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
  15 static struct commit_extra_header *read_commit_extra_header_lines(const char *buf, size_t len, const char **);
  16
  17 int save_commit_buffer = 1;
  18
  19 const char *commit_type = "commit";
  20
  21 struct commit *lookup_commit_reference_gently(const unsigned char *sha1,
  22                                               int quiet)
  23 {
  24         struct object *obj = deref_tag(parse_object(sha1), NULL, 0);
  25
  26         if (!obj)
  27                 return NULL;
  28         return object_as_type(obj, OBJ_COMMIT, quiet);
  29 }
  30
  31 struct commit *lookup_commit_reference(const unsigned char *sha1)
  32 {
  33         return lookup_commit_reference_gently(sha1, 0);
  34 }
  35
  36 struct commit *lookup_commit_or_die(const unsigned char *sha1, const char *ref_name)
  37 {
  38         struct commit *c = lookup_commit_reference(sha1);
  39         if (!c)
  40                 die(_("could not parse %s"), ref_name);
  41         if (hashcmp(sha1, c->object.sha1)) {
  42                 warning(_("%s %s is not a commit!"),
  43                         ref_name, sha1_to_hex(sha1));
  44         }
  45         return c;
  46 }
  47
  48 struct commit *lookup_commit(const unsigned char *sha1)
  49 {
  50         struct object *obj = lookup_object(sha1);
  51         if (!obj)
  52                 return create_object(sha1, alloc_commit_node());
  53         return object_as_type(obj, OBJ_COMMIT, 0);
  54 }
  55
  56 struct commit *lookup_commit_reference_by_name(const char *name)
  57 {
  58         unsigned char sha1[20];
  59         struct commit *commit;
  60
  61         if (get_sha1_committish(name, sha1))
  62                 return NULL;
  63         commit = lookup_commit_reference(sha1);
  64         if (parse_commit(commit))
  65                 return NULL;
  66         return commit;
  67 }
  68
  69 static unsigned long parse_commit_date(const char *buf, const char *tail)
  70 {
  71         const char *dateptr;
  72
  73         if (buf + 6 >= tail)
  74                 return 0;
  75         if (memcmp(buf, "author", 6))
  76                 return 0;
  77         while (buf < tail && *buf++ != '\n')
  78                 /* nada */;
  79         if (buf + 9 >= tail)
  80                 return 0;
  81         if (memcmp(buf, "committer", 9))
  82                 return 0;
  83         while (buf < tail && *buf++ != '>')
  84                 /* nada */;
  85         if (buf >= tail)
  86                 return 0;
  87         dateptr = buf;
  88         while (buf < tail && *buf++ != '\n')
  89                 /* nada */;
  90         if (buf >= tail)
  91                 return 0;
  92         /* dateptr < buf && buf[-1] == '\n', so strtoul will stop at buf-1 */
  93         return strtoul(dateptr, NULL, 10);
  94 }
  95
  96 static struct commit_graft **commit_graft;
  97 static int commit_graft_alloc, commit_graft_nr;
  98
  99 static const unsigned char *commit_graft_sha1_access(size_t index, void *table)
 100 {
 101         struct commit_graft **commit_graft_table = table;
 102         return commit_graft_table[index]->sha1;
 103 }
 104
 105 static int commit_graft_pos(const unsigned char *sha1)
 106 {
 107         return sha1_pos(sha1, commit_graft, commit_graft_nr,
 108                         commit_graft_sha1_access);
 109 }
 110
 111 int register_commit_graft(struct commit_graft *graft, int ignore_dups)
 112 {
 113         int pos = commit_graft_pos(graft->sha1);
 114
 115         if (0 <= pos) {
 116                 if (ignore_dups)
 117                         free(graft);
 118                 else {
 119                         free(commit_graft[pos]);
 120                         commit_graft[pos] = graft;
 121                 }
 122                 return 1;
 123         }
 124         pos = -pos - 1;
 125         ALLOC_GROW(commit_graft, commit_graft_nr + 1, commit_graft_alloc);
 126         commit_graft_nr++;
 127         if (pos < commit_graft_nr)
 128                 memmove(commit_graft + pos + 1,
 129                         commit_graft + pos,
 130                         (commit_graft_nr - pos - 1) *
 131                         sizeof(*commit_graft));
 132         commit_graft[pos] = graft;
 133         return 0;
 134 }
 135
 136 struct commit_graft *read_graft_line(char *buf, int len)
 137 {
 138         /* The format is just "Commit Parent1 Parent2 ...\n" */
 139         int i;
 140         struct commit_graft *graft = NULL;
 141
 142         while (len && isspace(buf[len-1]))
 143                 buf[--len] = '\0';
 144         if (buf[0] == '#' || buf[0] == '\0')
 145                 return NULL;
 146         if ((len + 1) % 41)
 147                 goto bad_graft_data;
 148         i = (len + 1) / 41 - 1;
 149         graft = xmalloc(sizeof(*graft) + 20 * i);
 150         graft->nr_parent = i;
 151         if (get_sha1_hex(buf, graft->sha1))
 152                 goto bad_graft_data;
 153         for (i = 40; i < len; i += 41) {
 154                 if (buf[i] != ' ')
 155                         goto bad_graft_data;
 156                 if (get_sha1_hex(buf + i + 1, graft->parent[i/41]))
 157                         goto bad_graft_data;
 158         }
 159         return graft;
 160
 161 bad_graft_data:
 162         error("bad graft data: %s", buf);
 163         free(graft);
 164         return NULL;
 165 }
 166
 167 static int read_graft_file(const char *graft_file)
 168 {
 169         FILE *fp = fopen(graft_file, "r");
 170         struct strbuf buf = STRBUF_INIT;
 171         if (!fp)
 172                 return -1;
 173         while (!strbuf_getwholeline(&buf, fp, '\n')) {
 174                 /* The format is just "Commit Parent1 Parent2 ...\n" */
 175                 struct commit_graft *graft = read_graft_line(buf.buf, buf.len);
 176                 if (!graft)
 177                         continue;
 178                 if (register_commit_graft(graft, 1))
 179                         error("duplicate graft data: %s", buf.buf);
 180         }
 181         fclose(fp);
 182         strbuf_release(&buf);
 183         return 0;
 184 }
 185
 186 static void prepare_commit_graft(void)
 187 {
 188         static int commit_graft_prepared;
 189         char *graft_file;
 190
 191         if (commit_graft_prepared)
 192                 return;
 193         graft_file = get_graft_file();
 194         read_graft_file(graft_file);
 195         /* make sure shallows are read */
 196         is_repository_shallow();
 197         commit_graft_prepared = 1;
 198 }
 199
 200 struct commit_graft *lookup_commit_graft(const unsigned char *sha1)
 201 {
 202         int pos;
 203         prepare_commit_graft();
 204         pos = commit_graft_pos(sha1);
 205         if (pos < 0)
 206                 return NULL;
 207         return commit_graft[pos];
 208 }
 209
 210 int for_each_commit_graft(each_commit_graft_fn fn, void *cb_data)
 211 {
 212         int i, ret;
 213         for (i = ret = 0; i < commit_graft_nr && !ret; i++)
 214                 ret = fn(commit_graft[i], cb_data);
 215         return ret;
 216 }
 217
 218 int unregister_shallow(const unsigned char *sha1)
 219 {
 220         int pos = commit_graft_pos(sha1);
 221         if (pos < 0)
 222                 return -1;
 223         if (pos + 1 < commit_graft_nr)
 224                 memmove(commit_graft + pos, commit_graft + pos + 1,
 225                                 sizeof(struct commit_graft *)
