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 #include "hook.h"
  25
  26 static struct commit_extra_header *read_commit_extra_header_lines(const char *buf, size_t len, const char **);
  27
  28 int save_commit_buffer = 1;
  29 int no_graft_file_deprecated_advice;
  30
  31 const char *commit_type = "commit";
  32
  33 struct commit *lookup_commit_reference_gently(struct repository *r,
  34                 const struct object_id *oid, int quiet)
  35 {
  36         struct object *obj = deref_tag(r,
  37                                        parse_object(r, oid),
  38                                        NULL, 0);
  39
  40         if (!obj)
  41                 return NULL;
  42         return object_as_type(obj, OBJ_COMMIT, quiet);
  43 }
  44
  45 struct commit *lookup_commit_reference(struct repository *r, const struct object_id *oid)
  46 {
  47         return lookup_commit_reference_gently(r, oid, 0);
  48 }
  49
  50 struct commit *lookup_commit_or_die(const struct object_id *oid, const char *ref_name)
  51 {
  52         struct commit *c = lookup_commit_reference(the_repository, oid);
  53         if (!c)
  54                 die(_("could not parse %s"), ref_name);
  55         if (!oideq(oid, &c->object.oid)) {
  56                 warning(_("%s %s is not a commit!"),
  57                         ref_name, oid_to_hex(oid));
  58         }
  59         return c;
  60 }
  61
  62 struct commit *lookup_commit_object(struct repository *r,
  63                                     const struct object_id *oid)
  64 {
  65         struct object *obj = parse_object(r, oid);
  66         return obj ? object_as_type(obj, OBJ_COMMIT, 0) : NULL;
  67
  68 }
  69
  70 struct commit *lookup_commit(struct repository *r, const struct object_id *oid)
  71 {
  72         struct object *obj = lookup_object(r, oid);
  73         if (!obj)
  74                 return create_object(r, oid, alloc_commit_node(r));
  75         return object_as_type(obj, OBJ_COMMIT, 0);
  76 }
  77
  78 struct commit *lookup_commit_reference_by_name(const char *name)
  79 {
  80         struct object_id oid;
  81         struct commit *commit;
  82
  83         if (get_oid_committish(name, &oid))
  84                 return NULL;
  85         commit = lookup_commit_reference(the_repository, &oid);
  86         if (parse_commit(commit))
  87                 return NULL;
  88         return commit;
  89 }
  90
  91 static timestamp_t parse_commit_date(const char *buf, const char *tail)
  92 {
  93         const char *dateptr;
  94
  95         if (buf + 6 >= tail)
  96                 return 0;
  97         if (memcmp(buf, "author", 6))
  98                 return 0;
  99         while (buf < tail && *buf++ != '\n')
 100                 /* nada */;
 101         if (buf + 9 >= tail)
 102                 return 0;
 103         if (memcmp(buf, "committer", 9))
 104                 return 0;
 105         while (buf < tail && *buf++ != '>')
 106                 /* nada */;
 107         if (buf >= tail)
 108                 return 0;
 109         dateptr = buf;
 110         while (buf < tail && *buf++ != '\n')
 111                 /* nada */;
 112         if (buf >= tail)
 113                 return 0;
 114         /* dateptr < buf && buf[-1] == '\n', so parsing will stop at buf-1 */
 115         return parse_timestamp(dateptr, NULL, 10);
 116 }
 117
 118 static const struct object_id *commit_graft_oid_access(size_t index, const void *table)
 119 {
 120         const struct commit_graft * const *commit_graft_table = table;
 121         return &commit_graft_table[index]->oid;
 122 }
 123
 124 int commit_graft_pos(struct repository *r, const struct object_id *oid)
 125 {
 126         return oid_pos(oid, r->parsed_objects->grafts,
 127                        r->parsed_objects->grafts_nr,
 128                        commit_graft_oid_access);
 129 }
 130
 131 static void unparse_commit(struct repository *r, const struct object_id *oid)
 132 {
 133         struct commit *c = lookup_commit(r, oid);
 134
 135         if (!c->object.parsed)
 136                 return;
 137         free_commit_list(c->parents);
 138         c->parents = NULL;
 139         c->object.parsed = 0;
 140 }
 141
 142 int register_commit_graft(struct repository *r, struct commit_graft *graft,
 143                           int ignore_dups)
 144 {
 145         int pos = commit_graft_pos(r, &graft->oid);
 146
 147         if (0 <= pos) {
 148                 if (ignore_dups)
 149                         free(graft);
 150                 else {
 151                         free(r->parsed_objects->grafts[pos]);
 152                         r->parsed_objects->grafts[pos] = graft;
 153                 }
 154                 return 1;
 155         }
 156         pos = -pos - 1;
 157         ALLOC_GROW(r->parsed_objects->grafts,
 158                    r->parsed_objects->grafts_nr + 1,
 159                    r->parsed_objects->grafts_alloc);
 160         r->parsed_objects->grafts_nr++;
 161         if (pos < r->parsed_objects->grafts_nr)
 162                 memmove(r->parsed_objects->grafts + pos + 1,
 163                         r->parsed_objects->grafts + pos,
 164                         (r->parsed_objects->grafts_nr - pos - 1) *
 165                         sizeof(*r->parsed_objects->grafts));
 166         r->parsed_objects->grafts[pos] = graft;
 167         unparse_commit(r, &graft->oid);
 168         return 0;
 169 }
 170
 171 struct commit_graft *read_graft_line(struct strbuf *line)
 172 {
 173         /* The format is just "Commit Parent1 Parent2 ...\n" */
 174         int i, phase;
 175         const char *tail = NULL;
 176         struct commit_graft *graft = NULL;
 177         struct object_id dummy_oid, *oid;
 178
 179         strbuf_rtrim(line);
 180         if (!line->len || line->buf[0] == '#')
 181                 return NULL;
 182         /*
 183          * phase 0 verifies line, counts hashes in line and allocates graft
 184          * phase 1 fills graft
 185          */
 186         for (phase = 0; phase < 2; phase++) {
 187                 oid = graft ? &graft->oid : &dummy_oid;
 188                 if (parse_oid_hex(line->buf, oid, &tail))
 189                         goto bad_graft_data;
 190                 for (i = 0; *tail != '\0'; i++) {
 191                         oid = graft ? &graft->parent[i] : &dummy_oid;
 192                         if (!isspace(*tail++) || parse_oid_hex(tail, oid, &tail))
 193                                 goto bad_graft_data;
 194                 }
 195                 if (!graft) {
 196                         graft = xmalloc(st_add(sizeof(*graft),
 197                                                st_mult(sizeof(struct object_id), i)));
 198                         graft->nr_parent = i;
 199                 }
 200         }
 201         return graft;
 202
 203 bad_graft_data:
 204         error("bad graft data: %s", line->buf);
 205         assert(!graft);
 206         return NULL;
 207 }
 208
 209 static int read_graft_file(struct repository *r, const char *graft_file)
 210 {
 211         FILE *fp = fopen_or_warn(graft_file, "r");
 212         struct strbuf buf = STRBUF_INIT;
 213         if (!fp)
 214                 return -1;
 215         if (!no_graft_file_deprecated_advice &&
 216             advice_enabled(ADVICE_GRAFT_FILE_DEPRECATED))
 217                 advise(_("Support for <GIT_DIR>/info/grafts is deprecated\n"
 218                          "and will be removed in a future Git version.\n"
 219                          "\n"
 220                          "Please use \"git replace --convert-graft-file\"\n"
 221                          "to convert the grafts into replace refs.\n"
 222                          "\n"
