commit.c

   1 #include "cache.h"
   2 #include "tag.h"
   3 #include "commit.h"
   4 #include "commit-graph.h"
   5 #include "pkt-line.h"
   6 #include "utf8.h"
   7 #include "diff.h"
   8 #include "revision.h"
   9 #include "notes.h"
  10 #include "gpg-interface.h"
  11 #include "mergesort.h"
  12 #include "commit-slab.h"
  13 #include "prio-queue.h"
  14 #include "sha1-lookup.h"
  15 #include "wt-status.h"
  16
  17 static struct commit_extra_header *read_commit_extra_header_lines(const char *buf, size_t len, const char **);
  18
  19 int save_commit_buffer = 1;
  20
  21 const char *commit_type = "commit";
  22
  23 struct commit *lookup_commit_reference_gently(const struct object_id *oid,
  24                                               int quiet)
  25 {
  26         struct object *obj = deref_tag(parse_object(oid), NULL, 0);
  27
  28         if (!obj)
  29                 return NULL;
  30         return object_as_type(obj, OBJ_COMMIT, quiet);
  31 }
  32
  33 struct commit *lookup_commit_reference(const struct object_id *oid)
  34 {
  35         return lookup_commit_reference_gently(oid, 0);
  36 }
  37
  38 struct commit *lookup_commit_or_die(const struct object_id *oid, const char *ref_name)
  39 {
  40         struct commit *c = lookup_commit_reference(oid);
  41         if (!c)
  42                 die(_("could not parse %s"), ref_name);
  43         if (oidcmp(oid, &c->object.oid)) {
  44                 warning(_("%s %s is not a commit!"),
  45                         ref_name, oid_to_hex(oid));
  46         }
  47         return c;
  48 }
  49
  50 struct commit *lookup_commit(const struct object_id *oid)
  51 {
  52         struct object *obj = lookup_object(oid->hash);
  53         if (!obj)
  54                 return create_object(oid->hash, alloc_commit_node());
  55         return object_as_type(obj, OBJ_COMMIT, 0);
  56 }
  57
  58 struct commit *lookup_commit_reference_by_name(const char *name)
  59 {
  60         struct object_id oid;
  61         struct commit *commit;
  62
  63         if (get_oid_committish(name, &oid))
  64                 return NULL;
  65         commit = lookup_commit_reference(&oid);
  66         if (parse_commit(commit))
  67                 return NULL;
  68         return commit;
  69 }
  70
  71 static timestamp_t parse_commit_date(const char *buf, const char *tail)
  72 {
  73         const char *dateptr;
  74
  75         if (buf + 6 >= tail)
  76                 return 0;
  77         if (memcmp(buf, "author", 6))
  78                 return 0;
  79         while (buf < tail && *buf++ != '\n')
  80                 /* nada */;
  81         if (buf + 9 >= tail)
  82                 return 0;
  83         if (memcmp(buf, "committer", 9))
  84                 return 0;
  85         while (buf < tail && *buf++ != '>')
  86                 /* nada */;
  87         if (buf >= tail)
  88                 return 0;
  89         dateptr = buf;
  90         while (buf < tail && *buf++ != '\n')
  91                 /* nada */;
  92         if (buf >= tail)
  93                 return 0;
  94         /* dateptr < buf && buf[-1] == '\n', so parsing will stop at buf-1 */
  95         return parse_timestamp(dateptr, NULL, 10);
  96 }
  97
  98 static struct commit_graft **commit_graft;
  99 static int commit_graft_alloc, commit_graft_nr;
 100
 101 static const unsigned char *commit_graft_sha1_access(size_t index, void *table)
 102 {
 103         struct commit_graft **commit_graft_table = table;
 104         return commit_graft_table[index]->oid.hash;
 105 }
 106
 107 static int commit_graft_pos(const unsigned char *sha1)
 108 {
 109         return sha1_pos(sha1, commit_graft, commit_graft_nr,
 110                         commit_graft_sha1_access);
 111 }
 112
 113 int register_commit_graft(struct commit_graft *graft, int ignore_dups)
 114 {
 115         int pos = commit_graft_pos(graft->oid.hash);
 116
 117         if (0 <= pos) {
 118                 if (ignore_dups)
 119                         free(graft);
 120                 else {
 121                         free(commit_graft[pos]);
 122                         commit_graft[pos] = graft;
 123                 }
 124                 return 1;
 125         }
 126         pos = -pos - 1;
 127         ALLOC_GROW(commit_graft, commit_graft_nr + 1, commit_graft_alloc);
 128         commit_graft_nr++;
 129         if (pos < commit_graft_nr)
 130                 MOVE_ARRAY(commit_graft + pos + 1, commit_graft + pos,
 131                            commit_graft_nr - pos - 1);
 132         commit_graft[pos] = graft;
 133         return 0;
 134 }
 135
 136 struct commit_graft *read_graft_line(struct strbuf *line)
 137 {
 138         /* The format is just "Commit Parent1 Parent2 ...\n" */
 139         int i, phase;
 140         const char *tail = NULL;
 141         struct commit_graft *graft = NULL;
 142         struct object_id dummy_oid, *oid;
 143
 144         strbuf_rtrim(line);
 145         if (!line->len || line->buf[0] == '#')
 146                 return NULL;
 147         /*
 148          * phase 0 verifies line, counts hashes in line and allocates graft
 149          * phase 1 fills graft
 150          */
 151         for (phase = 0; phase < 2; phase++) {
 152                 oid = graft ? &graft->oid : &dummy_oid;
 153                 if (parse_oid_hex(line->buf, oid, &tail))
 154                         goto bad_graft_data;
 155                 for (i = 0; *tail != '\0'; i++) {
 156                         oid = graft ? &graft->parent[i] : &dummy_oid;
 157                         if (!isspace(*tail++) || parse_oid_hex(tail, oid, &tail))
 158                                 goto bad_graft_data;
 159                 }
 160                 if (!graft) {
 161                         graft = xmalloc(st_add(sizeof(*graft),
 162                                                st_mult(sizeof(struct object_id), i)));
 163                         graft->nr_parent = i;
 164                 }
 165         }
 166         return graft;
 167
 168 bad_graft_data:
 169         error("bad graft data: %s", line->buf);
 170         assert(!graft);
 171         return NULL;
 172 }
 173
 174 static int read_graft_file(const char *graft_file)
 175 {
 176         FILE *fp = fopen_or_warn(graft_file, "r");
 177         struct strbuf buf = STRBUF_INIT;
 178         if (!fp)
 179                 return -1;
 180         while (!strbuf_getwholeline(&buf, fp, '\n')) {
 181                 /* The format is just "Commit Parent1 Parent2 ...\n" */
