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         struct object_id oid;
  59         struct commit *commit;
  60
  61         if (get_sha1_committish(name, oid.hash))
  62                 return NULL;
  63         commit = lookup_commit_reference(oid.hash);
  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]->oid.hash;
 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->oid.hash);
 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         const int entry_size = GIT_SHA1_HEXSZ + 1;
 142
 143         while (len && isspace(buf[len-1]))
 144                 buf[--len] = '\0';
 145         if (buf[0] == '#' || buf[0] == '\0')
 146                 return NULL;
 147         if ((len + 1) % entry_size)
 148                 goto bad_graft_data;
 149         i = (len + 1) / entry_size - 1;
 150         graft = xmalloc(sizeof(*graft) + GIT_SHA1_RAWSZ * i);
 151         graft->nr_parent = i;
 152         if (get_oid_hex(buf, &graft->oid))
 153                 goto bad_graft_data;
 154         for (i = GIT_SHA1_HEXSZ; i < len; i += entry_size) {
 155                 if (buf[i] != ' ')
 156                         goto bad_graft_data;
 157                 if (get_sha1_hex(buf + i + 1, graft->parent[i/entry_size].hash))
 158                         goto bad_graft_data;
 159         }
 160         return graft;
 161
 162 bad_graft_data:
 163         error("bad graft data: %s", buf);
 164         free(graft);
 165         return NULL;
 166 }
 167
 168 static int read_graft_file(const char *graft_file)
 169 {
 170         FILE *fp = fopen(graft_file, "r");
 171         struct strbuf buf = STRBUF_INIT;
 172         if (!fp)
 173                 return -1;
 174         while (!strbuf_getwholeline(&buf, fp, '\n')) {
 175                 /* The format is just "Commit Parent1 Parent2 ...\n" */
 176                 struct commit_graft *graft = read_graft_line(buf.buf, buf.len);
 177                 if (!graft)
 178                         continue;
 179                 if (register_commit_graft(graft, 1))
 180                         error("duplicate graft data: %s", buf.buf);
 181         }
 182         fclose(fp);
 183         strbuf_release(&buf);
 184         return 0;
 185 }
 186
 187 static void prepare_commit_graft(void)
 188 {
 189         static int commit_graft_prepared;
 190         char *graft_file;
 191
 192         if (commit_graft_prepared)
 193                 return;
 194         graft_file = get_graft_file();
 195         read_graft_file(graft_file);
 196         /* make sure shallows are read */
 197         is_repository_shallow();
 198         commit_graft_prepared = 1;
 199 }
 200
 201 struct commit_graft *lookup_commit_graft(const unsigned char *sha1)
 202 {
 203         int pos;
 204         prepare_commit_graft();
 205         pos = commit_graft_pos(sha1);
 206         if (pos < 0)
 207                 return NULL;
 208         return commit_graft[pos];
 209 }
 210
 211 int for_each_commit_graft(each_commit_graft_fn fn, void *cb_data)
 212 {
 213         int i, ret;
 214         for (i = ret = 0; i < commit_graft_nr && !ret; i++)
 215                 ret = fn(commit_graft[i], cb_data);
 216         return ret;
 217 }
 218
 219 int unregister_shallow(const unsigned char *sha1)
 220 {
 221         int pos = commit_graft_pos(sha1);
 222         if (pos < 0)
 223                 return -1;
 224         if (pos + 1 < commit_graft_nr)
 225                 memmove(commit_graft + pos, commit_graft + pos + 1,
 226                                 sizeof(struct commit_graft *)
 227                                 * (commit_graft_nr - pos - 1));
 228         commit_graft_nr--;
 229         return 0;
 230 }
 231
 232 struct commit_buffer {
 233         void *buffer;
 234         unsigned long size;
 235 };
 236 define_commit_slab(buffer_slab, struct commit_buffer);
 237 static struct buffer_slab buffer_slab = COMMIT_SLAB_INIT(1, buffer_slab);
 238
 239 void set_commit_buffer(struct commit *commit, void *buffer, unsigned long size)
 240 {
 241         struct commit_buffer *v = buffer_slab_at(&buffer_slab, commit);
 242         v->buffer = buffer;
 243         v->size = size;
 244 }
 245
 246 const void *get_cached_commit_buffer(const struct commit *commit, unsigned long *sizep)
 247 {
 248         struct commit_buffer *v = buffer_slab_at(&buffer_slab, commit);
 249         if (sizep)
 250                 *sizep = v->size;
 251         return v->buffer;
 252 }
 253
 254 const void *get_commit_buffer(const struct commit *commit, unsigned long *sizep)
 255 {
 256         const void *ret = get_cached_commit_buffer(commit, sizep);
 257         if (!ret) {
 258                 enum object_type type;
 259                 unsigned long size;
 260                 ret = read_sha1_file(commit->object.sha1, &type, &size);
 261                 if (!ret)
 262                         die("cannot read commit object %s",
 263                             sha1_to_hex(commit->object.sha1));
 264                 if (type != OBJ_COMMIT)
 265                         die("expected commit for %s, got %s",
 266                             sha1_to_hex(commit->object.sha1), typename(type));
 267                 if (sizep)
 268                         *sizep = size;
 269         }
 270         return ret;
 271 }
 272
 273 void unuse_commit_buffer(const struct commit *commit, const void *buffer)
 274 {
 275         struct commit_buffer *v = buffer_slab_at(&buffer_slab, commit);
 276         if (v->buffer != buffer)
 277                 free((void *)buffer);
 278 }
 279
 280 void free_commit_buffer(struct commit *commit)
 281 {
 282         struct commit_buffer *v = buffer_slab_at(&buffer_slab, commit);
 283         free(v->buffer);
 284         v->buffer = NULL;
 285         v->size = 0;
 286 }
 287
 288 const void *detach_commit_buffer(struct commit *commit, unsigned long *sizep)
 289 {
