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_gently(struct commit *item, int quiet_on_missing)
 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 quiet_on_missing ? -1 :
 370                         error("Could not read %s",
 371                              sha1_to_hex(item->object.sha1));
 372         if (type != OBJ_COMMIT) {
 373                 free(buffer);
 374                 return error("Object %s not a commit",
 375                              sha1_to_hex(item->object.sha1));
 376         }
 377         ret = parse_commit_buffer(item, buffer, size);
 378         if (save_commit_buffer && !ret) {
 379                 set_commit_buffer(item, buffer, size);
 380                 return 0;
 381         }
 382         free(buffer);
 383         return ret;
 384 }
 385
 386 void parse_commit_or_die(struct commit *item)
 387 {
 388         if (parse_commit(item))
 389                 die("unable to parse commit %s",
 390                     item ? sha1_to_hex(item->object.sha1) : "(null)");
 391 }
 392
 393 int find_commit_subject(const char *commit_buffer, const char **subject)
 394 {
 395         const char *eol;
 396         const char *p = commit_buffer;
 397
 398         while (*p && (*p != '\n' || p[1] != '\n'))
 399                 p++;
 400         if (*p) {
 401                 p += 2;
 402                 for (eol = p; *eol && *eol != '\n'; eol++)
 403                         ; /* do nothing */
 404         } else
 405                 eol = p;
 406
 407         *subject = p;
 408
 409         return eol - p;
 410 }
 411
 412 struct commit_list *commit_list_insert(struct commit *item, struct commit_list **list_p)
 413 {
 414         struct commit_list *new_list = xmalloc(sizeof(struct commit_list));
 415         new_list->item = item;
 416         new_list->next = *list_p;
 417         *list_p = new_list;
 418         return new_list;
 419 }
 420
 421 unsigned commit_list_count(const struct commit_list *l)
 422 {
 423         unsigned c = 0;
 424         for (; l; l = l->next )
 425                 c++;
 426         return c;
 427 }
 428
 429 struct commit_list *copy_commit_list(struct commit_list *list)
 430 {
 431         struct commit_list *head = NULL;
 432         struct commit_list **pp = &head;
 433         while (list) {
 434                 pp = commit_list_append(list->item, pp);
 435                 list = list->next;
 436         }
 437         return head;
 438 }
 439
 440 void free_commit_list(struct commit_list *list)
 441 {
 442         while (list) {
 443                 struct commit_list *temp = list;
 444                 list = temp->next;
 445                 free(temp);
 446         }
 447 }
 448
 449 struct commit_list * commit_list_insert_by_date(struct commit *item, struct commit_list **list)
 450 {
 451         struct commit_list **pp = list;
 452         struct commit_list *p;
 453         while ((p = *pp) != NULL) {
 454                 if (p->item->date < item->date) {
 455                         break;
 456                 }
 457                 pp = &p->next;
 458         }
 459         return commit_list_insert(item, pp);
 460 }
 461
 462 static int commit_list_compare_by_date(const void *a, const void *b)
 463 {
 464         unsigned long a_date = ((const struct commit_list *)a)->item->date;
 465         unsigned long b_date = ((const struct commit_list *)b)->item->date;
 466         if (a_date < b_date)
 467                 return 1;
 468         if (a_date > b_date)
 469                 return -1;
 470         return 0;
 471 }
 472
 473 static void *commit_list_get_next(const void *a)
 474 {
 475         return ((const struct commit_list *)a)->next;
 476 }
 477
 478 static void commit_list_set_next(void *a, void *next)
 479 {
 480         ((struct commit_list *)a)->next = next;
 481 }
 482
 483 void commit_list_sort_by_date(struct commit_list **list)
 484 {
 485         *list = llist_mergesort(*list, commit_list_get_next, commit_list_set_next,
 486                                 commit_list_compare_by_date);
 487 }
 488
 489 struct commit *pop_most_recent_commit(struct commit_list **list,
 490                                       unsigned int mark)
 491 {
 492         struct commit *ret = (*list)->item;
 493         struct commit_list *parents = ret->parents;
 494         struct commit_list *old = *list;
 495
 496         *list = (*list)->next;
 497         free(old);
 498
 499         while (parents) {
 500                 struct commit *commit = parents->item;
 501                 if (!parse_commit(commit) && !(commit->object.flags & mark)) {
 502                         commit->object.flags |= mark;
