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 int no_graft_file_deprecated_advice;
  29
  30 const char *commit_type = "commit";
  31
  32 struct commit *lookup_commit_reference_gently(struct repository *r,
  33                 const struct object_id *oid, int quiet)
  34 {
  35         struct object *obj = deref_tag(r,
  36                                        parse_object(r, oid),
  37                                        NULL, 0);
  38
  39         if (!obj)
  40                 return NULL;
  41         return object_as_type(obj, OBJ_COMMIT, quiet);
  42 }
  43
  44 struct commit *lookup_commit_reference(struct repository *r, const struct object_id *oid)
  45 {
  46         return lookup_commit_reference_gently(r, oid, 0);
  47 }
  48
  49 struct commit *lookup_commit_or_die(const struct object_id *oid, const char *ref_name)
  50 {
  51         struct commit *c = lookup_commit_reference(the_repository, oid);
  52         if (!c)
  53                 die(_("could not parse %s"), ref_name);
  54         if (!oideq(oid, &c->object.oid)) {
  55                 warning(_("%s %s is not a commit!"),
  56                         ref_name, oid_to_hex(oid));
  57         }
  58         return c;
  59 }
  60
  61 struct commit *lookup_commit(struct repository *r, const struct object_id *oid)
  62 {
  63         struct object *obj = lookup_object(r, oid);
  64         if (!obj)
  65                 return create_object(r, oid, alloc_commit_node(r));
  66         return object_as_type(obj, OBJ_COMMIT, 0);
  67 }
  68
  69 struct commit *lookup_commit_reference_by_name(const char *name)
  70 {
  71         struct object_id oid;
  72         struct commit *commit;
  73
  74         if (get_oid_committish(name, &oid))
  75                 return NULL;
  76         commit = lookup_commit_reference(the_repository, &oid);
  77         if (parse_commit(commit))
  78                 return NULL;
  79         return commit;
  80 }
  81
  82 static timestamp_t parse_commit_date(const char *buf, const char *tail)
  83 {
  84         const char *dateptr;
  85
  86         if (buf + 6 >= tail)
  87                 return 0;
  88         if (memcmp(buf, "author", 6))
  89                 return 0;
  90         while (buf < tail && *buf++ != '\n')
  91                 /* nada */;
  92         if (buf + 9 >= tail)
  93                 return 0;
  94         if (memcmp(buf, "committer", 9))
  95                 return 0;
  96         while (buf < tail && *buf++ != '>')
  97                 /* nada */;
  98         if (buf >= tail)
  99                 return 0;
 100         dateptr = buf;
 101         while (buf < tail && *buf++ != '\n')
 102                 /* nada */;
 103         if (buf >= tail)
 104                 return 0;
 105         /* dateptr < buf && buf[-1] == '\n', so parsing will stop at buf-1 */
 106         return parse_timestamp(dateptr, NULL, 10);
 107 }
 108
 109 static const struct object_id *commit_graft_oid_access(size_t index, const void *table)
 110 {
 111         const struct commit_graft * const *commit_graft_table = table;
 112         return &commit_graft_table[index]->oid;
 113 }
 114
 115 int commit_graft_pos(struct repository *r, const struct object_id *oid)
 116 {
 117         return oid_pos(oid, r->parsed_objects->grafts,
 118                        r->parsed_objects->grafts_nr,
 119                        commit_graft_oid_access);
 120 }
 121
 122 int register_commit_graft(struct repository *r, struct commit_graft *graft,
 123                           int ignore_dups)
 124 {
 125         int pos = commit_graft_pos(r, &graft->oid);
 126
 127         if (0 <= pos) {
 128                 if (ignore_dups)
 129                         free(graft);
 130                 else {
 131                         free(r->parsed_objects->grafts[pos]);
 132                         r->parsed_objects->grafts[pos] = graft;
 133                 }
 134                 return 1;
 135         }
 136         pos = -pos - 1;
 137         ALLOC_GROW(r->parsed_objects->grafts,
 138                    r->parsed_objects->grafts_nr + 1,
 139                    r->parsed_objects->grafts_alloc);
 140         r->parsed_objects->grafts_nr++;
 141         if (pos < r->parsed_objects->grafts_nr)
 142                 memmove(r->parsed_objects->grafts + pos + 1,
 143                         r->parsed_objects->grafts + pos,
 144                         (r->parsed_objects->grafts_nr - pos - 1) *
 145                         sizeof(*r->parsed_objects->grafts));
 146         r->parsed_objects->grafts[pos] = graft;
 147         return 0;
 148 }
 149
 150 struct commit_graft *read_graft_line(struct strbuf *line)
 151 {
 152         /* The format is just "Commit Parent1 Parent2 ...\n" */
 153         int i, phase;
 154         const char *tail = NULL;
 155         struct commit_graft *graft = NULL;
 156         struct object_id dummy_oid, *oid;
 157
 158         strbuf_rtrim(line);
 159         if (!line->len || line->buf[0] == '#')
 160                 return NULL;
 161         /*
 162          * phase 0 verifies line, counts hashes in line and allocates graft
 163          * phase 1 fills graft
 164          */
 165         for (phase = 0; phase < 2; phase++) {
 166                 oid = graft ? &graft->oid : &dummy_oid;
 167                 if (parse_oid_hex(line->buf, oid, &tail))
 168                         goto bad_graft_data;
 169                 for (i = 0; *tail != '\0'; i++) {
 170                         oid = graft ? &graft->parent[i] : &dummy_oid;
 171                         if (!isspace(*tail++) || parse_oid_hex(tail, oid, &tail))
 172                                 goto bad_graft_data;
 173                 }
 174                 if (!graft) {
 175                         graft = xmalloc(st_add(sizeof(*graft),
 176                                                st_mult(sizeof(struct object_id), i)));
 177                         graft->nr_parent = i;
 178                 }
 179         }
 180         return graft;
 181
 182 bad_graft_data:
 183         error("bad graft data: %s", line->buf);
 184         assert(!graft);
 185         return NULL;
 186 }
 187
 188 static int read_graft_file(struct repository *r, const char *graft_file)
 189 {
 190         FILE *fp = fopen_or_warn(graft_file, "r");
 191         struct strbuf buf = STRBUF_INIT;
 192         if (!fp)
 193                 return -1;
 194         if (!no_graft_file_deprecated_advice &&
 195             advice_enabled(ADVICE_GRAFT_FILE_DEPRECATED))
 196                 advise(_("Support for <GIT_DIR>/info/grafts is deprecated\n"
 197                          "and will be removed in a future Git version.\n"
 198                          "\n"
 199                          "Please use \"git replace --convert-graft-file\"\n"
 200                          "to convert the grafts into replace refs.\n"
 201                          "\n"
 202                          "Turn this message off by running\n"
 203                          "\"git config advice.graftFileDeprecated false\""));
 204         while (!strbuf_getwholeline(&buf, fp, '\n')) {
 205                 /* The format is just "Commit Parent1 Parent2 ...\n" */
 206                 struct commit_graft *graft = read_graft_line(&buf);
 207                 if (!graft)
 208                         continue;
 209                 if (register_commit_graft(r, graft, 1))
 210                         error("duplicate graft data: %s", buf.buf);
 211         }
 212         fclose(fp);
 213         strbuf_release(&buf);
 214         return 0;
 215 }
 216
 217 void prepare_commit_graft(struct repository *r)
 218 {
 219         char *graft_file;
 220
 221         if (r->parsed_objects->commit_graft_prepared)
 222                 return;
 223         if (!startup_info->have_repository)
 224                 return;
 225
 226         graft_file = get_graft_file(r);
 227         read_graft_file(r, graft_file);
 228         /* make sure shallows are read */
 229         is_repository_shallow(r);
 230         r->parsed_objects->commit_graft_prepared = 1;