 182                 struct commit_graft *graft = read_graft_line(&buf);
 183                 if (!graft)
 184                         continue;
 185                 if (register_commit_graft(graft, 1))
 186                         error("duplicate graft data: %s", buf.buf);
 187         }
 188         fclose(fp);
 189         strbuf_release(&buf);
 190         return 0;
 191 }
 192
 193 static void prepare_commit_graft(void)
 194 {
 195         static int commit_graft_prepared;
 196         char *graft_file;
 197
 198         if (commit_graft_prepared)
 199                 return;
 200         graft_file = get_graft_file();
 201         read_graft_file(graft_file);
 202         /* make sure shallows are read */
 203         is_repository_shallow();
 204         commit_graft_prepared = 1;
 205 }
 206
 207 struct commit_graft *lookup_commit_graft(const struct object_id *oid)
 208 {
 209         int pos;
 210         prepare_commit_graft();
 211         pos = commit_graft_pos(oid->hash);
 212         if (pos < 0)
 213                 return NULL;
 214         return commit_graft[pos];
 215 }
 216
 217 int for_each_commit_graft(each_commit_graft_fn fn, void *cb_data)
 218 {
 219         int i, ret;
 220         for (i = ret = 0; i < commit_graft_nr && !ret; i++)
 221                 ret = fn(commit_graft[i], cb_data);
 222         return ret;
 223 }
 224
 225 int unregister_shallow(const struct object_id *oid)
 226 {
 227         int pos = commit_graft_pos(oid->hash);
 228         if (pos < 0)
 229                 return -1;
 230         if (pos + 1 < commit_graft_nr)
 231                 MOVE_ARRAY(commit_graft + pos, commit_graft + pos + 1,
 232                            commit_graft_nr - pos - 1);
 233         commit_graft_nr--;
 234         return 0;
 235 }
 236
 237 struct commit_buffer {
 238         void *buffer;
 239         unsigned long size;
 240 };
 241 define_commit_slab(buffer_slab, struct commit_buffer);
 242 static struct buffer_slab buffer_slab = COMMIT_SLAB_INIT(1, buffer_slab);
 243
 244 void set_commit_buffer(struct commit *commit, void *buffer, unsigned long size)
 245 {
 246         struct commit_buffer *v = buffer_slab_at(&buffer_slab, commit);
 247         v->buffer = buffer;
 248         v->size = size;
 249 }
 250
 251 const void *get_cached_commit_buffer(const struct commit *commit, unsigned long *sizep)
 252 {
 253         struct commit_buffer *v = buffer_slab_peek(&buffer_slab, commit);
 254         if (!v) {
 255                 if (sizep)
 256                         *sizep = 0;
 257                 return NULL;
 258         }
 259         if (sizep)
 260                 *sizep = v->size;
 261         return v->buffer;
 262 }
 263
 264 const void *get_commit_buffer(const struct commit *commit, unsigned long *sizep)
 265 {
 266         const void *ret = get_cached_commit_buffer(commit, sizep);
 267         if (!ret) {
 268                 enum object_type type;
 269                 unsigned long size;
 270                 ret = read_object_file(&commit->object.oid, &type, &size);
 271                 if (!ret)
 272                         die("cannot read commit object %s",
 273                             oid_to_hex(&commit->object.oid));
 274                 if (type != OBJ_COMMIT)
 275                         die("expected commit for %s, got %s",
 276                             oid_to_hex(&commit->object.oid), type_name(type));
 277                 if (sizep)
 278                         *sizep = size;
 279         }
 280         return ret;
 281 }
 282
 283 void unuse_commit_buffer(const struct commit *commit, const void *buffer)
 284 {
 285         struct commit_buffer *v = buffer_slab_peek(&buffer_slab, commit);
 286         if (!(v && v->buffer == buffer))
 287                 free((void *)buffer);
 288 }
 289
 290 void free_commit_buffer(struct commit *commit)
 291 {
 292         struct commit_buffer *v = buffer_slab_peek(&buffer_slab, commit);
 293         if (v) {
 294                 FREE_AND_NULL(v->buffer);
 295                 v->size = 0;
 296         }
 297 }
 298
 299 const void *detach_commit_buffer(struct commit *commit, unsigned long *sizep)
 300 {
 301         struct commit_buffer *v = buffer_slab_peek(&buffer_slab, commit);
 302         void *ret;
 303
 304         if (!v) {
 305                 if (sizep)
 306                         *sizep = 0;
 307                 return NULL;
 308         }
 309         ret = v->buffer;
 310         if (sizep)
 311                 *sizep = v->size;
 312
 313         v->buffer = NULL;
 314         v->size = 0;
 315         return ret;
 316 }
 317
 318 int parse_commit_buffer(struct commit *item, const void *buffer, unsigned long size)
 319 {
 320         const char *tail = buffer;
 321         const char *bufptr = buffer;
 322         struct object_id parent;
 323         struct commit_list **pptr;
 324         struct commit_graft *graft;
 325         const int tree_entry_len = GIT_SHA1_HEXSZ + 5;
 326         const int parent_entry_len = GIT_SHA1_HEXSZ + 7;
 327
 328         if (item->object.parsed)
 329                 return 0;
 330         item->object.parsed = 1;
 331         tail += size;
 332         if (tail <= bufptr + tree_entry_len + 1 || memcmp(bufptr, "tree ", 5) ||
 333                         bufptr[tree_entry_len] != '\n')
 334                 return error("bogus commit object %s", oid_to_hex(&item->object.oid));
 335         if (get_sha1_hex(bufptr + 5, parent.hash) < 0)
 336                 return error("bad tree pointer in commit %s",
 337                              oid_to_hex(&item->object.oid));
 338         item->tree = lookup_tree(&parent);
 339         bufptr += tree_entry_len + 1; /* "tree " + "hex sha1" + "\n" */
 340         pptr = &item->parents;
 341
 342         graft = lookup_commit_graft(&item->object.oid);
 343         while (bufptr + parent_entry_len < tail && !memcmp(bufptr, "parent ", 7)) {
 344                 struct commit *new_parent;
 345
 346                 if (tail <= bufptr + parent_entry_len + 1 ||
 347                     get_sha1_hex(bufptr + 7, parent.hash) ||
 348                     bufptr[parent_entry_len] != '\n')
 349                         return error("bad parents in commit %s", oid_to_hex(&item->object.oid));
 350                 bufptr += parent_entry_len + 1;
 351                 /*
 352                  * The clone is shallow if nr_parent < 0, and we must
 353                  * not traverse its real parents even when we unhide them.
