commit.c

   1 #include "cache.h"
   2 #include "tag.h"
   3 #include "commit.h"
   4 #include "pkt-line.h"
   5 #include "utf8.h"
   6 #include "diff.h"
   7 #include "revision.h"
   8 #include "notes.h"
   9 #include "gpg-interface.h"
  10 #include "mergesort.h"
  11 #include "commit-slab.h"
  12 #include "prio-queue.h"
  13 #include "sha1-lookup.h"
  14 #include "wt-status.h"
  15 #include "advice.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         if (advice_graft_file_deprecated)
 181                 advise(_("Support for <GIT_DIR>/info/grafts is deprecated\n"
 182                          "and will be removed in a future Git version.\n"
 183                          "\n"
 184                          "Please use \"git replace --convert-graft-file\"\n"
 185                          "to convert the grafts into replace refs.\n"
 186                          "\n"
 187                          "Turn this message off by running\n"
 188                          "\"git config advice.graftFileDeprecated false\""));
 189         while (!strbuf_getwholeline(&buf, fp, '\n')) {
 190                 /* The format is just "Commit Parent1 Parent2 ...\n" */
 191                 struct commit_graft *graft = read_graft_line(&buf);
 192                 if (!graft)
 193                         continue;
 194                 if (register_commit_graft(graft, 1))
 195                         error("duplicate graft data: %s", buf.buf);
 196         }
 197         fclose(fp);
 198         strbuf_release(&buf);
 199         return 0;
 200 }
 201
 202 static void prepare_commit_graft(void)
 203 {
 204         static int commit_graft_prepared;
 205         char *graft_file;
 206
 207         if (commit_graft_prepared)
 208                 return;
 209         graft_file = get_graft_file();
 210         read_graft_file(graft_file);
 211         /* make sure shallows are read */
 212         is_repository_shallow();
 213         commit_graft_prepared = 1;
 214 }
 215
 216 struct commit_graft *lookup_commit_graft(const struct object_id *oid)
 217 {
 218         int pos;
 219         prepare_commit_graft();
 220         pos = commit_graft_pos(oid->hash);
 221         if (pos < 0)
 222                 return NULL;
 223         return commit_graft[pos];
 224 }
 225
 226 int for_each_commit_graft(each_commit_graft_fn fn, void *cb_data)
 227 {
 228         int i, ret;
 229         for (i = ret = 0; i < commit_graft_nr && !ret; i++)
 230                 ret = fn(commit_graft[i], cb_data);
 231         return ret;
 232 }
 233
 234 int unregister_shallow(const struct object_id *oid)
 235 {
 236         int pos = commit_graft_pos(oid->hash);
 237         if (pos < 0)
 238                 return -1;
 239         if (pos + 1 < commit_graft_nr)
 240                 MOVE_ARRAY(commit_graft + pos, commit_graft + pos + 1,
 241                            commit_graft_nr - pos - 1);
 242         commit_graft_nr--;
 243         return 0;
 244 }
 245
 246 struct commit_buffer {
 247         void *buffer;
 248         unsigned long size;
 249 };
 250 define_commit_slab(buffer_slab, struct commit_buffer);
 251 static struct buffer_slab buffer_slab = COMMIT_SLAB_INIT(1, buffer_slab);
 252
 253 void set_commit_buffer(struct commit *commit, void *buffer, unsigned long size)
 254 {
 255         struct commit_buffer *v = buffer_slab_at(&buffer_slab, commit);
 256         v->buffer = buffer;
 257         v->size = size;
 258 }
 259
 260 const void *get_cached_commit_buffer(const struct commit *commit, unsigned long *sizep)
 261 {
 262         struct commit_buffer *v = buffer_slab_peek(&buffer_slab, commit);
 263         if (!v) {
 264                 if (sizep)
 265                         *sizep = 0;
 266                 return NULL;
 267         }
 268         if (sizep)
 269                 *sizep = v->size;
 270         return v->buffer;
 271 }
 272
 273 const void *get_commit_buffer(const struct commit *commit, unsigned long *sizep)
 274 {
 275         const void *ret = get_cached_commit_buffer(commit, sizep);
 276         if (!ret) {
 277                 enum object_type type;
 278                 unsigned long size;
 279                 ret = read_object_file(&commit->object.oid, &type, &size);
 280                 if (!ret)
 281                         die("cannot read commit object %s",
 282                             oid_to_hex(&commit->object.oid));
 283                 if (type != OBJ_COMMIT)
 284                         die("expected commit for %s, got %s",
 285                             oid_to_hex(&commit->object.oid), type_name(type));
 286                 if (sizep)
 287                         *sizep = size;
 288         }
 289         return ret;
 290 }
 291
 292 void unuse_commit_buffer(const struct commit *commit, const void *buffer)
 293 {
 294         struct commit_buffer *v = buffer_slab_peek(&buffer_slab, commit);
 295         if (!(v && v->buffer == buffer))
 296                 free((void *)buffer);
 297 }
 298
 299 void free_commit_buffer(struct commit *commit)
 300 {
 301         struct commit_buffer *v = buffer_slab_peek(&buffer_slab, commit);
 302         if (v) {
 303                 FREE_AND_NULL(v->buffer);
 304                 v->size = 0;
 305         }
 306 }
 307
 308 const void *detach_commit_buffer(struct commit *commit, unsigned long *sizep)
 309 {
 310         struct commit_buffer *v = buffer_slab_peek(&buffer_slab, commit);
 311         void *ret;
 312
 313         if (!v) {
 314                 if (sizep)
 315                         *sizep = 0;
 316                 return NULL;
 317         }
 318         ret = v->buffer;
 319         if (sizep)
 320                 *sizep = v->size;
 321
 322         v->buffer = NULL;
 323         v->size = 0;
 324         return ret;
 325 }
 326
 327 int parse_commit_buffer(struct commit *item, const void *buffer, unsigned long size)
 328 {
 329         const char *tail = buffer;
 330         const char *bufptr = buffer;
 331         struct object_id parent;
 332         struct commit_list **pptr;
 333         struct commit_graft *graft;
 334         const int tree_entry_len = GIT_SHA1_HEXSZ + 5;
 335         const int parent_entry_len = GIT_SHA1_HEXSZ + 7;
 336
 337         if (item->object.parsed)
 338                 return 0;
 339         item->object.parsed = 1;
 340         tail += size;
 341         if (tail <= bufptr + tree_entry_len + 1 || memcmp(bufptr, "tree ", 5) ||
 342                         bufptr[tree_entry_len] != '\n')
 343                 return error("bogus commit object %s", oid_to_hex(&item->object.oid));
 344         if (get_sha1_hex(bufptr + 5, parent.hash) < 0)
 345                 return error("bad tree pointer in commit %s",
 346                              oid_to_hex(&item->object.oid));
 347         item->tree = lookup_tree(&parent);
 348         bufptr += tree_entry_len + 1; /* "tree " + "hex sha1" + "\n" */
 349         pptr = &item->parents;
 350
 351         graft = lookup_commit_graft(&item->object.oid);
 352         while (bufptr + parent_entry_len < tail && !memcmp(bufptr, "parent ", 7)) {
 353                 struct commit *new_parent;
 354
 355                 if (tail <= bufptr + parent_entry_len + 1 ||
 356                     get_sha1_hex(bufptr + 7, parent.hash) ||
 357                     bufptr[parent_entry_len] != '\n')
 358                         return error("bad parents in commit %s", oid_to_hex(&item->object.oid));
 359                 bufptr += parent_entry_len + 1;
 360                 /*
 361                  * The clone is shallow if nr_parent < 0, and we must
 362                  * not traverse its real parents even when we unhide them.