 231 }
 232
 233 struct commit_graft *lookup_commit_graft(struct repository *r, const struct object_id *oid)
 234 {
 235         int pos;
 236         prepare_commit_graft(r);
 237         pos = commit_graft_pos(r, oid);
 238         if (pos < 0)
 239                 return NULL;
 240         return r->parsed_objects->grafts[pos];
 241 }
 242
 243 int for_each_commit_graft(each_commit_graft_fn fn, void *cb_data)
 244 {
 245         int i, ret;
 246         for (i = ret = 0; i < the_repository->parsed_objects->grafts_nr && !ret; i++)
 247                 ret = fn(the_repository->parsed_objects->grafts[i], cb_data);
 248         return ret;
 249 }
 250
 251 struct commit_buffer {
 252         void *buffer;
 253         unsigned long size;
 254 };
 255 define_commit_slab(buffer_slab, struct commit_buffer);
 256
 257 struct buffer_slab *allocate_commit_buffer_slab(void)
 258 {
 259         struct buffer_slab *bs = xmalloc(sizeof(*bs));
 260         init_buffer_slab(bs);
 261         return bs;
 262 }
 263
 264 void free_commit_buffer_slab(struct buffer_slab *bs)
 265 {
 266         clear_buffer_slab(bs);
 267         free(bs);
 268 }
 269
 270 void set_commit_buffer(struct repository *r, struct commit *commit, void *buffer, unsigned long size)
 271 {
 272         struct commit_buffer *v = buffer_slab_at(
 273                 r->parsed_objects->buffer_slab, commit);
 274         v->buffer = buffer;
 275         v->size = size;
 276 }
 277
 278 const void *get_cached_commit_buffer(struct repository *r, const struct commit *commit, unsigned long *sizep)
 279 {
 280         struct commit_buffer *v = buffer_slab_peek(
 281                 r->parsed_objects->buffer_slab, commit);
 282         if (!v) {
 283                 if (sizep)
 284                         *sizep = 0;
 285                 return NULL;
 286         }
 287         if (sizep)
 288                 *sizep = v->size;
 289         return v->buffer;
 290 }
 291
 292 const void *repo_get_commit_buffer(struct repository *r,
 293                                    const struct commit *commit,
 294                                    unsigned long *sizep)
 295 {
 296         const void *ret = get_cached_commit_buffer(r, commit, sizep);
 297         if (!ret) {
 298                 enum object_type type;
 299                 unsigned long size;
 300                 ret = repo_read_object_file(r, &commit->object.oid, &type, &size);
 301                 if (!ret)
 302                         die("cannot read commit object %s",
 303                             oid_to_hex(&commit->object.oid));
 304                 if (type != OBJ_COMMIT)
 305                         die("expected commit for %s, got %s",
 306                             oid_to_hex(&commit->object.oid), type_name(type));
 307                 if (sizep)
 308                         *sizep = size;
 309         }
 310         return ret;
 311 }
 312
 313 void repo_unuse_commit_buffer(struct repository *r,
 314                               const struct commit *commit,
 315                               const void *buffer)
 316 {
 317         struct commit_buffer *v = buffer_slab_peek(
 318                 r->parsed_objects->buffer_slab, commit);
 319         if (!(v && v->buffer == buffer))
 320                 free((void *)buffer);
 321 }
 322
 323 void free_commit_buffer(struct parsed_object_pool *pool, struct commit *commit)
 324 {
 325         struct commit_buffer *v = buffer_slab_peek(
 326                 pool->buffer_slab, commit);
 327         if (v) {
 328                 FREE_AND_NULL(v->buffer);
 329                 v->size = 0;
 330         }
 331 }
 332
 333 static inline void set_commit_tree(struct commit *c, struct tree *t)
 334 {
 335         c->maybe_tree = t;
 336 }
 337
 338 struct tree *repo_get_commit_tree(struct repository *r,
 339                                   const struct commit *commit)
 340 {
 341         if (commit->maybe_tree || !commit->object.parsed)
 342                 return commit->maybe_tree;
 343
 344         if (commit_graph_position(commit) != COMMIT_NOT_FROM_GRAPH)
 345                 return get_commit_tree_in_graph(r, commit);
 346
 347         return NULL;
 348 }
 349
 350 struct object_id *get_commit_tree_oid(const struct commit *commit)
 351 {
 352         struct tree *tree = get_commit_tree(commit);
 353         return tree ? &tree->object.oid : NULL;
 354 }
 355
 356 void release_commit_memory(struct parsed_object_pool *pool, struct commit *c)
 357 {
 358         set_commit_tree(c, NULL);
 359         free_commit_buffer(pool, c);
 360         c->index = 0;
 361         free_commit_list(c->parents);
 362
 363         c->object.parsed = 0;
 364 }
 365
 366 const void *detach_commit_buffer(struct commit *commit, unsigned long *sizep)
 367 {
 368         struct commit_buffer *v = buffer_slab_peek(
 369                 the_repository->parsed_objects->buffer_slab, commit);
 370         void *ret;
 371
 372         if (!v) {
 373                 if (sizep)
 374                         *sizep = 0;
 375                 return NULL;
 376         }
 377         ret = v->buffer;
 378         if (sizep)
 379                 *sizep = v->size;
 380
 381         v->buffer = NULL;
 382         v->size = 0;
 383         return ret;
 384 }
 385
 386 int parse_commit_buffer(struct repository *r, struct commit *item, const void *buffer, unsigned long size, int check_graph)
 387 {
 388         const char *tail = buffer;
 389         const char *bufptr = buffer;
 390         struct object_id parent;
 391         struct commit_list **pptr;
 392         struct commit_graft *graft;
 393         const int tree_entry_len = the_hash_algo->hexsz + 5;
 394         const int parent_entry_len = the_hash_algo->hexsz + 7;
 395         struct tree *tree;
 396
 397         if (item->object.parsed)
 398                 return 0;
 399
 400         if (item->parents) {
 401                 /*
 402                  * Presumably this is leftover from an earlier failed parse;
 403                  * clear it out in preparation for us re-parsing (we'll hit the
 404                  * same error, but that's good, since it lets our caller know
 405                  * the result cannot be trusted.
 406                  */
 407                 free_commit_list(item->parents);
 408                 item->parents = NULL;
 409         }
 410
 411         tail += size;
 412         if (tail <= bufptr + tree_entry_len + 1 || memcmp(bufptr, "tree ", 5) ||
 413                         bufptr[tree_entry_len] != '\n')
 414                 return error("bogus commit object %s", oid_to_hex(&item->object.oid));
 415         if (get_oid_hex(bufptr + 5, &parent) < 0)
 416                 return error("bad tree pointer in commit %s",
 417                              oid_to_hex(&item->object.oid));
 418         tree = lookup_tree(r, &parent);
 419         if (!tree)
 420                 return error("bad tree pointer %s in commit %s",
 421                              oid_to_hex(&parent),
 422                              oid_to_hex(&item->object.oid));
 423         set_commit_tree(item, tree);
 424         bufptr += tree_entry_len + 1; /* "tree " + "hex sha1" + "\n" */
 425         pptr = &item->parents;
 426
 427         graft = lookup_commit_graft(r, &item->object.oid);
 428         if (graft)
 429                 r->parsed_objects->substituted_parent = 1;
 430         while (bufptr + parent_entry_len < tail && !memcmp(bufptr, "parent ", 7)) {
 431                 struct commit *new_parent;
 432
 433                 if (tail <= bufptr + parent_entry_len + 1 ||
 434                     get_oid_hex(bufptr + 7, &parent) ||
 435                     bufptr[parent_entry_len] != '\n')
 436                         return error("bad parents in commit %s", oid_to_hex(&item->object.oid));
 437                 bufptr += parent_entry_len + 1;
 438                 /*
 439                  * The clone is shallow if nr_parent < 0, and we must
 440                  * not traverse its real parents even when we unhide them.