 503                         commit_list_insert_by_date(commit, list);
 504                 }
 505                 parents = parents->next;
 506         }
 507         return ret;
 508 }
 509
 510 static void clear_commit_marks_1(struct commit_list **plist,
 511                                  struct commit *commit, unsigned int mark)
 512 {
 513         while (commit) {
 514                 struct commit_list *parents;
 515
 516                 if (!(mark & commit->object.flags))
 517                         return;
 518
 519                 commit->object.flags &= ~mark;
 520
 521                 parents = commit->parents;
 522                 if (!parents)
 523                         return;
 524
 525                 while ((parents = parents->next))
 526                         commit_list_insert(parents->item, plist);
 527
 528                 commit = commit->parents->item;
 529         }
 530 }
 531
 532 void clear_commit_marks_many(int nr, struct commit **commit, unsigned int mark)
 533 {
 534         struct commit_list *list = NULL;
 535
 536         while (nr--) {
 537                 commit_list_insert(*commit, &list);
 538                 commit++;
 539         }
 540         while (list)
 541                 clear_commit_marks_1(&list, pop_commit(&list), mark);
 542 }
 543
 544 void clear_commit_marks(struct commit *commit, unsigned int mark)
 545 {
 546         clear_commit_marks_many(1, &commit, mark);
 547 }
 548
 549 void clear_commit_marks_for_object_array(struct object_array *a, unsigned mark)
 550 {
 551         struct object *object;
 552         struct commit *commit;
 553         unsigned int i;
 554
 555         for (i = 0; i < a->nr; i++) {
 556                 object = a->objects[i].item;
 557                 commit = lookup_commit_reference_gently(object->sha1, 1);
 558                 if (commit)
 559                         clear_commit_marks(commit, mark);
 560         }
 561 }
 562
 563 struct commit *pop_commit(struct commit_list **stack)
 564 {
 565         struct commit_list *top = *stack;
 566         struct commit *item = top ? top->item : NULL;
 567
 568         if (top) {
 569                 *stack = top->next;
 570                 free(top);
 571         }
 572         return item;
 573 }
 574
 575 /*
 576  * Topological sort support
 577  */
 578
 579 /* count number of children that have not been emitted */
 580 define_commit_slab(indegree_slab, int);
 581
 582 /* record author-date for each commit object */
 583 define_commit_slab(author_date_slab, unsigned long);
 584
 585 static void record_author_date(struct author_date_slab *author_date,
 586                                struct commit *commit)
 587 {
 588         const char *buffer = get_commit_buffer(commit, NULL);
 589         struct ident_split ident;
 590         const char *ident_line;
 591         size_t ident_len;
 592         char *date_end;
 593         unsigned long date;
 594
 595         ident_line = find_commit_header(buffer, "author", &ident_len);
 596         if (!ident_line)
 597                 goto fail_exit; /* no author line */
 598         if (split_ident_line(&ident, ident_line, ident_len) ||
 599             !ident.date_begin || !ident.date_end)
 600                 goto fail_exit; /* malformed "author" line */
 601
 602         date = strtoul(ident.date_begin, &date_end, 10);
 603         if (date_end != ident.date_end)
 604                 goto fail_exit; /* malformed date */
 605         *(author_date_slab_at(author_date, commit)) = date;
 606
 607 fail_exit:
 608         unuse_commit_buffer(commit, buffer);
 609 }
 610
 611 static int compare_commits_by_author_date(const void *a_, const void *b_,
 612                                           void *cb_data)
 613 {
 614         const struct commit *a = a_, *b = b_;
 615         struct author_date_slab *author_date = cb_data;
 616         unsigned long a_date = *(author_date_slab_at(author_date, a));
 617         unsigned long b_date = *(author_date_slab_at(author_date, b));
 618
 619         /* newer commits with larger date first */
 620         if (a_date < b_date)
 621                 return 1;
 622         else if (a_date > b_date)
 623                 return -1;
 624         return 0;
 625 }
 626
 627 int compare_commits_by_commit_date(const void *a_, const void *b_, void *unused)
 628 {
 629         const struct commit *a = a_, *b = b_;
 630         /* newer commits with larger date first */
 631         if (a->date < b->date)
 632                 return 1;
 633         else if (a->date > b->date)
 634                 return -1;
 635         return 0;
 636 }
 637
 638 /*
 639  * Performs an in-place topological sort on the list supplied.
