commit.c

   1 #include "cache.h"
   2 #include "tag.h"
   3 #include "commit.h"
   4 #include "commit-graph.h"
   5 #include "repository.h"
   6 #include "object-store.h"
   7 #include "pkt-line.h"
   8 #include "utf8.h"
   9 #include "diff.h"
  10 #include "revision.h"
  11 #include "notes.h"
  12 #include "alloc.h"
  13 #include "gpg-interface.h"
  14 #include "mergesort.h"
  15 #include "commit-slab.h"
  16 #include "prio-queue.h"
  17 #include "sha1-lookup.h"
  18 #include "wt-status.h"
  19 #include "advice.h"
  20
  21 static struct commit_extra_header *read_commit_extra_header_lines(const char *buf, size_t len, const char **);
  22
  23 int save_commit_buffer = 1;
  24
  25 const char *commit_type = "commit";
  26
  27 struct commit *lookup_commit_reference_gently(const struct object_id *oid,
  28                                               int quiet)
  29 {
  30         struct object *obj = deref_tag(parse_object(oid), NULL, 0);
  31
  32         if (!obj)
  33                 return NULL;
  34         return object_as_type(obj, OBJ_COMMIT, quiet);
  35 }
  36
  37 struct commit *lookup_commit_reference(const struct object_id *oid)
  38 {
  39         return lookup_commit_reference_gently(oid, 0);
  40 }
  41
  42 struct commit *lookup_commit_or_die(const struct object_id *oid, const char *ref_name)
  43 {
  44         struct commit *c = lookup_commit_reference(oid);
  45         if (!c)
  46                 die(_("could not parse %s"), ref_name);
  47         if (oidcmp(oid, &c->object.oid)) {
  48                 warning(_("%s %s is not a commit!"),
  49                         ref_name, oid_to_hex(oid));
  50         }
  51         return c;
  52 }
  53
  54 struct commit *lookup_commit(const struct object_id *oid)
  55 {
  56         struct object *obj = lookup_object(oid->hash);
  57         if (!obj)
  58                 return create_object(the_repository, oid->hash,
  59                                      alloc_commit_node(the_repository));
  60         return object_as_type(obj, OBJ_COMMIT, 0);
  61 }
  62
  63 struct commit *lookup_commit_reference_by_name(const char *name)
  64 {
  65         struct object_id oid;
  66         struct commit *commit;
  67
  68         if (get_oid_committish(name, &oid))
  69                 return NULL;
  70         commit = lookup_commit_reference(&oid);
  71         if (parse_commit(commit))
  72                 return NULL;
  73         return commit;
  74 }
  75
  76 static timestamp_t parse_commit_date(const char *buf, const char *tail)
  77 {
  78         const char *dateptr;
  79
  80         if (buf + 6 >= tail)
  81                 return 0;
  82         if (memcmp(buf, "author", 6))
  83                 return 0;
  84         while (buf < tail && *buf++ != '\n')
  85                 /* nada */;
  86         if (buf + 9 >= tail)
  87                 return 0;
  88         if (memcmp(buf, "committer", 9))
  89                 return 0;
  90         while (buf < tail && *buf++ != '>')
  91                 /* nada */;
  92         if (buf >= tail)
  93                 return 0;
  94         dateptr = buf;
  95         while (buf < tail && *buf++ != '\n')
  96                 /* nada */;
  97         if (buf >= tail)
  98                 return 0;
  99         /* dateptr < buf && buf[-1] == '\n', so parsing will stop at buf-1 */
 100         return parse_timestamp(dateptr, NULL, 10);
 101 }
 102
 103 static const unsigned char *commit_graft_sha1_access(size_t index, void *table)
 104 {
 105         struct commit_graft **commit_graft_table = table;
 106         return commit_graft_table[index]->oid.hash;
 107 }
 108
 109 static int commit_graft_pos(struct repository *r, const unsigned char *sha1)
 110 {
 111         return sha1_pos(sha1, r->parsed_objects->grafts,
 112                         r->parsed_objects->grafts_nr,
 113                         commit_graft_sha1_access);
 114 }
 115
 116 int register_commit_graft(struct repository *r, struct commit_graft *graft,
 117                           int ignore_dups)
 118 {
 119         int pos = commit_graft_pos(r, graft->oid.hash);
 120
 121         if (0 <= pos) {
 122                 if (ignore_dups)
 123                         free(graft);
 124                 else {
 125                         free(r->parsed_objects->grafts[pos]);
 126                         r->parsed_objects->grafts[pos] = graft;
 127                 }
 128                 return 1;
 129         }
 130         pos = -pos - 1;
 131         ALLOC_GROW(r->parsed_objects->grafts,
 132                    r->parsed_objects->grafts_nr + 1,
 133                    r->parsed_objects->grafts_alloc);
 134         r->parsed_objects->grafts_nr++;
 135         if (pos < r->parsed_objects->grafts_nr)
 136                 memmove(r->parsed_objects->grafts + pos + 1,
 137                         r->parsed_objects->grafts + pos,
 138                         (r->parsed_objects->grafts_nr - pos - 1) *
 139                         sizeof(*r->parsed_objects->grafts));
 140         r->parsed_objects->grafts[pos] = graft;
 141         return 0;
 142 }
 143
 144 struct commit_graft *read_graft_line(struct strbuf *line)
 145 {
 146         /* The format is just "Commit Parent1 Parent2 ...\n" */
 147         int i, phase;
 148         const char *tail = NULL;
 149         struct commit_graft *graft = NULL;
 150         struct object_id dummy_oid, *oid;
 151
 152         strbuf_rtrim(line);
 153         if (!line->len || line->buf[0] == '#')
 154                 return NULL;
 155         /*
 156          * phase 0 verifies line, counts hashes in line and allocates graft
 157          * phase 1 fills graft
 158          */
 159         for (phase = 0; phase < 2; phase++) {
 160                 oid = graft ? &graft->oid : &dummy_oid;
 161                 if (parse_oid_hex(line->buf, oid, &tail))
 162                         goto bad_graft_data;
 163                 for (i = 0; *tail != '\0'; i++) {
 164                         oid = graft ? &graft->parent[i] : &dummy_oid;
 165                         if (!isspace(*tail++) || parse_oid_hex(tail, oid, &tail))
 166                                 goto bad_graft_data;
 167                 }
 168                 if (!graft) {
 169                         graft = xmalloc(st_add(sizeof(*graft),
 170                                                st_mult(sizeof(struct object_id), i)));
 171                         graft->nr_parent = i;
 172                 }
 173         }
 174         return graft;
 175
 176 bad_graft_data:
 177         error("bad graft data: %s", line->buf);
 178         assert(!graft);
 179         return NULL;
 180 }
 181
 182 static int read_graft_file(struct repository *r, const char *graft_file)
 183 {
 184         FILE *fp = fopen_or_warn(graft_file, "r");
 185         struct strbuf buf = STRBUF_INIT;
 186         if (!fp)
 187                 return -1;
 188         if (advice_graft_file_deprecated)
 189                 advise(_("Support for <GIT_DIR>/info/grafts is deprecated\n"
 190                          "and will be removed in a future Git version.\n"
 191                          "\n"
 192                          "Please use \"git replace --convert-graft-file\"\n"
 193                          "to convert the grafts into replace refs.\n"
 194                          "\n"
 195                          "Turn this message off by running\n"
 196                          "\"git config advice.graftFileDeprecated false\""));
 197         while (!strbuf_getwholeline(&buf, fp, '\n')) {
 198                 /* The format is just "Commit Parent1 Parent2 ...\n" */
 199                 struct commit_graft *graft = read_graft_line(&buf);