 354                  */
 355                 if (graft && (graft->nr_parent < 0 || grafts_replace_parents))
 356                         continue;
 357                 new_parent = lookup_commit(&parent);
 358                 if (new_parent)
 359                         pptr = &commit_list_insert(new_parent, pptr)->next;
 360         }
 361         if (graft) {
 362                 int i;
 363                 struct commit *new_parent;
 364                 for (i = 0; i < graft->nr_parent; i++) {
 365                         new_parent = lookup_commit(&graft->parent[i]);
 366                         if (!new_parent)
 367                                 continue;
 368                         pptr = &commit_list_insert(new_parent, pptr)->next;
 369                 }
 370         }
 371         item->date = parse_commit_date(bufptr, tail);
 372
 373         return 0;
 374 }
 375
 376 int parse_commit_gently(struct commit *item, int quiet_on_missing)
 377 {
 378         enum object_type type;
 379         void *buffer;
 380         unsigned long size;
 381         int ret;
 382
 383         if (!item)
 384                 return -1;
 385         if (item->object.parsed)
 386                 return 0;
 387         if (parse_commit_in_graph(item))
 388                 return 0;
 389         buffer = read_object_file(&item->object.oid, &type, &size);
 390         if (!buffer)
 391                 return quiet_on_missing ? -1 :
 392                         error("Could not read %s",
 393                              oid_to_hex(&item->object.oid));
 394         if (type != OBJ_COMMIT) {
 395                 free(buffer);
 396                 return error("Object %s not a commit",
 397                              oid_to_hex(&item->object.oid));
 398         }
 399         ret = parse_commit_buffer(item, buffer, size);
 400         if (save_commit_buffer && !ret) {
 401                 set_commit_buffer(item, buffer, size);
 402                 return 0;
 403         }
 404         free(buffer);
 405         return ret;
 406 }
 407
 408 void parse_commit_or_die(struct commit *item)
 409 {
 410         if (parse_commit(item))
 411                 die("unable to parse commit %s",
 412                     item ? oid_to_hex(&item->object.oid) : "(null)");
 413 }
 414
 415 int find_commit_subject(const char *commit_buffer, const char **subject)
 416 {
 417         const char *eol;
 418         const char *p = commit_buffer;
 419
 420         while (*p && (*p != '\n' || p[1] != '\n'))
 421                 p++;
 422         if (*p) {
 423                 p = skip_blank_lines(p + 2);
 424                 eol = strchrnul(p, '\n');
 425         } else
 426                 eol = p;
 427
 428         *subject = p;
 429
 430         return eol - p;
 431 }
 432
 433 struct commit_list *commit_list_insert(struct commit *item, struct commit_list **list_p)
 434 {
 435         struct commit_list *new_list = xmalloc(sizeof(struct commit_list));
 436         new_list->item = item;
 437         new_list->next = *list_p;
 438         *list_p = new_list;
 439         return new_list;
 440 }
 441
 442 unsigned commit_list_count(const struct commit_list *l)
 443 {
 444         unsigned c = 0;
 445         for (; l; l = l->next )
 446                 c++;
 447         return c;
 448 }
 449
 450 struct commit_list *copy_commit_list(struct commit_list *list)
 451 {
 452         struct commit_list *head = NULL;
 453         struct commit_list **pp = &head;
 454         while (list) {
 455                 pp = commit_list_append(list->item, pp);
 456                 list = list->next;
 457         }
 458         return head;
 459 }
 460
 461 void free_commit_list(struct commit_list *list)
 462 {
 463         while (list)
 464                 pop_commit(&list);
 465 }
 466
 467 struct commit_list * commit_list_insert_by_date(struct commit *item, struct commit_list **list)
 468 {
 469         struct commit_list **pp = list;
 470         struct commit_list *p;
 471         while ((p = *pp) != NULL) {
 472                 if (p->item->date < item->date) {
 473                         break;
 474                 }
 475                 pp = &p->next;
 476         }
 477         return commit_list_insert(item, pp);
 478 }
 479
 480 static int commit_list_compare_by_date(const void *a, const void *b)
 481 {
 482         timestamp_t a_date = ((const struct commit_list *)a)->item->date;
 483         timestamp_t b_date = ((const struct commit_list *)b)->item->date;
 484         if (a_date < b_date)
 485                 return 1;
 486         if (a_date > b_date)
 487                 return -1;
 488         return 0;
 489 }
 490
 491 static void *commit_list_get_next(const void *a)
 492 {
 493         return ((const struct commit_list *)a)->next;
 494 }
 495
 496 static void commit_list_set_next(void *a, void *next)
 497 {
 498         ((struct commit_list *)a)->next = next;
 499 }
 500
 501 void commit_list_sort_by_date(struct commit_list **list)
 502 {
 503         *list = llist_mergesort(*list, commit_list_get_next, commit_list_set_next,
 504                                 commit_list_compare_by_date);
 505 }
 506
 507 struct commit *pop_most_recent_commit(struct commit_list **list,
 508                                       unsigned int mark)
 509 {
 510         struct commit *ret = pop_commit(list);
 511         struct commit_list *parents = ret->parents;
 512
 513         while (parents) {
 514                 struct commit *commit = parents->item;
 515                 if (!parse_commit(commit) && !(commit->object.flags & mark)) {
 516                         commit->object.flags |= mark;
 517                         commit_list_insert_by_date(commit, list);
 518                 }
 519                 parents = parents->next;
 520         }
 521         return ret;
 522 }
 523
 524 static void clear_commit_marks_1(struct commit_list **plist,
 525                                  struct commit *commit, unsigned int mark)
 526 {
 527         while (commit) {
 528                 struct commit_list *parents;
 529
 530                 if (!(mark & commit->object.flags))
 531                         return;
 532
 533                 commit->object.flags &= ~mark;
 534
 535                 parents = commit->parents;
 536                 if (!parents)
 537                         return;
 538
 539                 while ((parents = parents->next))
 540                         commit_list_insert(parents->item, plist);
 541
 542                 commit = commit->parents->item;
 543         }
 544 }
 545
 546 void clear_commit_marks_many(int nr, struct commit **commit, unsigned int mark)
 547 {
 548         struct commit_list *list = NULL;
 549
 550         while (nr--) {
 551                 clear_commit_marks_1(&list, *commit, mark);
 552                 commit++;
 553         }
 554         while (list)
 555                 clear_commit_marks_1(&list, pop_commit(&list), mark);
 556 }
 557
 558 void clear_commit_marks(struct commit *commit, unsigned int mark)
 559 {
 560         clear_commit_marks_many(1, &commit, mark);
 561 }
 562
 563 struct commit *pop_commit(struct commit_list **stack)
 564 {
 565         struct commit_list *top = *stack;
 566         struct commit *item = top ? top->item : NULL;
 567
 568         if (top) {
 569                 *stack = top->next;
 570                 free(top);
 571         }
 572         return item;
 573 }
 574
 575 /*
 576  * Topological sort support
 577  */
 578
 579 /* count number of children that have not been emitted */
 580 define_commit_slab(indegree_slab, int);
 581
 582 /* record author-date for each commit object */
 583 define_commit_slab(author_date_slab, unsigned long);
 584
 585 static void record_author_date(struct author_date_slab *author_date,
 586                                struct commit *commit)
 587 {
 588         const char *buffer = get_commit_buffer(commit, NULL);
 589         struct ident_split ident;
 590         const char *ident_line;
 591         size_t ident_len;
 592         char *date_end;
 593         timestamp_t date;
 594
 595         ident_line = find_commit_header(buffer, "author", &ident_len);
 596         if (!ident_line)
 597                 goto fail_exit; /* no author line */
 598         if (split_ident_line(&ident, ident_line, ident_len) ||
 599             !ident.date_begin || !ident.date_end)
 600                 goto fail_exit; /* malformed "author" line */
 601
 602         date = parse_timestamp(ident.date_begin, &date_end, 10);
 603         if (date_end != ident.date_end)
 604                 goto fail_exit; /* malformed date */
 605         *(author_date_slab_at(author_date, commit)) = date;
 606
 607 fail_exit:
 608         unuse_commit_buffer(commit, buffer);
 609 }
 610
 611 static int compare_commits_by_author_date(const void *a_, const void *b_,
 612                                           void *cb_data)
 613 {
 614         const struct commit *a = a_, *b = b_;
 615         struct author_date_slab *author_date = cb_data;
 616         timestamp_t a_date = *(author_date_slab_at(author_date, a));
 617         timestamp_t b_date = *(author_date_slab_at(author_date, b));
 618
 619         /* newer commits with larger date first */
 620         if (a_date < b_date)
 621                 return 1;
 622         else if (a_date > b_date)
 623                 return -1;
 624         return 0;
 625 }
 626
 627 int compare_commits_by_commit_date(const void *a_, const void *b_, void *unused)
 628 {
 629         const struct commit *a = a_, *b = b_;
 630         /* newer commits with larger date first */
 631         if (a->date < b->date)
 632                 return 1;
 633         else if (a->date > b->date)
 634                 return -1;
 635         return 0;
 636 }
 637
 638 /*
 639  * Performs an in-place topological sort on the list supplied.