 223                          "Turn this message off by running\n"
 224                          "\"git config advice.graftFileDeprecated false\""));
 225         while (!strbuf_getwholeline(&buf, fp, '\n')) {
 226                 /* The format is just "Commit Parent1 Parent2 ...\n" */
 227                 struct commit_graft *graft = read_graft_line(&buf);
 228                 if (!graft)
 229                         continue;
 230                 if (register_commit_graft(r, graft, 1))
 231                         error("duplicate graft data: %s", buf.buf);
 232         }
 233         fclose(fp);
 234         strbuf_release(&buf);
 235         return 0;
 236 }
 237
 238 void prepare_commit_graft(struct repository *r)
 239 {
 240         char *graft_file;
 241
 242         if (r->parsed_objects->commit_graft_prepared)
 243                 return;
 244         if (!startup_info->have_repository)
 245                 return;
 246
 247         graft_file = get_graft_file(r);
 248         read_graft_file(r, graft_file);
 249         /* make sure shallows are read */
 250         is_repository_shallow(r);
 251         r->parsed_objects->commit_graft_prepared = 1;
 252 }
 253
 254 struct commit_graft *lookup_commit_graft(struct repository *r, const struct object_id *oid)
 255 {
 256         int pos;
 257         prepare_commit_graft(r);
 258         pos = commit_graft_pos(r, oid);
 259         if (pos < 0)
 260                 return NULL;
 261         return r->parsed_objects->grafts[pos];
 262 }
 263
 264 int for_each_commit_graft(each_commit_graft_fn fn, void *cb_data)
 265 {
 266         int i, ret;
 267         for (i = ret = 0; i < the_repository->parsed_objects->grafts_nr && !ret; i++)
 268                 ret = fn(the_repository->parsed_objects->grafts[i], cb_data);
 269         return ret;
 270 }
 271
 272 void reset_commit_grafts(struct repository *r)
 273 {
 274         int i;
 275
 276         for (i = 0; i < r->parsed_objects->grafts_nr; i++) {
 277                 unparse_commit(r, &r->parsed_objects->grafts[i]->oid);
 278                 free(r->parsed_objects->grafts[i]);
 279         }
 280         r->parsed_objects->grafts_nr = 0;
 281         r->parsed_objects->commit_graft_prepared = 0;
 282 }
 283
 284 struct commit_buffer {
 285         void *buffer;
 286         unsigned long size;
 287 };
 288 define_commit_slab(buffer_slab, struct commit_buffer);
 289
 290 struct buffer_slab *allocate_commit_buffer_slab(void)
 291 {
 292         struct buffer_slab *bs = xmalloc(sizeof(*bs));
 293         init_buffer_slab(bs);
 294         return bs;
 295 }
 296
 297 void free_commit_buffer_slab(struct buffer_slab *bs)
 298 {
 299         clear_buffer_slab(bs);
 300         free(bs);
 301 }
 302
 303 void set_commit_buffer(struct repository *r, struct commit *commit, void *buffer, unsigned long size)
 304 {
 305         struct commit_buffer *v = buffer_slab_at(
 306                 r->parsed_objects->buffer_slab, commit);
 307         v->buffer = buffer;
 308         v->size = size;
 309 }
 310
 311 const void *get_cached_commit_buffer(struct repository *r, const struct commit *commit, unsigned long *sizep)
 312 {
 313         struct commit_buffer *v = buffer_slab_peek(
 314                 r->parsed_objects->buffer_slab, commit);
 315         if (!v) {
 316                 if (sizep)
 317                         *sizep = 0;
 318                 return NULL;
 319         }
 320         if (sizep)
 321                 *sizep = v->size;
 322         return v->buffer;
 323 }
 324
 325 const void *repo_get_commit_buffer(struct repository *r,
 326                                    const struct commit *commit,
 327                                    unsigned long *sizep)
 328 {
 329         const void *ret = get_cached_commit_buffer(r, commit, sizep);
 330         if (!ret) {
 331                 enum object_type type;
 332                 unsigned long size;
 333                 ret = repo_read_object_file(r, &commit->object.oid, &type, &size);
 334                 if (!ret)
 335                         die("cannot read commit object %s",
 336                             oid_to_hex(&commit->object.oid));
 337                 if (type != OBJ_COMMIT)
 338                         die("expected commit for %s, got %s",
 339                             oid_to_hex(&commit->object.oid), type_name(type));
 340                 if (sizep)
 341                         *sizep = size;
 342         }
 343         return ret;
 344 }
 345
 346 void repo_unuse_commit_buffer(struct repository *r,
 347                               const struct commit *commit,
 348                               const void *buffer)
 349 {
 350         struct commit_buffer *v = buffer_slab_peek(
 351                 r->parsed_objects->buffer_slab, commit);
 352         if (!(v && v->buffer == buffer))
 353                 free((void *)buffer);
 354 }
 355
 356 void free_commit_buffer(struct parsed_object_pool *pool, struct commit *commit)
 357 {
 358         struct commit_buffer *v = buffer_slab_peek(
 359                 pool->buffer_slab, commit);
 360         if (v) {
 361                 FREE_AND_NULL(v->buffer);
 362                 v->size = 0;
 363         }
 364 }
 365
 366 static inline void set_commit_tree(struct commit *c, struct tree *t)
 367 {
 368         c->maybe_tree = t;
 369 }
 370
 371 struct tree *repo_get_commit_tree(struct repository *r,
 372                                   const struct commit *commit)
 373 {
 374         if (commit->maybe_tree || !commit->object.parsed)
 375                 return commit->maybe_tree;
 376
 377         if (commit_graph_position(commit) != COMMIT_NOT_FROM_GRAPH)
 378                 return get_commit_tree_in_graph(r, commit);
 379
 380         return NULL;
 381 }
 382
 383 struct object_id *get_commit_tree_oid(const struct commit *commit)
 384 {
 385         struct tree *tree = get_commit_tree(commit);
 386         return tree ? &tree->object.oid : NULL;
 387 }
 388
 389 void release_commit_memory(struct parsed_object_pool *pool, struct commit *c)
 390 {
 391         set_commit_tree(c, NULL);
 392         free_commit_buffer(pool, c);
 393         c->index = 0;
 394         free_commit_list(c->parents);
 395
 396         c->object.parsed = 0;
 397 }
 398
 399 const void *detach_commit_buffer(struct commit *commit, unsigned long *sizep)
 400 {
 401         struct commit_buffer *v = buffer_slab_peek(
 402                 the_repository->parsed_objects->buffer_slab, commit);
 403         void *ret;
 404
 405         if (!v) {
 406                 if (sizep)
 407                         *sizep = 0;
 408                 return NULL;
 409         }
 410         ret = v->buffer;
 411         if (sizep)
 412                 *sizep = v->size;
 413
 414         v->buffer = NULL;
 415         v->size = 0;
 416         return ret;
 417 }
 418
 419 int parse_commit_buffer(struct repository *r, struct commit *item, const void *buffer, unsigned long size, int check_graph)
 420 {
 421         const char *tail = buffer;
 422         const char *bufptr = buffer;
 423         struct object_id parent;
 424         struct commit_list **pptr;
 425         struct commit_graft *graft;
 426         const int tree_entry_len = the_hash_algo->hexsz + 5;
 427         const int parent_entry_len = the_hash_algo->hexsz + 7;
 428         struct tree *tree;
 429
 430         if (item->object.parsed)
 431                 return 0;
 432         /*
 433          * Presumably this is leftover from an earlier failed parse;
 434          * clear it out in preparation for us re-parsing (we'll hit the
 435          * same error, but that's good, since it lets our caller know
 436          * the result cannot be trusted.