 363                  */
 364                 if (graft && (graft->nr_parent < 0 || grafts_replace_parents))
 365                         continue;
 366                 new_parent = lookup_commit(&parent);
 367                 if (new_parent)
 368                         pptr = &commit_list_insert(new_parent, pptr)->next;
 369         }
 370         if (graft) {
 371                 int i;
 372                 struct commit *new_parent;
 373                 for (i = 0; i < graft->nr_parent; i++) {
 374                         new_parent = lookup_commit(&graft->parent[i]);
 375                         if (!new_parent)
 376                                 continue;
 377                         pptr = &commit_list_insert(new_parent, pptr)->next;
 378                 }
 379         }
 380         item->date = parse_commit_date(bufptr, tail);
 381
 382         return 0;
 383 }
 384
 385 int parse_commit_gently(struct commit *item, int quiet_on_missing)
 386 {
 387         enum object_type type;
 388         void *buffer;
 389         unsigned long size;
 390         int ret;
 391
 392         if (!item)
 393                 return -1;
 394         if (item->object.parsed)
 395                 return 0;
 396         buffer = read_object_file(&item->object.oid, &type, &size);
 397         if (!buffer)
 398                 return quiet_on_missing ? -1 :
 399                         error("Could not read %s",
 400                              oid_to_hex(&item->object.oid));
 401         if (type != OBJ_COMMIT) {
 402                 free(buffer);
 403                 return error("Object %s not a commit",
 404                              oid_to_hex(&item->object.oid));
 405         }
 406         ret = parse_commit_buffer(item, buffer, size);
 407         if (save_commit_buffer && !ret) {
 408                 set_commit_buffer(item, buffer, size);
 409                 return 0;
 410         }
 411         free(buffer);
 412         return ret;
 413 }
 414
 415 void parse_commit_or_die(struct commit *item)
 416 {
 417         if (parse_commit(item))
 418                 die("unable to parse commit %s",
 419                     item ? oid_to_hex(&item->object.oid) : "(null)");
 420 }
 421
 422 int find_commit_subject(const char *commit_buffer, const char **subject)
 423 {
 424         const char *eol;
 425         const char *p = commit_buffer;
 426
 427         while (*p && (*p != '\n' || p[1] != '\n'))
 428                 p++;
 429         if (*p) {
 430                 p = skip_blank_lines(p + 2);
 431                 eol = strchrnul(p, '\n');
 432         } else
 433                 eol = p;
 434
 435         *subject = p;
 436
 437         return eol - p;
 438 }
 439
 440 struct commit_list *commit_list_insert(struct commit *item, struct commit_list **list_p)
 441 {
 442         struct commit_list *new_list = xmalloc(sizeof(struct commit_list));
 443         new_list->item = item;
 444         new_list->next = *list_p;
 445         *list_p = new_list;
 446         return new_list;
 447 }
 448
 449 unsigned commit_list_count(const struct commit_list *l)
 450 {
 451         unsigned c = 0;
 452         for (; l; l = l->next )
 453                 c++;
 454         return c;
 455 }
 456
 457 struct commit_list *copy_commit_list(struct commit_list *list)
 458 {
 459         struct commit_list *head = NULL;
 460         struct commit_list **pp = &head;
 461         while (list) {
 462                 pp = commit_list_append(list->item, pp);
 463                 list = list->next;
 464         }
 465         return head;
 466 }
 467
 468 void free_commit_list(struct commit_list *list)
 469 {
 470         while (list)
 471                 pop_commit(&list);
 472 }
 473
 474 struct commit_list * commit_list_insert_by_date(struct commit *item, struct commit_list **list)
 475 {
 476         struct commit_list **pp = list;
 477         struct commit_list *p;
 478         while ((p = *pp) != NULL) {
 479                 if (p->item->date < item->date) {
 480                         break;
 481                 }
 482                 pp = &p->next;
 483         }
 484         return commit_list_insert(item, pp);
 485 }
 486
 487 static int commit_list_compare_by_date(const void *a, const void *b)
 488 {
 489         timestamp_t a_date = ((const struct commit_list *)a)->item->date;
 490         timestamp_t b_date = ((const struct commit_list *)b)->item->date;
 491         if (a_date < b_date)
 492                 return 1;
 493         if (a_date > b_date)
 494                 return -1;
 495         return 0;
 496 }
 497
 498 static void *commit_list_get_next(const void *a)
 499 {
 500         return ((const struct commit_list *)a)->next;
 501 }
 502
 503 static void commit_list_set_next(void *a, void *next)
 504 {
 505         ((struct commit_list *)a)->next = next;
 506 }
 507
 508 void commit_list_sort_by_date(struct commit_list **list)
 509 {
 510         *list = llist_mergesort(*list, commit_list_get_next, commit_list_set_next,
 511                                 commit_list_compare_by_date);
 512 }
 513
 514 struct commit *pop_most_recent_commit(struct commit_list **list,
 515                                       unsigned int mark)
 516 {
 517         struct commit *ret = pop_commit(list);
 518         struct commit_list *parents = ret->parents;
 519
 520         while (parents) {
 521                 struct commit *commit = parents->item;
 522                 if (!parse_commit(commit) && !(commit->object.flags & mark)) {
 523                         commit->object.flags |= mark;
 524                         commit_list_insert_by_date(commit, list);
 525                 }
 526                 parents = parents->next;
 527         }
 528         return ret;
 529 }
 530
 531 static void clear_commit_marks_1(struct commit_list **plist,
 532                                  struct commit *commit, unsigned int mark)
 533 {
 534         while (commit) {
 535                 struct commit_list *parents;
 536
 537                 if (!(mark & commit->object.flags))
 538                         return;
 539
 540                 commit->object.flags &= ~mark;
 541
 542                 parents = commit->parents;
 543                 if (!parents)
 544                         return;
 545
 546                 while ((parents = parents->next))
 547                         commit_list_insert(parents->item, plist);
 548
 549                 commit = commit->parents->item;
 550         }
 551 }
 552
 553 void clear_commit_marks_many(int nr, struct commit **commit, unsigned int mark)
 554 {
 555         struct commit_list *list = NULL;
 556
 557         while (nr--) {
 558                 clear_commit_marks_1(&list, *commit, mark);
 559                 commit++;
 560         }
 561         while (list)
 562                 clear_commit_marks_1(&list, pop_commit(&list), mark);
 563 }
 564
 565 void clear_commit_marks(struct commit *commit, unsigned int mark)