 441                  */
 442                 if (graft && (graft->nr_parent < 0 || grafts_replace_parents))
 443                         continue;
 444                 new_parent = lookup_commit(r, &parent);
 445                 if (!new_parent)
 446                         return error("bad parent %s in commit %s",
 447                                      oid_to_hex(&parent),
 448                                      oid_to_hex(&item->object.oid));
 449                 pptr = &commit_list_insert(new_parent, pptr)->next;
 450         }
 451         if (graft) {
 452                 int i;
 453                 struct commit *new_parent;
 454                 for (i = 0; i < graft->nr_parent; i++) {
 455                         new_parent = lookup_commit(r,
 456                                                    &graft->parent[i]);
 457                         if (!new_parent)
 458                                 return error("bad graft parent %s in commit %s",
 459                                              oid_to_hex(&graft->parent[i]),
 460                                              oid_to_hex(&item->object.oid));
 461                         pptr = &commit_list_insert(new_parent, pptr)->next;
 462                 }
 463         }
 464         item->date = parse_commit_date(bufptr, tail);
 465
 466         if (check_graph)
 467                 load_commit_graph_info(r, item);
 468
 469         item->object.parsed = 1;
 470         return 0;
 471 }
 472
 473 int repo_parse_commit_internal(struct repository *r,
 474                                struct commit *item,
 475                                int quiet_on_missing,
 476                                int use_commit_graph)
 477 {
 478         enum object_type type;
 479         void *buffer;
 480         unsigned long size;
 481         int ret;
 482
 483         if (!item)
 484                 return -1;
 485         if (item->object.parsed)
 486                 return 0;
 487         if (use_commit_graph && parse_commit_in_graph(r, item))
 488                 return 0;
 489         buffer = repo_read_object_file(r, &item->object.oid, &type, &size);
 490         if (!buffer)
 491                 return quiet_on_missing ? -1 :
 492                         error("Could not read %s",
 493                              oid_to_hex(&item->object.oid));
 494         if (type != OBJ_COMMIT) {
 495                 free(buffer);
 496                 return error("Object %s not a commit",
 497                              oid_to_hex(&item->object.oid));
 498         }
 499
 500         ret = parse_commit_buffer(r, item, buffer, size, 0);
 501         if (save_commit_buffer && !ret) {
 502                 set_commit_buffer(r, item, buffer, size);
 503                 return 0;
 504         }
 505         free(buffer);
 506         return ret;
 507 }
 508
 509 int repo_parse_commit_gently(struct repository *r,
 510                              struct commit *item, int quiet_on_missing)
 511 {
 512         return repo_parse_commit_internal(r, item, quiet_on_missing, 1);
 513 }
 514
 515 void parse_commit_or_die(struct commit *item)
 516 {
 517         if (parse_commit(item))
 518                 die("unable to parse commit %s",
 519                     item ? oid_to_hex(&item->object.oid) : "(null)");
 520 }
 521
 522 int find_commit_subject(const char *commit_buffer, const char **subject)
 523 {
 524         const char *eol;
 525         const char *p = commit_buffer;
 526
 527         while (*p && (*p != '\n' || p[1] != '\n'))
 528                 p++;
 529         if (*p) {
 530                 p = skip_blank_lines(p + 2);
 531                 eol = strchrnul(p, '\n');
 532         } else
 533                 eol = p;
 534
 535         *subject = p;
 536
 537         return eol - p;
 538 }
 539
 540 size_t commit_subject_length(const char *body)
 541 {
 542         const char *p = body;
 543         while (*p) {
 544                 const char *next = skip_blank_lines(p);
 545                 if (next != p)
 546                         break;
 547                 p = strchrnul(p, '\n');
 548                 if (*p)
 549                         p++;
 550         }
 551         return p - body;
 552 }
 553
 554 struct commit_list *commit_list_insert(struct commit *item, struct commit_list **list_p)
 555 {
 556         struct commit_list *new_list = xmalloc(sizeof(struct commit_list));
 557         new_list->item = item;
 558         new_list->next = *list_p;
 559         *list_p = new_list;
 560         return new_list;
 561 }
 562
 563 int commit_list_contains(struct commit *item, struct commit_list *list)
 564 {
 565         while (list) {
 566                 if (list->item == item)
 567                         return 1;
 568                 list = list->next;
 569         }
 570
 571         return 0;
 572 }
 573
 574 unsigned commit_list_count(const struct commit_list *l)
 575 {
 576         unsigned c = 0;
 577         for (; l; l = l->next )
 578                 c++;
 579         return c;
 580 }
 581
 582 struct commit_list *copy_commit_list(struct commit_list *list)
 583 {
 584         struct commit_list *head = NULL;
 585         struct commit_list **pp = &head;
 586         while (list) {
 587                 pp = commit_list_append(list->item, pp);
 588                 list = list->next;
 589         }
 590         return head;
 591 }
 592
 593 struct commit_list *reverse_commit_list(struct commit_list *list)
 594 {
 595         struct commit_list *next = NULL, *current, *backup;
 596         for (current = list; current; current = backup) {
 597                 backup = current->next;
 598                 current->next = next;
 599                 next = current;
 600         }
 601         return next;
 602 }
 603
 604 void free_commit_list(struct commit_list *list)
 605 {
 606         while (list)
 607                 pop_commit(&list);
 608 }
 609
 610 struct commit_list * commit_list_insert_by_date(struct commit *item, struct commit_list **list)
 611 {
 612         struct commit_list **pp = list;
 613         struct commit_list *p;
 614         while ((p = *pp) != NULL) {
 615                 if (p->item->date < item->date) {
 616                         break;
 617                 }
 618                 pp = &p->next;
 619         }
 620         return commit_list_insert(item, pp);
 621 }
 622
 623 static int commit_list_compare_by_date(const void *a, const void *b)
 624 {
 625         timestamp_t a_date = ((const struct commit_list *)a)->item->date;
 626         timestamp_t b_date = ((const struct commit_list *)b)->item->date;
 627         if (a_date < b_date)
 628                 return 1;
 629         if (a_date > b_date)
 630                 return -1;
 631         return 0;
 632 }
 633
 634 static void *commit_list_get_next(const void *a)
 635 {
 636         return ((const struct commit_list *)a)->next;
 637 }
 638
 639 static void commit_list_set_next(void *a, void *next)
 640 {
 641         ((struct commit_list *)a)->next = next;
 642 }
 643
 644 void commit_list_sort_by_date(struct commit_list **list)
 645 {
 646         *list = llist_mergesort(*list, commit_list_get_next, commit_list_set_next,
 647                                 commit_list_compare_by_date);
 648 }
 649
 650 struct commit *pop_most_recent_commit(struct commit_list **list,
 651                                       unsigned int mark)
 652 {
 653         struct commit *ret = pop_commit(list);
 654         struct commit_list *parents = ret->parents;
 655
 656         while (parents) {
 657                 struct commit *commit = parents->item;
 658                 if (!parse_commit(commit) && !(commit->object.flags & mark)) {
 659                         commit->object.flags |= mark;
 660                         commit_list_insert_by_date(commit, list);
 661                 }
 662                 parents = parents->next;
 663         }
 664         return ret;
 665 }
 666
 667 static void clear_commit_marks_1(struct commit_list **plist,
 668                                  struct commit *commit, unsigned int mark)
 669 {
 670         while (commit) {
 671                 struct commit_list *parents;
 672
