x7ff, 0xffff, 0x10ffff
1369         };
1370
1371         while (len) {
1372                 unsigned char c = *buf++;
1373                 int bytes, bad_offset;
1374                 unsigned int codepoint;
1375                 unsigned int min_val, max_val;
1376
1377                 len--;
1378                 offset++;
1379
1380                 /* Simple US-ASCII? No worries. */
1381                 if (c < 0x80)
1382                         continue;
1383
1384                 bad_offset = offset-1;
1385
1386                 /*
1387                  * Count how many more high bits set: that's how
1388                  * many more bytes this sequence should have.
1389                  */
1390                 bytes = 0;
1391                 while (c & 0x40) {
1392                         c <<= 1;
1393                         bytes++;
1394                 }
1395
1396                 /*
1397                  * Must be between 1 and 3 more bytes.  Longer sequences result in
1398                  * codepoints beyond U+10FFFF, which are guaranteed never to exist.
1399                  */
1400                 if (bytes < 1 || 3 < bytes)
1401                         return bad_offset;
1402
1403                 /* Do we *have* that many bytes? */
1404                 if (len < bytes)
1405                         return bad_offset;
1406
1407                 /*
1408                  * Place the encoded bits at the bottom of the value and compute the
1409                  * valid range.
1410                  */
1411                 codepoint = (c & 0x7f) >> bytes;
1412                 min_val = max_codepoint[bytes-1] + 1;
1413                 max_val = max_codepoint[bytes];
1414
1415                 offset += bytes;
1416                 len -= bytes;
1417
1418                 /* And verify that they are good continuation bytes */
1419                 do {
1420                         codepoint <<= 6;
1421                         codepoint |= *buf & 0x3f;
1422                         if ((*buf++ & 0xc0) != 0x80)
1423                                 return bad_offset;
1424                 } while (--bytes);
1425
1426                 /* Reject codepoints that are out of range for the sequence length. */
1427                 if (codepoint < min_val || codepoint > max_val)
1428                         return bad_offset;
1429                 /* Surrogates are only for UTF-16 and cannot be encoded in UTF-8. */
1430                 if ((codepoint & 0x1ff800) == 0xd800)
1431                         return bad_offset;
1432                 /* U+xxFFFE and U+xxFFFF are guaranteed non-characters. */
1433                 if ((codepoint & 0xfffe) == 0xfffe)
1434                         return bad_offset;
1435                 /* So are anything in the range U+FDD0..U+FDEF. */
1436                 if (codepoint >= 0xfdd0 && codepoint <= 0xfdef)
1437                         return bad_offset;
1438         }
1439         return -1;
1440 }
1441
1442 /*
1443  * This verifies that the buffer is in proper utf8 format.
1444  *
1445  * If it isn't, it assumes any non-utf8 characters are Latin1,
1446  * and does the conversion.
1447  */
1448 static int verify_utf8(struct strbuf *buf)
1449 {
1450         int ok = 1;
1451         long pos = 0;
1452
1453         for (;;) {
1454                 int bad;
1455                 unsigned char c;
1456                 unsigned char replace[2];
1457
1458                 bad = find_invalid_utf8(buf->buf + pos, buf->len - pos);
1459                 if (bad < 0)
1460                         return ok;
1461                 pos += bad;
1462                 ok = 0;
1463                 c = buf->buf[pos];
1464                 strbuf_remove(buf, pos, 1);
1465
1466                 /* We know 'c' must be in the range 128-255 */
1467                 replace[0] = 0xc0 + (c >> 6);
1468                 replace[1] = 0x80 + (c & 0x3f);
1469                 strbuf_insert(buf, pos, replace, 2);
1470                 pos += 2;
1471         }
1472 }
1473
1474 static const char commit_utf8_warn[] =
1475 N_("Warning: commit message did not conform to UTF-8.\n"
1476    "You may want to amend it after fixing the message, or set the config\n"
1477    "variable i18n.commitencoding to the encoding your project uses.\n");
1478
1479 int commit_tree_extended(const char *msg, size_t msg_len,
1480                          const struct object_id *tree,
1481                          struct commit_list *parents, struct object_id *ret,
1482                          const char *author, const char *committer,
1483                          const char *sign_commit,
1484                          struct commit_extra_header *extra)
1485 {
1486         int result;
1487         int encoding_is_utf8;
1488         struct strbuf buffer;
1489
1490         assert_oid_type(tree, OBJ_TREE);
1491
1492         if (memchr(msg, '\0', msg_len))
1493                 return error("a NUL byte in commit log message not allowed.");
1494
1495         /* Not having i18n.commitencoding is the same as having utf-8 */
1496         encoding_is_utf8 = is_encoding_utf8(git_commit_encoding);
1497
1498         strbuf_init(&buffer, 8192); /* should avoid reallocs for the headers */
1499         strbuf_addf(&buffer, "tree %s\n", oid_to_hex(tree));
1500
1501         /*
1502          * NOTE! This ordering means that the same exact tree merged with a
1503          * different order of parents will be a _different_ changeset even
1504          * if everything else stays the same.