 640  */
 641 void sort_in_topological_order(struct commit_list **list, enum rev_sort_order sort_order)
 642 {
 643         struct commit_list *next, *orig = *list;
 644         struct commit_list **pptr;
 645         struct indegree_slab indegree;
 646         struct prio_queue queue;
 647         struct commit *commit;
 648         struct author_date_slab author_date;
 649
 650         if (!orig)
 651                 return;
 652         *list = NULL;
 653
 654         init_indegree_slab(&indegree);
 655         memset(&queue, '\0', sizeof(queue));
 656
 657         switch (sort_order) {
 658         default: /* REV_SORT_IN_GRAPH_ORDER */
 659                 queue.compare = NULL;
 660                 break;
 661         case REV_SORT_BY_COMMIT_DATE:
 662                 queue.compare = compare_commits_by_commit_date;
 663                 break;
 664         case REV_SORT_BY_AUTHOR_DATE:
 665                 init_author_date_slab(&author_date);
 666                 queue.compare = compare_commits_by_author_date;
 667                 queue.cb_data = &author_date;
 668                 break;
 669         }
 670
 671         /* Mark them and clear the indegree */
 672         for (next = orig; next; next = next->next) {
 673                 struct commit *commit = next->item;
 674                 *(indegree_slab_at(&indegree, commit)) = 1;
 675                 /* also record the author dates, if needed */
 676                 if (sort_order == REV_SORT_BY_AUTHOR_DATE)
 677                         record_author_date(&author_date, commit);
 678         }
 679
 680         /* update the indegree */
 681         for (next = orig; next; next = next->next) {
 682                 struct commit_list *parents = next->item->parents;
 683                 while (parents) {
 684                         struct commit *parent = parents->item;
 685                         int *pi = indegree_slab_at(&indegree, parent);
 686
 687                         if (*pi)
 688                                 (*pi)++;
 689                         parents = parents->next;
 690                 }
 691         }
 692
 693         /*
 694          * find the tips
 695          *
 696          * tips are nodes not reachable from any other node in the list
 697          *
 698          * the tips serve as a starting set for the work queue.
 699          */
 700         for (next = orig; next; next = next->next) {
 701                 struct commit *commit = next->item;
 702
 703                 if (*(indegree_slab_at(&indegree, commit)) == 1)
 704                         prio_queue_put(&queue, commit);
 705         }
 706
 707         /*
 708          * This is unfortunate; the initial tips need to be shown
 709          * in the order given from the revision traversal machinery.
 710          */
 711         if (sort_order == REV_SORT_IN_GRAPH_ORDER)
 712                 prio_queue_reverse(&queue);
 713
 714         /* We no longer need the commit list */
 715         free_commit_list(orig);
 716
 717         pptr = list;
 718         *list = NULL;
 719         while ((commit = prio_queue_get(&queue)) != NULL) {
 720                 struct commit_list *parents;
 721
 722                 for (parents = commit->parents; parents ; parents = parents->next) {
 723                         struct commit *parent = parents->item;
 724                         int *pi = indegree_slab_at(&indegree, parent);
 725
 726                         if (!*pi)
 727                                 continue;
 728
 729                         /*
 730                          * parents are only enqueued for emission
 731                          * when all their children have been emitted thereby
 732                          * guaranteeing topological order.
 733                          */
 734                         if (--(*pi) == 1)
 735                                 prio_queue_put(&queue, parent);
 736                 }
 737                 /*
 738                  * all children of commit have already been
 739                  * emitted. we can emit it now.
 740                  */
 741                 *(indegree_slab_at(&indegree, commit)) = 0;
 742
 743                 pptr = &commit_list_insert(commit, pptr)->next;
 744         }
 745
 746         clear_indegree_slab(&indegree);
 747         clear_prio_queue(&queue);
 748         if (sort_order == REV_SORT_BY_AUTHOR_DATE)
 749                 clear_author_date_slab(&author_date);
 750 }
 751
 752 /* merge-base stuff */
 753
 754 /* Remember to update object flag allocation in object.h */
 755 #define PARENT1         (1u<<16)
 756 #define PARENT2         (1u<<17)
 757 #define STALE           (1u<<18)
 758 #define RESULT          (1u<<19)
 759
 760 static const unsigned all_flags = (PARENT1 | PARENT2 | STALE | RESULT);
 761
 762 static int queue_has_nonstale(struct prio_queue *queue)
 763 {
 764         int i;
 765         for (i = 0; i < queue->nr; i++) {
 766                 struct commit *commit = queue->array[i].data;
 767                 if (!(commit->object.flags & STALE))
 768                         return 1;
 769         }
 770         return 0;
 771 }
 772
