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 "hash-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 struct object_id *commit_graft_oid_access(size_t index, const void *table)
 109 {
 110         const struct commit_graft * const *commit_graft_table = table;
 111         return &commit_graft_table[index]->oid;
 112 }
 113
 114 int commit_graft_pos(struct repository *r, const struct object_id *oid)
 115 {
 116         return oid_pos(oid, r->parsed_objects->grafts,
 117                        r->parsed_objects->grafts_nr,
 118                        commit_graft_oid_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);
 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);
 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 size_t commit_subject_length(const char *body)
 539 {
 540         const char *p = body;
 541         while (*p) {
 542                 const char *next = skip_blank_lines(p);
 543                 if (next != p)
 544                         break;
 545                 p = strchrnul(p, '\n');
 546                 if (*p)
 547                         p++;
 548         }
 549         return p - body;
 550 }
 551
 552 struct commit_list *commit_list_insert(struct commit *item, struct commit_list **list_p)
 553 {
 554         struct commit_list *new_list = xmalloc(sizeof(struct commit_list));
 555         new_list->item = item;
 556         new_list->next = *list_p;
 557         *list_p = new_list;
 558         return new_list;
 559 }
 560
 561 int commit_list_contains(struct commit *item, struct commit_list *list)
 562 {
 563         while (list) {
 564                 if (list->item == item)
 565                         return 1;
 566                 list = list->next;
 567         }
 568
 569         return 0;
 570 }
 571
 572 unsigned commit_list_count(const struct commit_list *l)
 573 {
 574         unsigned c = 0;
 575         for (; l; l = l->next )
 576                 c++;
 577         return c;
 578 }
 579
 580 struct commit_list *copy_commit_list(struct commit_list *list)
 581 {
 582         struct commit_list *head = NULL;
 583         struct commit_list **pp = &head;
 584         while (list) {
 585                 pp = commit_list_append(list->item, pp);
 586                 list = list->next;
 587         }
 588         return head;
 589 }
 590
 591 struct commit_list *reverse_commit_list(struct commit_list *list)
 592 {
 593         struct commit_list *next = NULL, *current, *backup;
 594         for (current = list; current; current = backup) {
 595                 backup = current->next;
 596                 current->next = next;
 597                 next = current;
 598         }
 599         return next;
 600 }
 601
 602 void free_commit_list(struct commit_list *list)
 603 {
 604         while (list)
 605                 pop_commit(&list);
 606 }
 607
 608 struct commit_list * commit_list_insert_by_date(struct commit *item, struct commit_list **list)
 609 {
 610         struct commit_list **pp = list;
 611         struct commit_list *p;
 612         while ((p = *pp) != NULL) {
 613                 if (p->item->date < item->date) {
 614                         break;
 615                 }
 616                 pp = &p->next;
 617         }
 618         return commit_list_insert(item, pp);
 619 }
 620
 621 static int commit_list_compare_by_date(const void *a, const void *b)
 622 {
 623         timestamp_t a_date = ((const struct commit_list *)a)->item->date;
 624         timestamp_t b_date = ((const struct commit_list *)b)->item->date;
 625         if (a_date < b_date)
 626                 return 1;
 627         if (a_date > b_date)
 628                 return -1;
 629         return 0;
 630 }
 631
 632 static void *commit_list_get_next(const void *a)
 633 {
 634         return ((const struct commit_list *)a)->next;
 635 }
 636
 637 static void commit_list_set_next(void *a, void *next)
 638 {
 639         ((struct commit_list *)a)->next = next;
 640 }
 641
 642 void commit_list_sort_by_date(struct commit_list **list)
 643 {
 644         *list = llist_mergesort(*list, commit_list_get_next, commit_list_set_next,
 645                                 commit_list_compare_by_date);
 646 }
 647
 648 struct commit *pop_most_recent_commit(struct commit_list **list,
 649                                       unsigned int mark)
 650 {
 651         struct commit *ret = pop_commit(list);
 652         struct commit_list *parents = ret->parents;
 653
 654         while (parents) {
 655                 struct commit *commit = parents->item;
 656                 if (!parse_commit(commit) && !(commit->object.flags & mark)) {
 657                         commit->object.flags |= mark;
 658                         commit_list_insert_by_date(commit, list);
 659                 }
 660                 parents = parents->next;
 661         }
 662         return ret;
 663 }
 664
 665 static void clear_commit_marks_1(struct commit_list **plist,
 666                                  struct commit *commit, unsigned int mark)
 667 {
 668         while (commit) {
 669                 struct commit_list *parents;
 670
 671                 if (!(mark & commit->object.flags))
 672                         return;
 673
 674                 commit->object.flags &= ~mark;
 675
 676                 parents = commit->parents;
 677                 if (!parents)
 678                         return;
 679
 680                 while ((parents = parents->next))
 681                         commit_list_insert(parents->item, plist);
 682
 683                 commit = commit->parents->item;
 684         }
 685 }
 686
 687 void clear_commit_marks_many(int nr, struct commit **commit, unsigned int mark)
 688 {
 689         struct commit_list *list = NULL;
 690
 691         while (nr--) {
 692                 clear_commit_marks_1(&list, *commit, mark);
 693                 commit++;
 694         }
 695         while (list)
 696                 clear_commit_marks_1(&list, pop_commit(&list), mark);