 437          */
 438         free_commit_list(item->parents);
 439         item->parents = NULL;
 440
 441         tail += size;
 442         if (tail <= bufptr + tree_entry_len + 1 || memcmp(bufptr, "tree ", 5) ||
 443                         bufptr[tree_entry_len] != '\n')
 444                 return error("bogus commit object %s", oid_to_hex(&item->object.oid));
 445         if (get_oid_hex(bufptr + 5, &parent) < 0)
 446                 return error("bad tree pointer in commit %s",
 447                              oid_to_hex(&item->object.oid));
 448         tree = lookup_tree(r, &parent);
 449         if (!tree)
 450                 return error("bad tree pointer %s in commit %s",
 451                              oid_to_hex(&parent),
 452                              oid_to_hex(&item->object.oid));
 453         set_commit_tree(item, tree);
 454         bufptr += tree_entry_len + 1; /* "tree " + "hex sha1" + "\n" */
 455         pptr = &item->parents;
 456
 457         graft = lookup_commit_graft(r, &item->object.oid);
 458         if (graft)
 459                 r->parsed_objects->substituted_parent = 1;
 460         while (bufptr + parent_entry_len < tail && !memcmp(bufptr, "parent ", 7)) {
 461                 struct commit *new_parent;
 462
 463                 if (tail <= bufptr + parent_entry_len + 1 ||
 464                     get_oid_hex(bufptr + 7, &parent) ||
 465                     bufptr[parent_entry_len] != '\n')
 466                         return error("bad parents in commit %s", oid_to_hex(&item->object.oid));
 467                 bufptr += parent_entry_len + 1;
 468                 /*
 469                  * The clone is shallow if nr_parent < 0, and we must
 470                  * not traverse its real parents even when we unhide them.
 471                  */
 472                 if (graft && (graft->nr_parent < 0 || grafts_replace_parents))
 473                         continue;
 474                 new_parent = lookup_commit(r, &parent);
 475                 if (!new_parent)
 476                         return error("bad parent %s in commit %s",
 477                                      oid_to_hex(&parent),
 478                                      oid_to_hex(&item->object.oid));
 479                 pptr = &commit_list_insert(new_parent, pptr)->next;
 480         }
 481         if (graft) {
 482                 int i;
 483                 struct commit *new_parent;
 484                 for (i = 0; i < graft->nr_parent; i++) {
 485                         new_parent = lookup_commit(r,
 486                                                    &graft->parent[i]);
 487                         if (!new_parent)
 488                                 return error("bad graft parent %s in commit %s",
 489                                              oid_to_hex(&graft->parent[i]),
 490                                              oid_to_hex(&item->object.oid));
 491                         pptr = &commit_list_insert(new_parent, pptr)->next;
 492                 }
 493         }
 494         item->date = parse_commit_date(bufptr, tail);
 495
 496         if (check_graph)
 497                 load_commit_graph_info(r, item);
 498
 499         item->object.parsed = 1;
 500         return 0;
 501 }
 502
 503 int repo_parse_commit_internal(struct repository *r,
 504                                struct commit *item,
 505                                int quiet_on_missing,
 506                                int use_commit_graph)
 507 {
 508         enum object_type type;
 509         void *buffer;
 510         unsigned long size;
 511         int ret;
 512
 513         if (!item)
 514                 return -1;
 515         if (item->object.parsed)
 516                 return 0;
 517         if (use_commit_graph && parse_commit_in_graph(r, item))
 518                 return 0;
 519         buffer = repo_read_object_file(r, &item->object.oid, &type, &size);
 520         if (!buffer)
 521                 return quiet_on_missing ? -1 :
 522                         error("Could not read %s",
 523                              oid_to_hex(&item->object.oid));
 524         if (type != OBJ_COMMIT) {
 525                 free(buffer);
 526                 return error("Object %s not a commit",
 527                              oid_to_hex(&item->object.oid));
 528         }
 529
 530         ret = parse_commit_buffer(r, item, buffer, size, 0);
 531         if (save_commit_buffer && !ret) {
 532                 set_commit_buffer(r, item, buffer, size);
 533                 return 0;
 534         }
 535         free(buffer);
 536         return ret;
 537 }
 538
 539 int repo_parse_commit_gently(struct repository *r,
 540                              struct commit *item, int quiet_on_missing)
 541 {
 542         return repo_parse_commit_internal(r, item, quiet_on_missing, 1);
 543 }
 544
 545 void parse_commit_or_die(struct commit *item)
 546 {
 547         if (parse_commit(item))
 548                 die("unable to parse commit %s",
 549                     item ? oid_to_hex(&item->object.oid) : "(null)");
 550 }
 551
 552 int find_commit_subject(const char *commit_buffer, const char **subject)
 553 {
 554         const char *eol;
 555         const char *p = commit_buffer;
 556
 557         while (*p && (*p != '\n' || p[1] != '\n'))
 558                 p++;
 559         if (*p) {
 560                 p = skip_blank_lines(p + 2);
 561                 eol = strchrnul(p, '\n');
 562         } else
 563                 eol = p;
 564
 565         *subject = p;
 566
 567         return eol - p;
 568 }
 569
 570 size_t commit_subject_length(const char *body)
 571 {
 572         const char *p = body;
 573         while (*p) {
 574                 const char *next = skip_blank_lines(p);
 575                 if (next != p)
 576                         break;
 577                 p = strchrnul(p, '\n');
 578                 if (*p)
 579                         p++;
 580         }
 581         return p - body;
 582 }
 583
 584 struct commit_list *commit_list_insert(struct commit *item, struct commit_list **list_p)
 585 {
 586         struct commit_list *new_list = xmalloc(sizeof(struct commit_list));
 587         new_list->item = item;
 588         new_list->next = *list_p;
 589         *list_p = new_list;
 590         return new_list;
 591 }
 592
 593 int commit_list_contains(struct commit *item, struct commit_list *list)
 594 {
 595         while (list) {
 596                 if (list->item == item)
 597                         return 1;
 598                 list = list->next;
 599         }
 600
 601         return 0;
 602 }
 603
 604 unsigned commit_list_count(const struct commit_list *l)
 605 {
 606         unsigned c = 0;
 607         for (; l; l = l->next )
 608                 c++;
 609         return c;
 610 }
 611
 612 struct commit_list *copy_commit_list(struct commit_list *list)
 613 {
 614         struct commit_list *head = NULL;
 615         struct commit_list **pp = &head;
 616         while (list) {
 617                 pp = commit_list_append(list->item, pp);
 618                 list = list->next;
 619         }
 620         return head;
 621 }
 622
 623 struct commit_list *reverse_commit_list(struct commit_list *list)
 624 {
 625         struct commit_list *next = NULL, *current, *backup;
 626         for (current = list; current; current = backup) {
 627                 backup = current->next;
 628                 current->next = next;
 629                 next = current;
 630         }
 631         return next;
 632 }
 633
 634 void free_commit_list(struct commit_list *list)
 635 {
 636         while (list)
 637                 pop_commit(&list);
 638 }
 639
 640 struct commit_list * commit_list_insert_by_date(struct commit *item, struct commit_list **list)
 641 {
 642         struct commit_list **pp = list;
 643         struct commit_list *p;
 644         while ((p = *pp) != NULL) {
 645                 if (p->item->date < item->date) {
 646                         break;
 647                 }
 648                 pp = &p->next;
 649         }
 650         return commit_list_insert(item, pp);
 651 }
 652
 653 static int commit_list_compare_by_date(const struct commit_list *a,
 654                                        const struct commit_list *b)
 655 {
 656         timestamp_t a_date = a->item->date;
 657         timestamp_t b_date = b->item->date;
 658         if (a_date < b_date)
 659                 return 1;
 660         if (a_date > b_date)
 661                 return -1;
 662         return 0;
 663 }
 664
 665 DEFINE_LIST_SORT(static, commit_list_sort, struct commit_list, next);
 666
 667 void commit_list_sort_by_date(struct commit_list **list)
 668 {
 669         commit_list_sort(list, commit_list_compare_by_date);
 670 }
 671
 672 struct commit *pop_most_recent_commit(struct commit_list **list,
 673                                       unsigned int mark)
 674 {
 675         struct commit *ret = pop_commit(list);