 673                 if (!(mark & commit->object.flags))
 674                         return;
 675
 676                 commit->object.flags &= ~mark;
 677
 678                 parents = commit->parents;
 679                 if (!parents)
 680                         return;
 681
 682                 while ((parents = parents->next))
 683                         commit_list_insert(parents->item, plist);
 684
 685                 commit = commit->parents->item;
 686         }
 687 }
 688
 689 void clear_commit_marks_many(int nr, struct commit **commit, unsigned int mark)
 690 {
 691         struct commit_list *list = NULL;
 692
 693         while (nr--) {
 694                 clear_commit_marks_1(&list, *commit, mark);
 695                 commit++;
 696         }
 697         while (list)
 698                 clear_commit_marks_1(&list, pop_commit(&list), mark);
 699 }
 700
 701 void clear_commit_marks(struct commit *commit, unsigned int mark)
 702 {
 703         clear_commit_marks_many(1, &commit, mark);
 704 }
 705
 706 struct commit *pop_commit(struct commit_list **stack)
 707 {
 708         struct commit_list *top = *stack;
 709         struct commit *item = top ? top->item : NULL;
 710
 711         if (top) {
 712                 *stack = top->next;
 713                 free(top);
 714         }
 715         return item;
 716 }
 717
 718 /*
 719  * Topological sort support
 720  */
 721
 722 /* count number of children that have not been emitted */
 723 define_commit_slab(indegree_slab, int);
 724
 725 define_commit_slab(author_date_slab, timestamp_t);
 726
 727 void record_author_date(struct author_date_slab *author_date,
 728                         struct commit *commit)
 729 {
 730         const char *buffer = get_commit_buffer(commit, NULL);
 731         struct ident_split ident;
 732         const char *ident_line;
 733         size_t ident_len;
 734         char *date_end;
 735         timestamp_t date;
 736
 737         ident_line = find_commit_header(buffer, "author", &ident_len);
 738         if (!ident_line)
 739                 goto fail_exit; /* no author line */
 740         if (split_ident_line(&ident, ident_line, ident_len) ||
 741             !ident.date_begin || !ident.date_end)
 742                 goto fail_exit; /* malformed "author" line */
 743
 744         date = parse_timestamp(ident.date_begin, &date_end, 10);
 745         if (date_end != ident.date_end)
 746                 goto fail_exit; /* malformed date */
 747         *(author_date_slab_at(author_date, commit)) = date;
 748
 749 fail_exit:
 750         unuse_commit_buffer(commit, buffer);
 751 }
 752
 753 int compare_commits_by_author_date(const void *a_, const void *b_,
 754                                    void *cb_data)
 755 {
 756         const struct commit *a = a_, *b = b_;
 757         struct author_date_slab *author_date = cb_data;
 758         timestamp_t a_date = *(author_date_slab_at(author_date, a));
 759         timestamp_t b_date = *(author_date_slab_at(author_date, b));
 760
 761         /* newer commits with larger date first */
 762         if (a_date < b_date)
 763                 return 1;
 764         else if (a_date > b_date)
 765                 return -1;
 766         return 0;
 767 }
 768
 769 int compare_commits_by_gen_then_commit_date(const void *a_, const void *b_, void *unused)
 770 {
 771         const struct commit *a = a_, *b = b_;
 772         const timestamp_t generation_a = commit_graph_generation(a),
 773                           generation_b = commit_graph_generation(b);
 774
 775         /* newer commits first */
 776         if (generation_a < generation_b)
 777                 return 1;
 778         else if (generation_a > generation_b)
 779                 return -1;
 780
 781         /* use date as a heuristic when generations are equal */
 782         if (a->date < b->date)
 783                 return 1;
 784         else if (a->date > b->date)
 785                 return -1;
 786         return 0;
 787 }
 788
 789 int compare_commits_by_commit_date(const void *a_, const void *b_, void *unused)
 790 {
 791         const struct commit *a = a_, *b = b_;
 792         /* newer commits with larger date first */
 793         if (a->date < b->date)
 794                 return 1;
 795         else if (a->date > b->date)
 796                 return -1;
 797         return 0;
 798 }
 799
 800 /*
 801  * Performs an in-place topological sort on the list supplied.
 802  */
 803 void sort_in_topological_order(struct commit_list **list, enum rev_sort_order sort_order)
 804 {
 805         struct commit_list *next, *orig = *list;
 806         struct commit_list **pptr;
 807         struct indegree_slab indegree;
 808         struct prio_queue queue;
 809         struct commit *commit;
 810         struct author_date_slab author_date;
 811
 812         if (!orig)
 813                 return;
 814         *list = NULL;
 815
 816         init_indegree_slab(&indegree);
 817         memset(&queue, '\0', sizeof(queue));
 818
 819         switch (sort_order) {
 820         default: /* REV_SORT_IN_GRAPH_ORDER */
 821                 queue.compare = NULL;
 822                 break;
 823         case REV_SORT_BY_COMMIT_DATE:
 824                 queue.compare = compare_commits_by_commit_date;
 825                 break;
 826         case REV_SORT_BY_AUTHOR_DATE:
 827                 init_author_date_slab(&author_date);
 828                 queue.compare = compare_commits_by_author_date;
 829                 queue.cb_data = &author_date;
 830                 break;
 831         }
 832
 833         /* Mark them and clear the indegree */
 834         for (next = orig; next; next = next->next) {
 835                 struct commit *commit = next->item;
 836                 *(indegree_slab_at(&indegree, commit)) = 1;
 837                 /* also record the author dates, if needed */
 838                 if (sort_order == REV_SORT_BY_AUTHOR_DATE)
 839                         record_author_date(&author_date, commit);
 840         }
 841
 842         /* update the indegree */
 843         for (next = orig; next; next = next->next) {
 844                 struct commit_list *parents = next->item->parents;
 845                 while (parents) {
 846                         struct commit *parent = parents->item;
 847                         int *pi = indegree_slab_at(&indegree, parent);
 848
 849                         if (*pi)
 850                                 (*pi)++;
 851                         parents = parents->next;
 852                 }
 853         }
 854
 855         /*
 856          * find the tips
 857          *
 858          * tips are nodes not reachable from any other node in the list
 859          *
 860          * the tips serve as a starting set for the work queue.
 861          */
 862         for (next = orig; next; next = next->next) {
 863                 struct commit *commit = next->item;
 864
 865                 if (*(indegree_slab_at(&indegree, commit)) == 1)
 866                         prio_queue_put(&queue, commit);
 867         }
 868
 869         /*
 870          * This is unfortunate; the initial tips need to be shown
 871          * in the order given from the revision traversal machinery.
 872          */
 873         if (sort_order == REV_SORT_IN_GRAPH_ORDER)
 874                 prio_queue_reverse(&queue);
 875
 876         /* We no longer need the commit list */
 877         free_commit_list(orig);
 878
 879         pptr = list;
 880         *list = NULL;
 881         while ((commit = prio_queue_get(&queue)) != NULL) {
 882                 struct commit_list *parents;
 883
 884                 for (parents = commit->parents; parents ; parents = parents->next) {
 885                         struct commit *parent = parents->item;
 886                         int *pi = indegree_slab_at(&indegree, parent);
 887
 888                         if (!*pi)
 889                                 continue;
 890
 891                         /*
 892                          * parents are only enqueued for emission
 893                          * when all their children have been emitted thereby
 894                          * guaranteeing topological order.