 200                 if (!graft)
 201                         continue;
 202                 if (register_commit_graft(r, graft, 1))
 203                         error("duplicate graft data: %s", buf.buf);
 204         }
 205         fclose(fp);
 206         strbuf_release(&buf);
 207         return 0;
 208 }
 209
 210 static void prepare_commit_graft(struct repository *r)
 211 {
 212         char *graft_file;
 213
 214         if (r->parsed_objects->commit_graft_prepared)
 215                 return;
 216         if (!startup_info->have_repository)
 217                 return;
 218
 219         graft_file = get_graft_file(r);
 220         read_graft_file(r, graft_file);
 221         /* make sure shallows are read */
 222         is_repository_shallow(r);
 223         r->parsed_objects->commit_graft_prepared = 1;
 224 }
 225
 226 struct commit_graft *lookup_commit_graft(struct repository *r, const struct object_id *oid)
 227 {
 228         int pos;
 229         prepare_commit_graft(r);
 230         pos = commit_graft_pos(r, oid->hash);
 231         if (pos < 0)
 232                 return NULL;
 233         return r->parsed_objects->grafts[pos];
 234 }
 235
 236 int for_each_commit_graft(each_commit_graft_fn fn, void *cb_data)
 237 {
 238         int i, ret;
 239         for (i = ret = 0; i < the_repository->parsed_objects->grafts_nr && !ret; i++)
 240                 ret = fn(the_repository->parsed_objects->grafts[i], cb_data);
 241         return ret;
 242 }
 243
 244 int unregister_shallow(const struct object_id *oid)
 245 {
 246         int pos = commit_graft_pos(the_repository, oid->hash);
 247         if (pos < 0)
 248                 return -1;
 249         if (pos + 1 < the_repository->parsed_objects->grafts_nr)
 250                 MOVE_ARRAY(the_repository->parsed_objects->grafts + pos,
 251                            the_repository->parsed_objects->grafts + pos + 1,
 252                            the_repository->parsed_objects->grafts_nr - pos - 1);
 253         the_repository->parsed_objects->grafts_nr--;
 254         return 0;
 255 }
 256
 257 struct commit_buffer {
 258         void *buffer;
 259         unsigned long size;
 260 };
 261 define_commit_slab(buffer_slab, struct commit_buffer);
 262 static struct buffer_slab buffer_slab = COMMIT_SLAB_INIT(1, buffer_slab);
 263
 264 void set_commit_buffer(struct commit *commit, void *buffer, unsigned long size)
 265 {
 266         struct commit_buffer *v = buffer_slab_at(&buffer_slab, commit);
 267         v->buffer = buffer;
 268         v->size = size;
 269 }
 270
 271 const void *get_cached_commit_buffer(const struct commit *commit, unsigned long *sizep)
 272 {
 273         struct commit_buffer *v = buffer_slab_peek(&buffer_slab, commit);
 274         if (!v) {
 275                 if (sizep)
 276                         *sizep = 0;
 277                 return NULL;
 278         }
 279         if (sizep)
 280                 *sizep = v->size;
 281         return v->buffer;
 282 }
 283
 284 const void *get_commit_buffer(const struct commit *commit, unsigned long *sizep)
 285 {
 286         const void *ret = get_cached_commit_buffer(commit, sizep);
 287         if (!ret) {
 288                 enum object_type type;
 289                 unsigned long size;
 290                 ret = read_object_file(&commit->object.oid, &type, &size);
 291                 if (!ret)
 292                         die("cannot read commit object %s",
 293                             oid_to_hex(&commit->object.oid));
 294                 if (type != OBJ_COMMIT)
 295                         die("expected commit for %s, got %s",
 296                             oid_to_hex(&commit->object.oid), type_name(type));
 297                 if (sizep)
 298                         *sizep = size;
 299         }
 300         return ret;
 301 }
 302
 303 void unuse_commit_buffer(const struct commit *commit, const void *buffer)
 304 {
 305         struct commit_buffer *v = buffer_slab_peek(&buffer_slab, commit);
 306         if (!(v && v->buffer == buffer))
 307                 free((void *)buffer);
 308 }
 309
 310 void free_commit_buffer(struct commit *commit)
 311 {
 312         struct commit_buffer *v = buffer_slab_peek(&buffer_slab, commit);
 313         if (v) {
 314                 FREE_AND_NULL(v->buffer);
 315                 v->size = 0;
 316         }
 317 }
 318
 319 struct tree *get_commit_tree(const struct commit *commit)
 320 {
 321         if (commit->maybe_tree || !commit->object.parsed)
 322                 return commit->maybe_tree;
 323
 324         if (commit->graph_pos == COMMIT_NOT_FROM_GRAPH)
 325                 BUG("commit has NULL tree, but was not loaded from commit-graph");
 326
 327         return get_commit_tree_in_graph(commit);
 328 }
 329
 330 struct object_id *get_commit_tree_oid(const struct commit *commit)
 331 {
 332         return &get_commit_tree(commit)->object.oid;
 333 }
 334
 335 void release_commit_memory(struct commit *c)
 336 {
 337         c->maybe_tree = NULL;
 338         c->index = 0;
 339         free_commit_buffer(c);
 340         free_commit_list(c->parents);
 341         /* TODO: what about commit->util? */
 342
 343         c->object.parsed = 0;
 344 }
 345
 346 const void *detach_commit_buffer(struct commit *commit, unsigned long *sizep)
 347 {
 348         struct commit_buffer *v = buffer_slab_peek(&buffer_slab, commit);
 349         void *ret;
 350
 351         if (!v) {
 352                 if (sizep)
 353                         *sizep = 0;
 354                 return NULL;
 355         }
 356         ret = v->buffer;
 357         if (sizep)
 358                 *sizep = v->size;
 359
 360         v->buffer = NULL;
 361         v->size = 0;
 362         return ret;
 363 }
 364
 365 int parse_commit_buffer(struct commit *item, const void *buffer, unsigned long size, int check_graph)
 366 {
 367         const char *tail = buffer;
 368         const char *bufptr = buffer;
 369         struct object_id parent;
 370         struct commit_list **pptr;
 371         struct commit_graft *graft;
 372         const int tree_entry_len = the_hash_algo->hexsz + 5;
 373         const int parent_entry_len = the_hash_algo->hexsz + 7;
 374
 375         if (item->object.parsed)
 376                 return 0;
 377         item->object.parsed = 1;
 378         tail += size;
 379         if (tail <= bufptr + tree_entry_len + 1 || memcmp(bufptr, "tree ", 5) ||
 380                         bufptr[tree_entry_len] != '\n')
 381                 return error("bogus commit object %s", oid_to_hex(&item->object.oid));
 382         if (get_oid_hex(bufptr + 5, &parent) < 0)
 383                 return error("bad tree pointer in commit %s",
 384                              oid_to_hex(&item->object.oid));
 385         item->maybe_tree = lookup_tree(&parent);
 386         bufptr += tree_entry_len + 1; /* "tree " + "hex sha1" + "\n" */
 387         pptr = &item->parents;
 388
 389         graft = lookup_commit_graft(the_repository, &item->object.oid);
 390         while (bufptr + parent_entry_len < tail && !memcmp(bufptr, "parent ", 7)) {
 391                 struct commit *new_parent;
 392
 393                 if (tail <= bufptr + parent_entry_len + 1 ||
 394                     get_oid_hex(bufptr + 7, &parent) ||
 395                     bufptr[parent_entry_len] != '\n')
 396                         return error("bad parents in commit %s", oid_to_hex(&item->object.oid));
 397                 bufptr += parent_entry_len + 1;
 398                 /*
 399                  * The clone is shallow if nr_parent < 0, and we must
 400                  * not traverse its real parents even when we unhide them.