 697 }
 698
 699 void clear_commit_marks(struct commit *commit, unsigned int mark)
 700 {
 701         clear_commit_marks_many(1, &commit, mark);
 702 }
 703
 704 struct commit *pop_commit(struct commit_list **stack)
 705 {
 706         struct commit_list *top = *stack;
 707         struct commit *item = top ? top->item : NULL;
 708
 709         if (top) {
 710                 *stack = top->next;
 711                 free(top);
 712         }
 713         return item;
 714 }
 715
 716 /*
 717  * Topological sort support
 718  */
 719
 720 /* count number of children that have not been emitted */
 721 define_commit_slab(indegree_slab, int);
 722
 723 define_commit_slab(author_date_slab, timestamp_t);
 724
 725 void record_author_date(struct author_date_slab *author_date,
 726                         struct commit *commit)
 727 {
 728         const char *buffer = get_commit_buffer(commit, NULL);
 729         struct ident_split ident;
 730         const char *ident_line;
 731         size_t ident_len;
 732         char *date_end;
 733         timestamp_t date;
 734
 735         ident_line = find_commit_header(buffer, "author", &ident_len);
 736         if (!ident_line)
 737                 goto fail_exit; /* no author line */
 738         if (split_ident_line(&ident, ident_line, ident_len) ||
 739             !ident.date_begin || !ident.date_end)
 740                 goto fail_exit; /* malformed "author" line */
 741
 742         date = parse_timestamp(ident.date_begin, &date_end, 10);
 743         if (date_end != ident.date_end)
 744                 goto fail_exit; /* malformed date */
 745         *(author_date_slab_at(author_date, commit)) = date;
 746
 747 fail_exit:
 748         unuse_commit_buffer(commit, buffer);
 749 }
 750
 751 int compare_commits_by_author_date(const void *a_, const void *b_,
 752                                    void *cb_data)
 753 {
 754         const struct commit *a = a_, *b = b_;
 755         struct author_date_slab *author_date = cb_data;
 756         timestamp_t a_date = *(author_date_slab_at(author_date, a));
 757         timestamp_t b_date = *(author_date_slab_at(author_date, b));
 758
 759         /* newer commits with larger date first */
 760         if (a_date < b_date)
 761                 return 1;
 762         else if (a_date > b_date)
 763                 return -1;
 764         return 0;
 765 }
 766
 767 int compare_commits_by_gen_then_commit_date(const void *a_, const void *b_, void *unused)
 768 {
 769         const struct commit *a = a_, *b = b_;
 770         const timestamp_t generation_a = commit_graph_generation(a),
 771                           generation_b = commit_graph_generation(b);
 772
 773         /* newer commits first */
 774         if (generation_a < generation_b)
 775                 return 1;
 776         else if (generation_a > generation_b)
 777                 return -1;
 778
 779         /* use date as a heuristic when generations are equal */
 780         if (a->date < b->date)
 781                 return 1;
 782         else if (a->date > b->date)
 783                 return -1;
 784         return 0;
 785 }
 786
 787 int compare_commits_by_commit_date(const void *a_, const void *b_, void *unused)
 788 {
 789         const struct commit *a = a_, *b = b_;
 790         /* newer commits with larger date first */
 791         if (a->date < b->date)
 792                 return 1;
 793         else if (a->date > b->date)
 794                 return -1;
 795         return 0;
 796 }
 797
 798 /*
 799  * Performs an in-place topological sort on the list supplied.
 800  */
 801 void sort_in_topological_order(struct commit_list **list, enum rev_sort_order sort_order)
 802 {
 803         struct commit_list *next, *orig = *list;
 804         struct commit_list **pptr;
 805         struct indegree_slab indegree;
 806         struct prio_queue queue;
 807         struct commit *commit;
 808         struct author_date_slab author_date;
 809
 810         if (!orig)
 811                 return;
 812         *list = NULL;
 813
 814         init_indegree_slab(&indegree);
 815         memset(&queue, '\0', sizeof(queue));
 816
 817         switch (sort_order) {
 818         default: /* REV_SORT_IN_GRAPH_ORDER */
 819                 queue.compare = NULL;
 820                 break;
 821         case REV_SORT_BY_COMMIT_DATE:
 822                 queue.compare = compare_commits_by_commit_date;
 823                 break;
 824         case REV_SORT_BY_AUTHOR_DATE:
 825                 init_author_date_slab(&author_date);
 826                 queue.compare = compare_commits_by_author_date;
 827                 queue.cb_data = &author_date;
 828                 break;
 829         }
 830
 831         /* Mark them and clear the indegree */
 832         for (next = orig; next; next = next->next) {
 833                 struct commit *commit = next->item;
 834                 *(indegree_slab_at(&indegree, commit)) = 1;
 835                 /* also record the author dates, if needed */
 836                 if (sort_order == REV_SORT_BY_AUTHOR_DATE)
 837                         record_author_date(&author_date, commit);
 838         }
 839
 840         /* update the indegree */
 841         for (next = orig; next; next = next->next) {
 842                 struct commit_list *parents = next->item->parents;
 843                 while (parents) {
 844                         struct commit *parent = parents->item;
 845                         int *pi = indegree_slab_at(&indegree, parent);
 846
 847                         if (*pi)
 848                                 (*pi)++;
 849                         parents = parents->next;
 850                 }
 851         }
 852
 853         /*
 854          * find the tips
 855          *
 856          * tips are nodes not reachable from any other node in the list
 857          *
 858          * the tips serve as a starting set for the work queue.