 895                          */
 896                         if (--(*pi) == 1)
 897                                 prio_queue_put(&queue, parent);
 898                 }
 899                 /*
 900                  * all children of commit have already been
 901                  * emitted. we can emit it now.
 902                  */
 903                 *(indegree_slab_at(&indegree, commit)) = 0;
 904
 905                 pptr = &commit_list_insert(commit, pptr)->next;
 906         }
 907
 908         clear_indegree_slab(&indegree);
 909         clear_prio_queue(&queue);
 910         if (sort_order == REV_SORT_BY_AUTHOR_DATE)
 911                 clear_author_date_slab(&author_date);
 912 }
 913
 914 struct rev_collect {
 915         struct commit **commit;
 916         int nr;
 917         int alloc;
 918         unsigned int initial : 1;
 919 };
 920
 921 static void add_one_commit(struct object_id *oid, struct rev_collect *revs)
 922 {
 923         struct commit *commit;
 924
 925         if (is_null_oid(oid))
 926                 return;
 927
 928         commit = lookup_commit(the_repository, oid);
 929         if (!commit ||
 930             (commit->object.flags & TMP_MARK) ||
 931             parse_commit(commit))
 932                 return;
 933
 934         ALLOC_GROW(revs->commit, revs->nr + 1, revs->alloc);
 935         revs->commit[revs->nr++] = commit;
 936         commit->object.flags |= TMP_MARK;
 937 }
 938
 939 static int collect_one_reflog_ent(struct object_id *ooid, struct object_id *noid,
 940                                   const char *ident, timestamp_t timestamp,
 941                                   int tz, const char *message, void *cbdata)
 942 {
 943         struct rev_collect *revs = cbdata;
 944
 945         if (revs->initial) {
 946                 revs->initial = 0;
 947                 add_one_commit(ooid, revs);
 948         }
 949         add_one_commit(noid, revs);
 950         return 0;
 951 }
 952
 953 struct commit *get_fork_point(const char *refname, struct commit *commit)
 954 {
 955         struct object_id oid;
 956         struct rev_collect revs;
 957         struct commit_list *bases;
 958         int i;
 959         struct commit *ret = NULL;
 960         char *full_refname;
 961
 962         switch (dwim_ref(refname, strlen(refname), &oid, &full_refname, 0)) {
 963         case 0:
 964                 die("No such ref: '%s'", refname);
 965         case 1:
 966                 break; /* good */
 967         default:
 968                 die("Ambiguous refname: '%s'", refname);
 969         }
 970
 971         memset(&revs, 0, sizeof(revs));
 972         revs.initial = 1;
 973         for_each_reflog_ent(full_refname, collect_one_reflog_ent, &revs);
 974
 975         if (!revs.nr)
 976                 add_one_commit(&oid, &revs);
 977
 978         for (i = 0; i < revs.nr; i++)
 979                 revs.commit[i]->object.flags &= ~TMP_MARK;
 980
 981         bases = get_merge_bases_many(commit, revs.nr, revs.commit);
 982
 983         /*
 984          * There should be one and only one merge base, when we found
 985          * a common ancestor among reflog entries.
 986          */
 987         if (!bases || bases->next)
 988                 goto cleanup_return;
 989
 990         /* And the found one must be one of the reflog entries */
 991         for (i = 0; i < revs.nr; i++)
 992                 if (&bases->item->object == &revs.commit[i]->object)
 993                         break; /* found */
 994         if (revs.nr <= i)
 995                 goto cleanup_return;
 996
 997         ret = bases->item;
 998
 999 cleanup_return:
1000         free_commit_list(bases);
1001         free(full_refname);
1002         return ret;
1003 }
1004
1005 /*
1006  * Indexed by hash algorithm identifier.
1007  */
1008 static const char *gpg_sig_headers[] = {
1009         NULL,
1010         "gpgsig",
1011         "gpgsig-sha256",
1012 };
1013
1014 int sign_with_header(struct strbuf *buf, const char *keyid)
1015 {
1016         struct strbuf sig = STRBUF_INIT;
1017         int inspos, copypos;
1018         const char *eoh;
1019         const char *gpg_sig_header = gpg_sig_headers[hash_algo_by_ptr(the_hash_algo)];
1020         int gpg_sig_header_len = strlen(gpg_sig_header);
1021
1022         /* find the end of the header */
1023         eoh = strstr(buf->buf, "\n\n");
1024         if (!eoh)
1025                 inspos = buf->len;
1026         else
1027                 inspos = eoh - buf->buf + 1;
1028
1029         if (!keyid || !*keyid)
1030                 keyid = get_signing_key();
1031         if (sign_buffer(buf, &sig, keyid)) {
1032                 strbuf_release(&sig);
1033                 return -1;
1034         }
1035
1036         for (copypos = 0; sig.buf[copypos]; ) {
1037                 const char *bol = sig.buf + copypos;
1038                 const char *eol = strchrnul(bol, '\n');
1039                 int len = (eol - bol) + !!*eol;
1040
1041                 if (!copypos) {
1042                         strbuf_insert(buf, inspos, gpg_sig_header, gpg_sig_header_len);
1043                         inspos += gpg_sig_header_len;
1044                 }
1045                 strbuf_insertstr(buf, inspos++, " ");
1046                 strbuf_insert(buf, inspos, bol, len);
1047                 inspos += len;
1048                 copypos += len;
1049         }
1050         strbuf_release(&sig);
1051         return 0;
1052 }
1053
1054
1055
1056 int parse_signed_commit(const struct commit *commit,
1057                         struct strbuf *payload, struct strbuf *signature,
1058                         const struct git_hash_algo *algop)
1059 {
1060         unsigned long size;
1061         const char *buffer = get_commit_buffer(commit, &size);
1062         int ret = parse_buffer_signed_by_header(buffer, size, payload, signature, algop);
1063
1064         unuse_commit_buffer(commit, buffer);
1065         return ret;
1066 }
1067
1068 int parse_buffer_signed_by_header(const char *buffer,
1069                                   unsigned long size,
1070                                   struct strbuf *payload,
1071                                   struct strbuf *signature,
1072                                   const struct git_hash_algo *algop)
1073 {
1074         int in_signature = 0, saw_signature = 0, other_signature = 0;
1075         const char *line, *tail, *p;
1076         const char *gpg_sig_header = gpg_sig_headers[hash_algo_by_ptr(algop)];
1077
1078         line = buffer;
1079         tail = buffer + size;
1080         while (line < tail) {
1081                 const char *sig = NULL;
1082                 const char *next = memchr(line, '\n', tail - line);
1083
1084                 next = next ? next + 1 : tail;
1085                 if (in_signature && line[0] == ' ')
1086                         sig = line + 1;
1087                 else if (skip_prefix(line, gpg_sig_header, &p) &&
1088                          *p == ' ') {
1089                         sig = line + strlen(gpg_sig_header) + 1;
1090                         other_signature = 0;
1091                 }
1092                 else if (starts_with(line, "gpgsig"))
1093                         other_signature = 1;
1094                 else if (other_signature && line[0] != ' ')
1095                         other_signature = 0;
1096                 if (sig) {
1097                         strbuf_add(signature, sig, next - sig);
1098                         saw_signature = 1;
1099                         in_signature = 1;
1100                 } else {
1101                         if (*line == '\n')
1102                                 /* dump the whole remainder of the buffer */
1103                                 next = tail;
1104                         if (!other_signature)
1105                                 strbuf_add(payload, line, next - line);
1106                         in_signature = 0;
1107                 }
1108                 line = next;
1109         }
1110         return saw_signature;
1111 }
1112
1113 int remove_signature(struct strbuf *buf)
1114 {
1115         const char *line = buf->buf;
1116         const char *tail = buf->buf + buf->len;
1117         int in_signature = 0;
1118         struct sigbuf {
1119                 const char *start;