1505          */
1506         while (parents) {
1507                 struct commit *parent = pop_commit(&parents);
1508                 strbuf_addf(&buffer, "parent %s\n",
1509                             oid_to_hex(&parent->object.oid));
1510         }
1511
1512         /* Person/date information */
1513         if (!author)
1514                 author = git_author_info(IDENT_STRICT);
1515         strbuf_addf(&buffer, "author %s\n", author);
1516         if (!committer)
1517                 committer = git_committer_info(IDENT_STRICT);
1518         strbuf_addf(&buffer, "committer %s\n", committer);
1519         if (!encoding_is_utf8)
1520                 strbuf_addf(&buffer, "encoding %s\n", git_commit_encoding);
1521
1522         while (extra) {
1523                 add_extra_header(&buffer, extra);
1524                 extra = extra->next;
1525         }
1526         strbuf_addch(&buffer, '\n');
1527
1528         /* And add the comment */
1529         strbuf_add(&buffer, msg, msg_len);
1530
1531         /* And check the encoding */
1532         if (encoding_is_utf8 && !verify_utf8(&buffer))
1533                 fprintf(stderr, _(commit_utf8_warn));
1534
1535         if (sign_commit && do_sign_commit(&buffer, sign_commit)) {
1536                 result = -1;
1537                 goto out;
1538         }
1539
1540         result = write_object_file(buffer.buf, buffer.len, commit_type, ret);
1541 out:
1542         strbuf_release(&buffer);
1543         return result;
1544 }
1545
1546 define_commit_slab(merge_desc_slab, struct merge_remote_desc *);
1547 static struct merge_desc_slab merge_desc_slab = COMMIT_SLAB_INIT(1, merge_desc_slab);
1548
1549 struct merge_remote_desc *merge_remote_util(struct commit *commit)
1550 {
1551         return *merge_desc_slab_at(&merge_desc_slab, commit);
1552 }
1553
1554 void set_merge_remote_desc(struct commit *commit,
1555                            const char *name, struct object *obj)
1556 {
1557         struct merge_remote_desc *desc;
1558         FLEX_ALLOC_STR(desc, name, name);
1559         desc->obj = obj;
1560         *merge_desc_slab_at(&merge_desc_slab, commit) = desc;
1561 }
1562
1563 struct commit *get_merge_parent(const char *name)
1564 {
1565         struct object *obj;
1566         struct commit *commit;
1567         struct object_id oid;
1568         if (get_oid(name, &oid))
1569                 return NULL;
1570         obj = parse_object(the_repository, &oid);
1571         commit = (struct commit *)peel_to_type(name, 0, obj, OBJ_COMMIT);
1572         if (commit && !merge_remote_util(commit))
1573                 set_merge_remote_desc(commit, name, obj);
1574         return commit;
1575 }
1576
1577 /*
1578  * Append a commit to the end of the commit_list.
1579  *
1580  * next starts by pointing to the variable that holds the head of an
1581  * empty commit_list, and is updated to point to the "next" field of
1582  * the last item on the list as new commits are appended.