 859          */
 860         for (next = orig; next; next = next->next) {
 861                 struct commit *commit = next->item;
 862
 863                 if (*(indegree_slab_at(&indegree, commit)) == 1)
 864                         prio_queue_put(&queue, commit);
 865         }
 866
 867         /*
 868          * This is unfortunate; the initial tips need to be shown
 869          * in the order given from the revision traversal machinery.
 870          */
 871         if (sort_order == REV_SORT_IN_GRAPH_ORDER)
 872                 prio_queue_reverse(&queue);
 873
 874         /* We no longer need the commit list */
 875         free_commit_list(orig);
 876
 877         pptr = list;
 878         *list = NULL;
 879         while ((commit = prio_queue_get(&queue)) != NULL) {
 880                 struct commit_list *parents;
 881
 882                 for (parents = commit->parents; parents ; parents = parents->next) {
 883                         struct commit *parent = parents->item;
 884                         int *pi = indegree_slab_at(&indegree, parent);
 885
 886                         if (!*pi)
 887                                 continue;
 888
 889                         /*
 890                          * parents are only enqueued for emission
 891                          * when all their children have been emitted thereby
 892                          * guaranteeing topological order.
 893                          */
 894                         if (--(*pi) == 1)
 895                                 prio_queue_put(&queue, parent);
 896                 }
 897                 /*
 898                  * all children of commit have already been
 899                  * emitted. we can emit it now.
 900                  */
 901                 *(indegree_slab_at(&indegree, commit)) = 0;
 902
 903                 pptr = &commit_list_insert(commit, pptr)->next;
 904         }
 905
 906         clear_indegree_slab(&indegree);
 907         clear_prio_queue(&queue);
 908         if (sort_order == REV_SORT_BY_AUTHOR_DATE)
 909                 clear_author_date_slab(&author_date);
 910 }
 911
 912 struct rev_collect {
 913         struct commit **commit;
 914         int nr;
 915         int alloc;
 916         unsigned int initial : 1;
 917 };
 918
 919 static void add_one_commit(struct object_id *oid, struct rev_collect *revs)
 920 {
 921         struct commit *commit;
 922
 923         if (is_null_oid(oid))
 924                 return;
 925
 926         commit = lookup_commit(the_repository, oid);
 927         if (!commit ||
 928             (commit->object.flags & TMP_MARK) ||
 929             parse_commit(commit))
 930                 return;
 931
 932         ALLOC_GROW(revs->commit, revs->nr + 1, revs->alloc);
 933         revs->commit[revs->nr++] = commit;
 934         commit->object.flags |= TMP_MARK;
 935 }
 936
 937 static int collect_one_reflog_ent(struct object_id *ooid, struct object_id *noid,
 938                                   const char *ident, timestamp_t timestamp,
 939                                   int tz, const char *message, void *cbdata)
 940 {
 941         struct rev_collect *revs = cbdata;
 942
 943         if (revs->initial) {
 944                 revs->initial = 0;
 945                 add_one_commit(ooid, revs);
 946         }
 947         add_one_commit(noid, revs);
 948         return 0;
 949 }
 950
 951 struct commit *get_fork_point(const char *refname, struct commit *commit)
 952 {
 953         struct object_id oid;
 954         struct rev_collect revs;
 955         struct commit_list *bases;
 956         int i;
 957         struct commit *ret = NULL;
 958         char *full_refname;
 959
 960         switch (dwim_ref(refname, strlen(refname), &oid, &full_refname, 0)) {
 961         case 0:
 962                 die("No such ref: '%s'", refname);
 963         case 1:
 964                 break; /* good */
 965         default:
 966                 die("Ambiguous refname: '%s'", refname);
 967         }
 968
 969         memset(&revs, 0, sizeof(revs));
 970         revs.initial = 1;
 971         for_each_reflog_ent(full_refname, collect_one_reflog_ent, &revs);
 972
 973         if (!revs.nr)
 974                 add_one_commit(&oid, &revs);
 975
 976         for (i = 0; i < revs.nr; i++)
 977                 revs.commit[i]->object.flags &= ~TMP_MARK;
 978
 979         bases = get_merge_bases_many(commit, revs.nr, revs.commit);
 980
 981         /*
 982          * There should be one and only one merge base, when we found
 983          * a common ancestor among reflog entries.
 984          */
 985         if (!bases || bases->next)
 986                 goto cleanup_return;
 987
 988         /* And the found one must be one of the reflog entries */
 989         for (i = 0; i < revs.nr; i++)
 990                 if (&bases->item->object == &revs.commit[i]->object)
 991                         break; /* found */
 992         if (revs.nr <= i)
 993                 goto cleanup_return;
 994
 995         ret = bases->item;
 996
 997 cleanup_return:
 998         free_commit_list(bases);
 999         free(full_refname);
1000         return ret;
1001 }
1002
1003 /*
1004  * Indexed by hash algorithm identifier.