1120                 const char *end;
1121         } sigs[2], *sigp = &sigs[0];
1122         int i;
1123         const char *orig_buf = buf->buf;
1124
1125         memset(sigs, 0, sizeof(sigs));
1126
1127         while (line < tail) {
1128                 const char *next = memchr(line, '\n', tail - line);
1129                 next = next ? next + 1 : tail;
1130
1131                 if (in_signature && line[0] == ' ')
1132                         sigp->end = next;
1133                 else if (starts_with(line, "gpgsig")) {
1134                         int i;
1135                         for (i = 1; i < GIT_HASH_NALGOS; i++) {
1136                                 const char *p;
1137                                 if (skip_prefix(line, gpg_sig_headers[i], &p) &&
1138                                     *p == ' ') {
1139                                         sigp->start = line;
1140                                         sigp->end = next;
1141                                         in_signature = 1;
1142                                 }
1143                         }
1144                 } else {
1145                         if (*line == '\n')
1146                                 /* dump the whole remainder of the buffer */
1147                                 next = tail;
1148                         if (in_signature && sigp - sigs != ARRAY_SIZE(sigs))
1149                                 sigp++;
1150                         in_signature = 0;
1151                 }
1152                 line = next;
1153         }
1154
1155         for (i = ARRAY_SIZE(sigs) - 1; i >= 0; i--)
1156                 if (sigs[i].start)
1157                         strbuf_remove(buf, sigs[i].start - orig_buf, sigs[i].end - sigs[i].start);
1158
1159         return sigs[0].start != NULL;
1160 }
1161
1162 static void handle_signed_tag(struct commit *parent, struct commit_extra_header ***tail)
1163 {
1164         struct merge_remote_desc *desc;
1165         struct commit_extra_header *mergetag;
1166         char *buf;
1167         unsigned long size;
1168         enum object_type type;
1169         struct strbuf payload = STRBUF_INIT;
1170         struct strbuf signature = STRBUF_INIT;
1171
1172         desc = merge_remote_util(parent);
1173         if (!desc || !desc->obj)
1174                 return;
1175         buf = read_object_file(&desc->obj->oid, &type, &size);
1176         if (!buf || type != OBJ_TAG)
1177                 goto free_return;
1178         if (!parse_signature(buf, size, &payload, &signature))
1179                 goto free_return;
1180         /*
1181          * We could verify this signature and either omit the tag when
1182          * it does not validate, but the integrator may not have the
1183          * public key of the signer of the tag being merged, while a
1184          * later auditor may have it while auditing, so let's not run
1185          * verify-signed-buffer here for now...
1186          *
1187          * if (verify_signed_buffer(buf, len, buf + len, size - len, ...))
1188          *      warn("warning: signed tag unverified.");
1189          */
1190         CALLOC_ARRAY(mergetag, 1);
1191         mergetag->key = xstrdup("mergetag");
1192         mergetag->value = buf;
1193         mergetag->len = size;
1194
1195         **tail = mergetag;
1196         *tail = &mergetag->next;
1197         strbuf_release(&payload);
1198         strbuf_release(&signature);
1199         return;
1200
1201 free_return:
1202         free(buf);
1203 }
1204
1205 int check_commit_signature(const struct commit *commit, struct signature_check *sigc)
1206 {
1207         struct strbuf payload = STRBUF_INIT;
1208         struct strbuf signature = STRBUF_INIT;
1209         int ret = 1;
1210
1211         sigc->result = 'N';
1212
1213         if (parse_signed_commit(commit, &payload, &signature, the_hash_algo) <= 0)
1214                 goto out;
1215         ret = check_signature(payload.buf, payload.len, signature.buf,
1216                 signature.len, sigc);
1217
1218  out:
1219         strbuf_release(&payload);
1220         strbuf_release(&signature);
1221
1222         return ret;
1223 }
1224
1225 void verify_merge_signature(struct commit *commit, int verbosity,
1226                             int check_trust)
1227 {
1228         char hex[GIT_MAX_HEXSZ + 1];
1229         struct signature_check signature_check;
1230         int ret;
1231         memset(&signature_check, 0, sizeof(signature_check));
1232
1233         ret = check_commit_signature(commit, &signature_check);
1234
1235         find_unique_abbrev_r(hex, &commit->object.oid, DEFAULT_ABBREV);
1236         switch (signature_check.result) {
1237         case 'G':
1238                 if (ret || (check_trust && signature_check.trust_level < TRUST_MARGINAL))
1239                         die(_("Commit %s has an untrusted GPG signature, "
1240                               "allegedly by %s."), hex, signature_check.signer);
1241                 break;
1242         case 'B':
1243                 die(_("Commit %s has a bad GPG signature "
1244                       "allegedly by %s."), hex, signature_check.signer);
1245         default: /* 'N' */
1246                 die(_("Commit %s does not have a GPG signature."), hex);
1247         }
1248         if (verbosity >= 0 && signature_check.result == 'G')
1249                 printf(_("Commit %s has a good GPG signature by %s\n"),
1250                        hex, signature_check.signer);
1251
1252         signature_check_clear(&signature_check);
1253 }
1254
1255 void append_merge_tag_headers(struct commit_list *parents,
1256                               struct commit_extra_header ***tail)
1257 {
1258         while (parents) {
1259                 struct commit *parent = parents->item;
1260                 handle_signed_tag(parent, tail);
1261                 parents = parents->next;
1262         }
1263 }
1264
1265 static void add_extra_header(struct strbuf *buffer,
1266                              struct commit_extra_header *extra)
1267 {
1268         strbuf_addstr(buffer, extra->key);
1269         if (extra->len)
1270                 strbuf_add_lines(buffer, " ", extra->value, extra->len);
1271         else
1272                 strbuf_addch(buffer, '\n');
1273 }
1274
1275 struct commit_extra_header *read_commit_extra_headers(struct commit *commit,
1276                                                       const char **exclude)
1277 {
1278         struct commit_extra_header *extra = NULL;
1279         unsigned long size;
1280         const char *buffer = get_commit_buffer(commit, &size);
1281         extra = read_commit_extra_header_lines(buffer, size, exclude);
1282         unuse_commit_buffer(commit, buffer);
1283         return extra;
1284 }
1285
1286 int for_each_mergetag(each_mergetag_fn fn, struct commit *commit, void *data)
1287 {
1288         struct commit_extra_header *extra, *to_free;
1289         int res = 0;
1290
1291         to_free = read_commit_extra_headers(commit, NULL);
1292         for (extra = to_free; !res && extra; extra = extra->next) {
1293                 if (strcmp(extra->key, "mergetag"))
1294                         continue; /* not a merge tag */
1295                 res = fn(commit, extra, data);
1296         }
1297         free_commit_extra_headers(to_free);
1298         return res;
1299 }
1300
1301 static inline int standard_header_field(const char *field, size_t len)
1302 {
1303         return ((len == 4 && !memcmp(field, "tree", 4)) ||
1304                 (len == 6 && !memcmp(field, "parent", 6)) ||
1305                 (len == 6 && !memcmp(field, "author", 6)) ||
1306                 (len == 9 && !memcmp(field, "committer", 9)) ||
1307                 (len == 8 && !memcmp(field, "encoding", 8)));
1308 }
1309
1310 static int excluded_header_field(const char *field, size_t len, const char **exclude)
1311 {
1312         if (!exclude)
1313                 return 0;
1314
1315         while (*exclude) {
1316                 size_t xlen = strlen(*exclude);
1317                 if (len == xlen && !memcmp(field, *exclude, xlen))
1318                         return 1;
1319                 exclude++;
1320         }
1321         return 0;
1322 }
1323
1324 static struct commit_extra_header *read_commit_extra_header_lines(
1325         const char *buffer, size_t size,