 226                                 * (commit_graft_nr - pos - 1));
 227         commit_graft_nr--;
 228         return 0;
 229 }
 230
 231 struct commit_buffer {
 232         void *buffer;
 233         unsigned long size;
 234 };
 235 define_commit_slab(buffer_slab, struct commit_buffer);
 236 static struct buffer_slab buffer_slab = COMMIT_SLAB_INIT(1, buffer_slab);
 237
 238 void set_commit_buffer(struct commit *commit, void *buffer, unsigned long size)
 239 {
 240         struct commit_buffer *v = buffer_slab_at(&buffer_slab, commit);
 241         v->buffer = buffer;
 242         v->size = size;
 243 }
 244
 245 const void *get_cached_commit_buffer(const struct commit *commit, unsigned long *sizep)
 246 {
 247         struct commit_buffer *v = buffer_slab_at(&buffer_slab, commit);
 248         if (sizep)
 249                 *sizep = v->size;
 250         return v->buffer;
 251 }
 252
 253 const void *get_commit_buffer(const struct commit *commit, unsigned long *sizep)
 254 {
 255         const void *ret = get_cached_commit_buffer(commit, sizep);
 256         if (!ret) {
 257                 enum object_type type;
 258                 unsigned long size;
 259                 ret = read_sha1_file(commit->object.sha1, &type, &size);
 260                 if (!ret)
 261                         die("cannot read commit object %s",
 262                             sha1_to_hex(commit->object.sha1));
 263                 if (type != OBJ_COMMIT)
 264                         die("expected commit for %s, got %s",
 265                             sha1_to_hex(commit->object.sha1), typename(type));
 266                 if (sizep)
 267                         *sizep = size;
 268         }
 269         return ret;
 270 }
 271
 272 void unuse_commit_buffer(const struct commit *commit, const void *buffer)
 273 {
 274         struct commit_buffer *v = buffer_slab_at(&buffer_slab, commit);
 275         if (v->buffer != buffer)
 276                 free((void *)buffer);
 277 }
 278
 279 void free_commit_buffer(struct commit *commit)
 280 {
 281         struct commit_buffer *v = buffer_slab_at(&buffer_slab, commit);
 282         free(v->buffer);
 283         v->buffer = NULL;
 284         v->size = 0;
 285 }
 286
 287 const void *detach_commit_buffer(struct commit *commit, unsigned long *sizep)
 288 {
 289         struct commit_buffer *v = buffer_slab_at(&buffer_slab, commit);
 290         void *ret;
 291
 292         ret = v->buffer;
 293         if (sizep)
 294                 *sizep = v->size;
 295
 296         v->buffer = NULL;
 297         v->size = 0;
 298         return ret;
 299 }
 300
 301 int parse_commit_buffer(struct commit *item, const void *buffer, unsigned long size)
 302 {
 303         const char *tail = buffer;
 304         const char *bufptr = buffer;
 305         unsigned char parent[20];
 306         struct commit_list **pptr;
 307         struct commit_graft *graft;
 308
 309         if (item->object.parsed)
 310                 return 0;
 311         item->object.parsed = 1;
 312         tail += size;
 313         if (tail <= bufptr + 46 || memcmp(bufptr, "tree ", 5) || bufptr[45] != '\n')
 314                 return error("bogus commit object %s", sha1_to_hex(item->object.sha1));
 315         if (get_sha1_hex(bufptr + 5, parent) < 0)
 316                 return error("bad tree pointer in commit %s",
 317                              sha1_to_hex(item->object.sha1));
 318         item->tree = lookup_tree(parent);
 319         bufptr += 46; /* "tree " + "hex sha1" + "\n" */
 320         pptr = &item->parents;
 321
 322         graft = lookup_commit_graft(item->object.sha1);
 323         while (bufptr + 48 < tail && !memcmp(bufptr, "parent ", 7)) {
 324                 struct commit *new_parent;
 325
 326                 if (tail <= bufptr + 48 ||
 327                     get_sha1_hex(bufptr + 7, parent) ||
 328                     bufptr[47] != '\n')
 329                         return error("bad parents in commit %s", sha1_to_hex(item->object.sha1));
 330                 bufptr += 48;
 331                 /*
 332                  * The clone is shallow if nr_parent < 0, and we must
 333                  * not traverse its real parents even when we unhide them.
 334                  */
 335                 if (graft && (graft->nr_parent < 0 || grafts_replace_parents))
 336                         continue;
 337                 new_parent = lookup_commit(parent);
 338                 if (new_parent)
 339                         pptr = &commit_list_insert(new_parent, pptr)->next;
 340         }
 341         if (graft) {
 342                 int i;
 343                 struct commit *new_parent;
 344                 for (i = 0; i < graft->nr_parent; i++) {
 345                         new_parent = lookup_commit(graft->parent[i]);
 346                         if (!new_parent)
 347                                 continue;
 348                         pptr = &commit_list_insert(new_parent, pptr)->next;
 349                 }
 350         }
 351         item->date = parse_commit_date(bufptr, tail);
 352
 353         return 0;
 354 }
 355
 356 int parse_commit(struct commit *item)
 357 {
 358         enum object_type type;
 359         void *buffer;
 360         unsigned long size;
 361         int ret;
 362
 363         if (!item)
 364                 return -1;
 365         if (item->object.parsed)
 366                 return 0;
 367         buffer = read_sha1_file(item->object.sha1, &type, &size);
 368         if (!buffer)
 369                 return error("Could not read %s",
 370                              sha1_to_hex(item->object.sha1));
 371         if (type != OBJ_COMMIT) {
 372                 free(buffer);
 373                 return error("Object %s not a commit",
 374                              sha1_to_hex(item->object.sha1));
 375         }
 376         ret = parse_commit_buffer(item, buffer, size);
 377         if (save_commit_buffer && !ret) {
 378                 set_commit_buffer(item, buffer, size);
 379                 return 0;
 380         }
 381         free(buffer);
 382         return ret;
 383 }
 384
 385 void parse_commit_or_die(struct commit *item)
 386 {
 387         if (parse_commit(item))
 388                 die("unable to parse commit %s",
 389                     item ? sha1_to_hex(item->object.sha1) : "(null)");
 390 }
 391
 392 int find_commit_subject(const char *commit_buffer, const char **subject)
 393 {
 394         const char *eol;
 395         const char *p = commit_buffer;
 396
 397         while (*p && (*p != '\n' || p[1] != '\n'))
 398                 p++;
 399         if (*p) {
 400                 p += 2;
 401                 for (eol = p; *eol && *eol != '\n'; eol++)
 402                         ; /* do nothing */
 403         } else
 404                 eol = p;
 405
 406         *subject = p;
 407
 408         return eol - p;
 409 }
 410
 411 struct commit_list *commit_list_insert(struct commit *item, struct commit_list **list_p)
 412 {
 413         struct commit_list *new_list = xmalloc(sizeof(struct commit_list));
 414         new_list->item = item;
 415         new_list->next = *list_p;
 416         *list_p = new_list;
 417         return new_list;
 418 }
 419
 420 unsigned commit_list_count(const struct commit_list *l)
 421 {
 422         unsigned c = 0;
 423         for (; l; l = l->next )
 424                 c++;
 425         return c;
 426 }
 427
 428 struct commit_list *copy_commit_list(struct commit_list *list)
 429 {
 430         struct commit_list *head = NULL;