1005  */
1006 static const char *gpg_sig_headers[] = {
1007         NULL,
1008         "gpgsig",
1009         "gpgsig-sha256",
1010 };
1011
1012 int sign_with_header(struct strbuf *buf, const char *keyid)
1013 {
1014         struct strbuf sig = STRBUF_INIT;
1015         int inspos, copypos;
1016         const char *eoh;
1017         const char *gpg_sig_header = gpg_sig_headers[hash_algo_by_ptr(the_hash_algo)];
1018         int gpg_sig_header_len = strlen(gpg_sig_header);
1019
1020         /* find the end of the header */
1021         eoh = strstr(buf->buf, "\n\n");
1022         if (!eoh)
1023                 inspos = buf->len;
1024         else
1025                 inspos = eoh - buf->buf + 1;
1026
1027         if (!keyid || !*keyid)
1028                 keyid = get_signing_key();
1029         if (sign_buffer(buf, &sig, keyid)) {
1030                 strbuf_release(&sig);
1031                 return -1;
1032         }
1033
1034         for (copypos = 0; sig.buf[copypos]; ) {
1035                 const char *bol = sig.buf + copypos;
1036                 const char *eol = strchrnul(bol, '\n');
1037                 int len = (eol - bol) + !!*eol;
1038
1039                 if (!copypos) {
1040                         strbuf_insert(buf, inspos, gpg_sig_header, gpg_sig_header_len);
1041                         inspos += gpg_sig_header_len;
1042                 }
1043                 strbuf_insertstr(buf, inspos++, " ");
1044                 strbuf_insert(buf, inspos, bol, len);
1045                 inspos += len;
1046                 copypos += len;
1047         }
1048         strbuf_release(&sig);
1049         return 0;
1050 }
1051
1052
1053
1054 int parse_signed_commit(const struct commit *commit,
1055                         struct strbuf *payload, struct strbuf *signature,
1056                         const struct git_hash_algo *algop)
1057 {
1058         unsigned long size;
1059         const char *buffer = get_commit_buffer(commit, &size);
1060         int ret = parse_buffer_signed_by_header(buffer, size, payload, signature, algop);
1061
1062         unuse_commit_buffer(commit, buffer);
1063         return ret;
1064 }
1065
1066 int parse_buffer_signed_by_header(const char *buffer,
1067                                   unsigned long size,
1068                                   struct strbuf *payload,
1069                                   struct strbuf *signature,
1070                                   const struct git_hash_algo *algop)
1071 {
1072         int in_signature = 0, saw_signature = 0, other_signature = 0;
1073         const char *line, *tail, *p;
1074         const char *gpg_sig_header = gpg_sig_headers[hash_algo_by_ptr(algop)];
1075
1076         line = buffer;
1077         tail = buffer + size;
1078         while (line < tail) {
1079                 const char *sig = NULL;
1080                 const char *next = memchr(line, '\n', tail - line);
1081
1082                 next = next ? next + 1 : tail;
1083                 if (in_signature && line[0] == ' ')
1084                         sig = line + 1;
1085                 else if (skip_prefix(line, gpg_sig_header, &p) &&
1086                          *p == ' ') {
1087                         sig = line + strlen(gpg_sig_header) + 1;
1088                         other_signature = 0;
1089                 }
1090                 else if (starts_with(line, "gpgsig"))
1091                         other_signature = 1;
1092                 else if (other_signature && line[0] != ' ')
1093                         other_signature = 0;
1094                 if (sig) {
1095                         strbuf_add(signature, sig, next - sig);
1096                         saw_signature = 1;
1097                         in_signature = 1;
1098                 } else {
1099                         if (*line == '\n')
1100                                 /* dump the whole remainder of the buffer */
1101                                 next = tail;
1102                         if (!other_signature)
1103                                 strbuf_add(payload, line, next - line);
1104                         in_signature = 0;
1105                 }
1106                 line = next;
1107         }
1108         return saw_signature;
1109 }
1110
1111 int remove_signature(struct strbuf *buf)
1112 {
1113         const char *line = buf->buf;
1114         const char *tail = buf->buf + buf->len;
1115         int in_signature = 0;
1116         struct sigbuf {
1117                 const char *start;
1118                 const char *end;
1119         } sigs[2], *sigp = &sigs[0];
1120         int i;
1121         const char *orig_buf = buf->buf;
1122
1123         memset(sigs, 0, sizeof(sigs));
1124
1125         while (line < tail) {
1126                 const char *next = memchr(line, '\n', tail - line);
1127                 next = next ? next + 1 : tail;
1128
1129                 if (in_signature && line[0] == ' ')
1130                         sigp->end = next;
1131                 else if (starts_with(line, "gpgsig")) {
1132                         int i;
1133                         for (i = 1; i < GIT_HASH_NALGOS; i++) {
1134                                 const char *p;
1135                                 if (skip_prefix(line, gpg_sig_headers[i], &p) &&
1136                                     *p == ' ') {
1137                                         sigp->start = line;
1138                                         sigp->end = next;
1139                                         in_signature = 1;
1140                                 }
1141                         }
1142                 } else {
1143                         if (*line == '\n')
1144                                 /* dump the whole remainder of the buffer */
1145                                 next = tail;
1146                         if (in_signature && sigp - sigs != ARRAY_SIZE(sigs))
1147                                 sigp++;
1148                         in_signature = 0;
1149                 }
1150                 line = next;
1151         }
1152
1153         for (i = ARRAY_SIZE(sigs) - 1; i >= 0; i--)
1154                 if (sigs[i].start)
1155                         strbuf_remove(buf, sigs[i].start - orig_buf, sigs[i].end - sigs[i].start);
1156
1157         return sigs[0].start != NULL;
1158 }
1159
1160 static void handle_signed_tag(struct commit *parent, struct commit_extra_header ***tail)
1161 {
1162         struct merge_remote_desc *desc;
1163         struct commit_extra_header *mergetag;
1164         char *buf;
1165         unsigned long size;
1166         enum object_type type;
1167         struct strbuf payload = STRBUF_INIT;
1168         struct strbuf signature = STRBUF_INIT;
1169
1170         desc = merge_remote_util(parent);
1171         if (!desc || !desc->obj)
1172                 return;
1173         buf = read_object_file(&desc->obj->oid, &type, &size);
1174         if (!buf || type != OBJ_TAG)
1175                 goto free_return;
1176         if (!parse_signature(buf, size, &payload, &signature))
1177                 goto free_return;
1178         /*
1179          * We could verify this signature and either omit the tag when
1180          * it does not validate, but the integrator may not have the
1181          * public key of the signer of the tag he is merging, while a
1182          * later auditor may have it while auditing, so let's not run
1183          * verify-signed-buffer here for now...