1326         const char **exclude)
1327 {
1328         struct commit_extra_header *extra = NULL, **tail = &extra, *it = NULL;
1329         const char *line, *next, *eof, *eob;
1330         struct strbuf buf = STRBUF_INIT;
1331
1332         for (line = buffer, eob = line + size;
1333              line < eob && *line != '\n';
1334              line = next) {
1335                 next = memchr(line, '\n', eob - line);
1336                 next = next ? next + 1 : eob;
1337                 if (*line == ' ') {
1338                         /* continuation */
1339                         if (it)
1340                                 strbuf_add(&buf, line + 1, next - (line + 1));
1341                         continue;
1342                 }
1343                 if (it)
1344                         it->value = strbuf_detach(&buf, &it->len);
1345                 strbuf_reset(&buf);
1346                 it = NULL;
1347
1348                 eof = memchr(line, ' ', next - line);
1349                 if (!eof)
1350                         eof = next;
1351                 else if (standard_header_field(line, eof - line) ||
1352                          excluded_header_field(line, eof - line, exclude))
1353                         continue;
1354
1355                 CALLOC_ARRAY(it, 1);
1356                 it->key = xmemdupz(line, eof-line);
1357                 *tail = it;
1358                 tail = &it->next;
1359                 if (eof + 1 < next)
1360                         strbuf_add(&buf, eof + 1, next - (eof + 1));
1361         }
1362         if (it)
1363                 it->value = strbuf_detach(&buf, &it->len);
1364         return extra;
1365 }
1366
1367 void free_commit_extra_headers(struct commit_extra_header *extra)
1368 {
1369         while (extra) {
1370                 struct commit_extra_header *next = extra->next;
1371                 free(extra->key);
1372                 free(extra->value);
1373                 free(extra);
1374                 extra = next;
1375         }
1376 }
1377
1378 int commit_tree(const char *msg, size_t msg_len, const struct object_id *tree,
1379                 struct commit_list *parents, struct object_id *ret,
1380                 const char *author, const char *sign_commit)
1381 {
1382         struct commit_extra_header *extra = NULL, **tail = &extra;
1383         int result;
1384
1385         append_merge_tag_headers(parents, &tail);
1386         result = commit_tree_extended(msg, msg_len, tree, parents, ret, author,
1387                                       NULL, sign_commit, extra);
1388         free_commit_extra_headers(extra);
1389         return result;
1390 }
1391
1392 static int find_invalid_utf8(const char *buf, int len)
1393 {
1394         int offset = 0;
1395         static const unsigned int max_codepoint[] = {
1396                 0x7f, 0x7ff, 0xffff, 0x10ffff
1397         };
1398
1399         while (len) {
1400                 unsigned char c = *buf++;
1401                 int bytes, bad_offset;
1402                 unsigned int codepoint;
1403                 unsigned int min_val, max_val;
1404
1405                 len--;
1406                 offset++;
1407
1408                 /* Simple US-ASCII? No worries. */
1409                 if (c < 0x80)
1410                         continue;
1411
1412                 bad_offset = offset-1;
1413
1414                 /*
1415                  * Count how many more high bits set: that's how
1416                  * many more bytes this sequence should have.
1417                  */
1418                 bytes = 0;
1419                 while (c & 0x40) {
1420                         c <<= 1;
1421                         bytes++;
1422                 }
1423
1424                 /*
1425                  * Must be between 1 and 3 more bytes.  Longer sequences result in
1426                  * codepoints beyond U+10FFFF, which are guaranteed never to exist.
1427                  */
1428                 if (bytes < 1 || 3 < bytes)
1429                         return bad_offset;
1430
1431                 /* Do we *have* that many bytes? */
1432                 if (len < bytes)
1433                         return bad_offset;
1434
1435                 /*
1436                  * Place the encoded bits at the bottom of the value and compute the
1437                  * valid range.
1438                  */
1439                 codepoint = (c & 0x7f) >> bytes;
1440                 min_val = max_codepoint[bytes-1] + 1;
1441                 max_val = max_codepoint[bytes];
1442
1443                 offset += bytes;
1444                 len -= bytes;
1445
1446                 /* And verify that they are good continuation bytes */
1447                 do {
1448                         codepoint <<= 6;
1449                         codepoint |= *buf & 0x3f;
1450                         if ((*buf++ & 0xc0) != 0x80)
1451                                 return bad_offset;
1452                 } while (--bytes);
1453
1454                 /* Reject codepoints that are out of range for the sequence length. */
1455                 if (codepoint < min_val || codepoint > max_val)
1456                         return bad_offset;
1457                 /* Surrogates are only for UTF-16 and cannot be encoded in UTF-8. */
1458                 if ((codepoint & 0x1ff800) == 0xd800)
1459                         return bad_offset;
1460                 /* U+xxFFFE and U+xxFFFF are guaranteed non-characters. */
1461                 if ((codepoint & 0xfffe) == 0xfffe)
1462                         return bad_offset;
1463                 /* So are anything in the range U+FDD0..U+FDEF. */
1464                 if (codepoint >= 0xfdd0 && codepoint <= 0xfdef)
1465                         return bad_offset;
1466         }
1467         return -1;
1468 }
1469
1470 /*
1471  * This verifies that the buffer is in proper utf8 format.
1472  *
1473  * If it isn't, it assumes any non-utf8 characters are Latin1,
1474  * and does the conversion.
1475  */
1476 static int verify_utf8(struct strbuf *buf)
1477 {
1478         int ok = 1;
1479         long pos = 0;
1480
1481         for (;;) {
1482                 int bad;
1483                 unsigned char c;
1484                 unsigned char replace[2];
1485
1486                 bad = find_invalid_utf8(buf->buf + pos, buf->len - pos);
1487                 if (bad < 0)
1488                         return ok;
1489                 pos += bad;
1490                 ok = 0;
1491                 c = buf->buf[pos];
1492                 strbuf_remove(buf, pos, 1);
1493
1494                 /* We know 'c' must be in the range 128-255 */
1495                 replace[0] = 0xc0 + (c >> 6);
1496                 replace[1] = 0x80 + (c & 0x3f);
1497                 strbuf_insert(buf, pos, replace, 2);
1498                 pos += 2;
1499         }
1500 }
1501
1502 static const char commit_utf8_warn[] =
1503 N_("Warning: commit message did not conform to UTF-8.\n"
1504    "You may want to amend it after fixing the message, or set the config\n"
1505    "variable i18n.commitencoding to the encoding your project uses.\n");
1506
1507 int commit_tree_extended(const char *msg, size_t msg_len,
1508                          const struct object_id *tree,
1509                          struct commit_list *parents, struct object_id *ret,
1510                          const char *author, const char *committer,
1511                          const char *sign_commit,
1512                          struct commit_extra_header *extra)
1513 {
1514         int result;
1515         int encoding_is_utf8;
1516         struct strbuf buffer;
1517
1518         assert_oid_type(tree, OBJ_TREE);
1519
1520         if (memchr(msg, '\0', msg_len))
1521                 return error("a NUL byte in commit log message not allowed.");
1522
1523         /* Not having i18n.commitencoding is the same as having utf-8 */
1524         encoding_is_utf8 = is_encoding_utf8(git_commit_encoding);
1525
1526         strbuf_init(&buffer, 8192); /* should avoid reallocs for the headers */
1527         strbuf_addf(&buffer, "tree %s\n", oid_to_hex(tree));
1528
1529         /*
1530          * NOTE! This ordering means that the same exact tree merged with a
1531          * different order of parents will be a _different_ changeset even
1532          * if everything else stays the same.