 640  */
 641 void sort_in_topological_order(struct commit_list **list, enum rev_sort_order sort_order)
 642 {
 643         struct commit_list *next, *orig = *list;
 644         struct commit_list **pptr;
 645         struct indegree_slab indegree;
 646         struct prio_queue queue;
 647         struct commit *commit;
 648         struct author_date_slab author_date;
 649
 650         if (!orig)
 651                 return;
 652         *list = NULL;
 653
 654         init_indegree_slab(&indegree);
 655         memset(&queue, '\0', sizeof(queue));
 656
 657         switch (sort_order) {
 658         default: /* REV_SORT_IN_GRAPH_ORDER */
 659                 queue.compare = NULL;
 660                 break;
 661         case REV_SORT_BY_COMMIT_DATE:
 662                 queue.compare = compare_commits_by_commit_date;
 663                 break;
 664         case REV_SORT_BY_AUTHOR_DATE:
 665                 init_author_date_slab(&author_date);
 666                 queue.compare = compare_commits_by_author_date;
 667                 queue.cb_data = &author_date;
 668                 break;
 669         }
 670
 671         /* Mark them and clear the indegree */
 672         for (next = orig; next; next = next->next) {
 673                 struct commit *commit = next->item;
 674                 *(indegree_slab_at(&indegree, commit)) = 1;
 675                 /* also record the author dates, if needed */
 676                 if (sort_order == REV_SORT_BY_AUTHOR_DATE)
 677                         record_author_date(&author_date, commit);
 678         }
 679
 680         /* update the indegree */
 681         for (next = orig; next; next = next->next) {
 682                 struct commit_list *parents = next->item->parents;
 683                 while (parents) {
 684                         struct commit *parent = parents->item;
 685                         int *pi = indegree_slab_at(&indegree, parent);
 686
 687                         if (*pi)
 688                                 (*pi)++;
 689                         parents = parents->next;
 690                 }
 691         }
 692
 693         /*
 694          * find the tips
 695          *
 696          * tips are nodes not reachable from any other node in the list
 697          *
 698          * the tips serve as a starting set for the work queue.
 699          */
 700         for (next = orig; next; next = next->next) {
 701                 struct commit *commit = next->item;
 702
 703                 if (*(indegree_slab_at(&indegree, commit)) == 1)
 704                         prio_queue_put(&queue, commit);
 705         }
 706
 707         /*
 708          * This is unfortunate; the initial tips need to be shown
 709          * in the order given from the revision traversal machinery.
 710          */
 711         if (sort_order == REV_SORT_IN_GRAPH_ORDER)
 712                 prio_queue_reverse(&queue);
 713
 714         /* We no longer need the commit list */
 715         free_commit_list(orig);
 716
 717         pptr = list;
 718         *list = NULL;
 719         while ((commit = prio_queue_get(&queue)) != NULL) {
 720                 struct commit_list *parents;
 721
 722                 for (parents = commit->parents; parents ; parents = parents->next) {
 723                         struct commit *parent = parents->item;
 724                         int *pi = indegree_slab_at(&indegree, parent);
 725
 726                         if (!*pi)
 727                                 continue;
 728
 729                         /*
 730                          * parents are only enqueued for emission
 731                          * when all their children have been emitted thereby
 732                          * guaranteeing topological order.
 733                          */
 734                         if (--(*pi) == 1)
 735                                 prio_queue_put(&queue, parent);
 736                 }
 737                 /*
 738                  * all children of commit have already been
 739                  * emitted. we can emit it now.
 740                  */
 741                 *(indegree_slab_at(&indegree, commit)) = 0;
 742
 743                 pptr = &commit_list_insert(commit, pptr)->next;
 744         }
 745
 746         clear_indegree_slab(&indegree);
 747         clear_prio_queue(&queue);
 748         if (sort_order == REV_SORT_BY_AUTHOR_DATE)
 749                 clear_author_date_slab(&author_date);
 750 }
 751
 752 /* merge-base stuff */
 753
 754 /* Remember to update object flag allocation in object.h */
 755 #define PARENT1         (1u<<16)
 756 #define PARENT2         (1u<<17)
 757 #define STALE           (1u<<18)
 758 #define RESULT          (1u<<19)
 759
 760 static const unsigned all_flags = (PARENT1 | PARENT2 | STALE | RESULT);
 761
 762 static int queue_has_nonstale(struct prio_queue *queue)
 763 {
 764         int i;
 765         for (i = 0; i < queue->nr; i++) {
 766                 struct commit *commit = queue->array[i].data;
 767                 if (!(commit->object.flags & STALE))
 768                         return 1;
 769         }
 770         return 0;
 771 }
 772
 773 /* all input commits in one and twos[] must have been parsed! */
 774 static struct commit_list *paint_down_to_common(struct commit *one, int n, struct commit **twos)
 775 {
 776         struct prio_queue queue = { compare_commits_by_commit_date };
 777         struct commit_list *result = NULL;
 778         int i;
 779
 780         one->object.flags |= PARENT1;
 781         if (!n) {
 782                 commit_list_append(one, &result);
 783                 return result;
 784         }
 785         prio_queue_put(&queue, one);
 786
 787         for (i = 0; i < n; i++) {
 788                 twos[i]->object.flags |= PARENT2;
 789                 prio_queue_put(&queue, twos[i]);
 790         }
 791
 792         while (queue_has_nonstale(&queue)) {
 793                 struct commit *commit = prio_queue_get(&queue);
 794                 struct commit_list *parents;
 795                 int flags;
 796
 797                 flags = commit->object.flags & (PARENT1 | PARENT2 | STALE);
 798                 if (flags == (PARENT1 | PARENT2)) {
 799                         if (!(commit->object.flags & RESULT)) {
 800                                 commit->object.flags |= RESULT;
 801                                 commit_list_insert_by_date(commit, &result);
 802                         }
 803                         /* Mark parents of a found merge stale */
 804                         flags |= STALE;
 805                 }
 806                 parents = commit->parents;
 807                 while (parents) {
 808                         struct commit *p = parents->item;
 809                         parents = parents->next;
 810                         if ((p->object.flags & flags) == flags)
 811                                 continue;
 812                         if (parse_commit(p))
 813                                 return NULL;
 814                         p->object.flags |= flags;
 815                         prio_queue_put(&queue, p);
 816                 }
 817         }
 818
 819         clear_prio_queue(&queue);
 820         return result;
 821 }
 822
 823 static struct commit_list *merge_bases_many(struct commit *one, int n, struct commit **twos)
 824 {
 825         struct commit_list *list = NULL;
 826         struct commit_list *result = NULL;
 827         int i;
 828
 829         for (i = 0; i < n; i++) {
 830                 if (one == twos[i])
 831                         /*
 832                          * We do not mark this even with RESULT so we do not
 833                          * have to clean it up.