1184          *
1185          * if (verify_signed_buffer(buf, len, buf + len, size - len, ...))
1186          *      warn("warning: signed tag unverified.");
1187          */
1188         CALLOC_ARRAY(mergetag, 1);
1189         mergetag->key = xstrdup("mergetag");
1190         mergetag->value = buf;
1191         mergetag->len = size;
1192
1193         **tail = mergetag;
1194         *tail = &mergetag->next;
1195         strbuf_release(&payload);
1196         strbuf_release(&signature);
1197         return;
1198
1199 free_return:
1200         free(buf);
1201 }
1202
1203 int check_commit_signature(const struct commit *commit, struct signature_check *sigc)
1204 {
1205         struct strbuf payload = STRBUF_INIT;
1206         struct strbuf signature = STRBUF_INIT;
1207         int ret = 1;
1208
1209         sigc->result = 'N';
1210
1211         if (parse_signed_commit(commit, &payload, &signature, the_hash_algo) <= 0)
1212                 goto out;
1213         ret = check_signature(payload.buf, payload.len, signature.buf,
1214                 signature.len, sigc);
1215
1216  out:
1217         strbuf_release(&payload);
1218         strbuf_release(&signature);
1219
1220         return ret;
1221 }
1222
1223 void verify_merge_signature(struct commit *commit, int verbosity,
1224                             int check_trust)
1225 {
1226         char hex[GIT_MAX_HEXSZ + 1];
1227         struct signature_check signature_check;
1228         int ret;
1229         memset(&signature_check, 0, sizeof(signature_check));
1230
1231         ret = check_commit_signature(commit, &signature_check);
1232
1233         find_unique_abbrev_r(hex, &commit->object.oid, DEFAULT_ABBREV);
1234         switch (signature_check.result) {
1235         case 'G':
1236                 if (ret || (check_trust && signature_check.trust_level < TRUST_MARGINAL))
1237                         die(_("Commit %s has an untrusted GPG signature, "
1238                               "allegedly by %s."), hex, signature_check.signer);
1239                 break;
1240         case 'B':
1241                 die(_("Commit %s has a bad GPG signature "
1242                       "allegedly by %s."), hex, signature_check.signer);
1243         default: /* 'N' */
1244                 die(_("Commit %s does not have a GPG signature."), hex);
1245         }
1246         if (verbosity >= 0 && signature_check.result == 'G')
1247                 printf(_("Commit %s has a good GPG signature by %s\n"),
1248                        hex, signature_check.signer);
1249
1250         signature_check_clear(&signature_check);
1251 }
1252
1253 void append_merge_tag_headers(struct commit_list *parents,
1254                               struct commit_extra_header ***tail)
1255 {
1256         while (parents) {
1257                 struct commit *parent = parents->item;
1258                 handle_signed_tag(parent, tail);
1259                 parents = parents->next;
1260         }
1261 }
1262
1263 static void add_extra_header(struct strbuf *buffer,
1264                              struct commit_extra_header *extra)
1265 {
1266         strbuf_addstr(buffer, extra->key);
1267         if (extra->len)
1268                 strbuf_add_lines(buffer, " ", extra->value, extra->len);
1269         else
1270                 strbuf_addch(buffer, '\n');
1271 }
1272
1273 struct commit_extra_header *read_commit_extra_headers(struct commit *commit,
1274                                                       const char **exclude)
1275 {
1276         struct commit_extra_header *extra = NULL;
1277         unsigned long size;
1278         const char *buffer = get_commit_buffer(commit, &size);
1279         extra = read_commit_extra_header_lines(buffer, size, exclude);
1280         unuse_commit_buffer(commit, buffer);
1281         return extra;
1282 }
1283
1284 int for_each_mergetag(each_mergetag_fn fn, struct commit *commit, void *data)
1285 {
1286         struct commit_extra_header *extra, *to_free;
1287         int res = 0;
1288
1289         to_free = read_commit_extra_headers(commit, NULL);
1290         for (extra = to_free; !res && extra; extra = extra->next) {
1291                 if (strcmp(extra->key, "mergetag"))
1292                         continue; /* not a merge tag */
1293                 res = fn(commit, extra, data);
1294         }
1295         free_commit_extra_headers(to_free);
1296         return res;
1297 }
1298
1299 static inline int standard_header_field(const char *field, size_t len)
1300 {
1301         return ((len == 4 && !memcmp(field, "tree", 4)) ||
1302                 (len == 6 && !memcmp(field, "parent", 6)) ||
1303                 (len == 6 && !memcmp(field, "author", 6)) ||
1304                 (len == 9 && !memcmp(field, "committer", 9)) ||
1305                 (len == 8 && !memcmp(field, "encoding", 8)));
1306 }
1307