 773 /* all input commits in one and twos[] must have been parsed! */
 774 static struct commit_list *paint_down_to_common(struct commit *one, int n, struct commit **twos)
 775 {
 776         struct prio_queue queue = { compare_commits_by_commit_date };
 777         struct commit_list *result = NULL;
 778         int i;
 779
 780         one->object.flags |= PARENT1;
 781         if (!n) {
 782                 commit_list_append(one, &result);
 783                 return result;
 784         }
 785         prio_queue_put(&queue, one);
 786
 787         for (i = 0; i < n; i++) {
 788                 twos[i]->object.flags |= PARENT2;
 789                 prio_queue_put(&queue, twos[i]);
 790         }
 791
 792         while (queue_has_nonstale(&queue)) {
 793                 struct commit *commit = prio_queue_get(&queue);
 794                 struct commit_list *parents;
 795                 int flags;
 796
 797                 flags = commit->object.flags & (PARENT1 | PARENT2 | STALE);
 798                 if (flags == (PARENT1 | PARENT2)) {
 799                         if (!(commit->object.flags & RESULT)) {
 800                                 commit->object.flags |= RESULT;
 801                                 commit_list_insert_by_date(commit, &result);
 802                         }
 803                         /* Mark parents of a found merge stale */
 804                         flags |= STALE;
 805                 }
 806                 parents = commit->parents;
 807                 while (parents) {
 808                         struct commit *p = parents->item;
 809                         parents = parents->next;
 810                         if ((p->object.flags & flags) == flags)
 811                                 continue;
 812                         if (parse_commit(p))
 813                                 return NULL;
 814                         p->object.flags |= flags;
 815                         prio_queue_put(&queue, p);
 816                 }
 817         }
 818
 819         clear_prio_queue(&queue);
 820         return result;
 821 }
 822
 823 static struct commit_list *merge_bases_many(struct commit *one, int n, struct commit **twos)
 824 {
 825         struct commit_list *list = NULL;
 826         struct commit_list *result = NULL;
 827         int i;
 828
 829         for (i = 0; i < n; i++) {
 830                 if (one == twos[i])
 831                         /*
 832                          * We do not mark this even with RESULT so we do not
 833                          * have to clean it up.
 834                          */
 835                         return commit_list_insert(one, &result);
 836         }
 837
 838         if (parse_commit(one))
 839                 return NULL;
 840         for (i = 0; i < n; i++) {
 841                 if (parse_commit(twos[i]))
 842                         return NULL;
 843         }
 844
 845         list = paint_down_to_common(one, n, twos);
 846
 847         while (list) {
 848                 struct commit_list *next = list->next;
 849                 if (!(list->item->object.flags & STALE))
 850                         commit_list_insert_by_date(list->item, &result);
 851                 free(list);
 852                 list = next;
 853         }
 854         return result;
 855 }
 856
 857 struct commit_list *get_octopus_merge_bases(struct commit_list *in)
 858 {
 859         struct commit_list *i, *j, *k, *ret = NULL;
 860
 861         if (!in)
 862                 return ret;
 863
 864         commit_list_insert(in->item, &ret);
 865
 866         for (i = in->next; i; i = i->next) {
 867                 struct commit_list *new = NULL, *end = NULL;
 868
 869                 for (j = ret; j; j = j->next) {
 870                         struct commit_list *bases;
 871                         bases = get_merge_bases(i->item, j->item);
 872                         if (!new)
 873                                 new = bases;
 874                         else
 875                                 end->next = bases;
 876                         for (k = bases; k; k = k->next)
 877                                 end = k;
 878                 }
 879                 ret = new;
 880         }
 881         return ret;
 882 }
 883
 884 static int remove_redundant(struct commit **array, int cnt)
 885 {
 886         /*
 887          * Some commit in the array may be an ancestor of
 888          * another commit.  Move such commit to the end of
 889          * the array, and return the number of commits that
 890          * are independent from each other.
 891          */
 892         struct commit **work;
 893         unsigned char *redundant;
 894         int *filled_index;
 895         int i, j, filled;
 896
 897         work = xcalloc(cnt, sizeof(*work));
 898         redundant = xcalloc(cnt, 1);
 899         filled_index = xmalloc(sizeof(*filled_index) * (cnt - 1));
 900
 901         for (i = 0; i < cnt; i++)
 902                 parse_commit(array[i]);
 903         for (i = 0; i < cnt; i++) {
 904                 struct commit_list *common;
 905
 906                 if (redundant[i])
 907                         continue;