 676         struct commit_list *parents = ret->parents;
 677
 678         while (parents) {
 679                 struct commit *commit = parents->item;
 680                 if (!parse_commit(commit) && !(commit->object.flags & mark)) {
 681                         commit->object.flags |= mark;
 682                         commit_list_insert_by_date(commit, list);
 683                 }
 684                 parents = parents->next;
 685         }
 686         return ret;
 687 }
 688
 689 static void clear_commit_marks_1(struct commit_list **plist,
 690                                  struct commit *commit, unsigned int mark)
 691 {
 692         while (commit) {
 693                 struct commit_list *parents;
 694
 695                 if (!(mark & commit->object.flags))
 696                         return;
 697
 698                 commit->object.flags &= ~mark;
 699
 700                 parents = commit->parents;
 701                 if (!parents)
 702                         return;
 703
 704                 while ((parents = parents->next))
 705                         commit_list_insert(parents->item, plist);
 706
 707                 commit = commit->parents->item;
 708         }
 709 }
 710
 711 void clear_commit_marks_many(int nr, struct commit **commit, unsigned int mark)
 712 {
 713         struct commit_list *list = NULL;
 714
 715         while (nr--) {
 716                 clear_commit_marks_1(&list, *commit, mark);
 717                 commit++;
 718         }
 719         while (list)
 720                 clear_commit_marks_1(&list, pop_commit(&list), mark);
 721 }
 722
 723 void clear_commit_marks(struct commit *commit, unsigned int mark)
 724 {
 725         clear_commit_marks_many(1, &commit, mark);
 726 }
 727
 728 struct commit *pop_commit(struct commit_list **stack)
 729 {
 730         struct commit_list *top = *stack;
 731         struct commit *item = top ? top->item : NULL;
 732
 733         if (top) {
 734                 *stack = top->next;
 735                 free(top);
 736         }
 737         return item;
 738 }
 739
 740 /*
 741  * Topological sort support
 742  */
 743
 744 /* count number of children that have not been emitted */
 745 define_commit_slab(indegree_slab, int);
 746
 747 define_commit_slab(author_date_slab, timestamp_t);
 748
 749 void record_author_date(struct author_date_slab *author_date,
 750                         struct commit *commit)
 751 {
 752         const char *buffer = get_commit_buffer(commit, NULL);
 753         struct ident_split ident;
 754         const char *ident_line;
 755         size_t ident_len;
 756         char *date_end;
 757         timestamp_t date;
 758
 759         ident_line = find_commit_header(buffer, "author", &ident_len);
 760         if (!ident_line)
 761                 goto fail_exit; /* no author line */
 762         if (split_ident_line(&ident, ident_line, ident_len) ||
 763             !ident.date_begin || !ident.date_end)
 764                 goto fail_exit; /* malformed "author" line */
 765
 766         date = parse_timestamp(ident.date_begin, &date_end, 10);
 767         if (date_end != ident.date_end)
 768                 goto fail_exit; /* malformed date */
 769         *(author_date_slab_at(author_date, commit)) = date;
 770
 771 fail_exit:
 772         unuse_commit_buffer(commit, buffer);
 773 }
 774
 775 int compare_commits_by_author_date(const void *a_, const void *b_,
 776                                    void *cb_data)
 777 {
 778         const struct commit *a = a_, *b = b_;
 779         struct author_date_slab *author_date = cb_data;
 780         timestamp_t a_date = *(author_date_slab_at(author_date, a));
 781         timestamp_t b_date = *(author_date_slab_at(author_date, b));
 782
 783         /* newer commits with larger date first */
 784         if (a_date < b_date)
 785                 return 1;
 786         else if (a_date > b_date)
 787                 return -1;
 788         return 0;
 789 }
 790
 791 int compare_commits_by_gen_then_commit_date(const void *a_, const void *b_, void *unused)
 792 {
 793         const struct commit *a = a_, *b = b_;
 794         const timestamp_t generation_a = commit_graph_generation(a),
 795                           generation_b = commit_graph_generation(b);
 796
 797         /* newer commits first */
 798         if (generation_a < generation_b)
 799                 return 1;
 800         else if (generation_a > generation_b)
 801                 return -1;
 802
 803         /* use date as a heuristic when generations are equal */
 804         if (a->date < b->date)
 805                 return 1;
 806         else if (a->date > b->date)
 807                 return -1;
 808         return 0;
 809 }
 810
 811 int compare_commits_by_commit_date(const void *a_, const void *b_, void *unused)
 812 {
 813         const struct commit *a = a_, *b = b_;
 814         /* newer commits with larger date first */
 815         if (a->date < b->date)
 816                 return 1;
 817         else if (a->date > b->date)
 818                 return -1;
 819         return 0;
 820 }
 821
 822 /*
 823  * Performs an in-place topological sort on the list supplied.
 824  */
 825 void sort_in_topological_order(struct commit_list **list, enum rev_sort_order sort_order)
 826 {
 827         struct commit_list *next, *orig = *list;
 828         struct commit_list **pptr;
 829         struct indegree_slab indegree;
 830         struct prio_queue queue;
 831         struct commit *commit;
 832         struct author_date_slab author_date;
 833
 834         if (!orig)
 835                 return;
 836         *list = NULL;
 837
 838         init_indegree_slab(&indegree);
 839         memset(&queue, '\0', sizeof(queue));
 840
 841         switch (sort_order) {
 842         default: /* REV_SORT_IN_GRAPH_ORDER */
 843                 queue.compare = NULL;
 844                 break;
 845         case REV_SORT_BY_COMMIT_DATE:
 846                 queue.compare = compare_commits_by_commit_date;
 847                 break;
 848         case REV_SORT_BY_AUTHOR_DATE:
 849                 init_author_date_slab(&author_date);
 850                 queue.compare = compare_commits_by_author_date;
 851                 queue.cb_data = &author_date;
 852                 break;
 853         }
 854
 855         /* Mark them and clear the indegree */
 856         for (next = orig; next; next = next->next) {
 857                 struct commit *commit = next->item;
 858                 *(indegree_slab_at(&indegree, commit)) = 1;
 859                 /* also record the author dates, if needed */
 860                 if (sort_order == REV_SORT_BY_AUTHOR_DATE)
 861                         record_author_date(&author_date, commit);
 862         }
 863
 864         /* update the indegree */
 865         for (next = orig; next; next = next->next) {
 866                 struct commit_list *parents = next->item->parents;
 867                 while (parents) {
 868                         struct commit *parent = parents->item;
 869                         int *pi = indegree_slab_at(&indegree, parent);
 870
 871                         if (*pi)
 872                                 (*pi)++;
 873                         parents = parents->next;
 874                 }
 875         }
 876
 877         /*
 878          * find the tips
 879          *
 880          * tips are nodes not reachable from any other node in the list
 881          *
 882          * the tips serve as a starting set for the work queue.
 883          */
 884         for (next = orig; next; next = next->next) {
 885                 struct commit *commit = next->item;
 886
 887                 if (*(indegree_slab_at(&indegree, commit)) == 1)
 888                         prio_queue_put(&queue, commit);
 889         }
 890
 891         /*
 892          * This is unfortunate; the initial tips need to be shown
 893          * in the order given from the revision traversal machinery.
 894          */
 895         if (sort_order == REV_SORT_IN_GRAPH_ORDER)
 896                 prio_queue_reverse(&queue);
 897
 898         /* We no longer need the commit list */
 899         free_commit_list(orig);
 900
 901         pptr = list;
 902         *list = NULL;
 903         while ((commit = prio_queue_get(&queue)) != NULL) {
 904                 struct commit_list *parents;
 905
 906                 for (parents = commit->parents; parents ; parents = parents->next) {
 907                         struct commit *parent = parents->item;
 908                         int *pi = indegree_slab_at(&indegree, parent);
 909
 910                         if (!*pi)
 911                                 continue;
 912
 913                         /*
 914                          * parents are only enqueued for emission
 915                          * when all their children have been emitted thereby
 916                          * guaranteeing topological order.