 401                  */
 402                 if (graft && (graft->nr_parent < 0 || grafts_replace_parents))
 403                         continue;
 404                 new_parent = lookup_commit(&parent);
 405                 if (new_parent)
 406                         pptr = &commit_list_insert(new_parent, pptr)->next;
 407         }
 408         if (graft) {
 409                 int i;
 410                 struct commit *new_parent;
 411                 for (i = 0; i < graft->nr_parent; i++) {
 412                         new_parent = lookup_commit(&graft->parent[i]);
 413                         if (!new_parent)
 414                                 continue;
 415                         pptr = &commit_list_insert(new_parent, pptr)->next;
 416                 }
 417         }
 418         item->date = parse_commit_date(bufptr, tail);
 419
 420         if (check_graph)
 421                 load_commit_graph_info(item);
 422
 423         return 0;
 424 }
 425
 426 int parse_commit_internal(struct commit *item, int quiet_on_missing, int use_commit_graph)
 427 {
 428         enum object_type type;
 429         void *buffer;
 430         unsigned long size;
 431         int ret;
 432
 433         if (!item)
 434                 return -1;
 435         if (item->object.parsed)
 436                 return 0;
 437         if (use_commit_graph && parse_commit_in_graph(item))
 438                 return 0;
 439         buffer = read_object_file(&item->object.oid, &type, &size);
 440         if (!buffer)
 441                 return quiet_on_missing ? -1 :
 442                         error("Could not read %s",
 443                              oid_to_hex(&item->object.oid));
 444         if (type != OBJ_COMMIT) {
 445                 free(buffer);
 446                 return error("Object %s not a commit",
 447                              oid_to_hex(&item->object.oid));
 448         }
 449
 450         ret = parse_commit_buffer(item, buffer, size, 0);
 451         if (save_commit_buffer && !ret) {
 452                 set_commit_buffer(item, buffer, size);
 453                 return 0;
 454         }
 455         free(buffer);
 456         return ret;
 457 }
 458
 459 int parse_commit_gently(struct commit *item, int quiet_on_missing)
 460 {
 461         return parse_commit_internal(item, quiet_on_missing, 1);
 462 }
 463
 464 void parse_commit_or_die(struct commit *item)
 465 {
 466         if (parse_commit(item))
 467                 die("unable to parse commit %s",
 468                     item ? oid_to_hex(&item->object.oid) : "(null)");
 469 }
 470
 471 int find_commit_subject(const char *commit_buffer, const char **subject)
 472 {
 473         const char *eol;
 474         const char *p = commit_buffer;
 475
 476         while (*p && (*p != '\n' || p[1] != '\n'))
 477                 p++;
 478         if (*p) {
 479                 p = skip_blank_lines(p + 2);
 480                 eol = strchrnul(p, '\n');
 481         } else
 482                 eol = p;
 483
 484         *subject = p;
 485
 486         return eol - p;
 487 }
 488
 489 struct commit_list *commit_list_insert(struct commit *item, struct commit_list **list_p)
 490 {
 491         struct commit_list *new_list = xmalloc(sizeof(struct commit_list));
 492         new_list->item = item;
 493         new_list->next = *list_p;
 494         *list_p = new_list;
 495         return new_list;
 496 }
 497
 498 unsigned commit_list_count(const struct commit_list *l)
 499 {
 500         unsigned c = 0;
 501         for (; l; l = l->next )
 502                 c++;
 503         return c;
 504 }
 505
 506 struct commit_list *copy_commit_list(struct commit_list *list)
 507 {
 508         struct commit_list *head = NULL;
 509         struct commit_list **pp = &head;
 510         while (list) {
 511                 pp = commit_list_append(list->item, pp);
 512                 list = list->next;
 513         }
 514         return head;
 515 }
 516
 517 void free_commit_list(struct commit_list *list)
 518 {
 519         while (list)
 520                 pop_commit(&list);
 521 }
 522
 523 struct commit_list * commit_list_insert_by_date(struct commit *item, struct commit_list **list)
 524 {
 525         struct commit_list **pp = list;
 526         struct commit_list *p;
 527         while ((p = *pp) != NULL) {
 528                 if (p->item->date < item->date) {
 529                         break;
 530                 }
 531                 pp = &p->next;
 532         }
 533         return commit_list_insert(item, pp);
 534 }
 535
 536 static int commit_list_compare_by_date(const void *a, const void *b)
 537 {
 538         timestamp_t a_date = ((const struct commit_list *)a)->item->date;
 539         timestamp_t b_date = ((const struct commit_list *)b)->item->date;
 540         if (a_date < b_date)
 541                 return 1;
 542         if (a_date > b_date)
 543                 return -1;
 544         return 0;
 545 }
 546
 547 static void *commit_list_get_next(const void *a)
 548 {
 549         return ((const struct commit_list *)a)->next;
 550 }
 551
 552 static void commit_list_set_next(void *a, void *next)
 553 {
 554         ((struct commit_list *)a)->next = next;
 555 }
 556
 557 void commit_list_sort_by_date(struct commit_list **list)
 558 {
 559         *list = llist_mergesort(*list, commit_list_get_next, commit_list_set_next,
 560                                 commit_list_compare_by_date);
 561 }
 562
 563 struct commit *pop_most_recent_commit(struct commit_list **list,
 564                                       unsigned int mark)
 565 {
 566         struct commit *ret = pop_commit(list);
 567         struct commit_list *parents = ret->parents;
 568
 569         while (parents) {
 570                 struct commit *commit = parents->item;
 571                 if (!parse_commit(commit) && !(commit->object.flags & mark)) {
 572                         commit->object.flags |= mark;
 573                         commit_list_insert_by_date(commit, list);
 574                 }
 575                 parents = parents->next;
 576         }
 577         return ret;
 578 }
 579
 580 static void clear_commit_marks_1(struct commit_list **plist,
 581                                  struct commit *commit, unsigned int mark)
 582 {
 583         while (commit) {
 584                 struct commit_list *parents;
 585
 586                 if (!(mark & commit->object.flags))
 587                         return;
 588
 589                 commit->object.flags &= ~mark;
 590
 591                 parents = commit->parents;
 592                 if (!parents)
 593                         return;
 594
 595                 while ((parents = parents->next))
 596                         commit_list_insert(parents->item, plist);
 597
 598                 commit = commit->parents->item;
 599         }
 600 }
 601
 602 void clear_commit_marks_many(int nr, struct commit **commit, unsigned int mark)
 603 {
 604         struct commit_list *list = NULL;
 605
 606         while (nr--) {
 607                 clear_commit_marks_1(&list, *commit, mark);
 608                 commit++;
 609         }
 610         while (list)
 611                 clear_commit_marks_1(&list, pop_commit(&list), mark);
 612 }
 613
 614 void clear_commit_marks(struct commit *commit, unsigned int mark)
 615 {
 616         clear_commit_marks_many(1, &commit, mark);
 617 }
 618
 619 struct commit *pop_commit(struct commit_list **stack)
 620 {
 621         struct commit_list *top = *stack;
 622         struct commit *item = top ? top->item : NULL;
 623
 624         if (top) {
 625                 *stack = top->next;
 626                 free(top);
 627         }
 628         return item;
 629 }
 630
 631 /*
 632  * Topological sort support
 633  */
 634
 635 /* count number of children that have not been emitted */
 636 define_commit_slab(indegree_slab, int);
 637
 638 /* record author-date for each commit object */
 639 define_commit_slab(author_date_slab, unsigned long);
 640
 641 static void record_author_date(struct author_date_slab *author_date,
 642                                struct commit *commit)
 643 {
 644         const char *buffer = get_commit_buffer(commit, NULL);
 645         struct ident_split ident;
 646         const char *ident_line;
 647         size_t ident_len;
 648         char *date_end;
 649         timestamp_t date;
 650
 651         ident_line = find_commit_header(buffer, "author", &ident_len);
 652         if (!ident_line)
 653                 goto fail_exit; /* no author line */
 654         if (split_ident_line(&ident, ident_line, ident_len) ||
 655             !ident.date_begin || !ident.date_end)
 656                 goto fail_exit; /* malformed "author" line */
 657
 658         date = parse_timestamp(ident.date_begin, &date_end, 10);
 659         if (date_end != ident.date_end)
 660                 goto fail_exit; /* malformed date */
 661         *(author_date_slab_at(author_date, commit)) = date;
 662
 663 fail_exit:
 664         unuse_commit_buffer(commit, buffer);
 665 }
 666
 667 static int compare_commits_by_author_date(const void *a_, const void *b_,
 668                                           void *cb_data)
 669 {
 670         const struct commit *a = a_, *b = b_;
 671         struct author_date_slab *author_date = cb_data;
 672         timestamp_t a_date = *(author_date_slab_at(author_date, a));
 673         timestamp_t b_date = *(author_date_slab_at(author_date, b));
 674
 675         /* newer commits with larger date first */
 676         if (a_date < b_date)
 677                 return 1;
 678         else if (a_date > b_date)
 679                 return -1;
 680         return 0;
 681 }
 682
 683 int compare_commits_by_gen_then_commit_date(const void *a_, const void *b_, void *unused)
 684 {
 685         const struct commit *a = a_, *b = b_;
 686
 687         /* newer commits first */
 688         if (a->generation < b->generation)
 689                 return 1;
 690         else if (a->generation > b->generation)
 691                 return -1;
 692
 693         /* use date as a heuristic when generations are equal */
 694         if (a->date < b->date)
 695                 return 1;
 696         else if (a->date > b->date)
 697                 return -1;
 698         return 0;
 699 }
 700
 701 int compare_commits_by_commit_date(const void *a_, const void *b_, void *unused)
 702 {
 703         const struct commit *a = a_, *b = b_;
 704         /* newer commits with larger date first */
 705         if (a->date < b->date)
 706                 return 1;
 707         else if (a->date > b->date)
 708                 return -1;
 709         return 0;
 710 }
 711
 712 /*
 713  * Performs an in-place topological sort on the list supplied.