 908                 for (j = filled = 0; j < cnt; j++) {
 909                         if (i == j || redundant[j])
 910                                 continue;
 911                         filled_index[filled] = j;
 912                         work[filled++] = array[j];
 913                 }
 914                 common = paint_down_to_common(array[i], filled, work);
 915                 if (array[i]->object.flags & PARENT2)
 916                         redundant[i] = 1;
 917                 for (j = 0; j < filled; j++)
 918                         if (work[j]->object.flags & PARENT1)
 919                                 redundant[filled_index[j]] = 1;
 920                 clear_commit_marks(array[i], all_flags);
 921                 for (j = 0; j < filled; j++)
 922                         clear_commit_marks(work[j], all_flags);
 923                 free_commit_list(common);
 924         }
 925
 926         /* Now collect the result */
 927         memcpy(work, array, sizeof(*array) * cnt);
 928         for (i = filled = 0; i < cnt; i++)
 929                 if (!redundant[i])
 930                         array[filled++] = work[i];
 931         for (j = filled, i = 0; i < cnt; i++)
 932                 if (redundant[i])
 933                         array[j++] = work[i];
 934         free(work);
 935         free(redundant);
 936         free(filled_index);
 937         return filled;
 938 }
 939
 940 static struct commit_list *get_merge_bases_many_0(struct commit *one,
 941                                                   int n,
 942                                                   struct commit **twos,
 943                                                   int cleanup)
 944 {
 945         struct commit_list *list;
 946         struct commit **rslt;
 947         struct commit_list *result;
 948         int cnt, i;
 949
 950         result = merge_bases_many(one, n, twos);
 951         for (i = 0; i < n; i++) {
 952                 if (one == twos[i])
 953                         return result;
 954         }
 955         if (!result || !result->next) {
 956                 if (cleanup) {
 957                         clear_commit_marks(one, all_flags);
 958                         clear_commit_marks_many(n, twos, all_flags);
 959                 }
 960                 return result;
 961         }
 962
 963         /* There are more than one */
 964         cnt = commit_list_count(result);
 965         rslt = xcalloc(cnt, sizeof(*rslt));
 966         for (list = result, i = 0; list; list = list->next)
 967                 rslt[i++] = list->item;
 968         free_commit_list(result);
 969
 970         clear_commit_marks(one, all_flags);
 971         clear_commit_marks_many(n, twos, all_flags);
 972
 973         cnt = remove_redundant(rslt, cnt);
 974         result = NULL;
 975         for (i = 0; i < cnt; i++)
 976                 commit_list_insert_by_date(rslt[i], &result);
 977         free(rslt);
 978         return result;
 979 }
 980
 981 struct commit_list *get_merge_bases_many(struct commit *one,
 982                                          int n,
 983                                          struct commit **twos)
 984 {
 985         return get_merge_bases_many_0(one, n, twos, 1);
 986 }
 987
 988 struct commit_list *get_merge_bases_many_dirty(struct commit *one,
 989                                                int n,
 990                                                struct commit **twos)
 991 {
 992         return get_merge_bases_many_0(one, n, twos, 0);
 993 }
 994
 995 struct commit_list *get_merge_bases(struct commit *one, struct commit *two)
 996 {
 997         return get_merge_bases_many_0(one, 1, &two, 1);
 998 }
 999
1000 /*
1001  * Is "commit" a descendant of one of the elements on the "with_commit" list?
1002  */
1003 int is_descendant_of(struct commit *commit, struct commit_list *with_commit)
1004 {
1005         if (!with_commit)
1006                 return 1;
1007         while (with_commit) {
1008                 struct commit *other;
1009
1010                 other = with_commit->item;
1011                 with_commit = with_commit->next;
1012                 if (in_merge_bases(other, commit))
1013                         return 1;
1014         }
1015         return 0;
1016 }
1017
1018 /*
1019  * Is "commit" an ancestor of one of the "references"?
1020  */
1021 int in_merge_bases_many(struct commit *commit, int nr_reference, struct commit **reference)
1022 {
1023         struct commit_list *bases;
1024         int ret = 0, i;
1025
1026         if (parse_commit(commit))
1027                 return ret;
1028         for (i = 0; i < nr_reference; i++)
1029                 if (parse_commit(reference[i]))
1030                         return ret;
1031
1032         bases = paint_down_to_common(commit, nr_reference, reference);
1033         if (commit->object.flags & PARENT2)
1034                 ret = 1;
1035         clear_commit_marks(commit, all_flags);
1036         clear_commit_marks_many(nr_reference, reference, all_flags);
1037         free_commit_list(bases);
1038         return ret;
1039 }
1040
1041 /*
1042  * Is "commit" an ancestor of (i.e. reachable from) the "reference"?