 290         struct commit_buffer *v = buffer_slab_at(&buffer_slab, commit);
 291         void *ret;
 292
 293         ret = v->buffer;
 294         if (sizep)
 295                 *sizep = v->size;
 296
 297         v->buffer = NULL;
 298         v->size = 0;
 299         return ret;
 300 }
 301
 302 int parse_commit_buffer(struct commit *item, const void *buffer, unsigned long size)
 303 {
 304         const char *tail = buffer;
 305         const char *bufptr = buffer;
 306         struct object_id parent;
 307         struct commit_list **pptr;
 308         struct commit_graft *graft;
 309         const int tree_entry_len = GIT_SHA1_HEXSZ + 5;
 310         const int parent_entry_len = GIT_SHA1_HEXSZ + 7;
 311
 312         if (item->object.parsed)
 313                 return 0;
 314         item->object.parsed = 1;
 315         tail += size;
 316         if (tail <= bufptr + tree_entry_len + 1 || memcmp(bufptr, "tree ", 5) ||
 317                         bufptr[tree_entry_len] != '\n')
 318                 return error("bogus commit object %s", sha1_to_hex(item->object.sha1));
 319         if (get_sha1_hex(bufptr + 5, parent.hash) < 0)
 320                 return error("bad tree pointer in commit %s",
 321                              sha1_to_hex(item->object.sha1));
 322         item->tree = lookup_tree(parent.hash);
 323         bufptr += tree_entry_len + 1; /* "tree " + "hex sha1" + "\n" */
 324         pptr = &item->parents;
 325
 326         graft = lookup_commit_graft(item->object.sha1);
 327         while (bufptr + parent_entry_len < tail && !memcmp(bufptr, "parent ", 7)) {
 328                 struct commit *new_parent;
 329
 330                 if (tail <= bufptr + parent_entry_len + 1 ||
 331                     get_sha1_hex(bufptr + 7, parent.hash) ||
 332                     bufptr[parent_entry_len] != '\n')
 333                         return error("bad parents in commit %s", sha1_to_hex(item->object.sha1));
 334                 bufptr += parent_entry_len + 1;
 335                 /*
 336                  * The clone is shallow if nr_parent < 0, and we must
 337                  * not traverse its real parents even when we unhide them.
 338                  */
 339                 if (graft && (graft->nr_parent < 0 || grafts_replace_parents))
 340                         continue;
 341                 new_parent = lookup_commit(parent.hash);
 342                 if (new_parent)
 343                         pptr = &commit_list_insert(new_parent, pptr)->next;
 344         }
 345         if (graft) {
 346                 int i;
 347                 struct commit *new_parent;
 348                 for (i = 0; i < graft->nr_parent; i++) {
 349                         new_parent = lookup_commit(graft->parent[i].hash);
 350                         if (!new_parent)
 351                                 continue;
 352                         pptr = &commit_list_insert(new_parent, pptr)->next;
 353                 }
 354         }
 355         item->date = parse_commit_date(bufptr, tail);
 356
 357         return 0;
 358 }
 359
 360 int parse_commit_gently(struct commit *item, int quiet_on_missing)
 361 {
 362         enum object_type type;
 363         void *buffer;
 364         unsigned long size;
 365         int ret;
 366
 367         if (!item)
 368                 return -1;
 369         if (item->object.parsed)
 370                 return 0;
 371         buffer = read_sha1_file(item->object.sha1, &type, &size);
 372         if (!buffer)
 373                 return quiet_on_missing ? -1 :
 374                         error("Could not read %s",
 375                              sha1_to_hex(item->object.sha1));
 376         if (type != OBJ_COMMIT) {
 377                 free(buffer);
 378                 return error("Object %s not a commit",
 379                              sha1_to_hex(item->object.sha1));
 380         }
 381         ret = parse_commit_buffer(item, buffer, size);
 382         if (save_commit_buffer && !ret) {
 383                 set_commit_buffer(item, buffer, size);
 384                 return 0;
 385         }
 386         free(buffer);
 387         return ret;
 388 }
 389
 390 void parse_commit_or_die(struct commit *item)
 391 {
 392         if (parse_commit(item))
 393                 die("unable to parse commit %s",
 394                     item ? sha1_to_hex(item->object.sha1) : "(null)");
 395 }
 396
 397 int find_commit_subject(const char *commit_buffer, const char **subject)
 398 {
 399         const char *eol;
 400         const char *p = commit_buffer;
 401
 402         while (*p && (*p != '\n' || p[1] != '\n'))
 403                 p++;
 404         if (*p) {
 405                 p += 2;
 406                 for (eol = p; *eol && *eol != '\n'; eol++)
 407                         ; /* do nothing */
 408         } else
 409                 eol = p;
 410
 411         *subject = p;
 412
 413         return eol - p;
 414 }
 415
 416 struct commit_list *commit_list_insert(struct commit *item, struct commit_list **list_p)
 417 {
 418         struct commit_list *new_list = xmalloc(sizeof(struct commit_list));
 419         new_list->item = item;
 420         new_list->next = *list_p;
 421         *list_p = new_list;
 422         return new_list;
 423 }
 424
 425 unsigned commit_list_count(const struct commit_list *l)
 426 {
 427         unsigned c = 0;
 428         for (; l; l = l->next )
 429                 c++;
 430         return c;
 431 }
 432
 433 struct commit_list *copy_commit_list(struct commit_list *list)
 434 {
 435         struct commit_list *head = NULL;
 436         struct commit_list **pp = &head;
 437         while (list) {
 438                 pp = commit_list_append(list->item, pp);
 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 *buffer = get_commit_buffer(commit, NULL);
 593         struct ident_split ident;
 594         const char *ident_line;
 595         size_t ident_len;