1308 static int excluded_header_field(const char *field, size_t len, const char **exclude)
1309 {
1310         if (!exclude)
1311                 return 0;
1312
1313         while (*exclude) {
1314                 size_t xlen = strlen(*exclude);
1315                 if (len == xlen && !memcmp(field, *exclude, xlen))
1316                         return 1;
1317                 exclude++;
1318         }
1319         return 0;
1320 }
1321
1322 static struct commit_extra_header *read_commit_extra_header_lines(
1323         const char *buffer, size_t size,
1324         const char **exclude)
1325 {
1326         struct commit_extra_header *extra = NULL, **tail = &extra, *it = NULL;
1327         const char *line, *next, *eof, *eob;
1328         struct strbuf buf = STRBUF_INIT;
1329
1330         for (line = buffer, eob = line + size;
1331              line < eob && *line != '\n';
1332              line = next) {
1333                 next = memchr(line, '\n', eob - line);
1334                 next = next ? next + 1 : eob;
1335                 if (*line == ' ') {
1336                         /* continuation */
1337                         if (it)
1338                                 strbuf_add(&buf, line + 1, next - (line + 1));
1339                         continue;
1340                 }
1341                 if (it)
1342                         it->value = strbuf_detach(&buf, &it->len);
1343                 strbuf_reset(&buf);
1344                 it = NULL;
1345
1346                 eof = memchr(line, ' ', next - line);
1347                 if (!eof)
1348                         eof = next;
1349                 else if (standard_header_field(line, eof - line) ||
1350                          excluded_header_field(line, eof - line, exclude))
1351                         continue;
1352
1353                 CALLOC_ARRAY(it, 1);
1354                 it->key = xmemdupz(line, eof-line);
1355                 *tail = it;
1356                 tail = &it->next;
1357                 if (eof + 1 < next)
1358                         strbuf_add(&buf, eof + 1, next - (eof + 1));
1359         }
1360         if (it)
1361                 it->value = strbuf_detach(&buf, &it->len);
1362         return extra;
1363 }
1364
1365 void free_commit_extra_headers(struct commit_extra_header *extra)
1366 {
1367         while (extra) {
1368                 struct commit_extra_header *next = extra->next;
1369                 free(extra->key);
1370                 free(extra->value);
1371                 free(extra);
1372                 extra = next;
1373         }
1374 }
1375
1376 int commit_tree(const char *msg, size_t msg_len, const struct object_id *tree,
1377                 struct commit_list *parents, struct object_id *ret,
1378                 const char *author, const char *sign_commit)
1379 {
1380         struct commit_extra_header *extra = NULL, **tail = &extra;
1381         int result;
1382
1383         append_merge_tag_headers(parents, &tail);
1384         result = commit_tree_extended(msg, msg_len, tree, parents, ret, author,
1385                                       NULL, sign_commit, extra);
1386         free_commit_extra_headers(extra);
1387         return result;
1388 }
1389
1390 static int find_invalid_utf8(const char *buf, int len)
1391 {
1392         int offset = 0;
1393         static const unsigned int max_codepoint[] = {
1394                 0x7f, 0x7ff, 0xffff, 0x10ffff
1395         };
1396
1397         while (len) {
1398                 unsigned char c = *buf++;
1399                 int bytes, bad_offset;
1400                 unsigned int codepoint;
1401                 unsigned int min_val, max_val;
1402
1403                 len--;
1404                 offset++;
1405
1406                 /* Simple US-ASCII? No worries. */
1407                 if (c < 0x80)
1408                         continue;
1409
1410                 bad_offset = offset-1;
1411
1412                 /*
1413                  * Count how many more high bits set: that's how
1414                  * many more bytes this sequence should have.
1415                  */
1416                 bytes = 0;
1417                 while (c & 0x40) {
1418                         c <<= 1;
1419                         bytes++;
1420                 }
1421
1422                 /*
1423                  * Must be between 1 and 3 more bytes.  Longer sequences result in
1424                  * codepoints beyond U+10FFFF, which are guaranteed never to exist.
1425                  */
1426                 if (bytes < 1 || 3 < bytes)
1427                         return bad_offset;
1428
1429                 /* Do we *have* that many bytes? */
1430                 if (len < bytes)
1431                         return bad_offset;
1432
1433                 /*
1434                  * Place the encoded bits at the bottom of the value and compute the
1435                  * valid range.