 566 {
 567         clear_commit_marks_many(1, &commit, mark);
 568 }
 569
 570 struct commit *pop_commit(struct commit_list **stack)
 571 {
 572         struct commit_list *top = *stack;
 573         struct commit *item = top ? top->item : NULL;
 574
 575         if (top) {
 576                 *stack = top->next;
 577                 free(top);
 578         }
 579         return item;
 580 }
 581
 582 /*
 583  * Topological sort support
 584  */
 585
 586 /* count number of children that have not been emitted */
 587 define_commit_slab(indegree_slab, int);
 588
 589 /* record author-date for each commit object */
 590 define_commit_slab(author_date_slab, unsigned long);
 591
 592 static void record_author_date(struct author_date_slab *author_date,
 593                                struct commit *commit)
 594 {
 595         const char *buffer = get_commit_buffer(commit, NULL);
 596         struct ident_split ident;
 597         const char *ident_line;
 598         size_t ident_len;
 599         char *date_end;
 600         timestamp_t date;
 601
 602         ident_line = find_commit_header(buffer, "author", &ident_len);
 603         if (!ident_line)
 604                 goto fail_exit; /* no author line */
 605         if (split_ident_line(&ident, ident_line, ident_len) ||
 606             !ident.date_begin || !ident.date_end)
 607                 goto fail_exit; /* malformed "author" line */
 608
 609         date = parse_timestamp(ident.date_begin, &date_end, 10);
 610         if (date_end != ident.date_end)
 611                 goto fail_exit; /* malformed date */
 612         *(author_date_slab_at(author_date, commit)) = date;
 613
 614 fail_exit:
 615         unuse_commit_buffer(commit, buffer);
 616 }
 617
 618 static int compare_commits_by_author_date(const void *a_, const void *b_,
 619                                           void *cb_data)
 620 {
 621         const struct commit *a = a_, *b = b_;
 622         struct author_date_slab *author_date = cb_data;
 623         timestamp_t a_date = *(author_date_slab_at(author_date, a));
 624         timestamp_t b_date = *(author_date_slab_at(author_date, b));
 625
 626         /* newer commits with larger date first */
 627         if (a_date < b_date)
 628                 return 1;
 629         else if (a_date > b_date)
 630                 return -1;
 631         return 0;
 632 }
 633
 634 int compare_commits_by_commit_date(const void *a_, const void *b_, void *unused)
 635 {
 636         const struct commit *a = a_, *b = b_;
 637         /* newer commits with larger date first */
 638         if (a->date < b->date)
 639                 return 1;
 640         else if (a->date > b->date)
 641                 return -1;
 642         return 0;
 643 }
 644
 645 /*
 646  * Performs an in-place topological sort on the list supplied.
 647  */
 648 void sort_in_topological_order(struct commit_list **list, enum rev_sort_order sort_order)
 649 {
 650         struct commit_list *next, *orig = *list;
 651         struct commit_list **pptr;
 652         struct indegree_slab indegree;
 653         struct prio_queue queue;
 654         struct commit *commit;
 655         struct author_date_slab author_date;
 656
 657         if (!orig)
 658                 return;
 659         *list = NULL;
 660
 661         init_indegree_slab(&indegree);
 662         memset(&queue, '\0', sizeof(queue));
 663
 664         switch (sort_order) {
 665         default: /* REV_SORT_IN_GRAPH_ORDER */
 666                 queue.compare = NULL;
 667                 break;
 668         case REV_SORT_BY_COMMIT_DATE:
 669                 queue.compare = compare_commits_by_commit_date;
 670                 break;
 671         case REV_SORT_BY_AUTHOR_DATE:
 672                 init_author_date_slab(&author_date);
 673                 queue.compare = compare_commits_by_author_date;
 674                 queue.cb_data = &author_date;
 675                 break;
 676         }
 677
 678         /* Mark them and clear the indegree */
 679         for (next = orig; next; next = next->next) {
 680                 struct commit *commit = next->item;
 681                 *(indegree_slab_at(&indegree, commit)) = 1;
 682                 /* also record the author dates, if needed */
 683                 if (sort_order == REV_SORT_BY_AUTHOR_DATE)
 684                         record_author_date(&author_date, commit);
 685         }
 686
 687         /* update the indegree */
 688         for (next = orig; next; next = next->next) {
 689                 struct commit_list *parents = next->item->parents;
 690                 while (parents) {
 691                         struct commit *parent = parents->item;
 692                         int *pi = indegree_slab_at(&indegree, parent);
 693
 694                         if (*pi)
 695                                 (*pi)++;
 696                         parents = parents->next;
 697                 }
 698         }
 699
 700         /*
 701          * find the tips
 702          *
 703          * tips are nodes not reachable from any other node in the list
 704          *
 705          * the tips serve as a starting set for the work queue.
 706          */
 707         for (next = orig; next; next = next->next) {
 708                 struct commit *commit = next->item;
 709
 710                 if (*(indegree_slab_at(&indegree, commit)) == 1)
 711                         prio_queue_put(&queue, commit);
 712         }
 713
 714         /*
 715          * This is unfortunate; the initial tips need to be shown
 716          * in the order given from the revision traversal machinery.
 717          */
 718         if (sort_order == REV_SORT_IN_GRAPH_ORDER)
 719                 prio_queue_reverse(&queue);
 720
 721         /* We no longer need the commit list */
 722         free_commit_list(orig);
 723
 724         pptr = list;
 725         *list = NULL;
 726         while ((commit = prio_queue_get(&queue)) != NULL) {
 727                 struct commit_list *parents;
 728
 729                 for (parents = commit->parents; parents ; parents = parents->next) {
 730                         struct commit *parent = parents->item;
 731                         int *pi = indegree_slab_at(&indegree, parent);
 732
 733                         if (!*pi)
 734                                 continue;
 735
 736                         /*
 737                          * parents are only enqueued for emission
 738                          * when all their children have been emitted thereby
 739                          * guaranteeing topological order.
 740                          */
 741                         if (--(*pi) == 1)
 742                                 prio_queue_put(&queue, parent);
 743                 }
 744                 /*
 745                  * all children of commit have already been
 746                  * emitted. we can emit it now.