 596         char *date_end;
 597         unsigned long date;
 598
 599         ident_line = find_commit_header(buffer, "author", &ident_len);
 600         if (!ident_line)
 601                 goto fail_exit; /* no author line */
 602         if (split_ident_line(&ident, ident_line, ident_len) ||
 603             !ident.date_begin || !ident.date_end)
 604                 goto fail_exit; /* malformed "author" line */
 605
 606         date = strtoul(ident.date_begin, &date_end, 10);
 607         if (date_end != ident.date_end)
 608                 goto fail_exit; /* malformed date */
 609         *(author_date_slab_at(author_date, commit)) = date;
 610
 611 fail_exit:
 612         unuse_commit_buffer(commit, buffer);
 613 }
 614
 615 static int compare_commits_by_author_date(const void *a_, const void *b_,
 616                                           void *cb_data)
 617 {
 618         const struct commit *a = a_, *b = b_;
 619         struct author_date_slab *author_date = cb_data;
 620         unsigned long a_date = *(author_date_slab_at(author_date, a));
 621         unsigned long b_date = *(author_date_slab_at(author_date, b));
 622
 623         /* newer commits with larger date first */
 624         if (a_date < b_date)
 625                 return 1;
 626         else if (a_date > b_date)
 627                 return -1;
 628         return 0;
 629 }
 630
 631 int compare_commits_by_commit_date(const void *a_, const void *b_, void *unused)
 632 {
 633         const struct commit *a = a_, *b = b_;
 634         /* newer commits with larger date first */
 635         if (a->date < b->date)
 636                 return 1;
 637         else if (a->date > b->date)
 638                 return -1;
 639         return 0;
 640 }
 641
 642 /*
 643  * Performs an in-place topological sort on the list supplied.
 644  */
 645 void sort_in_topological_order(struct commit_list **list, enum rev_sort_order sort_order)
 646 {
 647         struct commit_list *next, *orig = *list;
 648         struct commit_list **pptr;
 649         struct indegree_slab indegree;
 650         struct prio_queue queue;
 651         struct commit *commit;
 652         struct author_date_slab author_date;
 653
 654         if (!orig)
 655                 return;
 656         *list = NULL;
 657
 658         init_indegree_slab(&indegree);
 659         memset(&queue, '\0', sizeof(queue));
 660
 661         switch (sort_order) {
 662         default: /* REV_SORT_IN_GRAPH_ORDER */
 663                 queue.compare = NULL;
 664                 break;
 665         case REV_SORT_BY_COMMIT_DATE:
 666                 queue.compare = compare_commits_by_commit_date;
 667                 break;
 668         case REV_SORT_BY_AUTHOR_DATE:
 669                 init_author_date_slab(&author_date);
 670                 queue.compare = compare_commits_by_author_date;
 671                 queue.cb_data = &author_date;
 672                 break;
 673         }
 674
 675         /* Mark them and clear the indegree */
 676         for (next = orig; next; next = next->next) {
 677                 struct commit *commit = next->item;
 678                 *(indegree_slab_at(&indegree, commit)) = 1;
 679                 /* also record the author dates, if needed */
 680                 if (sort_order == REV_SORT_BY_AUTHOR_DATE)
 681                         record_author_date(&author_date, commit);
 682         }
 683
 684         /* update the indegree */
 685         for (next = orig; next; next = next->next) {
 686                 struct commit_list *parents = next->item->parents;
 687                 while (parents) {
 688                         struct commit *parent = parents->item;
 689                         int *pi = indegree_slab_at(&indegree, parent);
 690
 691                         if (*pi)
 692                                 (*pi)++;
 693                         parents = parents->next;
 694                 }
 695         }
 696
 697         /*
 698          * find the tips
 699          *
 700          * tips are nodes not reachable from any other node in the list
 701          *
 702          * the tips serve as a starting set for the work queue.
 703          */
 704         for (next = orig; next; next = next->next) {
 705                 struct commit *commit = next->item;
 706
 707                 if (*(indegree_slab_at(&indegree, commit)) == 1)
 708                         prio_queue_put(&queue, commit);
 709         }
 710
 711         /*
 712          * This is unfortunate; the initial tips need to be shown
 713          * in the order given from the revision traversal machinery.
 714          */
 715         if (sort_order == REV_SORT_IN_GRAPH_ORDER)
 716                 prio_queue_reverse(&queue);
 717
 718         /* We no longer need the commit list */
 719         free_commit_list(orig);
 720
 721         pptr = list;
 722         *list = NULL;
 723         while ((commit = prio_queue_get(&queue)) != NULL) {
 724                 struct commit_list *parents;
 725
 726                 for (parents = commit->parents; parents ; parents = parents->next) {
 727                         struct commit *parent = parents->item;
 728                         int *pi = indegree_slab_at(&indegree, parent);
 729
 730                         if (!*pi)
 731                                 continue;
 732
 733                         /*
 734                          * parents are only enqueued for emission
 735                          * when all their children have been emitted thereby
 736                          * guaranteeing topological order.
 737                          */
 738                         if (--(*pi) == 1)
 739                                 prio_queue_put(&queue, parent);
 740                 }
 741                 /*
 742                  * all children of commit have already been
 743                  * emitted. we can emit it now.