1043  */
1044 int in_merge_bases(struct commit *commit, struct commit *reference)
1045 {
1046         return in_merge_bases_many(commit, 1, &reference);
1047 }
1048
1049 struct commit_list *reduce_heads(struct commit_list *heads)
1050 {
1051         struct commit_list *p;
1052         struct commit_list *result = NULL, **tail = &result;
1053         struct commit **array;
1054         int num_head, i;
1055
1056         if (!heads)
1057                 return NULL;
1058
1059         /* Uniquify */
1060         for (p = heads; p; p = p->next)
1061                 p->item->object.flags &= ~STALE;
1062         for (p = heads, num_head = 0; p; p = p->next) {
1063                 if (p->item->object.flags & STALE)
1064                         continue;
1065                 p->item->object.flags |= STALE;
1066                 num_head++;
1067         }
1068         array = xcalloc(num_head, sizeof(*array));
1069         for (p = heads, i = 0; p; p = p->next) {
1070                 if (p->item->object.flags & STALE) {
1071                         array[i++] = p->item;
1072                         p->item->object.flags &= ~STALE;
1073                 }
1074         }
1075         num_head = remove_redundant(array, num_head);
1076         for (i = 0; i < num_head; i++)
1077                 tail = &commit_list_insert(array[i], tail)->next;
1078         return result;
1079 }
1080
1081 static const char gpg_sig_header[] = "gpgsig";
1082 static const int gpg_sig_header_len = sizeof(gpg_sig_header) - 1;
1083
1084 static int do_sign_commit(struct strbuf *buf, const char *keyid)
1085 {
1086         struct strbuf sig = STRBUF_INIT;
1087         int inspos, copypos;
1088
1089         /* find the end of the header */
1090         inspos = strstr(buf->buf, "\n\n") - buf->buf + 1;
1091
1092         if (!keyid || !*keyid)
1093                 keyid = get_signing_key();
1094         if (sign_buffer(buf, &sig, keyid)) {
1095                 strbuf_release(&sig);
1096                 return -1;
1097         }
1098
1099         for (copypos = 0; sig.buf[copypos]; ) {
1100                 const char *bol = sig.buf + copypos;
1101                 const char *eol = strchrnul(bol, '\n');
1102                 int len = (eol - bol) + !!*eol;
1103
1104                 if (!copypos) {
1105                         strbuf_insert(buf, inspos, gpg_sig_header, gpg_sig_header_len);
1106                         inspos += gpg_sig_header_len;
1107                 }
1108                 strbuf_insert(buf, inspos++, " ", 1);
1109                 strbuf_insert(buf, inspos, bol, len);
1110                 inspos += len;
1111                 copypos += len;
1112         }
1113         strbuf_release(&sig);
1114         return 0;
1115 }
1116
1117 int parse_signed_commit(const struct commit *commit,
1118                         struct strbuf *payload, struct strbuf *signature)
1119 {
1120
1121         unsigned long size;
1122         const char *buffer = get_commit_buffer(commit, &size);
1123         int in_signature, saw_signature = -1;
1124         const char *line, *tail;
1125
1126         line = buffer;
1127         tail = buffer + size;
1128         in_signature = 0;
1129         saw_signature = 0;
1130         while (line < tail) {
1131                 const char *sig = NULL;
1132                 const char *next = memchr(line, '\n', tail - line);
1133
1134                 next = next ? next + 1 : tail;
1135                 if (in_signature && line[0] == ' ')
1136                         sig = line + 1;
1137                 else if (starts_with(line, gpg_sig_header) &&
1138                          line[gpg_sig_header_len] == ' ')
1139                         sig = line + gpg_sig_header_len + 1;
1140                 if (sig) {
1141                         strbuf_add(signature, sig, next - sig);
1142                         saw_signature = 1;
1143                         in_signature = 1;
1144                 } else {
1145                         if (*line == '\n')
1146                                 /* dump the whole remainder of the buffer */
1147                                 next = tail;
1148                         strbuf_add(payload, line, next - line);
1149                         in_signature = 0;
1150                 }
1151                 line = next;
1152         }
1153         unuse_commit_buffer(commit, buffer);
1154         return saw_signature;
1155 }
1156
1157 int remove_signature(struct strbuf *buf)
1158 {
1159         const char *line = buf->buf;
1160         const char *tail = buf->buf + buf->len;
1161         int in_signature = 0;
1162         const char *sig_start = NULL;
1163         const char *sig_end = NULL;
1164
1165         while (line < tail) {
1166                 const char *next = memchr(line, '\n', tail - line);
1167                 next = next ? next + 1 : tail;
1168
1169                 if (in_signature && line[0] == ' ')
1170                         sig_end = next;
1171                 else if (starts_with(line, gpg_sig_header) &&
1172                          line[gpg_sig_header_len] == ' ') {
1173                         sig_start = line;
1174                         sig_end = next;
1175                         in_signature = 1;
1176                 } else {
1177                         if (*line == '\n')
1178                                 /* dump the whole remainder of the buffer */
1179                                 next = tail;
1180                         in_signature = 0;
1181                 }
1182                 line = next;
1183         }
1184
1185         if (sig_start)
1186                 strbuf_remove(buf, sig_start - buf->buf, sig_end - sig_start);
1187
1188         return sig_start != NULL;
1189 }
1190
1191 static void handle_signed_tag(struct commit *parent, struct commit_extra_header ***tail)
1192 {
1193         struct merge_remote_desc *desc;
1194         struct commit_extra_header *mergetag;
1195         char *buf;
1196         unsigned long size, len;
1197         enum object_type type;
1198
1199         desc = merge_remote_util(parent);
1200         if (!desc || !desc->obj)
1201                 return;
1202         buf = read_sha1_file(desc->obj->sha1, &type, &size);
1203         if (!buf || type != OBJ_TAG)
1204                 goto free_return;
1205         len = parse_signature(buf, size);
1206         if (size == len)
1207                 goto free_return;
1208         /*
1209          * We could verify this signature and either omit the tag when
1210          * it does not validate, but the integrator may not have the
1211          * public key of the signer of the tag he is merging, while a
1212          * later auditor may have it while auditing, so let's not run
1213          * verify-signed-buffer here for now...