 747                  */
 748                 *(indegree_slab_at(&indegree, commit)) = 0;
 749
 750                 pptr = &commit_list_insert(commit, pptr)->next;
 751         }
 752
 753         clear_indegree_slab(&indegree);
 754         clear_prio_queue(&queue);
 755         if (sort_order == REV_SORT_BY_AUTHOR_DATE)
 756                 clear_author_date_slab(&author_date);
 757 }
 758
 759 /* merge-base stuff */
 760
 761 /* Remember to update object flag allocation in object.h */
 762 #define PARENT1         (1u<<16)
 763 #define PARENT2         (1u<<17)
 764 #define STALE           (1u<<18)
 765 #define RESULT          (1u<<19)
 766
 767 static const unsigned all_flags = (PARENT1 | PARENT2 | STALE | RESULT);
 768
 769 static int queue_has_nonstale(struct prio_queue *queue)
 770 {
 771         int i;
 772         for (i = 0; i < queue->nr; i++) {
 773                 struct commit *commit = queue->array[i].data;
 774                 if (!(commit->object.flags & STALE))
 775                         return 1;
 776         }
 777         return 0;
 778 }
 779
 780 /* all input commits in one and twos[] must have been parsed! */
 781 static struct commit_list *paint_down_to_common(struct commit *one, int n, struct commit **twos)
 782 {
 783         struct prio_queue queue = { compare_commits_by_commit_date };
 784         struct commit_list *result = NULL;
 785         int i;
 786
 787         one->object.flags |= PARENT1;
 788         if (!n) {
 789                 commit_list_append(one, &result);
 790                 return result;
 791         }
 792         prio_queue_put(&queue, one);
 793
 794         for (i = 0; i < n; i++) {
 795                 twos[i]->object.flags |= PARENT2;
 796                 prio_queue_put(&queue, twos[i]);
 797         }
 798
 799         while (queue_has_nonstale(&queue)) {
 800                 struct commit *commit = prio_queue_get(&queue);
 801                 struct commit_list *parents;
 802                 int flags;
 803
 804                 flags = commit->object.flags & (PARENT1 | PARENT2 | STALE);
 805                 if (flags == (PARENT1 | PARENT2)) {
 806                         if (!(commit->object.flags & RESULT)) {
 807                                 commit->object.flags |= RESULT;
 808                                 commit_list_insert_by_date(commit, &result);
 809                         }
 810                         /* Mark parents of a found merge stale */
 811                         flags |= STALE;
 812                 }
 813                 parents = commit->parents;
 814                 while (parents) {
 815                         struct commit *p = parents->item;
 816                         parents = parents->next;
 817                         if ((p->object.flags & flags) == flags)
 818                                 continue;
 819                         if (parse_commit(p))
 820                                 return NULL;
 821                         p->object.flags |= flags;
 822                         prio_queue_put(&queue, p);
 823                 }
 824         }
 825
 826         clear_prio_queue(&queue);
 827         return result;
 828 }
 829
 830 static struct commit_list *merge_bases_many(struct commit *one, int n, struct commit **twos)
 831 {
 832         struct commit_list *list = NULL;
 833         struct commit_list *result = NULL;
 834         int i;
 835
 836         for (i = 0; i < n; i++) {
 837                 if (one == twos[i])
 838                         /*
 839                          * We do not mark this even with RESULT so we do not
 840                          * have to clean it up.
 841                          */
 842                         return commit_list_insert(one, &result);
 843         }
 844
 845         if (parse_commit(one))
 846                 return NULL;
 847         for (i = 0; i < n; i++) {
 848                 if (parse_commit(twos[i]))
 849                         return NULL;
 850         }
 851
 852         list = paint_down_to_common(one, n, twos);
 853
 854         while (list) {
 855                 struct commit *commit = pop_commit(&list);
 856                 if (!(commit->object.flags & STALE))
 857                         commit_list_insert_by_date(commit, &result);
 858         }
 859         return result;
 860 }
 861
 862 struct commit_list *get_octopus_merge_bases(struct commit_list *in)
 863 {
 864         struct commit_list *i, *j, *k, *ret = NULL;
 865
 866         if (!in)
 867                 return ret;
 868
 869         commit_list_insert(in->item, &ret);
 870
 871         for (i = in->next; i; i = i->next) {
 872                 struct commit_list *new_commits = NULL, *end = NULL;
 873
 874                 for (j = ret; j; j = j->next) {
 875                         struct commit_list *bases;
 876                         bases = get_merge_bases(i->item, j->item);
 877                         if (!new_commits)
 878                                 new_commits = bases;
 879                         else
 880                                 end->next = bases;
 881                         for (k = bases; k; k = k->next)
 882                                 end = k;
 883                 }
 884                 ret = new_commits;
 885         }
 886         return ret;
 887 }
 888
 889 static int remove_redundant(struct commit **array, int cnt)
 890 {
 891         /*
 892          * Some commit in the array may be an ancestor of
 893          * another commit.  Move such commit to the end of
 894          * the array, and return the number of commits that
 895          * are independent from each other.