 917                          */
 918                         if (--(*pi) == 1)
 919                                 prio_queue_put(&queue, parent);
 920                 }
 921                 /*
 922                  * all children of commit have already been
 923                  * emitted. we can emit it now.
 924                  */
 925                 *(indegree_slab_at(&indegree, commit)) = 0;
 926
 927                 pptr = &commit_list_insert(commit, pptr)->next;
 928         }
 929
 930         clear_indegree_slab(&indegree);
 931         clear_prio_queue(&queue);
 932         if (sort_order == REV_SORT_BY_AUTHOR_DATE)
 933                 clear_author_date_slab(&author_date);
 934 }
 935
 936 struct rev_collect {
 937         struct commit **commit;
 938         int nr;
 939         int alloc;
 940         unsigned int initial : 1;
 941 };
 942
 943 static void add_one_commit(struct object_id *oid, struct rev_collect *revs)
 944 {
 945         struct commit *commit;
 946
 947         if (is_null_oid(oid))
 948                 return;
 949
 950         commit = lookup_commit(the_repository, oid);
 951         if (!commit ||
 952             (commit->object.flags & TMP_MARK) ||
 953             parse_commit(commit))
 954                 return;
 955
 956         ALLOC_GROW(revs->commit, revs->nr + 1, revs->alloc);
 957         revs->commit[revs->nr++] = commit;
 958         commit->object.flags |= TMP_MARK;
 959 }
 960
 961 static int collect_one_reflog_ent(struct object_id *ooid, struct object_id *noid,
 962                                   const char *ident UNUSED,
 963                                   timestamp_t timestamp UNUSED, int tz UNUSED,
 964                                   const char *message UNUSED, void *cbdata)
 965 {
 966         struct rev_collect *revs = cbdata;
 967
 968         if (revs->initial) {
 969                 revs->initial = 0;
 970                 add_one_commit(ooid, revs);
 971         }
 972         add_one_commit(noid, revs);
 973         return 0;
 974 }
 975
 976 struct commit *get_fork_point(const char *refname, struct commit *commit)
 977 {
 978         struct object_id oid;
 979         struct rev_collect revs;
 980         struct commit_list *bases;
 981         int i;
 982         struct commit *ret = NULL;
 983         char *full_refname;
 984
 985         switch (dwim_ref(refname, strlen(refname), &oid, &full_refname, 0)) {
 986         case 0:
 987                 die("No such ref: '%s'", refname);
 988         case 1:
 989                 break; /* good */
 990         default:
 991                 die("Ambiguous refname: '%s'", refname);
 992         }
 993
 994         memset(&revs, 0, sizeof(revs));
 995         revs.initial = 1;
 996         for_each_reflog_ent(full_refname, collect_one_reflog_ent, &revs);
 997
 998         if (!revs.nr)
 999                 add_one_commit(&oid, &revs);
1000
1001         for (i = 0; i < revs.nr; i++)
1002                 revs.commit[i]->object.flags &= ~TMP_MARK;
1003
1004         bases = get_merge_bases_many(commit, revs.nr, revs.commit);
1005
1006         /*
1007          * There should be one and only one merge base, when we found
1008          * a common ancestor among reflog entries.
1009          */
1010         if (!bases || bases->next)
1011                 goto cleanup_return;
1012
1013         /* And the found one must be one of the reflog entries */
1014         for (i = 0; i < revs.nr; i++)
1015                 if (&bases->item->object == &revs.commit[i]->object)
1016                         break; /* found */
1017         if (revs.nr <= i)
1018                 goto cleanup_return;
1019
1020         ret = bases->item;
1021
1022 cleanup_return:
1023         free_commit_list(bases);
1024         free(full_refname);
1025         return ret;
1026 }
1027
1028 /*
1029  * Indexed by hash algorithm identifier.
1030  */
1031 static const char *gpg_sig_headers[] = {
1032         NULL,
1033         "gpgsig",
1034         "gpgsig-sha256",
1035 };
1036
1037 int sign_with_header(struct strbuf *buf, const char *keyid)
1038 {
1039         struct strbuf sig = STRBUF_INIT;
1040         int inspos, copypos;
1041         const char *eoh;
1042         const char *gpg_sig_header = gpg_sig_headers[hash_algo_by_ptr(the_hash_algo)];
1043         int gpg_sig_header_len = strlen(gpg_sig_header);
1044
1045         /* find the end of the header */
1046         eoh = strstr(buf->buf, "\n\n");
1047         if (!eoh)
1048                 inspos = buf->len;
1049         else
1050                 inspos = eoh - buf->buf + 1;
1051
1052         if (!keyid || !*keyid)
1053                 keyid = get_signing_key();
1054         if (sign_buffer(buf, &sig, keyid)) {
1055                 strbuf_release(&sig);
1056                 return -1;
1057         }
1058
1059         for (copypos = 0; sig.buf[copypos]; ) {
1060                 const char *bol = sig.buf + copypos;
1061                 const char *eol = strchrnul(bol, '\n');
1062                 int len = (eol - bol) + !!*eol;
1063
1064                 if (!copypos) {
1065                         strbuf_insert(buf, inspos, gpg_sig_header, gpg_sig_header_len);
1066                         inspos += gpg_sig_header_len;
1067                 }
1068                 strbuf_insertstr(buf, inspos++, " ");
1069                 strbuf_insert(buf, inspos, bol, len);
1070                 inspos += len;
1071                 copypos += len;
1072         }
1073         strbuf_release(&sig);
1074         return 0;
1075 }
1076
1077
1078
1079 int parse_signed_commit(const struct commit *commit,
1080                         struct strbuf *payload, struct strbuf *signature,
1081                         const struct git_hash_algo *algop)
1082 {
1083         unsigned long size;
1084         const char *buffer = get_commit_buffer(commit, &size);
1085         int ret = parse_buffer_signed_by_header(buffer, size, payload, signature, algop);
1086
1087         unuse_commit_buffer(commit, buffer);
1088         return ret;
1089 }
1090
1091 int parse_buffer_signed_by_header(const char *buffer,
1092                                   unsigned long size,
1093                                   struct strbuf *payload,
1094                                   struct strbuf *signature,
1095                                   const struct git_hash_algo *algop)
1096 {
1097         int in_signature = 0, saw_signature = 0, other_signature = 0;
1098         const char *line, *tail, *p;
1099         const char *gpg_sig_header = gpg_sig_headers[hash_algo_by_ptr(algop)];
1100
1101         line = buffer;
1102         tail = buffer + size;
1103         while (line < tail) {
1104                 const char *sig = NULL;
1105                 const char *next = memchr(line, '\n', tail - line);
1106
1107                 next = next ? next + 1 : tail;
1108                 if (in_signature && line[0] == ' ')
1109                         sig = line + 1;
1110                 else if (skip_prefix(line, gpg_sig_header, &p) &&
1111                          *p == ' ') {
1112                         sig = line + strlen(gpg_sig_header) + 1;
1113                         other_signature = 0;
1114                 }
1115                 else if (starts_with(line, "gpgsig"))
1116                         other_signature = 1;
1117                 else if (other_signature && line[0] != ' ')
1118                         other_signature = 0;
1119                 if (sig) {
1120                         strbuf_add(signature, sig, next - sig);
1121                         saw_signature = 1;
1122                         in_signature = 1;
1123                 } else {
1124                         if (*line == '\n')
1125                                 /* dump the whole remainder of the buffer */
1126                                 next = tail;
1127                         if (!other_signature)
1128                                 strbuf_add(payload, line, next - line);
1129                         in_signature = 0;
1130                 }
1131                 line = next;
1132         }
1133         return saw_signature;
1134 }
1135
1136 int remove_signature(struct strbuf *buf)
1137 {
1138         const char *line = buf->buf;
1139         const char *tail = buf->buf + buf->len;
1140         int in_signature = 0;
1141         struct sigbuf {
1142                 const char *start;