1214          *
1215          * if (verify_signed_buffer(buf, len, buf + len, size - len, ...))
1216          *      warn("warning: signed tag unverified.");
1217          */
1218         mergetag = xcalloc(1, sizeof(*mergetag));
1219         mergetag->key = xstrdup("mergetag");
1220         mergetag->value = buf;
1221         mergetag->len = size;
1222
1223         **tail = mergetag;
1224         *tail = &mergetag->next;
1225         return;
1226
1227 free_return:
1228         free(buf);
1229 }
1230
1231 void check_commit_signature(const struct commit *commit, struct signature_check *sigc)
1232 {
1233         struct strbuf payload = STRBUF_INIT;
1234         struct strbuf signature = STRBUF_INIT;
1235         struct strbuf gpg_output = STRBUF_INIT;
1236         struct strbuf gpg_status = STRBUF_INIT;
1237         int status;
1238
1239         sigc->result = 'N';
1240
1241         if (parse_signed_commit(commit, &payload, &signature) <= 0)
1242                 goto out;
1243         status = verify_signed_buffer(payload.buf, payload.len,
1244                                       signature.buf, signature.len,
1245                                       &gpg_output, &gpg_status);
1246         if (status && !gpg_output.len)
1247                 goto out;
1248         sigc->payload = strbuf_detach(&payload, NULL);
1249         sigc->gpg_output = strbuf_detach(&gpg_output, NULL);
1250         sigc->gpg_status = strbuf_detach(&gpg_status, NULL);
1251         parse_gpg_output(sigc);
1252
1253  out:
1254         strbuf_release(&gpg_status);
1255         strbuf_release(&gpg_output);
1256         strbuf_release(&payload);
1257         strbuf_release(&signature);
1258 }
1259
1260
1261
1262 void append_merge_tag_headers(struct commit_list *parents,
1263                               struct commit_extra_header ***tail)
1264 {
1265         while (parents) {
1266                 struct commit *parent = parents->item;
1267                 handle_signed_tag(parent, tail);
1268                 parents = parents->next;
1269         }
1270 }
1271
1272 static void add_extra_header(struct strbuf *buffer,
1273                              struct commit_extra_header *extra)
1274 {
1275         strbuf_addstr(buffer, extra->key);
1276         if (extra->len)
1277                 strbuf_add_lines(buffer, " ", extra->value, extra->len);
1278         else
1279                 strbuf_addch(buffer, '\n');
1280 }
1281
1282 struct commit_extra_header *read_commit_extra_headers(struct commit *commit,
1283                                                       const char **exclude)
1284 {
1285         struct commit_extra_header *extra = NULL;
1286         unsigned long size;
1287         const char *buffer = get_commit_buffer(commit, &size);
1288         extra = read_commit_extra_header_lines(buffer, size, exclude);
1289         unuse_commit_buffer(commit, buffer);
1290         return extra;
1291 }
1292
1293 void for_each_mergetag(each_mergetag_fn fn, struct commit *commit, void *data)
1294 {
1295         struct commit_extra_header *extra, *to_free;
1296
1297         to_free = read_commit_extra_headers(commit, NULL);
1298         for (extra = to_free; extra; extra = extra->next) {
1299                 if (strcmp(extra->key, "mergetag"))
1300                         continue; /* not a merge tag */
1301                 fn(commit, extra, data);
1302         }
1303         free_commit_extra_headers(to_free);
1304 }
1305
1306 static inline int standard_header_field(const char *field, size_t len)
1307 {
1308         return ((len == 4 && !memcmp(field, "tree ", 5)) ||
1309                 (len == 6 && !memcmp(field, "parent ", 7)) ||
1310                 (len == 6 && !memcmp(field, "author ", 7)) ||
1311                 (len == 9 && !memcmp(field, "committer ", 10)) ||
1312                 (len == 8 && !memcmp(field, "encoding ", 9)));
1313 }
1314
1315 static int excluded_header_field(const char *field, size_t len, const char **exclude)
1316 {
1317         if (!exclude)
1318                 return 0;
1319
1320         while (*exclude) {
1321                 size_t xlen = strlen(*exclude);
1322                 if (len == xlen &&
1323                     !memcmp(field, *exclude, xlen) && field[xlen] == ' ')
1324                         return 1;
1325                 exclude++;
1326         }
1327         return 0;
1328 }
1329
1330 static struct commit_extra_header *read_commit_extra_header_lines(
1331         const char *buffer, size_t size,
1332         const char **exclude)
1333 {
1334         struct commit_extra_header *extra = NULL, **tail = &extra, *it = NULL;
1335         const char *line, *next, *eof, *eob;
1336         struct strbuf buf = STRBUF_INIT;
1337
1338         for (line = buffer, eob = line + size;