 896          */
 897         struct commit **work;
 898         unsigned char *redundant;
 899         int *filled_index;
 900         int i, j, filled;
 901
 902         work = xcalloc(cnt, sizeof(*work));
 903         redundant = xcalloc(cnt, 1);
 904         ALLOC_ARRAY(filled_index, cnt - 1);
 905
 906         for (i = 0; i < cnt; i++)
 907                 parse_commit(array[i]);
 908         for (i = 0; i < cnt; i++) {
 909                 struct commit_list *common;
 910
 911                 if (redundant[i])
 912                         continue;
 913                 for (j = filled = 0; j < cnt; j++) {
 914                         if (i == j || redundant[j])
 915                                 continue;
 916                         filled_index[filled] = j;
 917                         work[filled++] = array[j];
 918                 }
 919                 common = paint_down_to_common(array[i], filled, work);
 920                 if (array[i]->object.flags & PARENT2)
 921                         redundant[i] = 1;
 922                 for (j = 0; j < filled; j++)
 923                         if (work[j]->object.flags & PARENT1)
 924                                 redundant[filled_index[j]] = 1;
 925                 clear_commit_marks(array[i], all_flags);
 926                 clear_commit_marks_many(filled, work, all_flags);
 927                 free_commit_list(common);
 928         }
 929
 930         /* Now collect the result */
 931         COPY_ARRAY(work, array, cnt);
 932         for (i = filled = 0; i < cnt; i++)
 933                 if (!redundant[i])
 934                         array[filled++] = work[i];
 935         for (j = filled, i = 0; i < cnt; i++)
 936                 if (redundant[i])
 937                         array[j++] = work[i];
 938         free(work);
 939         free(redundant);
 940         free(filled_index);
 941         return filled;
 942 }
 943
 944 static struct commit_list *get_merge_bases_many_0(struct commit *one,
 945                                                   int n,
 946                                                   struct commit **twos,
 947                                                   int cleanup)
 948 {
 949         struct commit_list *list;
 950         struct commit **rslt;
 951         struct commit_list *result;
 952         int cnt, i;
 953
 954         result = merge_bases_many(one, n, twos);
 955         for (i = 0; i < n; i++) {
 956                 if (one == twos[i])
 957                         return result;
 958         }
 959         if (!result || !result->next) {
 960                 if (cleanup) {
 961                         clear_commit_marks(one, all_flags);
 962                         clear_commit_marks_many(n, twos, all_flags);
 963                 }
 964                 return result;
 965         }
 966
 967         /* There are more than one */
 968         cnt = commit_list_count(result);
 969         rslt = xcalloc(cnt, sizeof(*rslt));
 970         for (list = result, i = 0; list; list = list->next)
 971                 rslt[i++] = list->item;
 972         free_commit_list(result);
 973
 974         clear_commit_marks(one, all_flags);
 975         clear_commit_marks_many(n, twos, all_flags);
 976
 977         cnt = remove_redundant(rslt, cnt);
 978         result = NULL;
 979         for (i = 0; i < cnt; i++)
 980                 commit_list_insert_by_date(rslt[i], &result);
 981         free(rslt);
 982         return result;
 983 }
 984
 985 struct commit_list *get_merge_bases_many(struct commit *one,
 986                                          int n,
 987                                          struct commit **twos)
 988 {
 989         return get_merge_bases_many_0(one, n, twos, 1);
 990 }
 991
 992 struct commit_list *get_merge_bases_many_dirty(struct commit *one,
 993                                                int n,
 994                                                struct commit **twos)
 995 {
 996         return get_merge_bases_many_0(one, n, twos, 0);
 997 }
 998
 999 struct commit_list *get_merge_bases(struct commit *one, struct commit *two)
1000 {
1001         return get_merge_bases_many_0(one, 1, &two, 1);
1002 }
1003
1004 /*
1005  * Is "commit" a descendant of one of the elements on the "with_commit" list?
1006  */
1007 int is_descendant_of(struct commit *commit, struct commit_list *with_commit)
1008 {
1009         if (!with_commit)
1010                 return 1;
1011         while (with_commit) {
1012                 struct commit *other;
1013
1014                 other = with_commit->item;
1015                 with_commit = with_commit->next;
1016                 if (in_merge_bases(other, commit))
1017                         return 1;
1018         }
1019         return 0;
1020 }
1021
1022 /*
1023  * Is "commit" an ancestor of one of the "references"?
1024  */
1025 int in_merge_bases_many(struct commit *commit, int nr_reference, struct commit **reference)
1026 {
1027         struct commit_list *bases;
1028         int ret = 0, i;
1029
1030         if (parse_commit(commit))
1031                 return ret;
1032         for (i = 0; i < nr_reference; i++)
1033                 if (parse_commit(reference[i]))
1034                         return ret;
1035
1036         bases = paint_down_to_common(commit, nr_reference, reference);
1037         if (commit->object.flags & PARENT2)
1038                 ret = 1;
1039         clear_commit_marks(commit, all_flags);
1040         clear_commit_marks_many(nr_reference, reference, all_flags);
1041         free_commit_list(bases);
1042         return ret;
1043 }
1044
1045 /*
1046  * Is "commit" an ancestor of (i.e. reachable from) the "reference"?
1047  */
1048 int in_merge_bases(struct commit *commit, struct commit *reference)
1049 {
1050         return in_merge_bases_many(commit, 1, &reference);
1051 }
1052
1053 struct commit_list *reduce_heads(struct commit_list *heads)
1054 {
1055         struct commit_list *p;
1056         struct commit_list *result = NULL, **tail = &result;
1057         struct commit **array;
1058         int num_head, i;
1059
1060         if (!heads)
1061                 return NULL;
1062
1063         /* Uniquify */
1064         for (p = heads; p; p = p->next)
1065                 p->item->object.flags &= ~STALE;
1066         for (p = heads, num_head = 0; p; p = p->next) {
1067                 if (p->item->object.flags & STALE)
1068                         continue;
1069                 p->item->object.flags |= STALE;
1070                 num_head++;
1071         }
1072         array = xcalloc(num_head, sizeof(*array));
1073         for (p = heads, i = 0; p; p = p->next) {
1074                 if (p->item->object.flags & STALE) {
1075                         array[i++] = p->item;
1076                         p->item->object.flags &= ~STALE;
1077                 }
1078         }
1079         num_head = remove_redundant(array, num_head);
1080         for (i = 0; i < num_head; i++)
1081                 tail = &commit_list_insert(array[i], tail)->next;
1082         free(array);
1083         return result;
1084 }
1085
1086 void reduce_heads_replace(struct commit_list **heads)
1087 {
1088         struct commit_list *result = reduce_heads(*heads);
1089         free_commit_list(*heads);
1090         *heads = result;
1091 }
1092
1093 static const char gpg_sig_header[] = "gpgsig";
1094 static const int gpg_sig_header_len = sizeof(gpg_sig_header) - 1;
1095
1096 static int do_sign_commit(struct strbuf *buf, const char *keyid)
1097 {
1098         struct strbuf sig = STRBUF_INIT;