 744                  */
 745                 *(indegree_slab_at(&indegree, commit)) = 0;
 746
 747                 pptr = &commit_list_insert(commit, pptr)->next;
 748         }
 749
 750         clear_indegree_slab(&indegree);
 751         clear_prio_queue(&queue);
 752         if (sort_order == REV_SORT_BY_AUTHOR_DATE)
 753                 clear_author_date_slab(&author_date);
 754 }
 755
 756 /* merge-base stuff */
 757
 758 /* Remember to update object flag allocation in object.h */
 759 #define PARENT1         (1u<<16)
 760 #define PARENT2         (1u<<17)
 761 #define STALE           (1u<<18)
 762 #define RESULT          (1u<<19)
 763
 764 static const unsigned all_flags = (PARENT1 | PARENT2 | STALE | RESULT);
 765
 766 static int queue_has_nonstale(struct prio_queue *queue)
 767 {
 768         int i;
 769         for (i = 0; i < queue->nr; i++) {
 770                 struct commit *commit = queue->array[i].data;
 771                 if (!(commit->object.flags & STALE))
 772                         return 1;
 773         }
 774         return 0;
 775 }
 776
 777 /* all input commits in one and twos[] must have been parsed! */
 778 static struct commit_list *paint_down_to_common(struct commit *one, int n, struct commit **twos)
 779 {
 780         struct prio_queue queue = { compare_commits_by_commit_date };
 781         struct commit_list *result = NULL;
 782         int i;
 783
 784         one->object.flags |= PARENT1;
 785         if (!n) {
 786                 commit_list_append(one, &result);
 787                 return result;
 788         }
 789         prio_queue_put(&queue, one);
 790
 791         for (i = 0; i < n; i++) {
 792                 twos[i]->object.flags |= PARENT2;
 793                 prio_queue_put(&queue, twos[i]);
 794         }
 795
 796         while (queue_has_nonstale(&queue)) {
 797                 struct commit *commit = prio_queue_get(&queue);
 798                 struct commit_list *parents;
 799                 int flags;
 800
 801                 flags = commit->object.flags & (PARENT1 | PARENT2 | STALE);
 802                 if (flags == (PARENT1 | PARENT2)) {
 803                         if (!(commit->object.flags & RESULT)) {
 804                                 commit->object.flags |= RESULT;
 805                                 commit_list_insert_by_date(commit, &result);
 806                         }
 807                         /* Mark parents of a found merge stale */
 808                         flags |= STALE;
 809                 }
 810                 parents = commit->parents;
 811                 while (parents) {
 812                         struct commit *p = parents->item;
 813                         parents = parents->next;
 814                         if ((p->object.flags & flags) == flags)
 815                                 continue;
 816                         if (parse_commit(p))
 817                                 return NULL;
 818                         p->object.flags |= flags;
 819                         prio_queue_put(&queue, p);
 820                 }
 821         }
 822
 823         clear_prio_queue(&queue);
 824         return result;
 825 }
 826
 827 static struct commit_list *merge_bases_many(struct commit *one, int n, struct commit **twos)
 828 {
 829         struct commit_list *list = NULL;
 830         struct commit_list *result = NULL;
 831         int i;
 832
 833         for (i = 0; i < n; i++) {
 834                 if (one == twos[i])
 835                         /*
 836                          * We do not mark this even with RESULT so we do not
 837                          * have to clean it up.
 838                          */
 839                         return commit_list_insert(one, &result);
 840         }
 841
 842         if (parse_commit(one))
 843                 return NULL;
 844         for (i = 0; i < n; i++) {
 845                 if (parse_commit(twos[i]))
 846                         return NULL;
 847         }
 848
 849         list = paint_down_to_common(one, n, twos);
 850
 851         while (list) {
 852                 struct commit_list *next = list->next;
 853                 if (!(list->item->object.flags & STALE))
 854                         commit_list_insert_by_date(list->item, &result);
 855                 free(list);
 856                 list = next;
 857         }
 858         return result;
 859 }
 860
 861 struct commit_list *get_octopus_merge_bases(struct commit_list *in)
 862 {
 863         struct commit_list *i, *j, *k, *ret = NULL;
 864
 865         if (!in)
 866                 return ret;
 867
 868         commit_list_insert(in->item, &ret);
 869
 870         for (i = in->next; i; i = i->next) {
 871                 struct commit_list *new = NULL, *end = NULL;
 872
 873                 for (j = ret; j; j = j->next) {
 874                         struct commit_list *bases;
 875                         bases = get_merge_bases(i->item, j->item);
 876                         if (!new)
 877                                 new = bases;
 878                         else
 879                                 end->next = bases;
 880                         for (k = bases; k; k = k->next)
 881                                 end = k;
 882                 }
 883                 ret = new;
 884         }
 885         return ret;
 886 }
 887
 888 static int remove_redundant(struct commit **array, int cnt)
 889 {
 890         /*
 891          * Some commit in the array may be an ancestor of
 892          * another commit.  Move such commit to the end of
 893          * the array, and return the number of commits that
 894          * are independent from each other.