 714  */
 715 void sort_in_topological_order(struct commit_list **list, enum rev_sort_order sort_order)
 716 {
 717         struct commit_list *next, *orig = *list;
 718         struct commit_list **pptr;
 719         struct indegree_slab indegree;
 720         struct prio_queue queue;
 721         struct commit *commit;
 722         struct author_date_slab author_date;
 723
 724         if (!orig)
 725                 return;
 726         *list = NULL;
 727
 728         init_indegree_slab(&indegree);
 729         memset(&queue, '\0', sizeof(queue));
 730
 731         switch (sort_order) {
 732         default: /* REV_SORT_IN_GRAPH_ORDER */
 733                 queue.compare = NULL;
 734                 break;
 735         case REV_SORT_BY_COMMIT_DATE:
 736                 queue.compare = compare_commits_by_commit_date;
 737                 break;
 738         case REV_SORT_BY_AUTHOR_DATE:
 739                 init_author_date_slab(&author_date);
 740                 queue.compare = compare_commits_by_author_date;
 741                 queue.cb_data = &author_date;
 742                 break;
 743         }
 744
 745         /* Mark them and clear the indegree */
 746         for (next = orig; next; next = next->next) {
 747                 struct commit *commit = next->item;
 748                 *(indegree_slab_at(&indegree, commit)) = 1;
 749                 /* also record the author dates, if needed */
 750                 if (sort_order == REV_SORT_BY_AUTHOR_DATE)
 751                         record_author_date(&author_date, commit);
 752         }
 753
 754         /* update the indegree */
 755         for (next = orig; next; next = next->next) {
 756                 struct commit_list *parents = next->item->parents;
 757                 while (parents) {
 758                         struct commit *parent = parents->item;
 759                         int *pi = indegree_slab_at(&indegree, parent);
 760
 761                         if (*pi)
 762                                 (*pi)++;
 763                         parents = parents->next;
 764                 }
 765         }
 766
 767         /*
 768          * find the tips
 769          *
 770          * tips are nodes not reachable from any other node in the list
 771          *
 772          * the tips serve as a starting set for the work queue.
 773          */
 774         for (next = orig; next; next = next->next) {
 775                 struct commit *commit = next->item;
 776
 777                 if (*(indegree_slab_at(&indegree, commit)) == 1)
 778                         prio_queue_put(&queue, commit);
 779         }
 780
 781         /*
 782          * This is unfortunate; the initial tips need to be shown
 783          * in the order given from the revision traversal machinery.
 784          */
 785         if (sort_order == REV_SORT_IN_GRAPH_ORDER)
 786                 prio_queue_reverse(&queue);
 787
 788         /* We no longer need the commit list */
 789         free_commit_list(orig);
 790
 791         pptr = list;
 792         *list = NULL;
 793         while ((commit = prio_queue_get(&queue)) != NULL) {
 794                 struct commit_list *parents;
 795
 796                 for (parents = commit->parents; parents ; parents = parents->next) {
 797                         struct commit *parent = parents->item;
 798                         int *pi = indegree_slab_at(&indegree, parent);
 799
 800                         if (!*pi)
 801                                 continue;
 802
 803                         /*
 804                          * parents are only enqueued for emission
 805                          * when all their children have been emitted thereby
 806                          * guaranteeing topological order.
 807                          */
 808                         if (--(*pi) == 1)
 809                                 prio_queue_put(&queue, parent);
 810                 }
 811                 /*
 812                  * all children of commit have already been
 813                  * emitted. we can emit it now.
 814                  */
 815                 *(indegree_slab_at(&indegree, commit)) = 0;
 816
 817                 pptr = &commit_list_insert(commit, pptr)->next;
 818         }
 819
 820         clear_indegree_slab(&indegree);
 821         clear_prio_queue(&queue);
 822         if (sort_order == REV_SORT_BY_AUTHOR_DATE)
 823                 clear_author_date_slab(&author_date);
 824 }
 825
 826 /* merge-base stuff */
 827
 828 /* Remember to update object flag allocation in object.h */
 829 #define PARENT1         (1u<<16)
 830 #define PARENT2         (1u<<17)
 831 #define STALE           (1u<<18)
 832 #define RESULT          (1u<<19)
 833
 834 static const unsigned all_flags = (PARENT1 | PARENT2 | STALE | RESULT);
 835
 836 static int queue_has_nonstale(struct prio_queue *queue)
 837 {
 838         int i;
 839         for (i = 0; i < queue->nr; i++) {
 840                 struct commit *commit = queue->array[i].data;
 841                 if (!(commit->object.flags & STALE))
 842                         return 1;
 843         }
 844         return 0;
 845 }
 846
 847 /* all input commits in one and twos[] must have been parsed! */
 848 static struct commit_list *paint_down_to_common(struct commit *one, int n,
 849                                                 struct commit **twos,
 850                                                 int min_generation)
 851 {
 852         struct prio_queue queue = { compare_commits_by_gen_then_commit_date };
 853         struct commit_list *result = NULL;
 854         int i;
 855         uint32_t last_gen = GENERATION_NUMBER_INFINITY;
 856
 857         one->object.flags |= PARENT1;
 858         if (!n) {
 859                 commit_list_append(one, &result);
 860                 return result;
 861         }
 862         prio_queue_put(&queue, one);
 863
 864         for (i = 0; i < n; i++) {
 865                 twos[i]->object.flags |= PARENT2;
 866                 prio_queue_put(&queue, twos[i]);
 867         }
 868
 869         while (queue_has_nonstale(&queue)) {
 870                 struct commit *commit = prio_queue_get(&queue);
 871                 struct commit_list *parents;
 872                 int flags;
 873
 874                 if (commit->generation > last_gen)
 875                         BUG("bad generation skip %8x > %8x at %s",
 876                             commit->generation, last_gen,
 877                             oid_to_hex(&commit->object.oid));
 878                 last_gen = commit->generation;
 879
 880                 if (commit->generation < min_generation)
 881                         break;
 882
 883                 flags = commit->object.flags & (PARENT1 | PARENT2 | STALE);
 884                 if (flags == (PARENT1 | PARENT2)) {
 885                         if (!(commit->object.flags & RESULT)) {
 886                                 commit->object.flags |= RESULT;
 887                                 commit_list_insert_by_date(commit, &result);
 888                         }
 889                         /* Mark parents of a found merge stale */
 890                         flags |= STALE;
 891                 }
 892                 parents = commit->parents;
 893                 while (parents) {
 894                         struct commit *p = parents->item;
 895                         parents = parents->next;
 896                         if ((p->object.flags & flags) == flags)
 897                                 continue;
 898                         if (parse_commit(p))
 899                                 return NULL;
 900                         p->object.flags |= flags;
 901                         prio_queue_put(&queue, p);
 902                 }
 903         }
 904
 905         clear_prio_queue(&queue);
 906         return result;
 907 }
 908
 909 static struct commit_list *merge_bases_many(struct commit *one, int n, struct commit **twos)
 910 {
 911         struct commit_list *list = NULL;
 912         struct commit_list *result = NULL;
 913         int i;
 914
 915         for (i = 0; i < n; i++) {
 916                 if (one == twos[i])
 917                         /*
 918                          * We do not mark this even with RESULT so we do not
 919                          * have to clean it up.