 431         struct commit_list **pp = &head;
 432         while (list) {
 433                 struct commit_list *new;
 434                 new = xmalloc(sizeof(struct commit_list));
 435                 new->item = list->item;
 436                 new->next = NULL;
 437                 *pp = new;
 438                 pp = &new->next;
 439                 list = list->next;
 440         }
 441         return head;
 442 }
 443
 444 void free_commit_list(struct commit_list *list)
 445 {
 446         while (list) {
 447                 struct commit_list *temp = list;
 448                 list = temp->next;
 449                 free(temp);
 450         }
 451 }
 452
 453 struct commit_list * commit_list_insert_by_date(struct commit *item, struct commit_list **list)
 454 {
 455         struct commit_list **pp = list;
 456         struct commit_list *p;
 457         while ((p = *pp) != NULL) {
 458                 if (p->item->date < item->date) {
 459                         break;
 460                 }
 461                 pp = &p->next;
 462         }
 463         return commit_list_insert(item, pp);
 464 }
 465
 466 static int commit_list_compare_by_date(const void *a, const void *b)
 467 {
 468         unsigned long a_date = ((const struct commit_list *)a)->item->date;
 469         unsigned long b_date = ((const struct commit_list *)b)->item->date;
 470         if (a_date < b_date)
 471                 return 1;
 472         if (a_date > b_date)
 473                 return -1;
 474         return 0;
 475 }
 476
 477 static void *commit_list_get_next(const void *a)
 478 {
 479         return ((const struct commit_list *)a)->next;
 480 }
 481
 482 static void commit_list_set_next(void *a, void *next)
 483 {
 484         ((struct commit_list *)a)->next = next;
 485 }
 486
 487 void commit_list_sort_by_date(struct commit_list **list)
 488 {
 489         *list = llist_mergesort(*list, commit_list_get_next, commit_list_set_next,
 490                                 commit_list_compare_by_date);
 491 }
 492
 493 struct commit *pop_most_recent_commit(struct commit_list **list,
 494                                       unsigned int mark)
 495 {
 496         struct commit *ret = (*list)->item;
 497         struct commit_list *parents = ret->parents;
 498         struct commit_list *old = *list;
 499
 500         *list = (*list)->next;
 501         free(old);
 502
 503         while (parents) {
 504                 struct commit *commit = parents->item;
 505                 if (!parse_commit(commit) && !(commit->object.flags & mark)) {
 506                         commit->object.flags |= mark;
 507                         commit_list_insert_by_date(commit, list);
 508                 }
 509                 parents = parents->next;
 510         }
 511         return ret;
 512 }
 513
 514 static void clear_commit_marks_1(struct commit_list **plist,
 515                                  struct commit *commit, unsigned int mark)
 516 {
 517         while (commit) {
 518                 struct commit_list *parents;
 519
 520                 if (!(mark & commit->object.flags))
 521                         return;
 522
 523                 commit->object.flags &= ~mark;
 524
 525                 parents = commit->parents;
 526                 if (!parents)
 527                         return;
 528
 529                 while ((parents = parents->next))
 530                         commit_list_insert(parents->item, plist);
 531
 532                 commit = commit->parents->item;
 533         }
 534 }
 535
 536 void clear_commit_marks_many(int nr, struct commit **commit, unsigned int mark)
 537 {
 538         struct commit_list *list = NULL;
 539
 540         while (nr--) {
 541                 commit_list_insert(*commit, &list);
 542                 commit++;
 543         }
 544         while (list)
 545                 clear_commit_marks_1(&list, pop_commit(&list), mark);
 546 }
 547
 548 void clear_commit_marks(struct commit *commit, unsigned int mark)
 549 {
 550         clear_commit_marks_many(1, &commit, mark);
 551 }
 552
 553 void clear_commit_marks_for_object_array(struct object_array *a, unsigned mark)
 554 {
 555         struct object *object;
 556         struct commit *commit;
 557         unsigned int i;
 558
 559         for (i = 0; i < a->nr; i++) {
 560                 object = a->objects[i].item;
 561                 commit = lookup_commit_reference_gently(object->sha1, 1);
 562                 if (commit)
 563                         clear_commit_marks(commit, mark);
 564         }
 565 }
 566
 567 struct commit *pop_commit(struct commit_list **stack)
 568 {
 569         struct commit_list *top = *stack;
 570         struct commit *item = top ? top->item : NULL;
 571
 572         if (top) {
 573                 *stack = top->next;
 574                 free(top);
 575         }
 576         return item;
 577 }
 578
 579 /*
 580  * Topological sort support
 581  */
 582
 583 /* count number of children that have not been emitted */
 584 define_commit_slab(indegree_slab, int);
 585
 586 /* record author-date for each commit object */
 587 define_commit_slab(author_date_slab, unsigned long);
 588
 589 static void record_author_date(struct author_date_slab *author_date,
 590                                struct commit *commit)
 591 {
 592         const char *buf, *line_end, *ident_line;
 593         const char *buffer = get_commit_buffer(commit, NULL);
 594         struct ident_split ident;
 595         char *date_end;
 596         unsigned long date;
 597
 598         for (buf = buffer; buf; buf = line_end + 1) {
 599                 line_end = strchrnul(buf, '\n');
 600                 ident_line = skip_prefix(buf, "author ");
 601                 if (!ident_line) {
 602                         if (!line_end[0] || line_end[1] == '\n')
 603                                 return; /* end of header */
 604                         continue;
 605                 }
 606                 if (split_ident_line(&ident,
 607                                      ident_line, line_end - ident_line) ||
 608                     !ident.date_begin || !ident.date_end)
 609                         goto fail_exit; /* malformed "author" line */
 610                 break;
 611         }
 612
 613         date = strtoul(ident.date_begin, &date_end, 10);
 614         if (date_end != ident.date_end)
 615                 goto fail_exit; /* malformed date */
 616         *(author_date_slab_at(author_date, commit)) = date;
 617
 618 fail_exit:
 619         unuse_commit_buffer(commit, buffer);
 620 }
 621
 622 static int compare_commits_by_author_date(const void *a_, const void *b_,
 623                                           void *cb_data)
 624 {
 625         const struct commit *a = a_, *b = b_;
 626         struct author_date_slab *author_date = cb_data;
 627         unsigned long a_date = *(author_date_slab_at(author_date, a));
 628         unsigned long b_date = *(author_date_slab_at(author_date, b));
 629
 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 int compare_commits_by_commit_date(const void *a_, const void *b_, void *unused)
 639 {
 640         const struct commit *a = a_, *b = b_;
 641         /* newer commits with larger date first */
 642         if (a->date < b->date)
 643                 return 1;
 644         else if (a->date > b->date)
 645                 return -1;
 646         return 0;
 647 }
 648
 649 /*
 650  * Performs an in-place topological sort on the list supplied.