1436                  */
1437                 codepoint = (c & 0x7f) >> bytes;
1438                 min_val = max_codepoint[bytes-1] + 1;
1439                 max_val = max_codepoint[bytes];
1440
1441                 offset += bytes;
1442                 len -= bytes;
1443
1444                 /* And verify that they are good continuation bytes */
1445                 do {
1446                         codepoint <<= 6;
1447                         codepoint |= *buf & 0x3f;
1448                         if ((*buf++ & 0xc0) != 0x80)
1449                                 return bad_offset;
1450                 } while (--bytes);
1451
1452                 /* Reject codepoints that are out of range for the sequence length. */
1453                 if (codepoint < min_val || codepoint > max_val)
1454                         return bad_offset;
1455                 /* Surrogates are only for UTF-16 and cannot be encoded in UTF-8. */
1456                 if ((codepoint & 0x1ff800) == 0xd800)
1457                         return bad_offset;
1458                 /* U+xxFFFE and U+xxFFFF are guaranteed non-characters. */
1459                 if ((codepoint & 0xfffe) == 0xfffe)
1460                         return bad_offset;
1461                 /* So are anything in the range U+FDD0..U+FDEF. */
1462                 if (codepoint >= 0xfdd0 && codepoint <= 0xfdef)
1463                         return bad_offset;
1464         }
1465         return -1;
1466 }
1467
1468 /*
1469  * This verifies that the buffer is in proper utf8 format.
1470  *
1471  * If it isn't, it assumes any non-utf8 characters are Latin1,
1472  * and does the conversion.
1473  */
1474 static int verify_utf8(struct strbuf *buf)
1475 {
1476         int ok = 1;
1477         long pos = 0;
1478
1479         for (;;) {
1480                 int bad;
1481                 unsigned char c;
1482                 unsigned char replace[2];
1483
1484                 bad = find_invalid_utf8(buf->buf + pos, buf->len - pos);
1485                 if (bad < 0)
1486                         return ok;
1487                 pos += bad;
1488                 ok = 0;
1489                 c = buf->buf[pos];
1490                 strbuf_remove(buf, pos, 1);
1491
1492                 /* We know 'c' must be in the range 128-255 */
1493                 replace[0] = 0xc0 + (c >> 6);
1494                 replace[1] = 0x80 + (c & 0x3f);
1495                 strbuf_insert(buf, pos, replace, 2);
1496                 pos += 2;
1497         }
1498 }
1499
1500 static const char commit_utf8_warn[] =
1501 N_("Warning: commit message did not conform to UTF-8.\n"
1502    "You may want to amend it after fixing the message, or set the config\n"
1503    "variable i18n.commitencoding to the encoding your project uses.\n");
1504
1505 int commit_tree_extended(const char *msg, size_t msg_len,
1506                          const struct object_id *tree,
1507                          struct commit_list *parents, struct object_id *ret,
1508                          const char *author, const char *committer,
1509                          const char *sign_commit,
1510                          struct commit_extra_header *extra)
1511 {
1512         int result;
1513         int encoding_is_utf8;
1514         struct strbuf buffer;
1515
1516         assert_oid_type(tree, OBJ_TREE);
1517
1518         if (memchr(msg, '\0', msg_len))
1519                 return error("a NUL byte in commit log message not allowed.");
1520
1521         /* Not having i18n.commitencoding is the same as having utf-8 */
1522         encoding_is_utf8 = is_encoding_utf8(git_commit_encoding);
1523
1524         strbuf_init(&buffer, 8192); /* should avoid reallocs for the headers */
1525         strbuf_addf(&buffer, "tree %s\n", oid_to_hex(tree));
1526
1527         /*
1528          * NOTE! This ordering means that the same exact tree merged with a
1529          * different order of parents will be a _different_ changeset even
1530          * if everything else stays the same.
1531          */
1532         while (parents) {
1533                 struct commit *parent = pop_commit(&parents);
1534                 strbuf_addf(&buffer, "parent %s\n",
1535                             oid_to_hex(&parent->object.oid));
1536         }
1537
1538         /* Person/date information */
1539         if (!author)
1540                 author = git_author_info(IDENT_STRICT);
1541         strbuf_addf(&buffer, "author %s\n", author);
1542         if (!committer)
1543                 committer = git_committer_info(IDENT_STRICT);
1544         strbuf_addf(&buffer, "committer %s\n", committer);
1545         if (!encoding_is_utf8)
1546                 strbuf_addf(&buffer, "encoding %s\n", git_commit_encoding);
1547
1548         while (extra) {
1549                 add_extra_header(&buffer, extra);
1550                 extra = extra->next;
1551         }
1552         strbuf_addch(&buffer, '\n');
1553
1554         /* And add the comment */
1555         strbuf_add(&buffer, msg, msg_len);
1556
1557         /* And check the encoding */
1558         if (encoding_is_utf8 && !verify_utf8(&buffer))
1559                 fprintf(stderr, _(commit_utf8_warn));
1560
1561         if (sign_commit && sign_with_header(&buffer, sign_commit)) {
1562                 result = -1;
1563                 goto out;
1564         }
1565
1566         result = write_object_file(buffer.buf, buffer.len, commit_type, ret);
1567 out:
1568         strbuf_release(&buffer);
1569         return result;
1570 }
1571
1572 define_commit_slab(merge_desc_slab, struct merge_remote_desc *);
1573 static struct merge_desc_slab merge_desc_slab = COMMIT_SLAB_INIT(1, merge_desc_slab);
1574
1575 struct merge_remote_desc *merge_remote_util(struct commit *commit)
1576 {
1577         return *merge_desc_slab_at(&merge_desc_slab, commit);
1578 }
1579
1580 void set_merge_remote_desc(struct commit *commit,
1581                            const char *name, struct object *obj)
1582 {
1583         struct merge_remote_desc *desc;
1584         FLEX_ALLOC_STR(desc, name, name);
1585         desc->obj = obj;
1586         *merge_desc_slab_at(&merge_desc_slab, commit) = desc;
1587 }
1588
1589 struct commit *get_merge_parent(const char *name)
1590 {
1591         struct object *obj;
1592         struct commit *commit;
1593         struct object_id oid;
1594         if (get_oid(name, &oid))
1595                 return NULL;
1596         obj = parse_object(the_repository, &oid);
1597         commit = (struct commit *)peel_to_type(name, 0, obj, OBJ_COMMIT);
1598         if (commit && !merge_remote_util(commit))
1599                 set_merge_remote_desc(commit, name, obj);
1600         return commit;
1601 }
1602
1603 /*
1604  * Append a commit to the end of the commit_list.