1099         int inspos, copypos;
1100         const char *eoh;
1101
1102         /* find the end of the header */
1103         eoh = strstr(buf->buf, "\n\n");
1104         if (!eoh)
1105                 inspos = buf->len;
1106         else
1107                 inspos = eoh - buf->buf + 1;
1108
1109         if (!keyid || !*keyid)
1110                 keyid = get_signing_key();
1111         if (sign_buffer(buf, &sig, keyid)) {
1112                 strbuf_release(&sig);
1113                 return -1;
1114         }
1115
1116         for (copypos = 0; sig.buf[copypos]; ) {
1117                 const char *bol = sig.buf + copypos;
1118                 const char *eol = strchrnul(bol, '\n');
1119                 int len = (eol - bol) + !!*eol;
1120
1121                 if (!copypos) {
1122                         strbuf_insert(buf, inspos, gpg_sig_header, gpg_sig_header_len);
1123                         inspos += gpg_sig_header_len;
1124                 }
1125                 strbuf_insert(buf, inspos++, " ", 1);
1126                 strbuf_insert(buf, inspos, bol, len);
1127                 inspos += len;
1128                 copypos += len;
1129         }
1130         strbuf_release(&sig);
1131         return 0;
1132 }
1133
1134 int parse_signed_commit(const struct commit *commit,
1135                         struct strbuf *payload, struct strbuf *signature)
1136 {
1137
1138         unsigned long size;
1139         const char *buffer = get_commit_buffer(commit, &size);
1140         int in_signature, saw_signature = -1;
1141         const char *line, *tail;
1142
1143         line = buffer;
1144         tail = buffer + size;
1145         in_signature = 0;
1146         saw_signature = 0;
1147         while (line < tail) {
1148                 const char *sig = NULL;
1149                 const char *next = memchr(line, '\n', tail - line);
1150
1151                 next = next ? next + 1 : tail;
1152                 if (in_signature && line[0] == ' ')
1153                         sig = line + 1;
1154                 else if (starts_with(line, gpg_sig_header) &&
1155                          line[gpg_sig_header_len] == ' ')
1156                         sig = line + gpg_sig_header_len + 1;
1157                 if (sig) {
1158                         strbuf_add(signature, sig, next - sig);
1159                         saw_signature = 1;
1160                         in_signature = 1;
1161                 } else {
1162                         if (*line == '\n')
1163                                 /* dump the whole remainder of the buffer */
1164                                 next = tail;
1165                         strbuf_add(payload, line, next - line);
1166                         in_signature = 0;
1167                 }
1168                 line = next;
1169         }
1170         unuse_commit_buffer(commit, buffer);
1171         return saw_signature;
1172 }
1173
1174 int remove_signature(struct strbuf *buf)
1175 {
1176         const char *line = buf->buf;
1177         const char *tail = buf->buf + buf->len;
1178         int in_signature = 0;
1179         const char *sig_start = NULL;
1180         const char *sig_end = NULL;
1181
1182         while (line < tail) {
1183                 const char *next = memchr(line, '\n', tail - line);
1184                 next = next ? next + 1 : tail;
1185
1186                 if (in_signature && line[0] == ' ')
1187                         sig_end = next;
1188                 else if (starts_with(line, gpg_sig_header) &&
1189                          line[gpg_sig_header_len] == ' ') {
1190                         sig_start = line;
1191                         sig_end = next;
1192                         in_signature = 1;
1193                 } else {
1194                         if (*line == '\n')
1195                                 /* dump the whole remainder of the buffer */
1196                                 next = tail;
1197                         in_signature = 0;
1198                 }
1199                 line = next;
1200         }
1201
1202         if (sig_start)
1203                 strbuf_remove(buf, sig_start - buf->buf, sig_end - sig_start);
1204
1205         return sig_start != NULL;
1206 }
1207
1208 static void handle_signed_tag(struct commit *parent, struct commit_extra_header ***tail)
1209 {
1210         struct merge_remote_desc *desc;
1211         struct commit_extra_header *mergetag;
1212         char *buf;
1213         unsigned long size, len;
1214         enum object_type type;
1215
1216         desc = merge_remote_util(parent);
1217         if (!desc || !desc->obj)
1218                 return;
1219         buf = read_object_file(&desc->obj->oid, &type, &size);
1220         if (!buf || type != OBJ_TAG)
1221                 goto free_return;
1222         len = parse_signature(buf, size);
1223         if (size == len)
1224                 goto free_return;
1225         /*
1226          * We could verify this signature and either omit the tag when
1227          * it does not validate, but the integrator may not have the
1228          * public key of the signer of the tag he is merging, while a
1229          * later auditor may have it while auditing, so let's not run
1230          * verify-signed-buffer here for now...
1231          *
1232          * if (verify_signed_buffer(buf, len, buf + len, size - len, ...))
1233          *      warn("warning: signed tag unverified.");
1234          */
1235         mergetag = xcalloc(1, sizeof(*mergetag));
1236         mergetag->key = xstrdup("mergetag");
1237         mergetag->value = buf;
1238         mergetag->len = size;
1239
1240         **tail = mergetag;
1241         *tail = &mergetag->next;
1242         return;
1243
1244 free_return:
1245         free(buf);
1246 }
1247
1248 int check_commit_signature(const struct commit *commit, struct signature_check *sigc)
1249 {
1250         struct strbuf payload = STRBUF_INIT;
1251         struct strbuf signature = STRBUF_INIT;
1252         int ret = 1;
1253
1254         sigc->result = 'N';
1255
1256         if (parse_signed_commit(commit, &payload, &signature) <= 0)
1257                 goto out;
1258         ret = check_signature(payload.buf, payload.len, signature.buf,
1259                 signature.len, sigc);
1260
1261  out:
1262         strbuf_release(&payload);
1263         strbuf_release(&signature);
1264
1265         return ret;
1266 }
1267
1268
1269
1270 void append_merge_tag_headers(struct commit_list *parents,
1271                               struct commit_extra_header ***tail)
1272 {
1273         while (parents) {
1274                 struct commit *parent = parents->item;
1275                 handle_signed_tag(parent, tail);
1276                 parents = parents->next;
1277         }
1278 }
1279
1280 static void add_extra_header(struct strbuf *buffer,
1281                              struct commit_extra_header *extra)
1282 {
1283         strbuf_addstr(buffer, extra->key);
1284         if (extra->len)
1285                 strbuf_add_lines(buffer, " ", extra->value, extra->len);
1286         else
1287                 strbuf_addch(buffer, '\n');
1288 }
1289
1290 struct commit_extra_header *read_commit_extra_headers(struct commit *commit,
1291                                                       const char **exclude)
1292 {
1293         struct commit_extra_header *extra = NULL;
1294         unsigned long size;
1295         const char *buffer = get_commit_buffer(commit, &size);
1296         extra = read_commit_extra_header_lines(buffer, size, exclude);
1297         unuse_commit_buffer(commit, buffer);