 651  */
 652 void sort_in_topological_order(struct commit_list **list, enum rev_sort_order sort_order)
 653 {
 654         struct commit_list *next, *orig = *list;
 655         struct commit_list **pptr;
 656         struct indegree_slab indegree;
 657         struct prio_queue queue;
 658         struct commit *commit;
 659         struct author_date_slab author_date;
 660
 661         if (!orig)
 662                 return;
 663         *list = NULL;
 664
 665         init_indegree_slab(&indegree);
 666         memset(&queue, '\0', sizeof(queue));
 667
 668         switch (sort_order) {
 669         default: /* REV_SORT_IN_GRAPH_ORDER */
 670                 queue.compare = NULL;
 671                 break;
 672         case REV_SORT_BY_COMMIT_DATE:
 673                 queue.compare = compare_commits_by_commit_date;
 674                 break;
 675         case REV_SORT_BY_AUTHOR_DATE:
 676                 init_author_date_slab(&author_date);
 677                 queue.compare = compare_commits_by_author_date;
 678                 queue.cb_data = &author_date;
 679                 break;
 680         }
 681
 682         /* Mark them and clear the indegree */
 683         for (next = orig; next; next = next->next) {
 684                 struct commit *commit = next->item;
 685                 *(indegree_slab_at(&indegree, commit)) = 1;
 686                 /* also record the author dates, if needed */
 687                 if (sort_order == REV_SORT_BY_AUTHOR_DATE)
 688                         record_author_date(&author_date, commit);
 689         }
 690
 691         /* update the indegree */
 692         for (next = orig; next; next = next->next) {
 693                 struct commit_list *parents = next->item->parents;
 694                 while (parents) {
 695                         struct commit *parent = parents->item;
 696                         int *pi = indegree_slab_at(&indegree, parent);
 697
 698                         if (*pi)
 699                                 (*pi)++;
 700                         parents = parents->next;
 701                 }
 702         }
 703
 704         /*
 705          * find the tips
 706          *
 707          * tips are nodes not reachable from any other node in the list
 708          *
 709          * the tips serve as a starting set for the work queue.
 710          */
 711         for (next = orig; next; next = next->next) {
 712                 struct commit *commit = next->item;
 713
 714                 if (*(indegree_slab_at(&indegree, commit)) == 1)
 715                         prio_queue_put(&queue, commit);
 716         }
 717
 718         /*
 719          * This is unfortunate; the initial tips need to be shown
 720          * in the order given from the revision traversal machinery.
 721          */
 722         if (sort_order == REV_SORT_IN_GRAPH_ORDER)
 723                 prio_queue_reverse(&queue);
 724
 725         /* We no longer need the commit list */
 726         free_commit_list(orig);
 727
 728         pptr = list;
 729         *list = NULL;
 730         while ((commit = prio_queue_get(&queue)) != NULL) {
 731                 struct commit_list *parents;
 732
 733                 for (parents = commit->parents; parents ; parents = parents->next) {
 734                         struct commit *parent = parents->item;
 735                         int *pi = indegree_slab_at(&indegree, parent);
 736
 737                         if (!*pi)
 738                                 continue;
 739
 740                         /*
 741                          * parents are only enqueued for emission
 742                          * when all their children have been emitted thereby
 743                          * guaranteeing topological order.
 744                          */
 745                         if (--(*pi) == 1)
 746                                 prio_queue_put(&queue, parent);
 747                 }
 748                 /*
 749                  * all children of commit have already been
 750                  * emitted. we can emit it now.
 751                  */
 752                 *(indegree_slab_at(&indegree, commit)) = 0;
 753
 754                 pptr = &commit_list_insert(commit, pptr)->next;
 755         }
 756
 757         clear_indegree_slab(&indegree);
 758         clear_prio_queue(&queue);
 759         if (sort_order == REV_SORT_BY_AUTHOR_DATE)
 760                 clear_author_date_slab(&author_date);
 761 }
 762
 763 /* merge-base stuff */
 764
 765 /* Remember to update object flag allocation in object.h */
 766 #define PARENT1         (1u<<16)
 767 #define PARENT2         (1u<<17)
 768 #define STALE           (1u<<18)
 769 #define RESULT          (1u<<19)
 770
 771 static const unsigned all_flags = (PARENT1 | PARENT2 | STALE | RESULT);
 772
 773 static struct commit *interesting(struct commit_list *list)
 774 {
 775         while (list) {
 776                 struct commit *commit = list->item;
 777                 list = list->next;
 778                 if (commit->object.flags & STALE)
 779                         continue;
 780                 return commit;
 781         }
 782         return NULL;
 783 }
 784
 785 /* all input commits in one and twos[] must have been parsed! */
 786 static struct commit_list *paint_down_to_common(struct commit *one, int n, struct commit **twos)
 787 {
 788         struct commit_list *list = NULL;
 789         struct commit_list *result = NULL;
 790         int i;
 791
 792         one->object.flags |= PARENT1;
 793         commit_list_insert_by_date(one, &list);
 794         if (!n)
 795                 return list;
 796         for (i = 0; i < n; i++) {
 797                 twos[i]->object.flags |= PARENT2;
 798                 commit_list_insert_by_date(twos[i], &list);
 799         }
 800
 801         while (interesting(list)) {
 802                 struct commit *commit;
 803                 struct commit_list *parents;
 804                 struct commit_list *next;
 805                 int flags;
 806
 807                 commit = list->item;
 808                 next = list->next;
 809                 free(list);
 810                 list = next;
 811
 812                 flags = commit->object.flags & (PARENT1 | PARENT2 | STALE);
 813                 if (flags == (PARENT1 | PARENT2)) {
 814                         if (!(commit->object.flags & RESULT)) {
 815                                 commit->object.flags |= RESULT;
 816                                 commit_list_insert_by_date(commit, &result);
 817                         }
 818                         /* Mark parents of a found merge stale */
 819                         flags |= STALE;
 820                 }
 821                 parents = commit->parents;
 822                 while (parents) {
 823                         struct commit *p = parents->item;
 824                         parents = parents->next;
 825                         if ((p->object.flags & flags) == flags)
 826                                 continue;
 827                         if (parse_commit(p))
 828                                 return NULL;
 829                         p->object.flags |= flags;
 830                         commit_list_insert_by_date(p, &list);
 831                 }
 832         }
 833
 834         free_commit_list(list);
 835         return result;
 836 }
 837
 838 static struct commit_list *merge_bases_many(struct commit *one, int n, struct commit **twos)
 839 {
 840         struct commit_list *list = NULL;
 841         struct commit_list *result = NULL;
 842         int i;
 843
 844         for (i = 0; i < n; i++) {
 845                 if (one == twos[i])
 846                         /*
 847                          * We do not mark this even with RESULT so we do not
 848                          * have to clean it up.