1533          */
1534         while (parents) {
1535                 struct commit *parent = pop_commit(&parents);
1536                 strbuf_addf(&buffer, "parent %s\n",
1537                             oid_to_hex(&parent->object.oid));
1538         }
1539
1540         /* Person/date information */
1541         if (!author)
1542                 author = git_author_info(IDENT_STRICT);
1543         strbuf_addf(&buffer, "author %s\n", author);
1544         if (!committer)
1545                 committer = git_committer_info(IDENT_STRICT);
1546         strbuf_addf(&buffer, "committer %s\n", committer);
1547         if (!encoding_is_utf8)
1548                 strbuf_addf(&buffer, "encoding %s\n", git_commit_encoding);
1549
1550         while (extra) {
1551                 add_extra_header(&buffer, extra);
1552                 extra = extra->next;
1553         }
1554         strbuf_addch(&buffer, '\n');
1555
1556         /* And add the comment */
1557         strbuf_add(&buffer, msg, msg_len);
1558
1559         /* And check the encoding */
1560         if (encoding_is_utf8 && !verify_utf8(&buffer))
1561                 fprintf(stderr, _(commit_utf8_warn));
1562
1563         if (sign_commit && sign_with_header(&buffer, sign_commit)) {
1564                 result = -1;
1565                 goto out;
1566         }
1567
1568         result = write_object_file(buffer.buf, buffer.len, commit_type, ret);
1569 out:
1570         strbuf_release(&buffer);
1571         return result;
1572 }
1573
1574 define_commit_slab(merge_desc_slab, struct merge_remote_desc *);
1575 static struct merge_desc_slab merge_desc_slab = COMMIT_SLAB_INIT(1, merge_desc_slab);
1576
1577 struct merge_remote_desc *merge_remote_util(struct commit *commit)
1578 {
1579         return *merge_desc_slab_at(&merge_desc_slab, commit);
1580 }
1581
1582 void set_merge_remote_desc(struct commit *commit,
1583                            const char *name, struct object *obj)
1584 {
1585         struct merge_remote_desc *desc;
1586         FLEX_ALLOC_STR(desc, name, name);
1587         desc->obj = obj;
1588         *merge_desc_slab_at(&merge_desc_slab, commit) = desc;
1589 }
1590
1591 struct commit *get_merge_parent(const char *name)
1592 {
1593         struct object *obj;
1594         struct commit *commit;
1595         struct object_id oid;
1596         if (get_oid(name, &oid))
1597                 return NULL;
1598         obj = parse_object(the_repository, &oid);
1599         commit = (struct commit *)peel_to_type(name, 0, obj, OBJ_COMMIT);
1600         if (commit && !merge_remote_util(commit))
1601                 set_merge_remote_desc(commit, name, obj);
1602         return commit;
1603 }
1604
1605 /*
1606  * Append a commit to the end of the commit_list.
1607  *
1608  * next starts by pointing to the variable that holds the head of an
1609  * empty commit_list, and is updated to point to the "next" field of
1610  * the last item on the list as new commits are appended.
1611  *
1612  * Usage example:
1613  *
1614  *     struct commit_list *list;
1615  *     struct commit_list **next = &list;
1616  *
1617  *     next = commit_list_append(c1, next);
1618  *     next = commit_list_append(c2, next);
1619  *     assert(commit_list_count(list) == 2);
1620  *     return list;
1621  */
1622 struct commit_list **commit_list_append(struct commit *commit,
1623                                         struct commit_list **next)
1624 {
1625         struct commit_list *new_commit = xmalloc(sizeof(struct commit_list));
1626         new_commit->item = commit;
1627         *next = new_commit;
1628         new_commit->next = NULL;
1629         return &new_commit->next;
1630 }
1631
1632 const char *find_commit_header(const char *msg, const char *key, size_t *out_len)
1633 {
1634         int key_len = strlen(key);
1635         const char *line = msg;
1636
1637         while (line) {
1638                 const char *eol = strchrnul(line, '\n');
1639
1640                 if (line == eol)
1641                         return NULL;
1642
1643                 if (eol - line > key_len &&
1644                     !strncmp(line, key, key_len) &&
1645                     line[key_len] == ' ') {
1646                         *out_len = eol - line - key_len - 1;
1647                         return line + key_len + 1;
1648                 }
1649                 line = *eol ? eol + 1 : NULL;
1650         }
1651         return NULL;
1652 }
1653
1654 /*
1655  * Inspect the given string and determine the true "end" of the log message, in
1656  * order to find where to put a new Signed-off-by trailer.  Ignored are
1657  * trailing comment lines and blank lines.  To support "git commit -s
1658  * --amend" on an existing commit, we also ignore "Conflicts:".  To
1659  * support "git commit -v", we truncate at cut lines.
1660  *
1661  * Returns the number of bytes from the tail to ignore, to be fed as
1662  * the second parameter to append_signoff().
1663  */
1664 size_t ignore_non_trailer(const char *buf, size_t len)
1665 {
1666         size_t boc = 0;
1667         size_t bol = 0;
1668         int in_old_conflicts_block = 0;
1669         size_t cutoff = wt_status_locate_end(buf, len);
1670
1671         while (bol < cutoff) {
1672                 const char *next_line = memchr(buf + bol, '\n', len - bol);
1673
1674                 if (!next_line)
1675                         next_line = buf + len;
1676                 else
1677                         next_line++;
1678
1679                 if (buf[bol] == comment_line_char || buf[bol] == '\n') {
1680                         /* is this the first of the run of comments? */
1681                         if (!boc)
1682                                 boc = bol;
1683                         /* otherwise, it is just continuing */
1684                 } else if (starts_with(buf + bol, "Conflicts:\n")) {
1685                         in_old_conflicts_block = 1;
1686                         if (!boc)
1687                                 boc = bol;
1688                 } else if (in_old_conflicts_block && buf[bol] == '\t') {
1689                         ; /* a pathname in the conflicts block */
1690                 } else if (boc) {
1691                         /* the previous was not trailing comment */
1692                         boc = 0;
1693                         in_old_conflicts_block = 0;
1694                 }
1695                 bol = next_line - buf;
1696         }
1697         return boc ? len - boc : len - cutoff;
1698 }
1699
1700 int run_commit_hook(int editor_is_used, const char *index_file,
1701                     const char *name, ...)
1702 {
1703         struct strvec hook_env = STRVEC_INIT;
1704         va_list args;
1705         int ret;
1706
1707         strvec_pushf(&hook_env, "GIT_INDEX_FILE=%s", index_file);
1708
1709         /*
1710          * Let the hook know that no editor will be launched.
1711          */
1712         if (!editor_is_used)
1713                 strvec_push(&hook_env, "GIT_EDITOR=:");
1714
1715         va_start(args, name);
1716         ret = run_hook_ve(hook_env.v, name, args);
1717         va_end(args);
1718         strvec_clear(&hook_env);
1719
1720         return ret;
1721 }