1143                 const char *end;
1144         } sigs[2], *sigp = &sigs[0];
1145         int i;
1146         const char *orig_buf = buf->buf;
1147
1148         memset(sigs, 0, sizeof(sigs));
1149
1150         while (line < tail) {
1151                 const char *next = memchr(line, '\n', tail - line);
1152                 next = next ? next + 1 : tail;
1153
1154                 if (in_signature && line[0] == ' ')
1155                         sigp->end = next;
1156                 else if (starts_with(line, "gpgsig")) {
1157                         int i;
1158                         for (i = 1; i < GIT_HASH_NALGOS; i++) {
1159                                 const char *p;
1160                                 if (skip_prefix(line, gpg_sig_headers[i], &p) &&
1161                                     *p == ' ') {
1162                                         sigp->start = line;
1163                                         sigp->end = next;
1164                                         in_signature = 1;
1165                                 }
1166                         }
1167                 } else {
1168                         if (*line == '\n')
1169                                 /* dump the whole remainder of the buffer */
1170                                 next = tail;
1171                         if (in_signature && sigp - sigs != ARRAY_SIZE(sigs))
1172                                 sigp++;
1173                         in_signature = 0;
1174                 }
1175                 line = next;
1176         }
1177
1178         for (i = ARRAY_SIZE(sigs) - 1; i >= 0; i--)
1179                 if (sigs[i].start)
1180                         strbuf_remove(buf, sigs[i].start - orig_buf, sigs[i].end - sigs[i].start);
1181
1182         return sigs[0].start != NULL;
1183 }
1184
1185 static void handle_signed_tag(struct commit *parent, struct commit_extra_header ***tail)
1186 {
1187         struct merge_remote_desc *desc;
1188         struct commit_extra_header *mergetag;
1189         char *buf;
1190         unsigned long size;
1191         enum object_type type;
1192         struct strbuf payload = STRBUF_INIT;
1193         struct strbuf signature = STRBUF_INIT;
1194
1195         desc = merge_remote_util(parent);
1196         if (!desc || !desc->obj)
1197                 return;
1198         buf = read_object_file(&desc->obj->oid, &type, &size);
1199         if (!buf || type != OBJ_TAG)
1200                 goto free_return;
1201         if (!parse_signature(buf, size, &payload, &signature))
1202                 goto free_return;
1203         /*
1204          * We could verify this signature and either omit the tag when
1205          * it does not validate, but the integrator may not have the
1206          * public key of the signer of the tag being merged, while a
1207          * later auditor may have it while auditing, so let's not run
1208          * verify-signed-buffer here for now...
1209          *
1210          * if (verify_signed_buffer(buf, len, buf + len, size - len, ...))
1211          *      warn("warning: signed tag unverified.");
1212          */
1213         CALLOC_ARRAY(mergetag, 1);
1214         mergetag->key = xstrdup("mergetag");
1215         mergetag->value = buf;
1216         mergetag->len = size;
1217
1218         **tail = mergetag;
1219         *tail = &mergetag->next;
1220         strbuf_release(&payload);
1221         strbuf_release(&signature);
1222         return;
1223
1224 free_return:
1225         free(buf);
1226 }
1227
1228 int check_commit_signature(const struct commit *commit, struct signature_check *sigc)
1229 {
1230         struct strbuf payload = STRBUF_INIT;
1231         struct strbuf signature = STRBUF_INIT;
1232         int ret = 1;
1233
1234         sigc->result = 'N';
1235
1236         if (parse_signed_commit(commit, &payload, &signature, the_hash_algo) <= 0)
1237                 goto out;
1238
1239         sigc->payload_type = SIGNATURE_PAYLOAD_COMMIT;
1240         sigc->payload = strbuf_detach(&payload, &sigc->payload_len);
1241         ret = check_signature(sigc, signature.buf, signature.len);
1242
1243  out:
1244         strbuf_release(&payload);
1245         strbuf_release(&signature);
1246
1247         return ret;
1248 }
1249
1250 void verify_merge_signature(struct commit *commit, int verbosity,
1251                             int check_trust)
1252 {
1253         char hex[GIT_MAX_HEXSZ + 1];
1254         struct signature_check signature_check;
1255         int ret;
1256         memset(&signature_check, 0, sizeof(signature_check));
1257
1258         ret = check_commit_signature(commit, &signature_check);
1259
1260         find_unique_abbrev_r(hex, &commit->object.oid, DEFAULT_ABBREV);
1261         switch (signature_check.result) {
1262         case 'G':
1263                 if (ret || (check_trust && signature_check.trust_level < TRUST_MARGINAL))
1264                         die(_("Commit %s has an untrusted GPG signature, "
1265                               "allegedly by %s."), hex, signature_check.signer);
1266                 break;
1267         case 'B':
1268                 die(_("Commit %s has a bad GPG signature "
1269                       "allegedly by %s."), hex, signature_check.signer);
1270         default: /* 'N' */
1271                 die(_("Commit %s does not have a GPG signature."), hex);
1272         }
1273         if (verbosity >= 0 && signature_check.result == 'G')
1274                 printf(_("Commit %s has a good GPG signature by %s\n"),
1275                        hex, signature_check.signer);
1276
1277         signature_check_clear(&signature_check);
1278 }
1279
1280 void append_merge_tag_headers(struct commit_list *parents,
1281                               struct commit_extra_header ***tail)
1282 {
1283         while (parents) {
1284                 struct commit *parent = parents->item;
1285                 handle_signed_tag(parent, tail);
1286                 parents = parents->next;
1287         }
1288 }
1289
1290 static void add_extra_header(struct strbuf *buffer,
1291                              struct commit_extra_header *extra)
1292 {
1293         strbuf_addstr(buffer, extra->key);
1294         if (extra->len)
1295                 strbuf_add_lines(buffer, " ", extra->value, extra->len);
1296         else
1297                 strbuf_addch(buffer, '\n');
1298 }
1299
1300 struct commit_extra_header *read_commit_extra_headers(struct commit *commit,
1301                                                       const char **exclude)
1302 {
1303         struct commit_extra_header *extra = NULL;
1304         unsigned long size;
1305         const char *buffer = get_commit_buffer(commit, &size);
1306         extra = read_commit_extra_header_lines(buffer, size, exclude);
1307         unuse_commit_buffer(commit, buffer);
1308         return extra;
1309 }
1310
1311 int for_each_mergetag(each_mergetag_fn fn, struct commit *commit, void *data)
1312 {
1313         struct commit_extra_header *extra, *to_free;
1314         int res = 0;
1315
1316         to_free = read_commit_extra_headers(commit, NULL);
1317         for (extra = to_free; !res && extra; extra = extra->next) {
1318                 if (strcmp(extra->key, "mergetag"))
1319                         continue; /* not a merge tag */
1320                 res = fn(commit, extra, data);
1321         }
1322         free_commit_extra_headers(to_free);
1323         return res;
1324 }
1325
1326 static inline int standard_header_field(const char *field, size_t len)
1327 {
1328         return ((len == 4 && !memcmp(field, "tree", 4)) ||
1329                 (len == 6 && !memcmp(field, "parent", 6)) ||
1330                 (len == 6 && !memcmp(field, "author", 6)) ||
1331                 (len == 9 && !memcmp(field, "committer", 9)) ||
1332                 (len == 8 && !memcmp(field, "encoding", 8)));
1333 }
1334
1335 static int excluded_header_field(const char *field, size_t len, const char **exclude)
1336 {
1337         if (!exclude)
1338                 return 0;
1339
1340         while (*exclude) {
1341                 size_t xlen = strlen(*exclude);
1342                 if (len == xlen && !memcmp(field, *exclude, xlen))
1343                         return 1;
1344                 exclude++;
1345         }
1346         return 0;
1347 }
1348
1349 static struct commit_extra_header *read_commit_extra_header_lines(