 849                          */
 850                         return commit_list_insert(one, &result);
 851         }
 852
 853         if (parse_commit(one))
 854                 return NULL;
 855         for (i = 0; i < n; i++) {
 856                 if (parse_commit(twos[i]))
 857                         return NULL;
 858         }
 859
 860         list = paint_down_to_common(one, n, twos);
 861
 862         while (list) {
 863                 struct commit_list *next = list->next;
 864                 if (!(list->item->object.flags & STALE))
 865                         commit_list_insert_by_date(list->item, &result);
 866                 free(list);
 867                 list = next;
 868         }
 869         return result;
 870 }
 871
 872 struct commit_list *get_octopus_merge_bases(struct commit_list *in)
 873 {
 874         struct commit_list *i, *j, *k, *ret = NULL;
 875
 876         if (!in)
 877                 return ret;
 878
 879         commit_list_insert(in->item, &ret);
 880
 881         for (i = in->next; i; i = i->next) {
 882                 struct commit_list *new = NULL, *end = NULL;
 883
 884                 for (j = ret; j; j = j->next) {
 885                         struct commit_list *bases;
 886                         bases = get_merge_bases(i->item, j->item, 1);
 887                         if (!new)
 888                                 new = bases;
 889                         else
 890                                 end->next = bases;
 891                         for (k = bases; k; k = k->next)
 892                                 end = k;
 893                 }
 894                 ret = new;
 895         }
 896         return ret;
 897 }
 898
 899 static int remove_redundant(struct commit **array, int cnt)
 900 {
 901         /*
 902          * Some commit in the array may be an ancestor of
 903          * another commit.  Move such commit to the end of
 904          * the array, and return the number of commits that
 905          * are independent from each other.
 906          */
 907         struct commit **work;
 908         unsigned char *redundant;
 909         int *filled_index;
 910         int i, j, filled;
 911
 912         work = xcalloc(cnt, sizeof(*work));
 913         redundant = xcalloc(cnt, 1);
 914         filled_index = xmalloc(sizeof(*filled_index) * (cnt - 1));
 915
 916         for (i = 0; i < cnt; i++)
 917                 parse_commit(array[i]);
 918         for (i = 0; i < cnt; i++) {
 919                 struct commit_list *common;
 920
 921                 if (redundant[i])
 922                         continue;
 923                 for (j = filled = 0; j < cnt; j++) {
 924                         if (i == j || redundant[j])
 925                                 continue;
 926                         filled_index[filled] = j;
 927                         work[filled++] = array[j];
 928                 }
 929                 common = paint_down_to_common(array[i], filled, work);
 930                 if (array[i]->object.flags & PARENT2)
 931                         redundant[i] = 1;
 932                 for (j = 0; j < filled; j++)
 933                         if (work[j]->object.flags & PARENT1)
 934                                 redundant[filled_index[j]] = 1;
 935                 clear_commit_marks(array[i], all_flags);
 936                 for (j = 0; j < filled; j++)
 937                         clear_commit_marks(work[j], all_flags);
 938                 free_commit_list(common);
 939         }
 940
 941         /* Now collect the result */
 942         memcpy(work, array, sizeof(*array) * cnt);
 943         for (i = filled = 0; i < cnt; i++)
 944                 if (!redundant[i])
 945                         array[filled++] = work[i];
 946         for (j = filled, i = 0; i < cnt; i++)
 947                 if (redundant[i])
 948                         array[j++] = work[i];
 949         free(work);
 950         free(redundant);
 951         free(filled_index);
 952         return filled;
 953 }
 954
 955 struct commit_list *get_merge_bases_many(struct commit *one,
 956                                          int n,
 957                                          struct commit **twos,
 958                                          int cleanup)
 959 {
 960         struct commit_list *list;
 961         struct commit **rslt;
 962         struct commit_list *result;
 963         int cnt, i;
 964
 965         result = merge_bases_many(one, n, twos);
 966         for (i = 0; i < n; i++) {
 967                 if (one == twos[i])
 968                         return result;
 969         }
 970         if (!result || !result->next) {
 971                 if (cleanup) {
 972                         clear_commit_marks(one, all_flags);
 973                         clear_commit_marks_many(n, twos, all_flags);
 974                 }
 975                 return result;
 976         }
 977
 978         /* There are more than one */
 979         cnt = 0;
 980         list = result;
 981         while (list) {
 982                 list = list->next;
 983                 cnt++;
 984         }
 985         rslt = xcalloc(cnt, sizeof(*rslt));
 986         for (list = result, i = 0; list; list = list->next)
 987                 rslt[i++] = list->item;
 988         free_commit_list(result);
 989
 990         clear_commit_marks(one, all_flags);
 991         clear_commit_marks_many(n, twos, all_flags);
 992
 993         cnt = remove_redundant(rslt, cnt);
 994         result = NULL;
 995         for (i = 0; i < cnt; i++)
 996                 commit_list_insert_by_date(rslt[i], &result);
 997         free(rslt);
 998         return result;
 999 }
1000
1001 struct commit_list *get_merge_bases(struct commit *one, struct commit *two,
1002                                     int cleanup)
1003 {
1004         return get_merge_bases_many(one, 1, &two, cleanup);
1005 }
1006
1007 /*
1008  * Is "commit" a descendant of one of the elements on the "with_commit" list?
1009  */
1010 int is_descendant_of(struct commit *commit, struct commit_list *with_commit)
1011 {
1012         if (!with_commit)
1013                 return 1;
1014         while (with_commit) {
1015                 struct commit *other;
1016
1017                 other = with_commit->item;
1018                 with_commit = with_commit->next;
1019                 if (in_merge_bases(other, commit))
1020                         return 1;
1021         }
1022         return 0;
1023 }
1024
1025 /*
1026  * Is "commit" an ancestor of one of the "references"?
1027  */
1028 int in_merge_bases_many(struct commit *commit, int nr_reference, struct commit **reference)
1029 {
1030         struct commit_list *bases;
1031         int ret = 0, i;
1032
1033         if (parse_commit(commit))
1034                 return ret;
1035         for (i = 0; i < nr_reference; i++)
1036                 if (parse_commit(reference[i]))
1037                         return ret;
1038
1039         bases = paint_down_to_common(commit, nr_reference, reference);
1040         if (commit->object.flags & PARENT2)
1041                 ret = 1;
1042         clear_commit_marks(commit, all_flags);
1043         clear_commit_marks_many(nr_reference, reference, all_flags);
1044         free_commit_list(bases);
1045         return ret;
1046 }
1047
1048 /*
1049  * Is "commit" an ancestor of (i.e. reachable from) the "reference"?