 834                          */
 835                         return commit_list_insert(one, &result);
 836         }
 837
 838         if (parse_commit(one))
 839                 return NULL;
 840         for (i = 0; i < n; i++) {
 841                 if (parse_commit(twos[i]))
 842                         return NULL;
 843         }
 844
 845         list = paint_down_to_common(one, n, twos);
 846
 847         while (list) {
 848                 struct commit *commit = pop_commit(&list);
 849                 if (!(commit->object.flags & STALE))
 850                         commit_list_insert_by_date(commit, &result);
 851         }
 852         return result;
 853 }
 854
 855 struct commit_list *get_octopus_merge_bases(struct commit_list *in)
 856 {
 857         struct commit_list *i, *j, *k, *ret = NULL;
 858
 859         if (!in)
 860                 return ret;
 861
 862         commit_list_insert(in->item, &ret);
 863
 864         for (i = in->next; i; i = i->next) {
 865                 struct commit_list *new_commits = NULL, *end = NULL;
 866
 867                 for (j = ret; j; j = j->next) {
 868                         struct commit_list *bases;
 869                         bases = get_merge_bases(i->item, j->item);
 870                         if (!new_commits)
 871                                 new_commits = bases;
 872                         else
 873                                 end->next = bases;
 874                         for (k = bases; k; k = k->next)
 875                                 end = k;
 876                 }
 877                 ret = new_commits;
 878         }
 879         return ret;
 880 }
 881
 882 static int remove_redundant(struct commit **array, int cnt)
 883 {
 884         /*
 885          * Some commit in the array may be an ancestor of
 886          * another commit.  Move such commit to the end of
 887          * the array, and return the number of commits that
 888          * are independent from each other.
 889          */
 890         struct commit **work;
 891         unsigned char *redundant;
 892         int *filled_index;
 893         int i, j, filled;
 894
 895         work = xcalloc(cnt, sizeof(*work));
 896         redundant = xcalloc(cnt, 1);
 897         ALLOC_ARRAY(filled_index, cnt - 1);
 898
 899         for (i = 0; i < cnt; i++)
 900                 parse_commit(array[i]);
 901         for (i = 0; i < cnt; i++) {
 902                 struct commit_list *common;
 903
 904                 if (redundant[i])
 905                         continue;
 906                 for (j = filled = 0; j < cnt; j++) {
 907                         if (i == j || redundant[j])
 908                                 continue;
 909                         filled_index[filled] = j;
 910                         work[filled++] = array[j];
 911                 }
 912                 common = paint_down_to_common(array[i], filled, work);
 913                 if (array[i]->object.flags & PARENT2)
 914                         redundant[i] = 1;
 915                 for (j = 0; j < filled; j++)
 916                         if (work[j]->object.flags & PARENT1)
 917                                 redundant[filled_index[j]] = 1;
 918                 clear_commit_marks(array[i], all_flags);
 919                 clear_commit_marks_many(filled, work, all_flags);
 920                 free_commit_list(common);
 921         }
 922
 923         /* Now collect the result */
 924         COPY_ARRAY(work, array, cnt);
 925         for (i = filled = 0; i < cnt; i++)
 926                 if (!redundant[i])
 927                         array[filled++] = work[i];
 928         for (j = filled, i = 0; i < cnt; i++)
 929                 if (redundant[i])
 930                         array[j++] = work[i];
 931         free(work);
 932         free(redundant);
 933         free(filled_index);
 934         return filled;
 935 }
 936
 937 static struct commit_list *get_merge_bases_many_0(struct commit *one,
 938                                                   int n,
 939                                                   struct commit **twos,
 940                                                   int cleanup)
 941 {
 942         struct commit_list *list;
 943         struct commit **rslt;
 944         struct commit_list *result;
 945         int cnt, i;
 946
 947         result = merge_bases_many(one, n, twos);
 948         for (i = 0; i < n; i++) {
 949                 if (one == twos[i])
 950                         return result;
 951         }
 952         if (!result || !result->next) {
 953                 if (cleanup) {
 954                         clear_commit_marks(one, all_flags);
 955                         clear_commit_marks_many(n, twos, all_flags);
 956                 }
 957                 return result;
 958         }
 959
 960         /* There are more than one */
 961         cnt = commit_list_count(result);
 962         rslt = xcalloc(cnt, sizeof(*rslt));
 963         for (list = result, i = 0; list; list = list->next)
 964                 rslt[i++] = list->item;
 965         free_commit_list(result);
 966
 967         clear_commit_marks(one, all_flags);
 968         clear_commit_marks_many(n, twos, all_flags);
 969
 970         cnt = remove_redundant(rslt, cnt);
 971         result = NULL;
 972         for (i = 0; i < cnt; i++)
 973                 commit_list_insert_by_date(rslt[i], &result);
 974         free(rslt);
 975         return result;
 976 }
 977
 978 struct commit_list *get_merge_bases_many(struct commit *one,
 979                                          int n,
 980                                          struct commit **twos)
 981 {
 982         return get_merge_bases_many_0(one, n, twos, 1);
 983 }
 984
 985 struct commit_list *get_merge_bases_many_dirty(struct commit *one,
 986                                                int n,
 987                                                struct commit **twos)
 988 {
 989         return get_merge_bases_many_0(one, n, twos, 0);
 990 }
 991
 992 struct commit_list *get_merge_bases(struct commit *one, struct commit *two)
 993 {
 994         return get_merge_bases_many_0(one, 1, &two, 1);
 995 }
 996
 997 /*
 998  * Is "commit" a descendant of one of the elements on the "with_commit" list?
 999  */
1000 int is_descendant_of(struct commit *commit, struct commit_list *with_commit)
1001 {
1002         if (!with_commit)
1003                 return 1;
1004         while (with_commit) {
1005                 struct commit *other;
1006
1007                 other = with_commit->item;
1008                 with_commit = with_commit->next;
1009                 if (in_merge_bases(other, commit))
1010                         return 1;
1011         }
1012         return 0;
1013 }
1014
1015 /*
1016  * Is "commit" an ancestor of one of the "references"?
1017  */
1018 int in_merge_bases_many(struct commit *commit, int nr_reference, struct commit **reference)
1019 {
1020         struct commit_list *bases;
1021         int ret = 0, i;
1022
1023         if (parse_commit(commit))
1024                 return ret;
1025         for (i = 0; i < nr_reference; i++)
1026                 if (parse_commit(reference[i]))
1027                         return ret;
1028
1029         bases = paint_down_to_common(commit, nr_reference, reference);
1030         if (commit->object.flags & PARENT2)
1031                 ret = 1;
1032         clear_commit_marks(commit, all_flags);
1033         clear_commit_marks_many(nr_reference, reference, all_flags);
1034         free_commit_list(bases);
1035         return ret;
1036 }
1037
1038 /*
1039  * Is "commit" an ancestor of (i.e. reachable from) the "reference"?