 920                          */
 921                         return commit_list_insert(one, &result);
 922         }
 923
 924         if (parse_commit(one))
 925                 return NULL;
 926         for (i = 0; i < n; i++) {
 927                 if (parse_commit(twos[i]))
 928                         return NULL;
 929         }
 930
 931         list = paint_down_to_common(one, n, twos, 0);
 932
 933         while (list) {
 934                 struct commit *commit = pop_commit(&list);
 935                 if (!(commit->object.flags & STALE))
 936                         commit_list_insert_by_date(commit, &result);
 937         }
 938         return result;
 939 }
 940
 941 struct commit_list *get_octopus_merge_bases(struct commit_list *in)
 942 {
 943         struct commit_list *i, *j, *k, *ret = NULL;
 944
 945         if (!in)
 946                 return ret;
 947
 948         commit_list_insert(in->item, &ret);
 949
 950         for (i = in->next; i; i = i->next) {
 951                 struct commit_list *new_commits = NULL, *end = NULL;
 952
 953                 for (j = ret; j; j = j->next) {
 954                         struct commit_list *bases;
 955                         bases = get_merge_bases(i->item, j->item);
 956                         if (!new_commits)
 957                                 new_commits = bases;
 958                         else
 959                                 end->next = bases;
 960                         for (k = bases; k; k = k->next)
 961                                 end = k;
 962                 }
 963                 ret = new_commits;
 964         }
 965         return ret;
 966 }
 967
 968 static int remove_redundant(struct commit **array, int cnt)
 969 {
 970         /*
 971          * Some commit in the array may be an ancestor of
 972          * another commit.  Move such commit to the end of
 973          * the array, and return the number of commits that
 974          * are independent from each other.
 975          */
 976         struct commit **work;
 977         unsigned char *redundant;
 978         int *filled_index;
 979         int i, j, filled;
 980
 981         work = xcalloc(cnt, sizeof(*work));
 982         redundant = xcalloc(cnt, 1);
 983         ALLOC_ARRAY(filled_index, cnt - 1);
 984
 985         for (i = 0; i < cnt; i++)
 986                 parse_commit(array[i]);
 987         for (i = 0; i < cnt; i++) {
 988                 struct commit_list *common;
 989                 uint32_t min_generation = array[i]->generation;
 990
 991                 if (redundant[i])
 992                         continue;
 993                 for (j = filled = 0; j < cnt; j++) {
 994                         if (i == j || redundant[j])
 995                                 continue;
 996                         filled_index[filled] = j;
 997                         work[filled++] = array[j];
 998
 999                         if (array[j]->generation < min_generation)
1000                                 min_generation = array[j]->generation;
1001                 }
1002                 common = paint_down_to_common(array[i], filled, work,
1003                                               min_generation);
1004                 if (array[i]->object.flags & PARENT2)
1005                         redundant[i] = 1;
1006                 for (j = 0; j < filled; j++)
1007                         if (work[j]->object.flags & PARENT1)
1008                                 redundant[filled_index[j]] = 1;
1009                 clear_commit_marks(array[i], all_flags);
1010                 clear_commit_marks_many(filled, work, all_flags);
1011                 free_commit_list(common);
1012         }
1013
1014         /* Now collect the result */
1015         COPY_ARRAY(work, array, cnt);
1016         for (i = filled = 0; i < cnt; i++)
1017                 if (!redundant[i])
1018                         array[filled++] = work[i];
1019         for (j = filled, i = 0; i < cnt; i++)
1020                 if (redundant[i])
1021                         array[j++] = work[i];
1022         free(work);
1023         free(redundant);
1024         free(filled_index);
1025         return filled;
1026 }
1027
1028 static struct commit_list *get_merge_bases_many_0(struct commit *one,
1029                                                   int n,
1030                                                   struct commit **twos,
1031                                                   int cleanup)
1032 {
1033         struct commit_list *list;
1034         struct commit **rslt;
1035         struct commit_list *result;
1036         int cnt, i;
1037
1038         result = merge_bases_many(one, n, twos);
1039         for (i = 0; i < n; i++) {
1040                 if (one == twos[i])
1041                         return result;
1042         }
1043         if (!result || !result->next) {
1044                 if (cleanup) {
1045                         clear_commit_marks(one, all_flags);
1046                         clear_commit_marks_many(n, twos, all_flags);
1047                 }
1048                 return result;
1049         }
1050
1051         /* There are more than one */
1052         cnt = commit_list_count(result);
1053         rslt = xcalloc(cnt, sizeof(*rslt));
1054         for (list = result, i = 0; list; list = list->next)
1055                 rslt[i++] = list->item;
1056         free_commit_list(result);
1057
1058         clear_commit_marks(one, all_flags);
1059         clear_commit_marks_many(n, twos, all_flags);
1060
1061         cnt = remove_redundant(rslt, cnt);
1062         result = NULL;
1063         for (i = 0; i < cnt; i++)
1064                 commit_list_insert_by_date(rslt[i], &result);
1065         free(rslt);
1066         return result;
1067 }
1068
1069 struct commit_list *get_merge_bases_many(struct commit *one,
1070                                          int n,
1071                                          struct commit **twos)
1072 {
1073         return get_merge_bases_many_0(one, n, twos, 1);
1074 }
1075
1076 struct commit_list *get_merge_bases_many_dirty(struct commit *one,
1077                                                int n,
1078                                                struct commit **twos)
1079 {
1080         return get_merge_bases_many_0(one, n, twos, 0);
1081 }
1082
1083 struct commit_list *get_merge_bases(struct commit *one, struct commit *two)
1084 {
1085         return get_merge_bases_many_0(one, 1, &two, 1);
1086 }
1087
1088 /*
1089  * Is "commit" a descendant of one of the elements on the "with_commit" list?
1090  */
1091 int is_descendant_of(struct commit *commit, struct commit_list *with_commit)
1092 {
1093         if (!with_commit)
1094                 return 1;
1095         while (with_commit) {
1096                 struct commit *other;
1097
1098                 other = with_commit->item;
1099                 with_commit = with_commit->next;
1100                 if (in_merge_bases(other, commit))
1101                         return 1;
1102         }
1103         return 0;
1104 }
1105
1106 /*
1107  * Is "commit" an ancestor of one of the "references"?
1108  */
1109 int in_merge_bases_many(struct commit *commit, int nr_reference, struct commit **reference)
1110 {
1111         struct commit_list *bases;
1112         int ret = 0, i;
1113         uint32_t min_generation = GENERATION_NUMBER_INFINITY;
1114
1115         if (parse_commit(commit))
1116                 return ret;
1117         for (i = 0; i < nr_reference; i++) {
1118                 if (parse_commit(reference[i]))
1119                         return ret;
1120                 if (reference[i]->generation < min_generation)
1121                         min_generation = reference[i]->generation;
1122         }
1123
1124         if (commit->generation > min_generation)
1125                 return ret;
1126
1127         bases = paint_down_to_common(commit, nr_reference, reference, commit->generation);
1128         if (commit->object.flags & PARENT2)
1129                 ret = 1;
1130         clear_commit_marks(commit, all_flags);
1131         clear_commit_marks_many(nr_reference, reference, all_flags);
1132         free_commit_list(bases);
1133         return ret;
1134 }
1135
1136 /*
1137  * Is "commit" an ancestor of (i.e. reachable from) the "reference"?