1605  *
1606  * next starts by pointing to the variable that holds the head of an
1607  * empty commit_list, and is updated to point to the "next" field of
1608  * the last item on the list as new commits are appended.
1609  *
1610  * Usage example:
1611  *
1612  *     struct commit_list *list;
1613  *     struct commit_list **next = &list;
1614  *
1615  *     next = commit_list_append(c1, next);
1616  *     next = commit_list_append(c2, next);
1617  *     assert(commit_list_count(list) == 2);
1618  *     return list;
1619  */
1620 struct commit_list **commit_list_append(struct commit *commit,
1621                                         struct commit_list **next)
1622 {
1623         struct commit_list *new_commit = xmalloc(sizeof(struct commit_list));
1624         new_commit->item = commit;
1625         *next = new_commit;
1626         new_commit->next = NULL;
1627         return &new_commit->next;
1628 }
1629
1630 const char *find_commit_header(const char *msg, const char *key, size_t *out_len)
1631 {
1632         int key_len = strlen(key);
1633         const char *line = msg;
1634
1635         while (line) {
1636                 const char *eol = strchrnul(line, '\n');
1637
1638                 if (line == eol)
1639                         return NULL;
1640
1641                 if (eol - line > key_len &&
1642                     !strncmp(line, key, key_len) &&
1643                     line[key_len] == ' ') {
1644                         *out_len = eol - line - key_len - 1;
1645                         return line + key_len + 1;
1646                 }
1647                 line = *eol ? eol + 1 : NULL;
1648         }
1649         return NULL;
1650 }
1651
1652 /*
1653  * Inspect the given string and determine the true "end" of the log message, in
1654  * order to find where to put a new Signed-off-by trailer.  Ignored are
1655  * trailing comment lines and blank lines.  To support "git commit -s
1656  * --amend" on an existing commit, we also ignore "Conflicts:".  To
1657  * support "git commit -v", we truncate at cut lines.
1658  *
1659  * Returns the number of bytes from the tail to ignore, to be fed as
1660  * the second parameter to append_signoff().
1661  */
1662 size_t ignore_non_trailer(const char *buf, size_t len)
1663 {
1664         size_t boc = 0;
1665         size_t bol = 0;
1666         int in_old_conflicts_block = 0;
1667         size_t cutoff = wt_status_locate_end(buf, len);
1668
1669         while (bol < cutoff) {
1670                 const char *next_line = memchr(buf + bol, '\n', len - bol);
1671
1672                 if (!next_line)
1673                         next_line = buf + len;
1674                 else
1675                         next_line++;
1676
1677                 if (buf[bol] == comment_line_char || buf[bol] == '\n') {
1678                         /* is this the first of the run of comments? */
1679                         if (!boc)
1680                                 boc = bol;
1681                         /* otherwise, it is just continuing */
1682                 } else if (starts_with(buf + bol, "Conflicts:\n")) {
1683                         in_old_conflicts_block = 1;
1684                         if (!boc)
1685                                 boc = bol;
1686                 } else if (in_old_conflicts_block && buf[bol] == '\t') {
1687                         ; /* a pathname in the conflicts block */
1688                 } else if (boc) {
1689                         /* the previous was not trailing comment */
1690                         boc = 0;
1691                         in_old_conflicts_block = 0;
1692                 }
1693                 bol = next_line - buf;
1694         }
1695         return boc ? len - boc : len - cutoff;
1696 }
1697
1698 int run_commit_hook(int editor_is_used, const char *index_file,
1699                     const char *name, ...)
1700 {
1701         struct strvec hook_env = STRVEC_INIT;
1702         va_list args;
1703         int ret;
1704
1705         strvec_pushf(&hook_env, "GIT_INDEX_FILE=%s", index_file);
1706
1707         /*
1708          * Let the hook know that no editor will be launched.
1709          */
1710         if (!editor_is_used)
1711                 strvec_push(&hook_env, "GIT_EDITOR=:");
1712
1713         va_start(args, name);
1714         ret = run_hook_ve(hook_env.v, name, args);
1715         va_end(args);
1716         strvec_clear(&hook_env);
1717
1718         return ret;
1719 }