1040  */
1041 int in_merge_bases(struct commit *commit, struct commit *reference)
1042 {
1043         return in_merge_bases_many(commit, 1, &reference);
1044 }
1045
1046 struct commit_list *reduce_heads(struct commit_list *heads)
1047 {
1048         struct commit_list *p;
1049         struct commit_list *result = NULL, **tail = &result;
1050         struct commit **array;
1051         int num_head, i;
1052
1053         if (!heads)
1054                 return NULL;
1055
1056         /* Uniquify */
1057         for (p = heads; p; p = p->next)
1058                 p->item->object.flags &= ~STALE;
1059         for (p = heads, num_head = 0; p; p = p->next) {
1060                 if (p->item->object.flags & STALE)
1061                         continue;
1062                 p->item->object.flags |= STALE;
1063                 num_head++;
1064         }
1065         array = xcalloc(num_head, sizeof(*array));
1066         for (p = heads, i = 0; p; p = p->next) {
1067                 if (p->item->object.flags & STALE) {
1068                         array[i++] = p->item;
1069                         p->item->object.flags &= ~STALE;
1070                 }
1071         }
1072         num_head = remove_redundant(array, num_head);
1073         for (i = 0; i < num_head; i++)
1074                 tail = &commit_list_insert(array[i], tail)->next;
1075         free(array);
1076         return result;
1077 }
1078
1079 void reduce_heads_replace(struct commit_list **heads)
1080 {
1081         struct commit_list *result = reduce_heads(*heads);
1082         free_commit_list(*heads);
1083         *heads = result;
1084 }
1085
1086 static const char gpg_sig_header[] = "gpgsig";
1087 static const int gpg_sig_header_len = sizeof(gpg_sig_header) - 1;
1088
1089 static int do_sign_commit(struct strbuf *buf, const char *keyid)
1090 {
1091         struct strbuf sig = STRBUF_INIT;
1092         int inspos, copypos;
1093         const char *eoh;
1094
1095         /* find the end of the header */
1096         eoh = strstr(buf->buf, "\n\n");
1097         if (!eoh)
1098                 inspos = buf->len;
1099         else
1100                 inspos = eoh - buf->buf + 1;
1101
1102         if (!keyid || !*keyid)
1103                 keyid = get_signing_key();
1104         if (sign_buffer(buf, &sig, keyid)) {
1105                 strbuf_release(&sig);
1106                 return -1;
1107         }
1108
1109         for (copypos = 0; sig.buf[copypos]; ) {
1110                 const char *bol = sig.buf + copypos;
1111                 const char *eol = strchrnul(bol, '\n');
1112                 int len = (eol - bol) + !!*eol;
1113
1114                 if (!copypos) {
1115                         strbuf_insert(buf, inspos, gpg_sig_header, gpg_sig_header_len);
1116                         inspos += gpg_sig_header_len;
1117                 }
1118                 strbuf_insert(buf, inspos++, " ", 1);
1119                 strbuf_insert(buf, inspos, bol, len);
1120                 inspos += len;
1121                 copypos += len;
1122         }
1123         strbuf_release(&sig);
1124         return 0;
1125 }
1126
1127 int parse_signed_commit(const struct commit *commit,
1128                         struct strbuf *payload, struct strbuf *signature)
1129 {
1130
1131         unsigned long size;
1132         const char *buffer = get_commit_buffer(commit, &size);
1133         int in_signature, saw_signature = -1;
1134         const char *line, *tail;
1135
1136         line = buffer;
1137         tail = buffer + size;
1138         in_signature = 0;
1139         saw_signature = 0;
1140         while (line < tail) {
1141                 const char *sig = NULL;
1142                 const char *next = memchr(line, '\n', tail - line);
1143
1144                 next = next ? next + 1 : tail;
1145                 if (in_signature && line[0] == ' ')
1146                         sig = line + 1;
1147                 else if (starts_with(line, gpg_sig_header) &&
1148                          line[gpg_sig_header_len] == ' ')
1149                         sig = line + gpg_sig_header_len + 1;
1150                 if (sig) {
1151                         strbuf_add(signature, sig, next - sig);
1152                         saw_signature = 1;
1153                         in_signature = 1;
1154                 } else {
1155                         if (*line == '\n')
1156                                 /* dump the whole remainder of the buffer */
1157                                 next = tail;
1158                         strbuf_add(payload, line, next - line);
1159                         in_signature = 0;
1160                 }
1161                 line = next;
1162         }
1163         unuse_commit_buffer(commit, buffer);
1164         return saw_signature;
1165 }
1166
1167 int remove_signature(struct strbuf *buf)
1168 {
1169         const char *line = buf->buf;
1170         const char *tail = buf->buf + buf->len;
1171         int in_signature = 0;
1172         const char *sig_start = NULL;
1173         const char *sig_end = NULL;
1174
1175         while (line < tail) {
1176                 const char *next = memchr(line, '\n', tail - line);
1177                 next = next ? next + 1 : tail;
1178
1179                 if (in_signature && line[0] == ' ')
1180                         sig_end = next;
1181                 else if (starts_with(line, gpg_sig_header) &&
1182                          line[gpg_sig_header_len] == ' ') {
1183                         sig_start = line;
1184                         sig_end = next;
1185                         in_signature = 1;
1186                 } else {
1187                         if (*line == '\n')
1188                                 /* dump the whole remainder of the buffer */
1189                                 next = tail;
1190                         in_signature = 0;
1191                 }
1192                 line = next;
1193         }
1194
1195         if (sig_start)
1196                 strbuf_remove(buf, sig_start - buf->buf, sig_end - sig_start);
1197
1198         return sig_start != NULL;
1199 }
1200
1201 static void handle_signed_tag(struct commit *parent, struct commit_extra_header ***tail)
1202 {
1203         struct merge_remote_desc *desc;
1204         struct commit_extra_header *mergetag;
1205         char *buf;
1206         unsigned long size, len;
1207         enum object_type type;
1208
1209         desc = merge_remote_util(parent);
1210         if (!desc || !desc->obj)
1211                 return;
1212         buf = read_object_file(&desc->obj->oid, &type, &size);
1213         if (!buf || type != OBJ_TAG)
1214                 goto free_return;
1215         len = parse_signature(buf, size);
1216         if (size == len)
1217                 goto free_return;
1218         /*
1219          * We could verify this signature and either omit the tag when
1220          * it does not validate, but the integrator may not have the
1221          * public key of the signer of the tag he is merging, while a
1222          * later auditor may have it while auditing, so let's not run
1223          * verify-signed-buffer here for now...
1224          *
1225          * if (verify_signed_buffer(buf, len, buf + len, size - len, ...))