1050  */
1051 int in_merge_bases(struct commit *commit, struct commit *reference)
1052 {
1053         return in_merge_bases_many(commit, 1, &reference);
1054 }
1055
1056 struct commit_list *reduce_heads(struct commit_list *heads)
1057 {
1058         struct commit_list *p;
1059         struct commit_list *result = NULL, **tail = &result;
1060         struct commit **array;
1061         int num_head, i;
1062
1063         if (!heads)
1064                 return NULL;
1065
1066         /* Uniquify */
1067         for (p = heads; p; p = p->next)
1068                 p->item->object.flags &= ~STALE;
1069         for (p = heads, num_head = 0; p; p = p->next) {
1070                 if (p->item->object.flags & STALE)
1071                         continue;
1072                 p->item->object.flags |= STALE;
1073                 num_head++;
1074         }
1075         array = xcalloc(num_head, sizeof(*array));
1076         for (p = heads, i = 0; p; p = p->next) {
1077                 if (p->item->object.flags & STALE) {
1078                         array[i++] = p->item;
1079                         p->item->object.flags &= ~STALE;
1080                 }
1081         }
1082         num_head = remove_redundant(array, num_head);
1083         for (i = 0; i < num_head; i++)
1084                 tail = &commit_list_insert(array[i], tail)->next;
1085         return result;
1086 }
1087
1088 static const char gpg_sig_header[] = "gpgsig";
1089 static const int gpg_sig_header_len = sizeof(gpg_sig_header) - 1;
1090
1091 static int do_sign_commit(struct strbuf *buf, const char *keyid)
1092 {
1093         struct strbuf sig = STRBUF_INIT;
1094         int inspos, copypos;
1095
1096         /* find the end of the header */
1097         inspos = strstr(buf->buf, "\n\n") - buf->buf + 1;
1098
1099         if (!keyid || !*keyid)
1100                 keyid = get_signing_key();
1101         if (sign_buffer(buf, &sig, keyid)) {
1102                 strbuf_release(&sig);
1103                 return -1;
1104         }
1105
1106         for (copypos = 0; sig.buf[copypos]; ) {
1107                 const char *bol = sig.buf + copypos;
1108                 const char *eol = strchrnul(bol, '\n');
1109                 int len = (eol - bol) + !!*eol;
1110
1111                 if (!copypos) {
1112                         strbuf_insert(buf, inspos, gpg_sig_header, gpg_sig_header_len);
1113                         inspos += gpg_sig_header_len;
1114                 }
1115                 strbuf_insert(buf, inspos++, " ", 1);
1116                 strbuf_insert(buf, inspos, bol, len);
1117                 inspos += len;
1118                 copypos += len;
1119         }
1120         strbuf_release(&sig);
1121         return 0;
1122 }
1123
1124 int parse_signed_commit(const struct commit *commit,
1125                         struct strbuf *payload, struct strbuf *signature)
1126 {
1127
1128         unsigned long size;
1129         const char *buffer = get_commit_buffer(commit, &size);
1130         int in_signature, saw_signature = -1;
1131         const char *line, *tail;
1132
1133         line = buffer;
1134         tail = buffer + size;
1135         in_signature = 0;
1136         saw_signature = 0;
1137         while (line < tail) {
1138                 const char *sig = NULL;
1139                 const char *next = memchr(line, '\n', tail - line);
1140
1141                 next = next ? next + 1 : tail;
1142                 if (in_signature && line[0] == ' ')
1143                         sig = line + 1;
1144                 else if (starts_with(line, gpg_sig_header) &&
1145                          line[gpg_sig_header_len] == ' ')
1146                         sig = line + gpg_sig_header_len + 1;
1147                 if (sig) {
1148                         strbuf_add(signature, sig, next - sig);
1149                         saw_signature = 1;
1150                         in_signature = 1;
1151                 } else {
1152                         if (*line == '\n')
1153                                 /* dump the whole remainder of the buffer */
1154                                 next = tail;
1155                         strbuf_add(payload, line, next - line);
1156                         in_signature = 0;
1157                 }
1158                 line = next;
1159         }
1160         unuse_commit_buffer(commit, buffer);
1161         return saw_signature;
1162 }
1163
1164 static void handle_signed_tag(struct commit *parent, struct commit_extra_header ***tail)
1165 {
1166         struct merge_remote_desc *desc;
1167         struct commit_extra_header *mergetag;
1168         char *buf;
1169         unsigned long size, len;
1170         enum object_type type;
1171
1172         desc = merge_remote_util(parent);
1173         if (!desc || !desc->obj)
1174                 return;
1175         buf = read_sha1_file(desc->obj->sha1, &type, &size);
1176         if (!buf || type != OBJ_TAG)
1177                 goto free_return;
1178         len = parse_signature(buf, size);
1179         if (size == len)
1180                 goto free_return;
1181         /*
1182          * We could verify this signature and either omit the tag when
1183          * it does not validate, but the integrator may not have the
1184          * public key of the signer of the tag he is merging, while a
1185          * later auditor may have it while auditing, so let's not run
1186          * verify-signed-buffer here for now...
1187          *
1188          * if (verify_signed_buffer(buf, len, buf + len, size - len, ...))