1350         const char *buffer, size_t size,
1351         const char **exclude)
1352 {
1353         struct commit_extra_header *extra = NULL, **tail = &extra, *it = NULL;
1354         const char *line, *next, *eof, *eob;
1355         struct strbuf buf = STRBUF_INIT;
1356
1357         for (line = buffer, eob = line + size;
1358              line < eob && *line != '\n';
1359              line = next) {
1360                 next = memchr(line, '\n', eob - line);
1361                 next = next ? next + 1 : eob;
1362                 if (*line == ' ') {
1363                         /* continuation */
1364                         if (it)
1365                                 strbuf_add(&buf, line + 1, next - (line + 1));
1366                         continue;
1367                 }
1368                 if (it)
1369                         it->value = strbuf_detach(&buf, &it->len);
1370                 strbuf_reset(&buf);
1371                 it = NULL;
1372
1373                 eof = memchr(line, ' ', next - line);
1374                 if (!eof)
1375                         eof = next;
1376                 else if (standard_header_field(line, eof - line) ||
1377                          excluded_header_field(line, eof - line, exclude))
1378                         continue;
1379
1380                 CALLOC_ARRAY(it, 1);
1381                 it->key = xmemdupz(line, eof-line);
1382                 *tail = it;
1383                 tail = &it->next;
1384                 if (eof + 1 < next)
1385                         strbuf_add(&buf, eof + 1, next - (eof + 1));
1386         }
1387         if (it)
1388                 it->value = strbuf_detach(&buf, &it->len);
1389         return extra;
1390 }
1391
1392 void free_commit_extra_headers(struct commit_extra_header *extra)
1393 {
1394         while (extra) {
1395                 struct commit_extra_header *next = extra->next;
1396                 free(extra->key);
1397                 free(extra->value);
1398                 free(extra);
1399                 extra = next;
1400         }
1401 }
1402
1403 int commit_tree(const char *msg, size_t msg_len, const struct object_id *tree,
1404                 struct commit_list *parents, struct object_id *ret,
1405                 const char *author, const char *sign_commit)
1406 {
1407         struct commit_extra_header *extra = NULL, **tail = &extra;
1408         int result;
1409
1410         append_merge_tag_headers(parents, &tail);
1411         result = commit_tree_extended(msg, msg_len, tree, parents, ret, author,
1412                                       NULL, sign_commit, extra);
1413         free_commit_extra_headers(extra);
1414         return result;
1415 }
1416
1417 static int find_invalid_utf8(const char *buf, int len)
1418 {
1419         int offset = 0;
1420         static const unsigned int max_codepoint[] = {
1421                 0x7f, 0x7ff, 0xffff, 0x10ffff
1422         };
1423
1424         while (len) {
1425                 unsigned char c = *buf++;
1426                 int bytes, bad_offset;
1427                 unsigned int codepoint;
1428                 unsigned int min_val, max_val;
1429
1430                 len--;
1431                 offset++;
1432
1433                 /* Simple US-ASCII? No worries. */
1434                 if (c < 0x80)
1435                         continue;
1436
1437                 bad_offset = offset-1;
1438
1439                 /*
1440                  * Count how many more high bits set: that's how
1441                  * many more bytes this sequence should have.
1442                  */
1443                 bytes = 0;
1444                 while (c & 0x40) {
1445                         c <<= 1;
1446                         bytes++;
1447                 }
1448
1449                 /*
1450                  * Must be between 1 and 3 more bytes.  Longer sequences result in
1451                  * codepoints beyond U+10FFFF, which are guaranteed never to exist.
1452                  */
1453                 if (bytes < 1 || 3 < bytes)
1454                         return bad_offset;
1455
1456                 /* Do we *have* that many bytes? */
1457                 if (len < bytes)
1458                         return bad_offset;
1459
1460                 /*
1461                  * Place the encoded bits at the bottom of the value and compute the
1462                  * valid range.
1463                  */
1464                 codepoint = (c & 0x7f) >> bytes;
1465                 min_val = max_codepoint[bytes-1] + 1;
1466                 max_val = max_codepoint[bytes];
1467
1468                 offset += bytes;
1469                 len -= bytes;
1470
1471                 /* And verify that they are good continuation bytes */
1472                 do {
1473                         codepoint <<= 6;
1474                         codepoint |= *buf & 0x3f;
1475                         if ((*buf++ & 0xc0) != 0x80)
1476                                 return bad_offset;
1477                 } while (--bytes);
1478
1479                 /* Reject codepoints that are out of range for the sequence length. */
1480                 if (codepoint < min_val || codepoint > max_val)
1481                         return bad_offset;
1482                 /* Surrogates are only for UTF-16 and cannot be encoded in UTF-8. */
1483                 if ((codepoint & 0x1ff800) == 0xd800)
1484                         return bad_offset;
1485                 /* U+xxFFFE and U+xxFFFF are guaranteed non-characters. */
1486                 if ((codepoint & 0xfffe) == 0xfffe)
1487                         return bad_offset;
1488                 /* So are anything in the range U+FDD0..U+FDEF. */
1489                 if (codepoint >= 0xfdd0 && codepoint <= 0xfdef)
1490                         return bad_offset;
1491         }
1492         return -1;
1493 }
1494
1495 /*
1496  * This verifies that the buffer is in proper utf8 format.
1497  *
1498  * If it isn't, it assumes any non-utf8 characters are Latin1,
1499  * and does the conversion.
1500  */
1501 static int verify_utf8(struct strbuf *buf)
1502 {
1503         int ok = 1;
1504         long pos = 0;
1505
1506         for (;;) {
1507                 int bad;
1508                 unsigned char c;
1509                 unsigned char replace[2];
1510
1511                 bad = find_invalid_utf8(buf->buf + pos, buf->len - pos);
1512                 if (bad < 0)
1513                         return ok;
1514                 pos += bad;
1515                 ok = 0;
1516                 c = buf->buf[pos];
1517                 strbuf_remove(buf, pos, 1);
1518
1519                 /* We know 'c' must be in the range 128-255 */
1520                 replace[0] = 0xc0 + (c >> 6);
1521                 replace[1] = 0x80 + (c & 0x3f);
1522                 strbuf_insert(buf, pos, replace, 2);
1523                 pos += 2;
1524         }
1525 }
1526
1527 static const char commit_utf8_warn[] =
1528 N_("Warning: commit message did not conform to UTF-8.\n"
1529    "You may want to amend it after fixing the message, or set the config\n"
1530    "variable i18n.commitEncoding to the encoding your project uses.\n");
1531
1532 int commit_tree_extended(const char *msg, size_t msg_len,
1533                          const struct object_id *tree,
1534                          struct commit_list *parents, struct object_id *ret,
1535                          const char *author, const char *committer,
1536                          const char *sign_commit,
1537                          struct commit_extra_header *extra)
1538 {
1539         int result;
1540         int encoding_is_utf8;
1541         struct strbuf buffer;
1542
1543         assert_oid_type(tree, OBJ_TREE);
1544
1545         if (memchr(msg, '\0', msg_len))
1546                 return error("a NUL byte in commit log message not allowed.");
1547
1548         /* Not having i18n.commitencoding is the same as having utf-8 */
1549         encoding_is_utf8 = is_encoding_utf8(git_commit_encoding);
1550
1551         strbuf_init(&buffer, 8192); /* should avoid reallocs for the headers */
1552         strbuf_addf(&buffer, "tree %s\n", oid_to_hex(tree));
1553
1554         /*
1555          * NOTE! This ordering means that the same exact tree merged with a
1556          * different order of parents will be a _different_ changeset even
1557          * if everything else stays the same.