1583  *
1584  * Usage example:
1585  *
1586  *     struct commit_list *list;
1587  *     struct commit_list **next = &list;
1588  *
1589  *     next = commit_list_append(c1, next);
1590  *     next = commit_list_append(c2, next);
1591  *     assert(commit_list_count(list) == 2);
1592  *     return list;
1593  */
1594 struct commit_list **commit_list_append(struct commit *commit,
1595                                         struct commit_list **next)
1596 {
1597         struct commit_list *new_commit = xmalloc(sizeof(struct commit_list));
1598         new_commit->item = commit;
1599         *next = new_commit;
1600         new_commit->next = NULL;
1601         return &new_commit->next;
1602 }
1603
1604 const char *find_commit_header(const char *msg, const char *key, size_t *out_len)
1605 {
1606         int key_len = strlen(key);
1607         const char *line = msg;
1608
1609         while (line) {
1610                 const char *eol = strchrnul(line, '\n');
1611
1612                 if (line == eol)
1613                         return NULL;
1614
1615                 if (eol - line > key_len &&
1616                     !strncmp(line, key, key_len) &&
1617                     line[key_len] == ' ') {
1618                         *out_len = eol - line - key_len - 1;
1619                         return line + key_len + 1;
1620                 }
1621                 line = *eol ? eol + 1 : NULL;
1622         }
1623         return NULL;
1624 }
1625
1626 /*
1627  * Inspect the given string and determine the true "end" of the log message, in
1628  * order to find where to put a new Signed-off-by trailer.  Ignored are
1629  * trailing comment lines and blank lines.  To support "git commit -s
1630  * --amend" on an existing commit, we also ignore "Conflicts:".  To
1631  * support "git commit -v", we truncate at cut lines.
1632  *
1633  * Returns the number of bytes from the tail to ignore, to be fed as
1634  * the second parameter to append_signoff().
1635  */
1636 size_t ignore_non_trailer(const char *buf, size_t len)
1637 {
1638         size_t boc = 0;
1639         size_t bol = 0;
1640         int in_old_conflicts_block = 0;
1641         size_t cutoff = wt_status_locate_end(buf, len);
1642
1643         while (bol < cutoff) {
1644                 const char *next_line = memchr(buf + bol, '\n', len - bol);
1645
1646                 if (!next_line)
1647                         next_line = buf + len;
1648                 else
1649                         next_line++;
1650
1651                 if (buf[bol] == comment_line_char || buf[bol] == '\n') {
1652                         /* is this the first of the run of comments? */
1653                         if (!boc)
1654                                 boc = bol;
1655                         /* otherwise, it is just continuing */
1656                 } else if (starts_with(buf + bol, "Conflicts:\n")) {
1657                         in_old_conflicts_block = 1;
1658                         if (!boc)
1659                                 boc = bol;
1660                 } else if (in_old_conflicts_block && buf[bol] == '\t') {
1661                         ; /* a pathname in the conflicts block */
1662                 } else if (boc) {
1663                         /* the previous was not trailing comment */
1664                         boc = 0;
1665                         in_old_conflicts_block = 0;
1666                 }
1667                 bol = next_line - buf;
1668         }
1669         return boc ? len - boc : len - cutoff;
1670 }
1671
1672 int run_commit_hook(int editor_is_used, const char *index_file,
1673                     const char *name, ...)
1674 {
1675         struct strvec hook_env = STRVEC_INIT;
1676         va_list args;
1677         int ret;
1678
1679         strvec_pushf(&hook_env, "GIT_INDEX_FILE=%s", index_file);
1680
1681         /*
1682          * Let the hook know that no editor will be launched.
1683          */
1684         if (!editor_is_used)
1685                 strvec_push(&hook_env, "GIT_EDITOR=:");
1686
1687         va_start(args, name);
1688         ret = run_hook_ve(hook_env.v, name, args);
1689         va_end(args);
1690         strvec_clear(&hook_env);
1691
1692         return ret;
1693 }