1189          *      warn("warning: signed tag unverified.");
1190          */
1191         mergetag = xcalloc(1, sizeof(*mergetag));
1192         mergetag->key = xstrdup("mergetag");
1193         mergetag->value = buf;
1194         mergetag->len = size;
1195
1196         **tail = mergetag;
1197         *tail = &mergetag->next;
1198         return;
1199
1200 free_return:
1201         free(buf);
1202 }
1203
1204 static struct {
1205         char result;
1206         const char *check;
1207 } sigcheck_gpg_status[] = {
1208         { 'G', "\n[GNUPG:] GOODSIG " },
1209         { 'B', "\n[GNUPG:] BADSIG " },
1210         { 'U', "\n[GNUPG:] TRUST_NEVER" },
1211         { 'U', "\n[GNUPG:] TRUST_UNDEFINED" },
1212 };
1213
1214 static void parse_gpg_output(struct signature_check *sigc)
1215 {
1216         const char *buf = sigc->gpg_status;
1217         int i;
1218
1219         /* Iterate over all search strings */
1220         for (i = 0; i < ARRAY_SIZE(sigcheck_gpg_status); i++) {
1221                 const char *found, *next;
1222
1223                 found = skip_prefix(buf, sigcheck_gpg_status[i].check + 1);
1224                 if (!found) {
1225                         found = strstr(buf, sigcheck_gpg_status[i].check);
1226                         if (!found)
1227                                 continue;
1228                         found += strlen(sigcheck_gpg_status[i].check);
1229                 }
1230                 sigc->result = sigcheck_gpg_status[i].result;
1231                 /* The trust messages are not followed by key/signer information */
1232                 if (sigc->result != 'U') {
1233                         sigc->key = xmemdupz(found, 16);
1234                         found += 17;
1235                         next = strchrnul(found, '\n');
1236                         sigc->signer = xmemdupz(found, next - found);
1237                 }
1238         }
1239 }
1240
1241 void 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         struct strbuf gpg_output = STRBUF_INIT;
1246         struct strbuf gpg_status = STRBUF_INIT;
1247         int status;
1248
1249         sigc->result = 'N';
1250
1251         if (parse_signed_commit(commit, &payload, &signature) <= 0)
1252                 goto out;
1253         status = verify_signed_buffer(payload.buf, payload.len,
1254                                       signature.buf, signature.len,
1255                                       &gpg_output, &gpg_status);
1256         if (status && !gpg_output.len)
1257                 goto out;
1258         sigc->gpg_output = strbuf_detach(&gpg_output, NULL);
1259         sigc->gpg_status = strbuf_detach(&gpg_status, NULL);
1260         parse_gpg_output(sigc);
1261
1262  out:
1263         strbuf_release(&gpg_status);
1264         strbuf_release(&gpg_output);
1265         strbuf_release(&payload);
1266         strbuf_release(&signature);
1267 }
1268
1269
1270
1271 void append_merge_tag_headers(struct commit_list *parents,
1272                               struct commit_extra_header ***tail)
1273 {
1274         while (parents) {
1275                 struct commit *parent = parents->item;
1276                 handle_signed_tag(parent, tail);
1277                 parents = parents->next;
1278         }
1279 }
1280
1281 static void add_extra_header(struct strbuf *buffer,
1282                              struct commit_extra_header *extra)
1283 {
1284         strbuf_addstr(buffer, extra->key);
1285         if (extra->len)
1286                 strbuf_add_lines(buffer, " ", extra->value, extra->len);
1287         else
1288                 strbuf_addch(buffer, '\n');
1289 }
1290
1291 struct commit_extra_header *read_commit_extra_headers(struct commit *commit,
1292                                                       const char **exclude)
1293 {
1294         struct commit_extra_header *extra = NULL;
1295         unsigned long size;
1296         const char *buffer = get_commit_buffer(commit, &size);
1297         extra = read_commit_extra_header_lines(buffer, size, exclude);
1298         unuse_commit_buffer(commit, buffer);
1299         return extra;
1300 }
1301
1302 static inline int standard_header_field(const char *field, size_t len)
1303 {
1304         return ((len == 4 && !memcmp(field, "tree ", 5)) ||
1305                 (len == 6 && !memcmp(field, "parent ", 7)) ||
1306                 (len == 6 && !memcmp(field, "author ", 7)) ||
1307                 (len == 9 && !memcmp(field, "committer ", 10)) ||
1308                 (len == 8 && !memcmp(field, "encoding ", 9)));
1309 }
1310
1311 static int excluded_header_field(const char *field, size_t len, const char **exclude)
1312 {
1313         if (!exclude)
1314                 return 0;
1315
1316         while (*exclude) {
1317                 size_t xlen = strlen(*exclude);
1318                 if (len == xlen &&
1319                     !memcmp(field, *exclude, xlen) && field[xlen] == ' ')
1320                         return 1;
1321                 exclude++;
1322         }
1323         return 0;
1324 }
1325
1326 static struct commit_extra_header *read_commit_extra_header_lines(
1327         const char *buffer, size_t size,
1328         const char **exclude)
1329 {
1330         struct commit_extra_header *extra = NULL, **tail = &extra, *it = NULL;
1331         const char *line, *next, *eof, *eob;
1332         struct strbuf buf = STRBUF_INIT;
1333
1334         for (line = buffer, eob = line + size;
1335              line < eob && *line != '\n';
1336              line = next) {
1337                 next = memchr(line, '\n', eob - line);
1338                 next = next ? next + 1 : eob;
1339                 if (*line == ' ') {
1340                         /* continuation */
1341                         if (it)
1342                                 strbuf_add(&buf, line + 1, next - (line + 1));
1343                         continue;
1344                 }
1345                 if (it)
1346                         it->value = strbuf_detach(&buf, &it->len);
1347                 strbuf_reset(&buf);
1348                 it = NULL;
1349
1350                 eof = strchr(line, ' ');
1351                 if (next <= eof)
1352                         eof = next;
1353
1354                 if (standard_header_field(line, eof - line) ||
1355                     excluded_header_field(line, eof - line, exclude))
1356                         continue;
1357
1358                 it = xcalloc(1, sizeof(*it));
1359                 it->key = xmemdupz(line, eof-line);
1360                 *tail = it;
1361                 tail = &it->next;
1362                 if (eof + 1 < next)
1363                         strbuf_add(&buf, eof + 1, next - (eof + 1));
1364         }
1365         if (it)
1366                 it->value = strbuf_detach(&buf, &it->len);
1367         return extra;
1368 }
1369
1370 void free_commit_extra_headers(struct commit_extra_header *extra)
1371 {
1372         while (extra) {
1373                 struct commit_extra_header *next = extra->next;
1374                 free(extra->key);
1375                 free(extra->value);
1376                 free(extra);
1377                 extra = next;
1378         }
1379 }
1380
1381 int commit_tree(const char *msg, size_t msg_len,
1382                 const unsigned char *tree,
1383                 struct commit_list *parents, unsigned char *ret,
1384                 const char *author, const char *sign_commit)
1385 {
1386         struct commit_extra_header *extra = NULL, **tail = &extra;
1387         int result;
1388
1389         append_merge_tag_headers(parents, &tail);
1390         result = commit_tree_extended(msg, msg_len, tree, parents, ret,
1391                                       author, sign_commit, extra);
1392         free_commit_extra_headers(extra);
1393         return result;
1394 }
1395
1396 static int find_invalid_utf8(const char *buf, int len)
1397 {
1398         int offset = 0;
1399         static const unsigned int max_codepoint[] = {
1400                 0x7f, 0x7ff, 0xffff, 0x10ffff
1401         };
1402
1403         while (len) {
1404                 unsigned char c = *buf++;
1405                 int bytes, bad_offset;
1406                 unsigned int codepoint;
1407                 unsigned int min_val, max_val;
1408
1409                 len--;
1410                 offset++;
1411
1412                 /* Simple US-ASCII? No worries. */
1413                 if (c < 0x80)
1414                         continue;
1415
1416                 bad_offset = offset-1;
1417
1418                 /*
1419                  * Count how many more high bits set: that's how
1420                  * many more bytes this sequence should have.
1421                  */
1422                 bytes = 0;
1423                 while (c & 0x40) {
1424                         c <<= 1;
1425                         bytes++;
1426                 }
1427
1428                 /*
1429                  * Must be between 1 and 3 more bytes.  Longer sequences result in
1430                  * codepoints beyond U+10FFFF, which are guaranteed never to exist.
1431                  */
1432                 if (bytes < 1 || 3 < bytes)
1433                         return bad_offset;
1434
1435                 /* Do we *have* that many bytes? */
1436                 if (len < bytes)
1437                         return bad_offset;
1438
1439                 /*
1440                  * Place the encoded bits at the bottom of the value and compute the
1441                  * valid range.