1138  */
1139 int in_merge_bases(struct commit *commit, struct commit *reference)
1140 {
1141         return in_merge_bases_many(commit, 1, &reference);
1142 }
1143
1144 struct commit_list *reduce_heads(struct commit_list *heads)
1145 {
1146         struct commit_list *p;
1147         struct commit_list *result = NULL, **tail = &result;
1148         struct commit **array;
1149         int num_head, i;
1150
1151         if (!heads)
1152                 return NULL;
1153
1154         /* Uniquify */
1155         for (p = heads; p; p = p->next)
1156                 p->item->object.flags &= ~STALE;
1157         for (p = heads, num_head = 0; p; p = p->next) {
1158                 if (p->item->object.flags & STALE)
1159                         continue;
1160                 p->item->object.flags |= STALE;
1161                 num_head++;
1162         }
1163         array = xcalloc(num_head, sizeof(*array));
1164         for (p = heads, i = 0; p; p = p->next) {
1165                 if (p->item->object.flags & STALE) {
1166                         array[i++] = p->item;
1167                         p->item->object.flags &= ~STALE;
1168                 }
1169         }
1170         num_head = remove_redundant(array, num_head);
1171         for (i = 0; i < num_head; i++)
1172                 tail = &commit_list_insert(array[i], tail)->next;
1173         free(array);
1174         return result;
1175 }
1176
1177 void reduce_heads_replace(struct commit_list **heads)
1178 {
1179         struct commit_list *result = reduce_heads(*heads);
1180         free_commit_list(*heads);
1181         *heads = result;
1182 }
1183
1184 static const char gpg_sig_header[] = "gpgsig";
1185 static const int gpg_sig_header_len = sizeof(gpg_sig_header) - 1;
1186
1187 static int do_sign_commit(struct strbuf *buf, const char *keyid)
1188 {
1189         struct strbuf sig = STRBUF_INIT;
1190         int inspos, copypos;
1191         const char *eoh;
1192
1193         /* find the end of the header */
1194         eoh = strstr(buf->buf, "\n\n");
1195         if (!eoh)
1196                 inspos = buf->len;
1197         else
1198                 inspos = eoh - buf->buf + 1;
1199
1200         if (!keyid || !*keyid)
1201                 keyid = get_signing_key();
1202         if (sign_buffer(buf, &sig, keyid)) {
1203                 strbuf_release(&sig);
1204                 return -1;
1205         }
1206
1207         for (copypos = 0; sig.buf[copypos]; ) {
1208                 const char *bol = sig.buf + copypos;
1209                 const char *eol = strchrnul(bol, '\n');
1210                 int len = (eol - bol) + !!*eol;
1211
1212                 if (!copypos) {
1213                         strbuf_insert(buf, inspos, gpg_sig_header, gpg_sig_header_len);
1214                         inspos += gpg_sig_header_len;
1215                 }
1216                 strbuf_insert(buf, inspos++, " ", 1);
1217                 strbuf_insert(buf, inspos, bol, len);
1218                 inspos += len;
1219                 copypos += len;
1220         }
1221         strbuf_release(&sig);
1222         return 0;
1223 }
1224
1225 int parse_signed_commit(const struct commit *commit,
1226                         struct strbuf *payload, struct strbuf *signature)
1227 {
1228
1229         unsigned long size;
1230         const char *buffer = get_commit_buffer(commit, &size);
1231         int in_signature, saw_signature = -1;
1232         const char *line, *tail;
1233
1234         line = buffer;
1235         tail = buffer + size;
1236         in_signature = 0;
1237         saw_signature = 0;
1238         while (line < tail) {
1239                 const char *sig = NULL;
1240                 const char *next = memchr(line, '\n', tail - line);
1241
1242                 next = next ? next + 1 : tail;
1243                 if (in_signature && line[0] == ' ')
1244                         sig = line + 1;
1245                 else if (starts_with(line, gpg_sig_header) &&
1246                          line[gpg_sig_header_len] == ' ')
1247                         sig = line + gpg_sig_header_len + 1;
1248                 if (sig) {
1249                         strbuf_add(signature, sig, next - sig);
1250                         saw_signature = 1;
1251                         in_signature = 1;
1252                 } else {
1253                         if (*line == '\n')
1254                                 /* dump the whole remainder of the buffer */
1255                                 next = tail;
1256                         strbuf_add(payload, line, next - line);
1257                         in_signature = 0;
1258                 }
1259                 line = next;
1260         }
1261         unuse_commit_buffer(commit, buffer);
1262         return saw_signature;
1263 }
1264
1265 int remove_signature(struct strbuf *buf)
1266 {
1267         const char *line = buf->buf;
1268         const char *tail = buf->buf + buf->len;
1269         int in_signature = 0;
1270         const char *sig_start = NULL;
1271         const char *sig_end = NULL;
1272
1273         while (line < tail) {
1274                 const char *next = memchr(line, '\n', tail - line);
1275                 next = next ? next + 1 : tail;
1276
1277                 if (in_signature && line[0] == ' ')
1278                         sig_end = next;
1279                 else if (starts_with(line, gpg_sig_header) &&
1280                          line[gpg_sig_header_len] == ' ') {
1281                         sig_start = line;
1282                         sig_end = next;
1283                         in_signature = 1;
1284                 } else {
1285                         if (*line == '\n')
1286                                 /* dump the whole remainder of the buffer */
1287                                 next = tail;
1288                         in_signature = 0;
1289                 }
1290                 line = next;
1291         }
1292
1293         if (sig_start)
1294                 strbuf_remove(buf, sig_start - buf->buf, sig_end - sig_start);
1295
1296         return sig_start != NULL;
1297 }
1298
1299 static void handle_signed_tag(struct commit *parent, struct commit_extra_header ***tail)
1300 {
1301         struct merge_remote_desc *desc;
1302         struct commit_extra_header *mergetag;
1303         char *buf;
1304         unsigned long size, len;
1305         enum object_type type;
1306
1307         desc = merge_remote_util(parent);
1308         if (!desc || !desc->obj)
1309                 return;
1310         buf = read_object_file(&desc->obj->oid, &type, &size);
1311         if (!buf || type != OBJ_TAG)
1312                 goto free_return;
1313         len = parse_signature(buf, size);
1314         if (size == len)
1315                 goto free_return;
1316         /*
1317          * We could verify this signature and either omit the tag when
1318          * it does not validate, but the integrator may not have the
1319          * public key of the signer of the tag he is merging, while a
1320          * later auditor may have it while auditing, so let's not run
1321          * verify-signed-buffer here for now...
1322          *
1323          * if (verify_signed_buffer(buf, len, buf + len, size - len, ...))