 895          */
 896         struct commit **work;
 897         unsigned char *redundant;
 898         int *filled_index;
 899         int i, j, filled;
 900
 901         work = xcalloc(cnt, sizeof(*work));
 902         redundant = xcalloc(cnt, 1);
 903         filled_index = xmalloc(sizeof(*filled_index) * (cnt - 1));
 904
 905         for (i = 0; i < cnt; i++)
 906                 parse_commit(array[i]);
 907         for (i = 0; i < cnt; i++) {
 908                 struct commit_list *common;
 909
 910                 if (redundant[i])
 911                         continue;
 912                 for (j = filled = 0; j < cnt; j++) {
 913                         if (i == j || redundant[j])
 914                                 continue;
 915                         filled_index[filled] = j;
 916                         work[filled++] = array[j];
 917                 }
 918                 common = paint_down_to_common(array[i], filled, work);
 919                 if (array[i]->object.flags & PARENT2)
 920                         redundant[i] = 1;
 921                 for (j = 0; j < filled; j++)
 922                         if (work[j]->object.flags & PARENT1)
 923                                 redundant[filled_index[j]] = 1;
 924                 clear_commit_marks(array[i], all_flags);
 925                 for (j = 0; j < filled; j++)
 926                         clear_commit_marks(work[j], all_flags);
 927                 free_commit_list(common);
 928         }
 929
 930         /* Now collect the result */
 931         memcpy(work, array, sizeof(*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         return result;
1083 }
1084
1085 static const char gpg_sig_header[] = "gpgsig";
1086 static const int gpg_sig_header_len = sizeof(gpg_sig_header) - 1;
1087
1088 static int do_sign_commit(struct strbuf *buf, const char *keyid)
1089 {
1090         struct strbuf sig = STRBUF_INIT;
1091         int inspos, copypos;
1092
1093         /* find the end of the header */
1094         inspos = strstr(buf->buf, "\n\n") - buf->buf + 1;
1095
1096         if (!keyid || !*keyid)
1097                 keyid = get_signing_key();
1098         if (sign_buffer(buf, &sig, keyid)) {
1099                 strbuf_release(&sig);
1100                 return -1;
1101         }
1102
1103         for (copypos = 0; sig.buf[copypos]; ) {
1104                 const char *bol = sig.buf + copypos;
1105                 const char *eol = strchrnul(bol, '\n');
1106                 int len = (eol - bol) + !!*eol;
1107
1108                 if (!copypos) {
1109                         strbuf_insert(buf, inspos, gpg_sig_header, gpg_sig_header_len);
1110                         inspos += gpg_sig_header_len;
1111                 }
1112                 strbuf_insert(buf, inspos++, " ", 1);
1113                 strbuf_insert(buf, inspos, bol, len);
1114                 inspos += len;
1115                 copypos += len;
1116         }
1117         strbuf_release(&sig);
1118         return 0;
1119 }
1120
1121 int parse_signed_commit(const struct commit *commit,
1122                         struct strbuf *payload, struct strbuf *signature)
1123 {
1124
1125         unsigned long size;
1126         const char *buffer = get_commit_buffer(commit, &size);
1127         int in_signature, saw_signature = -1;
1128         const char *line, *tail;
1129
1130         line = buffer;
1131         tail = buffer + size;
1132         in_signature = 0;
1133         saw_signature = 0;
1134         while (line < tail) {
1135                 const char *sig = NULL;
1136                 const char *next = memchr(line, '\n', tail - line);
1137
1138                 next = next ? next + 1 : tail;
1139                 if (in_signature && line[0] == ' ')
1140                         sig = line + 1;
1141                 else if (starts_with(line, gpg_sig_header) &&
1142                          line[gpg_sig_header_len] == ' ')
1143                         sig = line + gpg_sig_header_len + 1;
1144                 if (sig) {
1145                         strbuf_add(signature, sig, next - sig);
1146                         saw_signature = 1;
1147                         in_signature = 1;
1148                 } else {
1149                         if (*line == '\n')
1150                                 /* dump the whole remainder of the buffer */
1151                                 next = tail;
1152                         strbuf_add(payload, line, next - line);
1153                         in_signature = 0;
1154                 }
1155                 line = next;
1156         }
1157         unuse_commit_buffer(commit, buffer);
1158         return saw_signature;
1159 }
1160
1161 int remove_signature(struct strbuf *buf)
1162 {
1163         const char *line = buf->buf;
1164         const char *tail = buf->buf + buf->len;
1165         int in_signature = 0;
1166         const char *sig_start = NULL;
1167         const char *sig_end = NULL;
1168
1169         while (line < tail) {
1170                 const char *next = memchr(line, '\n', tail - line);
1171                 next = next ? next + 1 : tail;
1172
1173                 if (in_signature && line[0] == ' ')
1174                         sig_end = next;
1175                 else if (starts_with(line, gpg_sig_header) &&
1176                          line[gpg_sig_header_len] == ' ') {
1177                         sig_start = line;
1178                         sig_end = next;
1179                         in_signature = 1;
1180                 } else {
1181                         if (*line == '\n')
1182                                 /* dump the whole remainder of the buffer */
1183                                 next = tail;
1184                         in_signature = 0;
1185                 }
1186                 line = next;
1187         }
1188
1189         if (sig_start)
1190                 strbuf_remove(buf, sig_start - buf->buf, sig_end - sig_start);
1191
1192         return sig_start != NULL;
1193 }
1194
1195 static void handle_signed_tag(struct commit *parent, struct commit_extra_header ***tail)
1196 {
1197         struct merge_remote_desc *desc;
1198         struct commit_extra_header *mergetag;
1199         char *buf;
1200         unsigned long size, len;
1201         enum object_type type;
1202
1203         desc = merge_remote_util(parent);
1204         if (!desc || !desc->obj)
1205                 return;
1206         buf = read_sha1_file(desc->obj->sha1, &type, &size);
1207         if (!buf || type != OBJ_TAG)
1208                 goto free_return;
1209         len = parse_signature(buf, size);
1210         if (size == len)
1211                 goto free_return;
1212         /*
1213          * We could verify this signature and either omit the tag when
1214          * it does not validate, but the integrator may not have the
1215          * public key of the signer of the tag he is merging, while a
1216          * later auditor may have it while auditing, so let's not run
1217          * verify-signed-buffer here for now...
1218          *
1219          * if (verify_signed_buffer(buf, len, buf + len, size - len, ...))