1442                  */
1443                 codepoint = (c & 0x7f) >> bytes;
1444                 min_val = max_codepoint[bytes-1] + 1;
1445                 max_val = max_codepoint[bytes];
1446
1447                 offset += bytes;
1448                 len -= bytes;
1449
1450                 /* And verify that they are good continuation bytes */
1451                 do {
1452                         codepoint <<= 6;
1453                         codepoint |= *buf & 0x3f;
1454                         if ((*buf++ & 0xc0) != 0x80)
1455                                 return bad_offset;
1456                 } while (--bytes);
1457
1458                 /* Reject codepoints that are out of range for the sequence length. */
1459                 if (codepoint < min_val || codepoint > max_val)
1460                         return bad_offset;
1461                 /* Surrogates are only for UTF-16 and cannot be encoded in UTF-8. */
1462                 if ((codepoint & 0x1ff800) == 0xd800)
1463                         return bad_offset;
1464                 /* U+xxFFFE and U+xxFFFF are guaranteed non-characters. */
1465                 if ((codepoint & 0xfffe) == 0xfffe)
1466                         return bad_offset;
1467                 /* So are anything in the range U+FDD0..U+FDEF. */
1468                 if (codepoint >= 0xfdd0 && codepoint <= 0xfdef)
1469                         return bad_offset;
1470         }
1471         return -1;
1472 }
1473
1474 /*
1475  * This verifies that the buffer is in proper utf8 format.
1476  *
1477  * If it isn't, it assumes any non-utf8 characters are Latin1,
1478  * and does the conversion.
1479  */
1480 static int verify_utf8(struct strbuf *buf)
1481 {
1482         int ok = 1;
1483         long pos = 0;
1484
1485         for (;;) {
1486                 int bad;
1487                 unsigned char c;
1488                 unsigned char replace[2];
1489
1490                 bad = find_invalid_utf8(buf->buf + pos, buf->len - pos);
1491                 if (bad < 0)
1492                         return ok;
1493                 pos += bad;
1494                 ok = 0;
1495                 c = buf->buf[pos];
1496                 strbuf_remove(buf, pos, 1);
1497
1498                 /* We know 'c' must be in the range 128-255 */
1499                 replace[0] = 0xc0 + (c >> 6);
1500                 replace[1] = 0x80 + (c & 0x3f);
1501                 strbuf_insert(buf, pos, replace, 2);
1502                 pos += 2;
1503         }
1504 }
1505
1506 static const char commit_utf8_warn[] =
1507 "Warning: commit message did not conform to UTF-8.\n"
1508 "You may want to amend it after fixing the message, or set the config\n"
1509 "variable i18n.commitencoding to the encoding your project uses.\n";
1510
1511 int commit_tree_extended(const char *msg, size_t msg_len,
1512                          const unsigned char *tree,
1513                          struct commit_list *parents, unsigned char *ret,
1514                          const char *author, const char *sign_commit,
1515                          struct commit_extra_header *extra)
1516 {
1517         int result;
1518         int encoding_is_utf8;
1519         struct strbuf buffer;
1520
1521         assert_sha1_type(tree, OBJ_TREE);
1522
1523         if (memchr(msg, '\0', msg_len))
1524                 return error("a NUL byte in commit log message not allowed.");
1525
1526         /* Not having i18n.commitencoding is the same as having utf-8 */
1527         encoding_is_utf8 = is_encoding_utf8(git_commit_encoding);
1528
1529         strbuf_init(&buffer, 8192); /* should avoid reallocs for the headers */
1530         strbuf_addf(&buffer, "tree %s\n", sha1_to_hex(tree));
1531
1532         /*
1533          * NOTE! This ordering means that the same exact tree merged with a
1534          * different order of parents will be a _different_ changeset even
1535          * if everything else stays the same.
1536          */
1537         while (parents) {
1538                 struct commit_list *next = parents->next;
1539                 struct commit *parent = parents->item;
1540
1541                 strbuf_addf(&buffer, "parent %s\n",
1542                             sha1_to_hex(parent->object.sha1));
1543                 free(parents);
1544                 parents = next;
1545         }
1546
1547         /* Person/date information */
1548         if (!author)
1549                 author = git_author_info(IDENT_STRICT);
1550         strbuf_addf(&buffer, "author %s\n", author);
1551         strbuf_addf(&buffer, "committer %s\n", git_committer_info(IDENT_STRICT));
1552         if (!encoding_is_utf8)
1553                 strbuf_addf(&buffer, "encoding %s\n", git_commit_encoding);
1554
1555         while (extra) {
1556                 add_extra_header(&buffer, extra);
1557                 extra = extra->next;
1558         }
1559         strbuf_addch(&buffer, '\n');
1560
1561         /* And add the comment */
1562         strbuf_add(&buffer, msg, msg_len);
1563
1564         /* And check the encoding */
1565         if (encoding_is_utf8 && !verify_utf8(&buffer))
1566                 fprintf(stderr, commit_utf8_warn);
1567
1568         if (sign_commit && do_sign_commit(&buffer, sign_commit))
1569                 return -1;
1570
1571         result = write_sha1_file(buffer.buf, buffer.len, commit_type, ret);
1572         strbuf_release(&buffer);
1573         return result;
1574 }
1575
1576 struct commit *get_merge_parent(const char *name)
1577 {
1578         struct object *obj;
1579         struct commit *commit;
1580         unsigned char sha1[20];
1581         if (get_sha1(name, sha1))
1582                 return NULL;
1583         obj = parse_object(sha1);
1584         commit = (struct commit *)peel_to_type(name, 0, obj, OBJ_COMMIT);
1585         if (commit && !commit->util) {
1586                 struct merge_remote_desc *desc;
1587                 desc = xmalloc(sizeof(*desc));
1588                 desc->obj = obj;
1589                 desc->name = strdup(name);
1590                 commit->util = desc;
1591         }
1592         return commit;
1593 }
1594
1595 /*
1596  * Append a commit to the end of the commit_list.
1597  *
1598  * next starts by pointing to the variable that holds the head of an
1599  * empty commit_list, and is updated to point to the "next" field of
1600  * the last item on the list as new commits are appended.
1601  *
1602  * Usage example:
1603  *
1604  *     struct commit_list *list;
1605  *     struct commit_list **next = &list;
1606  *
1607  *     next = commit_list_append(c1, next);
1608  *     next = commit_list_append(c2, next);
1609  *     assert(commit_list_count(list) == 2);
1610  *     return list;
1611  */
1612 struct commit_list **commit_list_append(struct commit *commit,
1613                                         struct commit_list **next)
1614 {
1615         struct commit_list *new = xmalloc(sizeof(struct commit_list));
1616         new->item = commit;
1617         *next = new;
1618         new->next = NULL;
1619         return &new->next;
1620 }
1621
1622 void print_commit_list(struct commit_list *list,
1623                        const char *format_cur,
1624                        const char *format_last)
1625 {
1626         for ( ; list; list = list->next) {
1627                 const char *format = list->next ? format_cur : format_last;
1628                 printf(format, sha1_to_hex(list->item->object.sha1));
1629         }
1630 }