1298         return extra;
1299 }
1300
1301 int for_each_mergetag(each_mergetag_fn fn, struct commit *commit, void *data)
1302 {
1303         struct commit_extra_header *extra, *to_free;
1304         int res = 0;
1305
1306         to_free = read_commit_extra_headers(commit, NULL);
1307         for (extra = to_free; !res && extra; extra = extra->next) {
1308                 if (strcmp(extra->key, "mergetag"))
1309                         continue; /* not a merge tag */
1310                 res = fn(commit, extra, data);
1311         }
1312         free_commit_extra_headers(to_free);
1313         return res;
1314 }
1315
1316 static inline int standard_header_field(const char *field, size_t len)
1317 {
1318         return ((len == 4 && !memcmp(field, "tree", 4)) ||
1319                 (len == 6 && !memcmp(field, "parent", 6)) ||
1320                 (len == 6 && !memcmp(field, "author", 6)) ||
1321                 (len == 9 && !memcmp(field, "committer", 9)) ||
1322                 (len == 8 && !memcmp(field, "encoding", 8)));
1323 }
1324
1325 static int excluded_header_field(const char *field, size_t len, const char **exclude)
1326 {
1327         if (!exclude)
1328                 return 0;
1329
1330         while (*exclude) {
1331                 size_t xlen = strlen(*exclude);
1332                 if (len == xlen && !memcmp(field, *exclude, xlen))
1333                         return 1;
1334                 exclude++;
1335         }
1336         return 0;
1337 }
1338
1339 static struct commit_extra_header *read_commit_extra_header_lines(
1340         const char *buffer, size_t size,
1341         const char **exclude)
1342 {
1343         struct commit_extra_header *extra = NULL, **tail = &extra, *it = NULL;
1344         const char *line, *next, *eof, *eob;
1345         struct strbuf buf = STRBUF_INIT;
1346
1347         for (line = buffer, eob = line + size;
1348              line < eob && *line != '\n';
1349              line = next) {
1350                 next = memchr(line, '\n', eob - line);
1351                 next = next ? next + 1 : eob;
1352                 if (*line == ' ') {
1353                         /* continuation */
1354                         if (it)
1355                                 strbuf_add(&buf, line + 1, next - (line + 1));
1356                         continue;
1357                 }
1358                 if (it)
1359                         it->value = strbuf_detach(&buf, &it->len);
1360                 strbuf_reset(&buf);
1361                 it = NULL;
1362
1363                 eof = memchr(line, ' ', next - line);
1364                 if (!eof)
1365                         eof = next;
1366                 else if (standard_header_field(line, eof - line) ||
1367                          excluded_header_field(line, eof - line, exclude))
1368                         continue;
1369
1370                 it = xcalloc(1, sizeof(*it));
1371                 it->key = xmemdupz(line, eof-line);
1372                 *tail = it;
1373                 tail = &it->next;
1374                 if (eof + 1 < next)
1375                         strbuf_add(&buf, eof + 1, next - (eof + 1));
1376         }
1377         if (it)
1378                 it->value = strbuf_detach(&buf, &it->len);
1379         return extra;
1380 }
1381
1382 void free_commit_extra_headers(struct commit_extra_header *extra)
1383 {
1384         while (extra) {
1385                 struct commit_extra_header *next = extra->next;
1386                 free(extra->key);
1387                 free(extra->value);
1388                 free(extra);
1389                 extra = next;
1390         }
1391 }
1392
1393 int commit_tree(const char *msg, size_t msg_len, const struct object_id *tree,
1394                 struct commit_list *parents, struct object_id *ret,
1395                 const char *author, const char *sign_commit)
1396 {
1397         struct commit_extra_header *extra = NULL, **tail = &extra;
1398         int result;
1399
1400         append_merge_tag_headers(parents, &tail);
1401         result = commit_tree_extended(msg, msg_len, tree, parents, ret,
1402                                       author, sign_commit, extra);
1403         free_commit_extra_headers(extra);
1404         return result;
1405 }
1406
1407 static int find_invalid_utf8(const char *buf, int len)
1408 {
1409         int offset = 0;
1410         static const unsigned int max_codepoint[] = {
1411                 0x7f, 0x7ff, 0xffff, 0x10ffff
1412         };
1413
1414         while (len) {
1415                 unsigned char c = *buf++;
1416                 int bytes, bad_offset;
1417                 unsigned int codepoint;
1418                 unsigned int min_val, max_val;
1419
1420                 len--;
1421                 offset++;
1422
1423                 /* Simple US-ASCII? No worries. */
1424                 if (c < 0x80)
1425                         continue;
1426
1427                 bad_offset = offset-1;
1428
1429                 /*
1430                  * Count how many more high bits set: that's how
1431                  * many more bytes this sequence should have.
1432                  */
1433                 bytes = 0;
1434                 while (c & 0x40) {
1435                         c <<= 1;
1436                         bytes++;
1437                 }
1438
1439                 /*
1440                  * Must be between 1 and 3 more bytes.  Longer sequences result in
1441                  * codepoints beyond U+10FFFF, which are guaranteed never to exist.
1442                  */
1443                 if (bytes < 1 || 3 < bytes)
1444                         return bad_offset;
1445
1446                 /* Do we *have* that many bytes? */
1447                 if (len < bytes)
1448                         return bad_offset;
1449
1450                 /*
1451                  * Place the encoded bits at the bottom of the value and compute the
1452                  * valid range.
1453                  */
1454                 codepoint = (c & 0x7f) >> bytes;
1455                 min_val = max_codepoint[bytes-1] + 1;
1456                 max_val = max_codepoint[bytes];
1457
1458                 offset += bytes;
1459                 len -= bytes;
1460
1461                 /* And verify that they are good continuation bytes */
1462                 do {
1463                         codepoint <<= 6;
1464                         codepoint |= *buf & 0x3f;
1465                         if ((*buf++ & 0xc0) != 0x80)
1466                                 return bad_offset;
1467                 } while (--bytes);
1468
1469                 /* Reject codepoints that are out of range for the sequence length. */
1470                 if (codepoint < min_val || codepoint > max_val)
1471                         return bad_offset;
1472                 /* Surrogates are only for UTF-16 and cannot be encoded in UTF-8. */
1473                 if ((codepoint & 0x1ff800) == 0xd800)
1474                         return bad_offset;
1475                 /* U+xxFFFE and U+xxFFFF are guaranteed non-characters. */
1476                 if ((codepoint & 0xfffe) == 0xfffe)
1477                         return bad_offset;
1478                 /* So are anything in the range U+FDD0..U+FDEF. */
1479                 if (codepoint >= 0xfdd0 && codepoint <= 0xfdef)
1480                         return bad_offset;
1481         }
1482         return -1;
1483 }
1484
1485 /*
1486  * This verifies that the buffer is in proper utf8 format.
1487  *
1488  * If it isn't, it assumes any non-utf8 characters are Latin1,
1489  * and does the conversion.