1220          *      warn("warning: signed tag unverified.");
1221          */
1222         mergetag = xcalloc(1, sizeof(*mergetag));
1223         mergetag->key = xstrdup("mergetag");
1224         mergetag->value = buf;
1225         mergetag->len = size;
1226
1227         **tail = mergetag;
1228         *tail = &mergetag->next;
1229         return;
1230
1231 free_return:
1232         free(buf);
1233 }
1234
1235 void check_commit_signature(const struct commit *commit, struct signature_check *sigc)
1236 {
1237         struct strbuf payload = STRBUF_INIT;
1238         struct strbuf signature = STRBUF_INIT;
1239         struct strbuf gpg_output = STRBUF_INIT;
1240         struct strbuf gpg_status = STRBUF_INIT;
1241         int status;
1242
1243         sigc->result = 'N';
1244
1245         if (parse_signed_commit(commit, &payload, &signature) <= 0)
1246                 goto out;
1247         status = verify_signed_buffer(payload.buf, payload.len,
1248                                       signature.buf, signature.len,
1249                                       &gpg_output, &gpg_status);
1250         if (status && !gpg_output.len)
1251                 goto out;
1252         sigc->payload = strbuf_detach(&payload, NULL);
1253         sigc->gpg_output = strbuf_detach(&gpg_output, NULL);
1254         sigc->gpg_status = strbuf_detach(&gpg_status, NULL);
1255         parse_gpg_output(sigc);
1256
1257  out:
1258         strbuf_release(&gpg_status);
1259         strbuf_release(&gpg_output);
1260         strbuf_release(&payload);
1261         strbuf_release(&signature);
1262 }
1263
1264
1265
1266 void append_merge_tag_headers(struct commit_list *parents,
1267                               struct commit_extra_header ***tail)
1268 {
1269         while (parents) {
1270                 struct commit *parent = parents->item;
1271                 handle_signed_tag(parent, tail);
1272                 parents = parents->next;
1273         }
1274 }
1275
1276 static void add_extra_header(struct strbuf *buffer,
1277                              struct commit_extra_header *extra)
1278 {
1279         strbuf_addstr(buffer, extra->key);
1280         if (extra->len)
1281                 strbuf_add_lines(buffer, " ", extra->value, extra->len);
1282         else
1283                 strbuf_addch(buffer, '\n');
1284 }
1285
1286 struct commit_extra_header *read_commit_extra_headers(struct commit *commit,
1287                                                       const char **exclude)
1288 {
1289         struct commit_extra_header *extra = NULL;
1290         unsigned long size;
1291         const char *buffer = get_commit_buffer(commit, &size);
1292         extra = read_commit_extra_header_lines(buffer, size, exclude);
1293         unuse_commit_buffer(commit, buffer);
1294         return extra;
1295 }
1296
1297 void for_each_mergetag(each_mergetag_fn fn, struct commit *commit, void *data)
1298 {
1299         struct commit_extra_header *extra, *to_free;
1300
1301         to_free = read_commit_extra_headers(commit, NULL);
1302         for (extra = to_free; extra; extra = extra->next) {
1303                 if (strcmp(extra->key, "mergetag"))
1304                         continue; /* not a merge tag */
1305                 fn(commit, extra, data);
1306         }
1307         free_commit_extra_headers(to_free);
1308 }
1309
1310 static inline int standard_header_field(const char *field, size_t len)
1311 {
1312         return ((len == 4 && !memcmp(field, "tree ", 5)) ||
1313                 (len == 6 && !memcmp(field, "parent ", 7)) ||
1314                 (len == 6 && !memcmp(field, "author ", 7)) ||
1315                 (len == 9 && !memcmp(field, "committer ", 10)) ||
1316                 (len == 8 && !memcmp(field, "encoding ", 9)));
1317 }
1318
1319 static int excluded_header_field(const char *field, size_t len, const char **exclude)
1320 {
1321         if (!exclude)
1322                 return 0;
1323
1324         while (*exclude) {
1325                 size_t xlen = strlen(*exclude);
1326                 if (len == xlen &&
1327                     !memcmp(field, *exclude, xlen) && field[xlen] == ' ')
1328                         return 1;
1329                 exclude++;
1330         }
1331         return 0;
1332 }
1333
1334 static struct commit_extra_header *read_commit_extra_header_lines(
1335         const char *buffer, size_t size,
1336         const char **exclude)
1337 {
1338         struct commit_extra_header *extra = NULL, **tail = &extra, *it = NULL;
1339         const char *line, *next, *eof, *eob;
1340         struct strbuf buf = STRBUF_INIT;
1341
1342         for (line = buffer, eob = line + size;
1343              line < eob && *line != '\n';
1344              line = next) {
1345                 next = memchr(line, '\n', eob - line);
1346                 next = next ? next + 1 : eob;
1347                 if (*line == ' ') {
1348                         /* continuation */
1349                         if (it)
1350                                 strbuf_add(&buf, line + 1, next - (line + 1));
1351                         continue;
1352                 }
1353                 if (it)
1354                         it->value = strbuf_detach(&buf, &it->len);
1355                 strbuf_reset(&buf);
1356                 it = NULL;
1357
1358                 eof = strchr(line, ' ');
1359                 if (next <= eof)
1360                         eof = next;
1361
1362                 if (standard_header_field(line, eof - line) ||
1363                     excluded_header_field(line, eof - line, exclude))
1364                         continue;
1365
1366                 it = xcalloc(1, sizeof(*it));
1367                 it->key = xmemdupz(line, eof-line);
1368                 *tail = it;
1369                 tail = &it->next;
1370                 if (eof + 1 < next)
1371                         strbuf_add(&buf, eof + 1, next - (eof + 1));
1372         }
1373         if (it)
1374                 it->value = strbuf_detach(&buf, &it->len);
1375         return extra;
1376 }
1377
1378 void free_commit_extra_headers(struct commit_extra_header *extra)
1379 {
1380         while (extra) {
1381                 struct commit_extra_header *next = extra->next;
1382                 free(extra->key);
1383                 free(extra->value);
1384                 free(extra);
1385                 extra = next;
1386         }
1387 }
1388
1389 int commit_tree(const char *msg, size_t msg_len,
1390                 const unsigned char *tree,
1391                 struct commit_list *parents, unsigned char *ret,
1392                 const char *author, const char *sign_commit)
1393 {
1394         struct commit_extra_header *extra = NULL, **tail = &extra;
1395         int result;
1396
1397         append_merge_tag_headers(parents, &tail);
1398         result = commit_tree_extended(msg, msg_len, tree, parents, ret,
1399                                       author, sign_commit, extra);
1400         free_commit_extra_headers(extra);
1401         return result;
1402 }
1403
1404 static int find_invalid_utf8(const char *buf, int len)
1405 {
1406         int offset = 0;
1407         static const unsigned int max_codepoint[] = {
1408                 0x7f, 0x7ff, 0xffff, 0x10ffff
1409         };
1410
1411         while (len) {
1412                 unsigned char c = *buf++;
1413                 int bytes, bad_offset;
1414                 unsigned int codepoint;
1415                 unsigned int min_val, max_val;
1416
1417                 len--;
1418                 offset++;
1419
1420                 /* Simple US-ASCII? No worries. */
1421                 if (c < 0x80)
1422                         continue;
1423
1424                 bad_offset = offset-1;
1425
1426                 /*
1427                  * Count how many more high bits set: that's how
1428                  * many more bytes this sequence should have.