1558          */
1559         while (parents) {
1560                 struct commit *parent = pop_commit(&parents);
1561                 strbuf_addf(&buffer, "parent %s\n",
1562                             oid_to_hex(&parent->object.oid));
1563         }
1564
1565         /* Person/date information */
1566         if (!author)
1567                 author = git_author_info(IDENT_STRICT);
1568         strbuf_addf(&buffer, "author %s\n", author);
1569         if (!committer)
1570                 committer = git_committer_info(IDENT_STRICT);
1571         strbuf_addf(&buffer, "committer %s\n", committer);
1572         if (!encoding_is_utf8)
1573                 strbuf_addf(&buffer, "encoding %s\n", git_commit_encoding);
1574
1575         while (extra) {
1576                 add_extra_header(&buffer, extra);
1577                 extra = extra->next;
1578         }
1579         strbuf_addch(&buffer, '\n');
1580
1581         /* And add the comment */
1582         strbuf_add(&buffer, msg, msg_len);
1583
1584         /* And check the encoding */
1585         if (encoding_is_utf8 && !verify_utf8(&buffer))
1586                 fprintf(stderr, _(commit_utf8_warn));
1587
1588         if (sign_commit && sign_with_header(&buffer, sign_commit)) {
1589                 result = -1;
1590                 goto out;
1591         }
1592
1593         result = write_object_file(buffer.buf, buffer.len, OBJ_COMMIT, ret);
1594 out:
1595         strbuf_release(&buffer);
1596         return result;
1597 }
1598
1599 define_commit_slab(merge_desc_slab, struct merge_remote_desc *);
1600 static struct merge_desc_slab merge_desc_slab = COMMIT_SLAB_INIT(1, merge_desc_slab);
1601
1602 struct merge_remote_desc *merge_remote_util(struct commit *commit)
1603 {
1604         return *merge_desc_slab_at(&merge_desc_slab, commit);
1605 }
1606
1607 void set_merge_remote_desc(struct commit *commit,
1608                            const char *name, struct object *obj)
1609 {
1610         struct merge_remote_desc *desc;
1611         FLEX_ALLOC_STR(desc, name, name);
1612         desc->obj = obj;
1613         *merge_desc_slab_at(&merge_desc_slab, commit) = desc;
1614 }
1615
1616 struct commit *get_merge_parent(const char *name)
1617 {
1618         struct object *obj;
1619         struct commit *commit;
1620         struct object_id oid;
1621         if (get_oid(name, &oid))
1622                 return NULL;
1623         obj = parse_object(the_repository, &oid);
1624         commit = (struct commit *)peel_to_type(name, 0, obj, OBJ_COMMIT);
1625         if (commit && !merge_remote_util(commit))
1626                 set_merge_remote_desc(commit, name, obj);
1627         return commit;
1628 }
1629
1630 /*
1631  * Append a commit to the end of the commit_list.
1632  *
1633  * next starts by pointing to the variable that holds the head of an
1634  * empty commit_list, and is updated to point to the "next" field of
1635  * the last item on the list as new commits are appended.
1636  *
1637  * Usage example:
1638  *
1639  *     struct commit_list *list;
1640  *     struct commit_list **next = &list;
1641  *
1642  *     next = commit_list_append(c1, next);
1643  *     next = commit_list_append(c2, next);
1644  *     assert(commit_list_count(list) == 2);
1645  *     return list;
1646  */
1647 struct commit_list **commit_list_append(struct commit *commit,
1648                                         struct commit_list **next)
1649 {
1650         struct commit_list *new_commit = xmalloc(sizeof(struct commit_list));
1651         new_commit->item = commit;
1652         *next = new_commit;
1653         new_commit->next = NULL;
1654         return &new_commit->next;
1655 }
1656
1657 const char *find_header_mem(const char *msg, size_t len,
1658                         const char *key, size_t *out_len)
1659 {
1660         int key_len = strlen(key);
1661         const char *line = msg;
1662
1663         /*
1664          * NEEDSWORK: It's possible for strchrnul() to scan beyond the range
1665          * given by len. However, current callers are safe because they compute
1666          * len by scanning a NUL-terminated block of memory starting at msg.
1667          * Nonetheless, it would be better to ensure the function does not look
1668          * at msg beyond the len provided by the caller.
1669          */
1670         while (line && line < msg + len) {
1671                 const char *eol = strchrnul(line, '\n');
1672
1673                 if (line == eol)
1674                         return NULL;
1675
1676                 if (eol - line > key_len &&
1677                     !strncmp(line, key, key_len) &&
1678                     line[key_len] == ' ') {
1679                         *out_len = eol - line - key_len - 1;
1680                         return line + key_len + 1;
1681                 }
1682                 line = *eol ? eol + 1 : NULL;
1683         }
1684         return NULL;
1685 }
1686
1687 const char *find_commit_header(const char *msg, const char *key, size_t *out_len)
1688 {
1689         return find_header_mem(msg, strlen(msg), key, out_len);
1690 }
1691 /*
1692  * Inspect the given string and determine the true "end" of the log message, in
1693  * order to find where to put a new Signed-off-by trailer.  Ignored are
1694  * trailing comment lines and blank lines.  To support "git commit -s
1695  * --amend" on an existing commit, we also ignore "Conflicts:".  To
1696  * support "git commit -v", we truncate at cut lines.
1697  *
1698  * Returns the number of bytes from the tail to ignore, to be fed as
1699  * the second parameter to append_signoff().
1700  */
1701 size_t ignore_non_trailer(const char *buf, size_t len)
1702 {
1703         size_t boc = 0;
1704         size_t bol = 0;
1705         int in_old_conflicts_block = 0;
1706         size_t cutoff = wt_status_locate_end(buf, len);
1707
1708         while (bol < cutoff) {
1709                 const char *next_line = memchr(buf + bol, '\n', len - bol);
1710
1711                 if (!next_line)
1712                         next_line = buf + len;
1713                 else
1714                         next_line++;
1715
1716                 if (buf[bol] == comment_line_char || buf[bol] == '\n') {
1717                         /* is this the first of the run of comments? */
1718                         if (!boc)
1719                                 boc = bol;
1720                         /* otherwise, it is just continuing */
1721                 } else if (starts_with(buf + bol, "Conflicts:\n")) {
1722                         in_old_conflicts_block = 1;
1723                         if (!boc)
1724                                 boc = bol;
1725                 } else if (in_old_conflicts_block && buf[bol] == '\t') {
1726                         ; /* a pathname in the conflicts block */
1727                 } else if (boc) {
1728                         /* the previous was not trailing comment */
1729                         boc = 0;
1730                         in_old_conflicts_block = 0;
1731                 }
1732                 bol = next_line - buf;
1733         }
1734         return boc ? len - boc : len - cutoff;
1735 }
1736
1737 int run_commit_hook(int editor_is_used, const char *index_file,
1738                     int *invoked_hook, const char *name, ...)
1739 {
1740         struct run_hooks_opt opt = RUN_HOOKS_OPT_INIT;
1741         va_list args;
1742         const char *arg;
1743
1744         strvec_pushf(&opt.env, "GIT_INDEX_FILE=%s", index_file);
1745
1746         /*
1747          * Let the hook know that no editor will be launched.
1748          */
1749         if (!editor_is_used)
1750                 strvec_push(&opt.env, "GIT_EDITOR=:");
1751
1752         va_start(args, name);
1753         while ((arg = va_arg(args, const char *)))
1754                 strvec_push(&opt.args, arg);
1755         va_end(args);
1756
1757         opt.invoked_hook = invoked_hook;
1758         return run_hooks_opt(name, &opt);
1759 }