1324          *      warn("warning: signed tag unverified.");
1325          */
1326         mergetag = xcalloc(1, sizeof(*mergetag));
1327         mergetag->key = xstrdup("mergetag");
1328         mergetag->value = buf;
1329         mergetag->len = size;
1330
1331         **tail = mergetag;
1332         *tail = &mergetag->next;
1333         return;
1334
1335 free_return:
1336         free(buf);
1337 }
1338
1339 int check_commit_signature(const struct commit *commit, struct signature_check *sigc)
1340 {
1341         struct strbuf payload = STRBUF_INIT;
1342         struct strbuf signature = STRBUF_INIT;
1343         int ret = 1;
1344
1345         sigc->result = 'N';
1346
1347         if (parse_signed_commit(commit, &payload, &signature) <= 0)
1348                 goto out;
1349         ret = check_signature(payload.buf, payload.len, signature.buf,
1350                 signature.len, sigc);
1351
1352  out:
1353         strbuf_release(&payload);
1354         strbuf_release(&signature);
1355
1356         return ret;
1357 }
1358
1359
1360
1361 void append_merge_tag_headers(struct commit_list *parents,
1362                               struct commit_extra_header ***tail)
1363 {
1364         while (parents) {
1365                 struct commit *parent = parents->item;
1366                 handle_signed_tag(parent, tail);
1367                 parents = parents->next;
1368         }
1369 }
1370
1371 static void add_extra_header(struct strbuf *buffer,
1372                              struct commit_extra_header *extra)
1373 {
1374         strbuf_addstr(buffer, extra->key);
1375         if (extra->len)
1376                 strbuf_add_lines(buffer, " ", extra->value, extra->len);
1377         else
1378                 strbuf_addch(buffer, '\n');
1379 }
1380
1381 struct commit_extra_header *read_commit_extra_headers(struct commit *commit,
1382                                                       const char **exclude)
1383 {
1384         struct commit_extra_header *extra = NULL;
1385         unsigned long size;
1386         const char *buffer = get_commit_buffer(commit, &size);
1387         extra = read_commit_extra_header_lines(buffer, size, exclude);
1388         unuse_commit_buffer(commit, buffer);
1389         return extra;
1390 }
1391
1392 int for_each_mergetag(each_mergetag_fn fn, struct commit *commit, void *data)
1393 {
1394         struct commit_extra_header *extra, *to_free;
1395         int res = 0;
1396
1397         to_free = read_commit_extra_headers(commit, NULL);
1398         for (extra = to_free; !res && extra; extra = extra->next) {
1399                 if (strcmp(extra->key, "mergetag"))
1400                         continue; /* not a merge tag */
1401                 res = fn(commit, extra, data);
1402         }
1403         free_commit_extra_headers(to_free);
1404         return res;
1405 }
1406
1407 static inline int standard_header_field(const char *field, size_t len)
1408 {
1409         return ((len == 4 && !memcmp(field, "tree", 4)) ||
1410                 (len == 6 && !memcmp(field, "parent", 6)) ||
1411                 (len == 6 && !memcmp(field, "author", 6)) ||
1412                 (len == 9 && !memcmp(field, "committer", 9)) ||
1413                 (len == 8 && !memcmp(field, "encoding", 8)));
1414 }
1415
1416 static int excluded_header_field(const char *field, size_t len, const char **exclude)
1417 {
1418         if (!exclude)
1419                 return 0;
1420
1421         while (*exclude) {
1422                 size_t xlen = strlen(*exclude);
1423                 if (len == xlen && !memcmp(field, *exclude, xlen))
1424                         return 1;
1425                 exclude++;
1426         }
1427         return 0;
1428 }
1429
1430 static struct commit_extra_header *read_commit_extra_header_lines(
1431         const char *buffer, size_t size,
1432         const char **exclude)
1433 {
1434         struct commit_extra_header *extra = NULL, **tail = &extra, *it = NULL;
1435         const char *line, *next, *eof, *eob;
1436         struct strbuf buf = STRBUF_INIT;
1437
1438         for (line = buffer, eob = line + size;
1439              line < eob && *line != '\n';
1440              line = next) {
1441                 next = memchr(line, '\n', eob - line);
1442                 next = next ? next + 1 : eob;
1443                 if (*line == ' ') {
1444                         /* continuation */
1445                         if (it)
1446                                 strbuf_add(&buf, line + 1, next - (line + 1));
1447                         continue;
1448                 }
1449                 if (it)
1450                         it->value = strbuf_detach(&buf, &it->len);
1451                 strbuf_reset(&buf);
1452                 it = NULL;
1453
1454                 eof = memchr(line, ' ', next - line);
1455                 if (!eof)
1456                         eof = next;
1457                 else if (standard_header_field(line, eof - line) ||
1458                          excluded_header_field(line, eof - line, exclude))
1459                         continue;
1460
1461                 it = xcalloc(1, sizeof(*it));
1462                 it->key = xmemdupz(line, eof-line);
1463                 *tail = it;
1464                 tail = &it->next;
1465                 if (eof + 1 < next)
1466                         strbuf_add(&buf, eof + 1, next - (eof + 1));
1467         }
1468         if (it)
1469                 it->value = strbuf_detach(&buf, &it->len);
1470         return extra;
1471 }
1472
1473 void free_commit_extra_headers(struct commit_extra_header *extra)
1474 {
1475         while (extra) {
1476                 struct commit_extra_header *next = extra->next;
1477                 free(extra->key);
1478                 free(extra->value);
1479                 free(extra);
1480                 extra = next;
1481         }
1482 }
1483
1484 int commit_tree(const char *msg, size_t msg_len, const struct object_id *tree,
1485                 struct commit_list *parents, struct object_id *ret,
1486                 const char *author, const char *sign_commit)
1487 {
1488         struct commit_extra_header *extra = NULL, **tail = &extra;
1489         int result;
1490
1491         append_merge_tag_headers(parents, &tail);
1492         result = commit_tree_extended(msg, msg_len, tree, parents, ret,
1493                                       author, sign_commit, extra);
1494         free_commit_extra_headers(extra);
1495         return result;
1496 }
1497
1498 static int find_invalid_utf8(const char *buf, int len)
1499 {
1500         int offset = 0;
1501         static const unsigned int max_codepoint[] = {
1502                 0x7f, 0x7ff, 0xffff, 0x10ffff
1503         };
1504
1505         while (len) {
1506                 unsigned char c = *buf++;
1507                 int bytes, bad_offset;
1508                 unsigned int codepoint;
1509                 unsigned int min_val, max_val;
1510
1511                 len--;
1512                 offset++;
1513
1514                 /* Simple US-ASCII? No worries. */
1515                 if (c < 0x80)
1516                         continue;
1517
1518                 bad_offset = offset-1;
1519
1520                 /*
1521                  * Count how many more high bits set: that's how
1522                  * many more bytes this sequence should have.
1523                  */
1524                 bytes = 0;
1525                 while (c & 0x40) {
1526                         c <<= 1;
1527                         bytes++;
1528                 }
1529
1530                 /*
1531                  * Must be between 1 and 3 more bytes.  Longer sequences result in
1532                  * codepoints beyond U+10FFFF, which are guaranteed never to exist.
1533                  */
1534                 if (bytes < 1 || 3 < bytes)
1535                         return bad_offset;
1536
1537                 /* Do we *have* that many bytes? */
1538                 if (len < bytes)
1539                         return bad_offset;
1540
1541                 /*
1542                  * Place the encoded bits at the bottom of the value and compute the
1543                  * valid range.
1544                  */
1545                 codepoint = (c & 0x7f) >> bytes;
1546                 min_val = max_codepoint[bytes-1] + 1;
1547                 max_val = max_codepoint[bytes];
1548
1549                 offset += bytes;
1550                 len -= bytes;
1551
1552                 /* And verify that they are good continuation bytes */
1553                 do {
1554                         codepoint <<= 6;
1555                         codepoint |= *buf & 0x3f;
1556                         if ((*buf++ & 0xc0) != 0x80)
1557                                 return bad_offset;
1558                 } while (--bytes);
1559
1560                 /* Reject codepoints that are out of range for the sequence length. */
1561                 if (codepoint < min_val || codepoint > max_val)
1562                         return bad_offset;
1563                 /* Surrogates are only for UTF-16 and cannot be encoded in UTF-8. */
1564                 if ((codepoint & 0x1ff800) == 0xd800)
1565                         return bad_offset;
1566                 /* U+xxFFFE and U+xxFFFF are guaranteed non-characters. */
1567                 if ((codepoint & 0xfffe) == 0xfffe)
1568                         return bad_offset;
1569                 /* So are anything in the range U+FDD0..U+FDEF. */
1570                 if (codepoint >= 0xfdd0 && codepoint <= 0xfdef)
1571                         return bad_offset;
1572         }
1573         return -1;
1574 }
1575
1576 /*
1577  * This verifies that the buffer is in proper utf8 format.
1578  *
1579  * If it isn't, it assumes any non-utf8 characters are Latin1,
1580  * and does the conversion.