1226          *      warn("warning: signed tag unverified.");
1227          */
1228         mergetag = xcalloc(1, sizeof(*mergetag));
1229         mergetag->key = xstrdup("mergetag");
1230         mergetag->value = buf;
1231         mergetag->len = size;
1232
1233         **tail = mergetag;
1234         *tail = &mergetag->next;
1235         return;
1236
1237 free_return:
1238         free(buf);
1239 }
1240
1241 int check_commit_signature(const struct commit *commit, struct signature_check *sigc)
1242 {
1243         struct strbuf payload = STRBUF_INIT;
1244         struct strbuf signature = STRBUF_INIT;
1245         int ret = 1;
1246
1247         sigc->result = 'N';
1248
1249         if (parse_signed_commit(commit, &payload, &signature) <= 0)
1250                 goto out;
1251         ret = check_signature(payload.buf, payload.len, signature.buf,
1252                 signature.len, sigc);
1253
1254  out:
1255         strbuf_release(&payload);
1256         strbuf_release(&signature);
1257
1258         return ret;
1259 }
1260
1261
1262
1263 void append_merge_tag_headers(struct commit_list *parents,
1264                               struct commit_extra_header ***tail)
1265 {
1266         while (parents) {
1267                 struct commit *parent = parents->item;
1268                 handle_signed_tag(parent, tail);
1269                 parents = parents->next;
1270         }
1271 }
1272
1273 static void add_extra_header(struct strbuf *buffer,
1274                              struct commit_extra_header *extra)
1275 {
1276         strbuf_addstr(buffer, extra->key);
1277         if (extra->len)
1278                 strbuf_add_lines(buffer, " ", extra->value, extra->len);
1279         else
1280                 strbuf_addch(buffer, '\n');
1281 }
1282
1283 struct commit_extra_header *read_commit_extra_headers(struct commit *commit,
1284                                                       const char **exclude)
1285 {
1286         struct commit_extra_header *extra = NULL;
1287         unsigned long size;
1288         const char *buffer = get_commit_buffer(commit, &size);
1289         extra = read_commit_extra_header_lines(buffer, size, exclude);
1290         unuse_commit_buffer(commit, buffer);
1291         return extra;
1292 }
1293
1294 void for_each_mergetag(each_mergetag_fn fn, struct commit *commit, void *data)
1295 {
1296         struct commit_extra_header *extra, *to_free;
1297
1298         to_free = read_commit_extra_headers(commit, NULL);
1299         for (extra = to_free; extra; extra = extra->next) {
1300                 if (strcmp(extra->key, "mergetag"))
1301                         continue; /* not a merge tag */
1302                 fn(commit, extra, data);
1303         }
1304         free_commit_extra_headers(to_free);
1305 }
1306
1307 static inline int standard_header_field(const char *field, size_t len)
1308 {
1309         return ((len == 4 && !memcmp(field, "tree", 4)) ||
1310                 (len == 6 && !memcmp(field, "parent", 6)) ||
1311                 (len == 6 && !memcmp(field, "author", 6)) ||
1312                 (len == 9 && !memcmp(field, "committer", 9)) ||
1313                 (len == 8 && !memcmp(field, "encoding", 8)));
1314 }
1315
1316 static int excluded_header_field(const char *field, size_t len, const char **exclude)
1317 {
1318         if (!exclude)
1319                 return 0;
1320
1321         while (*exclude) {
1322                 size_t xlen = strlen(*exclude);
1323                 if (len == xlen && !memcmp(field, *exclude, xlen))
1324                         return 1;
1325                 exclude++;
1326         }
1327         return 0;
1328 }
1329
1330 static struct commit_extra_header *read_commit_extra_header_lines(
1331         const char *buffer, size_t size,
1332         const char **exclude)
1333 {
1334         struct commit_extra_header *extra = NULL, **tail = &extra, *it = NULL;
1335         const char *line, *next, *eof, *eob;
1336         struct strbuf buf = STRBUF_INIT;
1337
1338         for (line = buffer, eob = line + size;
1339              line < eob && *line != '\n';
1340              line = next) {
1341                 next = memchr(line, '\n', eob - line);
1342                 next = next ? next + 1 : eob;
1343                 if (*line == ' ') {
1344                         /* continuation */
1345                         if (it)
1346                                 strbuf_add(&buf, line + 1, next - (line + 1));
1347                         continue;
1348                 }
1349                 if (it)
1350                         it->value = strbuf_detach(&buf, &it->len);
1351                 strbuf_reset(&buf);
1352                 it = NULL;
1353
1354                 eof = memchr(line, ' ', next - line);
1355                 if (!eof)
1356                         eof = next;
1357                 else if (standard_header_field(line, eof - line) ||
1358                          excluded_header_field(line, eof - line, exclude))
1359                         continue;
1360
1361                 it = xcalloc(1, sizeof(*it));
1362                 it->key = xmemdupz(line, eof-line);
1363                 *tail = it;
1364                 tail = &it->next;
1365                 if (eof + 1 < next)
1366                         strbuf_add(&buf, eof + 1, next - (eof + 1));
1367         }
1368         if (it)
1369                 it->value = strbuf_detach(&buf, &it->len);
1370         return extra;
1371 }
1372
1373 void free_commit_extra_headers(struct commit_extra_header *extra)
1374 {
1375         while (extra) {
1376                 struct commit_extra_header *next = extra->next;
1377                 free(extra->key);
1378                 free(extra->value);
1379                 free(extra);
1380                 extra = next;
1381         }
1382 }
1383
1384 int commit_tree(const char *msg, size_t msg_len, const struct object_id *tree,
1385                 struct commit_list *parents, struct object_id *ret,
1386                 const char *author, const char *sign_commit)
1387 {
1388         struct commit_extra_header *extra = NULL, **tail = &extra;
1389         int result;
1390
1391         append_merge_tag_headers(parents, &tail);
1392         result = commit_tree_extended(msg, msg_len, tree, parents, ret,
1393                                       author, sign_commit, extra);
1394         free_commit_extra_headers(extra);
1395         return result;
1396 }
1397
1398 static int find_invalid_utf8(const char *buf, int len)
1399 {
1400         int offset = 0;
1401         static const unsigned int max_codepoint[] = {
1402                 0x7f, 0x7ff, 0xffff, 0x10ffff
1403         };
1404
1405         while (len) {
1406                 unsigned char c = *buf++;
1407                 int bytes, bad_offset;
1408                 unsigned int codepoint;
1409                 unsigned int min_val, max_val;
1410
1411                 len--;
1412                 offset++;
1413
1414                 /* Simple US-ASCII? No worries. */
1415                 if (c < 0x80)
1416                         continue;
1417
1418                 bad_offset = offset-1;
1419
1420                 /*
1421                  * Count how many more high bits set: that's how
1422                  * many more bytes this sequence should have.
1423                  */
1424                 bytes = 0;
1425                 while (c & 0x40) {
1426                         c <<= 1;
1427                         bytes++;
1428                 }
1429
1430                 /*
1431                  * Must be between 1 and 3 more bytes.  Longer sequences result in
1432                  * codepoints beyond U+10FFFF, which are guaranteed never to exist.
1433                  */
1434                 if (bytes < 1 || 3 < bytes)
1435                         return bad_offset;
1436
1437                 /* Do we *have* that many bytes? */
1438                 if (len < bytes)
1439                         return bad_offset;
1440
1441                 /*
1442                  * Place the encoded bits at the bottom of the value and compute the
1443                  * valid range.
1444                  */
1445                 codepoint = (c & 0x7f) >> bytes;
1446                 min_val = max_codepoint[bytes-1] + 1;
1447                 max_val = max_codepoint[bytes];
1448
1449                 offset += bytes;
1450                 len -= bytes;
1451
1452                 /* And verify that they are good continuation bytes */
1453                 do {
1454                         codepoint <<= 6;
1455                         codepoint |= *buf & 0x3f;
1456                         if ((*buf++ & 0xc0) != 0x80)
1457                                 return bad_offset;
1458                 } while (--bytes);
1459
1460                 /* Reject codepoints that are out of range for the sequence length. */
1461                 if (codepoint < min_val || codepoint > max_val)
1462                         return bad_offset;
1463                 /* Surrogates are only for UTF-16 and cannot be encoded in UTF-8. */
1464                 if ((codepoint & 0x1ff800) == 0xd800)
1465                         return bad_offset;
1466                 /* U+xxFFFE and U+xxFFFF are guaranteed non-characters. */
1467                 if ((codepoint & 0xfffe) == 0xfffe)
1468                         return bad_offset;
1469                 /* So are anything in the range U+FDD0..U+FDEF. */
1470                 if (codepoint >= 0xfdd0 && codepoint <= 0xfdef)
1471                         return bad_offset;
1472         }
1473         return -1;
1474 }
1475
1476 /*
1477  * This verifies that the buffer is in proper utf8 format.
1478  *
1479  * If it isn't, it assumes any non-utf8 characters are Latin1,
1480  * and does the conversion.