1429                  */
1430                 bytes = 0;
1431                 while (c & 0x40) {
1432                         c <<= 1;
1433                         bytes++;
1434                 }
1435
1436                 /*
1437                  * Must be between 1 and 3 more bytes.  Longer sequences result in
1438                  * codepoints beyond U+10FFFF, which are guaranteed never to exist.
1439                  */
1440                 if (bytes < 1 || 3 < bytes)
1441                         return bad_offset;
1442
1443                 /* Do we *have* that many bytes? */
1444                 if (len < bytes)
1445                         return bad_offset;
1446
1447                 /*
1448                  * Place the encoded bits at the bottom of the value and compute the
1449                  * valid range.
1450                  */
1451                 codepoint = (c & 0x7f) >> bytes;
1452                 min_val = max_codepoint[bytes-1] + 1;
1453                 max_val = max_codepoint[bytes];
1454
1455                 offset += bytes;
1456                 len -= bytes;
1457
1458                 /* And verify that they are good continuation bytes */
1459                 do {
1460                         codepoint <<= 6;
1461                         codepoint |= *buf & 0x3f;
1462                         if ((*buf++ & 0xc0) != 0x80)
1463                                 return bad_offset;
1464                 } while (--bytes);
1465
1466                 /* Reject codepoints that are out of range for the sequence length. */
1467                 if (codepoint < min_val || codepoint > max_val)
1468                         return bad_offset;
1469                 /* Surrogates are only for UTF-16 and cannot be encoded in UTF-8. */
1470                 if ((codepoint & 0x1ff800) == 0xd800)
1471                         return bad_offset;
1472                 /* U+xxFFFE and U+xxFFFF are guaranteed non-characters. */
1473                 if ((codepoint & 0xfffe) == 0xfffe)
1474                         return bad_offset;
1475                 /* So are anything in the range U+FDD0..U+FDEF. */
1476                 if (codepoint >= 0xfdd0 && codepoint <= 0xfdef)
1477                         return bad_offset;
1478         }
1479         return -1;
1480 }
1481
1482 /*
1483  * This verifies that the buffer is in proper utf8 format.
1484  *
1485  * If it isn't, it assumes any non-utf8 characters are Latin1,
1486  * and does the conversion.
1487  */
1488 static int verify_utf8(struct strbuf *buf)
1489 {
1490         int ok = 1;
1491         long pos = 0;
1492
1493         for (;;) {
1494                 int bad;
1495                 unsigned char c;
1496                 unsigned char replace[2];
1497
1498                 bad = find_invalid_utf8(buf->buf + pos, buf->len - pos);
1499                 if (bad < 0)
1500                         return ok;
1501                 pos += bad;
1502                 ok = 0;
1503                 c = buf->buf[pos];
1504                 strbuf_remove(buf, pos, 1);
1505
1506                 /* We know 'c' must be in the range 128-255 */
1507                 replace[0] = 0xc0 + (c >> 6);
1508                 replace[1] = 0x80 + (c & 0x3f);
1509                 strbuf_insert(buf, pos, replace, 2);
1510                 pos += 2;
1511         }
1512 }
1513
1514 static const char commit_utf8_warn[] =
1515 "Warning: commit message did not conform to UTF-8.\n"
1516 "You may want to amend it after fixing the message, or set the config\n"
1517 "variable i18n.commitencoding to the encoding your project uses.\n";
1518
1519 int commit_tree_extended(const char *msg, size_t msg_len,
1520                          const unsigned char *tree,
1521                          struct commit_list *parents, unsigned char *ret,
1522                          const char *author, const char *sign_commit,
1523                          struct commit_extra_header *extra)
1524 {
1525         int result;
1526         int encoding_is_utf8;
1527         struct strbuf buffer;
1528
1529         assert_sha1_type(tree, OBJ_TREE);
1530
1531         if (memchr(msg, '\0', msg_len))
1532                 return error("a NUL byte in commit log message not allowed.");
1533
1534         /* Not having i18n.commitencoding is the same as having utf-8 */
1535         encoding_is_utf8 = is_encoding_utf8(git_commit_encoding);
1536
1537         strbuf_init(&buffer, 8192); /* should avoid reallocs for the headers */
1538         strbuf_addf(&buffer, "tree %s\n", sha1_to_hex(tree));
1539
1540         /*
1541          * NOTE! This ordering means that the same exact tree merged with a
1542          * different order of parents will be a _different_ changeset even
1543          * if everything else stays the same.