1581  */
1582 static int verify_utf8(struct strbuf *buf)
1583 {
1584         int ok = 1;
1585         long pos = 0;
1586
1587         for (;;) {
1588                 int bad;
1589                 unsigned char c;
1590                 unsigned char replace[2];
1591
1592                 bad = find_invalid_utf8(buf->buf + pos, buf->len - pos);
1593                 if (bad < 0)
1594                         return ok;
1595                 pos += bad;
1596                 ok = 0;
1597                 c = buf->buf[pos];
1598                 strbuf_remove(buf, pos, 1);
1599
1600                 /* We know 'c' must be in the range 128-255 */
1601                 replace[0] = 0xc0 + (c >> 6);
1602                 replace[1] = 0x80 + (c & 0x3f);
1603                 strbuf_insert(buf, pos, replace, 2);
1604                 pos += 2;
1605         }
1606 }
1607
1608 static const char commit_utf8_warn[] =
1609 N_("Warning: commit message did not conform to UTF-8.\n"
1610    "You may want to amend it after fixing the message, or set the config\n"
1611    "variable i18n.commitencoding to the encoding your project uses.\n");
1612
1613 int commit_tree_extended(const char *msg, size_t msg_len,
1614                          const struct object_id *tree,
1615                          struct commit_list *parents, struct object_id *ret,
1616                          const char *author, const char *sign_commit,
1617                          struct commit_extra_header *extra)
1618 {
1619         int result;
1620         int encoding_is_utf8;
1621         struct strbuf buffer;
1622
1623         assert_oid_type(tree, OBJ_TREE);
1624
1625         if (memchr(msg, '\0', msg_len))
1626                 return error("a NUL byte in commit log message not allowed.");
1627
1628         /* Not having i18n.commitencoding is the same as having utf-8 */
1629         encoding_is_utf8 = is_encoding_utf8(git_commit_encoding);
1630
1631         strbuf_init(&buffer, 8192); /* should avoid reallocs for the headers */
1632         strbuf_addf(&buffer, "tree %s\n", oid_to_hex(tree));
1633
1634         /*
1635          * NOTE! This ordering means that the same exact tree merged with a
1636          * different order of parents will be a _different_ changeset even
1637          * if everything else stays the same.
1638          */
1639         while (parents) {
1640                 struct commit *parent = pop_commit(&parents);
1641                 strbuf_addf(&buffer, "parent %s\n",
1642                             oid_to_hex(&parent->object.oid));
1643         }
1644
1645         /* Person/date information */
1646         if (!author)
1647                 author = git_author_info(IDENT_STRICT);
1648         strbuf_addf(&buffer, "author %s\n", author);
1649         strbuf_addf(&buffer, "committer %s\n", git_committer_info(IDENT_STRICT));
1650         if (!encoding_is_utf8)
1651                 strbuf_addf(&buffer, "encoding %s\n", git_commit_encoding);
1652
1653         while (extra) {
1654                 add_extra_header(&buffer, extra);
1655                 extra = extra->next;
1656         }
1657         strbuf_addch(&buffer, '\n');
1658
1659         /* And add the comment */
1660         strbuf_add(&buffer, msg, msg_len);
1661
1662         /* And check the encoding */
1663         if (encoding_is_utf8 && !verify_utf8(&buffer))
1664                 fprintf(stderr, _(commit_utf8_warn));
1665
1666         if (sign_commit && do_sign_commit(&buffer, sign_commit)) {
1667                 result = -1;
1668                 goto out;
1669         }
1670
1671         result = write_object_file(buffer.buf, buffer.len, commit_type, ret);
1672 out:
1673         strbuf_release(&buffer);
1674         return result;
1675 }
1676
1677 define_commit_slab(merge_desc_slab, struct merge_remote_desc *);
1678 static struct merge_desc_slab merge_desc_slab = COMMIT_SLAB_INIT(1, merge_desc_slab);
1679
1680 struct merge_remote_desc *merge_remote_util(struct commit *commit)
1681 {
1682         return *merge_desc_slab_at(&merge_desc_slab, commit);
1683 }
1684
1685 void set_merge_remote_desc(struct commit *commit,
1686                            const char *name, struct object *obj)
1687 {
1688         struct merge_remote_desc *desc;
1689         FLEX_ALLOC_STR(desc, name, name);
1690         desc->obj = obj;
1691         *merge_desc_slab_at(&merge_desc_slab, commit) = desc;
1692 }
1693
1694 struct commit *get_merge_parent(const char *name)
1695 {
1696         struct object *obj;
1697         struct commit *commit;
1698         struct object_id oid;
1699         if (get_oid(name, &oid))
1700                 return NULL;
1701         obj = parse_object(&oid);
1702         commit = (struct commit *)peel_to_type(name, 0, obj, OBJ_COMMIT);
1703         if (commit && !merge_remote_util(commit))
1704                 set_merge_remote_desc(commit, name, obj);
1705         return commit;
1706 }
1707
1708 /*
1709  * Append a commit to the end of the commit_list.
1710  *
1711  * next starts by pointing to the variable that holds the head of an
1712  * empty commit_list, and is updated to point to the "next" field of
1713  * the last item on the list as new commits are appended.
1714  *
1715  * Usage example:
1716  *
1717  *     struct commit_list *list;
1718  *     struct commit_list **next = &list;
1719  *
1720  *     next = commit_list_append(c1, next);
1721  *     next = commit_list_append(c2, next);
1722  *     assert(commit_list_count(list) == 2);
1723  *     return list;
1724  */
1725 struct commit_list **commit_list_append(struct commit *commit,
1726                                         struct commit_list **next)
1727 {
1728         struct commit_list *new_commit = xmalloc(sizeof(struct commit_list));
1729         new_commit->item = commit;
1730         *next = new_commit;
1731         new_commit->next = NULL;
1732         return &new_commit->next;
1733 }
1734
1735 const char *find_commit_header(const char *msg, const char *key, size_t *out_len)
1736 {
1737         int key_len = strlen(key);
1738         const char *line = msg;
1739
1740         while (line) {
1741                 const char *eol = strchrnul(line, '\n');
1742
1743                 if (line == eol)
1744                         return NULL;
1745
1746                 if (eol - line > key_len &&
1747                     !strncmp(line, key, key_len) &&
1748                     line[key_len] == ' ') {
1749                         *out_len = eol - line - key_len - 1;
1750                         return line + key_len + 1;
1751                 }
1752                 line = *eol ? eol + 1 : NULL;
1753         }
1754         return NULL;
1755 }
1756
1757 /*
1758  * Inspect the given string and determine the true "end" of the log message, in
1759  * order to find where to put a new Signed-off-by: line.  Ignored are
1760  * trailing comment lines and blank lines.  To support "git commit -s
1761  * --amend" on an existing commit, we also ignore "Conflicts:".  To
1762  * support "git commit -v", we truncate at cut lines.
1763  *
1764  * Returns the number of bytes from the tail to ignore, to be fed as
1765  * the second parameter to append_signoff().
1766  */
1767 int ignore_non_trailer(const char *buf, size_t len)
1768 {
1769         int boc = 0;
1770         int bol = 0;
1771         int in_old_conflicts_block = 0;
1772         size_t cutoff = wt_status_locate_end(buf, len);
1773
1774         while (bol < cutoff) {
1775                 const char *next_line = memchr(buf + bol, '\n', len - bol);
1776
1777                 if (!next_line)
1778                         next_line = buf + len;
1779                 else
1780                         next_line++;
1781
1782                 if (buf[bol] == comment_line_char || buf[bol] == '\n') {
1783                         /* is this the first of the run of comments? */
1784                         if (!boc)
1785                                 boc = bol;
1786                         /* otherwise, it is just continuing */
1787                 } else if (starts_with(buf + bol, "Conflicts:\n")) {
1788                         in_old_conflicts_block = 1;
1789                         if (!boc)
1790                                 boc = bol;
1791                 } else if (in_old_conflicts_block && buf[bol] == '\t') {
1792                         ; /* a pathname in the conflicts block */
1793                 } else if (boc) {
1794                         /* the previous was not trailing comment */
1795                         boc = 0;
1796                         in_old_conflicts_block = 0;
1797                 }
1798                 bol = next_line - buf;
1799         }
1800         return boc ? len - boc : len - cutoff;
1801 }