1481  */
1482 static int verify_utf8(struct strbuf *buf)
1483 {
1484         int ok = 1;
1485         long pos = 0;
1486
1487         for (;;) {
1488                 int bad;
1489                 unsigned char c;
1490                 unsigned char replace[2];
1491
1492                 bad = find_invalid_utf8(buf->buf + pos, buf->len - pos);
1493                 if (bad < 0)
1494                         return ok;
1495                 pos += bad;
1496                 ok = 0;
1497                 c = buf->buf[pos];
1498                 strbuf_remove(buf, pos, 1);
1499
1500                 /* We know 'c' must be in the range 128-255 */
1501                 replace[0] = 0xc0 + (c >> 6);
1502                 replace[1] = 0x80 + (c & 0x3f);
1503                 strbuf_insert(buf, pos, replace, 2);
1504                 pos += 2;
1505         }
1506 }
1507
1508 static const char commit_utf8_warn[] =
1509 N_("Warning: commit message did not conform to UTF-8.\n"
1510    "You may want to amend it after fixing the message, or set the config\n"
1511    "variable i18n.commitencoding to the encoding your project uses.\n");
1512
1513 int commit_tree_extended(const char *msg, size_t msg_len,
1514                          const struct object_id *tree,
1515                          struct commit_list *parents, struct object_id *ret,
1516                          const char *author, const char *sign_commit,
1517                          struct commit_extra_header *extra)
1518 {
1519         int result;
1520         int encoding_is_utf8;
1521         struct strbuf buffer;
1522
1523         assert_oid_type(tree, OBJ_TREE);
1524
1525         if (memchr(msg, '\0', msg_len))
1526                 return error("a NUL byte in commit log message not allowed.");
1527
1528         /* Not having i18n.commitencoding is the same as having utf-8 */
1529         encoding_is_utf8 = is_encoding_utf8(git_commit_encoding);
1530
1531         strbuf_init(&buffer, 8192); /* should avoid reallocs for the headers */
1532         strbuf_addf(&buffer, "tree %s\n", oid_to_hex(tree));
1533
1534         /*
1535          * NOTE! This ordering means that the same exact tree merged with a
1536          * different order of parents will be a _different_ changeset even
1537          * if everything else stays the same.
1538          */
1539         while (parents) {
1540                 struct commit *parent = pop_commit(&parents);
1541                 strbuf_addf(&buffer, "parent %s\n",
1542                             oid_to_hex(&parent->object.oid));
1543         }
1544
1545         /* Person/date information */
1546         if (!author)
1547                 author = git_author_info(IDENT_STRICT);
1548         strbuf_addf(&buffer, "author %s\n", author);
1549         strbuf_addf(&buffer, "committer %s\n", git_committer_info(IDENT_STRICT));
1550         if (!encoding_is_utf8)
1551                 strbuf_addf(&buffer, "encoding %s\n", git_commit_encoding);
1552
1553         while (extra) {
1554                 add_extra_header(&buffer, extra);
1555                 extra = extra->next;
1556         }
1557         strbuf_addch(&buffer, '\n');
1558
1559         /* And add the comment */
1560         strbuf_add(&buffer, msg, msg_len);
1561
1562         /* And check the encoding */
1563         if (encoding_is_utf8 && !verify_utf8(&buffer))
1564                 fprintf(stderr, _(commit_utf8_warn));
1565
1566         if (sign_commit && do_sign_commit(&buffer, sign_commit)) {
1567                 result = -1;
1568                 goto out;
1569         }
1570
1571         result = write_object_file(buffer.buf, buffer.len, commit_type, ret);
1572 out:
1573         strbuf_release(&buffer);
1574         return result;
1575 }
1576
1577 void set_merge_remote_desc(struct commit *commit,
1578                            const char *name, struct object *obj)
1579 {
1580         struct merge_remote_desc *desc;
1581         FLEX_ALLOC_STR(desc, name, name);
1582         desc->obj = obj;
1583         commit->util = desc;
1584 }
1585
1586 struct commit *get_merge_parent(const char *name)
1587 {
1588         struct object *obj;
1589         struct commit *commit;
1590         struct object_id oid;
1591         if (get_oid(name, &oid))
1592                 return NULL;
1593         obj = parse_object(&oid);
1594         commit = (struct commit *)peel_to_type(name, 0, obj, OBJ_COMMIT);
1595         if (commit && !commit->util)
1596                 set_merge_remote_desc(commit, name, obj);
1597         return commit;
1598 }
1599
1600 /*
1601  * Append a commit to the end of the commit_list.
1602  *
1603  * next starts by pointing to the variable that holds the head of an
1604  * empty commit_list, and is updated to point to the "next" field of
1605  * the last item on the list as new commits are appended.
1606  *
1607  * Usage example:
1608  *
1609  *     struct commit_list *list;
1610  *     struct commit_list **next = &list;
1611  *
1612  *     next = commit_list_append(c1, next);
1613  *     next = commit_list_append(c2, next);
1614  *     assert(commit_list_count(list) == 2);
1615  *     return list;
1616  */
1617 struct commit_list **commit_list_append(struct commit *commit,
1618                                         struct commit_list **next)
1619 {
1620         struct commit_list *new_commit = xmalloc(sizeof(struct commit_list));
1621         new_commit->item = commit;
1622         *next = new_commit;
1623         new_commit->next = NULL;
1624         return &new_commit->next;
1625 }
1626
1627 const char *find_commit_header(const char *msg, const char *key, size_t *out_len)
1628 {
1629         int key_len = strlen(key);
1630         const char *line = msg;
1631
1632         while (line) {
1633                 const char *eol = strchrnul(line, '\n');
1634
1635                 if (line == eol)
1636                         return NULL;
1637
1638                 if (eol - line > key_len &&
1639                     !strncmp(line, key, key_len) &&
1640                     line[key_len] == ' ') {
1641                         *out_len = eol - line - key_len - 1;
1642                         return line + key_len + 1;
1643                 }
1644                 line = *eol ? eol + 1 : NULL;
1645         }
1646         return NULL;
1647 }
1648
1649 /*
1650  * Inspect the given string and determine the true "end" of the log message, in
1651  * order to find where to put a new Signed-off-by: line.  Ignored are
1652  * trailing comment lines and blank lines.  To support "git commit -s
1653  * --amend" on an existing commit, we also ignore "Conflicts:".  To
1654  * support "git commit -v", we truncate at cut lines.
1655  *
1656  * Returns the number of bytes from the tail to ignore, to be fed as
1657  * the second parameter to append_signoff().
1658  */
1659 int ignore_non_trailer(const char *buf, size_t len)
1660 {
1661         int boc = 0;
1662         int bol = 0;
1663         int in_old_conflicts_block = 0;
1664         size_t cutoff = wt_status_locate_end(buf, len);
1665
1666         while (bol < cutoff) {
1667                 const char *next_line = memchr(buf + bol, '\n', len - bol);
1668
1669                 if (!next_line)
1670                         next_line = buf + len;
1671                 else
1672                         next_line++;
1673
1674                 if (buf[bol] == comment_line_char || buf[bol] == '\n') {
1675                         /* is this the first of the run of comments? */
1676                         if (!boc)
1677                                 boc = bol;
1678                         /* otherwise, it is just continuing */
1679                 } else if (starts_with(buf + bol, "Conflicts:\n")) {
1680                         in_old_conflicts_block = 1;
1681                         if (!boc)
1682                                 boc = bol;
1683                 } else if (in_old_conflicts_block && buf[bol] == '\t') {
1684                         ; /* a pathname in the conflicts block */
1685                 } else if (boc) {
1686                         /* the previous was not trailing comment */
1687                         boc = 0;
1688                         in_old_conflicts_block = 0;
1689                 }
1690                 bol = next_line - buf;
1691         }
1692         return boc ? len - boc : len - cutoff;
1693 }