1490  */
1491 static int verify_utf8(struct strbuf *buf)
1492 {
1493         int ok = 1;
1494         long pos = 0;
1495
1496         for (;;) {
1497                 int bad;
1498                 unsigned char c;
1499                 unsigned char replace[2];
1500
1501                 bad = find_invalid_utf8(buf->buf + pos, buf->len - pos);
1502                 if (bad < 0)
1503                         return ok;
1504                 pos += bad;
1505                 ok = 0;
1506                 c = buf->buf[pos];
1507                 strbuf_remove(buf, pos, 1);
1508
1509                 /* We know 'c' must be in the range 128-255 */
1510                 replace[0] = 0xc0 + (c >> 6);
1511                 replace[1] = 0x80 + (c & 0x3f);
1512                 strbuf_insert(buf, pos, replace, 2);
1513                 pos += 2;
1514         }
1515 }
1516
1517 static const char commit_utf8_warn[] =
1518 N_("Warning: commit message did not conform to UTF-8.\n"
1519    "You may want to amend it after fixing the message, or set the config\n"
1520    "variable i18n.commitencoding to the encoding your project uses.\n");
1521
1522 int commit_tree_extended(const char *msg, size_t msg_len,
1523                          const struct object_id *tree,
1524                          struct commit_list *parents, struct object_id *ret,
1525                          const char *author, const char *sign_commit,
1526                          struct commit_extra_header *extra)
1527 {
1528         int result;
1529         int encoding_is_utf8;
1530         struct strbuf buffer;
1531
1532         assert_oid_type(tree, OBJ_TREE);
1533
1534         if (memchr(msg, '\0', msg_len))
1535                 return error("a NUL byte in commit log message not allowed.");
1536
1537         /* Not having i18n.commitencoding is the same as having utf-8 */
1538         encoding_is_utf8 = is_encoding_utf8(git_commit_encoding);
1539
1540         strbuf_init(&buffer, 8192); /* should avoid reallocs for the headers */
1541         strbuf_addf(&buffer, "tree %s\n", oid_to_hex(tree));
1542
1543         /*
1544          * NOTE! This ordering means that the same exact tree merged with a
1545          * different order of parents will be a _different_ changeset even
1546          * if everything else stays the same.
1547          */
1548         while (parents) {
1549                 struct commit *parent = pop_commit(&parents);
1550                 strbuf_addf(&buffer, "parent %s\n",
1551                             oid_to_hex(&parent->object.oid));
1552         }
1553
1554         /* Person/date information */
1555         if (!author)
1556                 author = git_author_info(IDENT_STRICT);
1557         strbuf_addf(&buffer, "author %s\n", author);
1558         strbuf_addf(&buffer, "committer %s\n", git_committer_info(IDENT_STRICT));
1559         if (!encoding_is_utf8)
1560                 strbuf_addf(&buffer, "encoding %s\n", git_commit_encoding);
1561
1562         while (extra) {
1563                 add_extra_header(&buffer, extra);
1564                 extra = extra->next;
1565         }
1566         strbuf_addch(&buffer, '\n');
1567
1568         /* And add the comment */
1569         strbuf_add(&buffer, msg, msg_len);
1570
1571         /* And check the encoding */
1572         if (encoding_is_utf8 && !verify_utf8(&buffer))
1573                 fprintf(stderr, _(commit_utf8_warn));
1574
1575         if (sign_commit && do_sign_commit(&buffer, sign_commit)) {
1576                 result = -1;
1577                 goto out;
1578         }
1579
1580         result = write_object_file(buffer.buf, buffer.len, commit_type, ret);
1581 out:
1582         strbuf_release(&buffer);
1583         return result;
1584 }
1585
1586 void set_merge_remote_desc(struct commit *commit,
1587                            const char *name, struct object *obj)
1588 {
1589         struct merge_remote_desc *desc;
1590         FLEX_ALLOC_STR(desc, name, name);
1591         desc->obj = obj;
1592         commit->util = desc;
1593 }
1594
1595 struct commit *get_merge_parent(const char *name)
1596 {
1597         struct object *obj;
1598         struct commit *commit;
1599         struct object_id oid;
1600         if (get_oid(name, &oid))
1601                 return NULL;
1602         obj = parse_object(&oid);
1603         commit = (struct commit *)peel_to_type(name, 0, obj, OBJ_COMMIT);
1604         if (commit && !commit->util)
1605                 set_merge_remote_desc(commit, name, obj);
1606         return commit;
1607 }
1608
1609 /*
1610  * Append a commit to the end of the commit_list.
1611  *
1612  * next starts by pointing to the variable that holds the head of an
1613  * empty commit_list, and is updated to point to the "next" field of
1614  * the last item on the list as new commits are appended.
1615  *
1616  * Usage example:
1617  *
1618  *     struct commit_list *list;
1619  *     struct commit_list **next = &list;
1620  *
1621  *     next = commit_list_append(c1, next);
1622  *     next = commit_list_append(c2, next);
1623  *     assert(commit_list_count(list) == 2);
1624  *     return list;
1625  */
1626 struct commit_list **commit_list_append(struct commit *commit,
1627                                         struct commit_list **next)
1628 {
1629         struct commit_list *new_commit = xmalloc(sizeof(struct commit_list));
1630         new_commit->item = commit;
1631         *next = new_commit;
1632         new_commit->next = NULL;
1633         return &new_commit->next;
1634 }
1635
1636 const char *find_commit_header(const char *msg, const char *key, size_t *out_len)
1637 {
1638         int key_len = strlen(key);
1639         const char *line = msg;
1640
1641         while (line) {
1642                 const char *eol = strchrnul(line, '\n');
1643
1644                 if (line == eol)
1645                         return NULL;
1646
1647                 if (eol - line > key_len &&
1648                     !strncmp(line, key, key_len) &&
1649                     line[key_len] == ' ') {
1650                         *out_len = eol - line - key_len - 1;
1651                         return line + key_len + 1;
1652                 }
1653                 line = *eol ? eol + 1 : NULL;
1654         }
1655         return NULL;
1656 }
1657
1658 /*
1659  * Inspect the given string and determine the true "end" of the log message, in
1660  * order to find where to put a new Signed-off-by: line.  Ignored are
1661  * trailing comment lines and blank lines.  To support "git commit -s
1662  * --amend" on an existing commit, we also ignore "Conflicts:".  To
1663  * support "git commit -v", we truncate at cut lines.
1664  *
1665  * Returns the number of bytes from the tail to ignore, to be fed as
1666  * the second parameter to append_signoff().
1667  */
1668 int ignore_non_trailer(const char *buf, size_t len)
1669 {
1670         int boc = 0;
1671         int bol = 0;
1672         int in_old_conflicts_block = 0;
1673         size_t cutoff = wt_status_locate_end(buf, len);
1674
1675         while (bol < cutoff) {
1676                 const char *next_line = memchr(buf + bol, '\n', len - bol);
1677
1678                 if (!next_line)
1679                         next_line = buf + len;
1680                 else
1681                         next_line++;
1682
1683                 if (buf[bol] == comment_line_char || buf[bol] == '\n') {
1684                         /* is this the first of the run of comments? */
1685                         if (!boc)
1686                                 boc = bol;
1687                         /* otherwise, it is just continuing */
1688                 } else if (starts_with(buf + bol, "Conflicts:\n")) {
1689                         in_old_conflicts_block = 1;
1690                         if (!boc)
1691                                 boc = bol;
1692                 } else if (in_old_conflicts_block && buf[bol] == '\t') {
1693                         ; /* a pathname in the conflicts block */
1694                 } else if (boc) {
1695                         /* the previous was not trailing comment */
1696                         boc = 0;
1697                         in_old_conflicts_block = 0;
1698                 }
1699                 bol = next_line - buf;
1700         }
1701         return boc ? len - boc : len - cutoff;
1702 }