1544          */
1545         while (parents) {
1546                 struct commit_list *next = parents->next;
1547                 struct commit *parent = parents->item;
1548
1549                 strbuf_addf(&buffer, "parent %s\n",
1550                             sha1_to_hex(parent->object.sha1));
1551                 free(parents);
1552                 parents = next;
1553         }
1554
1555         /* Person/date information */
1556         if (!author)
1557                 author = git_author_info(IDENT_STRICT);
1558         strbuf_addf(&buffer, "author %s\n", author);
1559         strbuf_addf(&buffer, "committer %s\n", git_committer_info(IDENT_STRICT));
1560         if (!encoding_is_utf8)
1561                 strbuf_addf(&buffer, "encoding %s\n", git_commit_encoding);
1562
1563         while (extra) {
1564                 add_extra_header(&buffer, extra);
1565                 extra = extra->next;
1566         }
1567         strbuf_addch(&buffer, '\n');
1568
1569         /* And add the comment */
1570         strbuf_add(&buffer, msg, msg_len);
1571
1572         /* And check the encoding */
1573         if (encoding_is_utf8 && !verify_utf8(&buffer))
1574                 fprintf(stderr, commit_utf8_warn);
1575
1576         if (sign_commit && do_sign_commit(&buffer, sign_commit))
1577                 return -1;
1578
1579         result = write_sha1_file(buffer.buf, buffer.len, commit_type, ret);
1580         strbuf_release(&buffer);
1581         return result;
1582 }
1583
1584 struct commit *get_merge_parent(const char *name)
1585 {
1586         struct object *obj;
1587         struct commit *commit;
1588         struct object_id oid;
1589         if (get_sha1(name, oid.hash))
1590                 return NULL;
1591         obj = parse_object(oid.hash);
1592         commit = (struct commit *)peel_to_type(name, 0, obj, OBJ_COMMIT);
1593         if (commit && !commit->util) {
1594                 struct merge_remote_desc *desc;
1595                 desc = xmalloc(sizeof(*desc));
1596                 desc->obj = obj;
1597                 desc->name = strdup(name);
1598                 commit->util = desc;
1599         }
1600         return commit;
1601 }
1602
1603 /*
1604  * Append a commit to the end of the commit_list.
1605  *
1606  * next starts by pointing to the variable that holds the head of an
1607  * empty commit_list, and is updated to point to the "next" field of
1608  * the last item on the list as new commits are appended.
1609  *
1610  * Usage example:
1611  *
1612  *     struct commit_list *list;
1613  *     struct commit_list **next = &list;
1614  *
1615  *     next = commit_list_append(c1, next);
1616  *     next = commit_list_append(c2, next);
1617  *     assert(commit_list_count(list) == 2);
1618  *     return list;
1619  */
1620 struct commit_list **commit_list_append(struct commit *commit,
1621                                         struct commit_list **next)
1622 {
1623         struct commit_list *new = xmalloc(sizeof(struct commit_list));
1624         new->item = commit;
1625         *next = new;
1626         new->next = NULL;
1627         return &new->next;
1628 }
1629
1630 void print_commit_list(struct commit_list *list,
1631                        const char *format_cur,
1632                        const char *format_last)
1633 {
1634         for ( ; list; list = list->next) {
1635                 const char *format = list->next ? format_cur : format_last;
1636                 printf(format, sha1_to_hex(list->item->object.sha1));
1637         }
1638 }
1639
1640 const char *find_commit_header(const char *msg, const char *key, size_t *out_len)
1641 {
1642         int key_len = strlen(key);
1643         const char *line = msg;
1644
1645         while (line) {
1646                 const char *eol = strchrnul(line, '\n');
1647
1648                 if (line == eol)
1649                         return NULL;
1650
1651                 if (eol - line > key_len &&
1652                     !strncmp(line, key, key_len) &&
1653                     line[key_len] == ' ') {
1654                         *out_len = eol - line - key_len - 1;
1655                         return line + key_len + 1;
1656                 }
1657                 line = *eol ? eol + 1 : NULL;
1658         }
1659         return NULL;
1660 }
1661
1662 /*
1663  * Inspect sb and determine the true "end" of the log message, in
1664  * order to find where to put a new Signed-off-by: line.  Ignored are
1665  * trailing comment lines and blank lines, and also the traditional
1666  * "Conflicts:" block that is not commented out, so that we can use
1667  * "git commit -s --amend" on an existing commit that forgot to remove
1668  * it.
1669  *
1670  * Returns the number of bytes from the tail to ignore, to be fed as
1671  * the second parameter to append_signoff().
1672  */
1673 int ignore_non_trailer(struct strbuf *sb)
1674 {
1675         int boc = 0;
1676         int bol = 0;
1677         int in_old_conflicts_block = 0;
1678
1679         while (bol < sb->len) {
1680                 char *next_line;
1681
1682                 if (!(next_line = memchr(sb->buf + bol, '\n', sb->len - bol)))
1683                         next_line = sb->buf + sb->len;
1684                 else
1685                         next_line++;
1686
1687                 if (sb->buf[bol] == comment_line_char || sb->buf[bol] == '\n') {
1688                         /* is this the first of the run of comments? */
1689                         if (!boc)
1690                                 boc = bol;
1691                         /* otherwise, it is just continuing */
1692                 } else if (starts_with(sb->buf + bol, "Conflicts:\n")) {
1693                         in_old_conflicts_block = 1;
1694                         if (!boc)
1695                                 boc = bol;
1696                 } else if (in_old_conflicts_block && sb->buf[bol] == '\t') {
1697                         ; /* a pathname in the conflicts block */
1698                 } else if (boc) {
1699                         /* the previous was not trailing comment */
1700                         boc = 0;
1701                         in_old_conflicts_block = 0;
1702                 }
1703                 bol = next_line - sb->buf;
1704         }
1705         return boc ? sb->len - boc : 0;
1706 }