1339              line < eob && *line != '\n';
1340              line = next) {
1341                 next = memchr(line, '\n', eob - line);
1342                 next = next ? next + 1 : eob;
1343                 if (*line == ' ') {
1344                         /* continuation */
1345                         if (it)
1346                                 strbuf_add(&buf, line + 1, next - (line + 1));
1347                         continue;
1348                 }
1349                 if (it)
1350                         it->value = strbuf_detach(&buf, &it->len);
1351                 strbuf_reset(&buf);
1352                 it = NULL;
1353
1354                 eof = strchr(line, ' ');
1355                 if (next <= eof)
1356                         eof = next;
1357
1358                 if (standard_header_field(line, eof - line) ||
1359                     excluded_header_field(line, eof - line, exclude))
1360                         continue;
1361
1362                 it = xcalloc(1, sizeof(*it));
1363                 it->key = xmemdupz(line, eof-line);
1364                 *tail = it;
1365                 tail = &it->next;
1366                 if (eof + 1 < next)
1367                         strbuf_add(&buf, eof + 1, next - (eof + 1));
1368         }
1369         if (it)
1370                 it->value = strbuf_detach(&buf, &it->len);
1371         return extra;
1372 }
1373
1374 void free_commit_extra_headers(struct commit_extra_header *extra)
1375 {
1376         while (extra) {
1377                 struct commit_extra_header *next = extra->next;
1378                 free(extra->key);
1379                 free(extra->value);
1380                 free(extra);
1381                 extra = next;
1382         }
1383 }
1384
1385 int commit_tree(const char *msg, size_t msg_len,
1386                 const unsigned char *tree,
1387                 struct commit_list *parents, unsigned char *ret,
1388                 const char *author, const char *sign_commit)
1389 {
1390         struct commit_extra_header *extra = NULL, **tail = &extra;
1391         int result;
1392
1393         append_merge_tag_headers(parents, &tail);
1394         result = commit_tree_extended(msg, msg_len, tree, parents, ret,
1395                                       author, sign_commit, extra);
1396         free_commit_extra_headers(extra);
1397         return result;
1398 }
1399
1400 static int find_invalid_utf8(const char *buf, int len)
1401 {
1402         int offset = 0;
1403         static const unsigned int max_codepoint[] = {
1404                 0x7f, 0x7ff, 0xffff, 0x10ffff
1405         };
1406
1407         while (len) {
1408                 unsigned char c = *buf++;
1409                 int bytes, bad_offset;
1410                 unsigned int codepoint;
1411                 unsigned int min_val, max_val;
1412
1413                 len--;
1414                 offset++;
1415
1416                 /* Simple US-ASCII? No worries. */
1417                 if (c < 0x80)
1418                         continue;
1419
1420                 bad_offset = offset-1;
1421
1422                 /*
1423                  * Count how many more high bits set: that's how
1424                  * many more bytes this sequence should have.
1425                  */
1426                 bytes = 0;
1427                 while (c & 0x40) {
1428                         c <<= 1;
1429                         bytes++;
1430                 }
1431
1432                 /*
1433                  * Must be between 1 and 3 more bytes.  Longer sequences result in
1434                  * codepoints beyond U+10FFFF, which are guaranteed never to exist.
1435                  */
1436                 if (bytes < 1 || 3 < bytes)
1437                         return bad_offset;
1438
1439                 /* Do we *have* that many bytes? */
1440                 if (len < bytes)
1441                         return bad_offset;
1442
1443                 /*
1444                  * Place the encoded bits at the bottom of the value and compute the
1445                  * valid range.
1446                  */
1447                 codepoint = (c & 0x7f) >> bytes;
1448                 min_val = max_codepoint[bytes-1] + 1;
1449                 max_val = max_codepoint[bytes];
1450
1451                 offset += bytes;
1452                 len -= bytes;
1453
1454                 /* And verify that they are good continuation bytes */
1455                 do {
1456                         codepoint <<= 6;
1457                         codepoint |= *buf & 0x3f;
1458                         if ((*buf++ & 0xc0) != 0x80)
1459                                 return bad_offset;
1460                 } while (--bytes);
1461
1462                 /* Reject codepoints that are out of range for the sequence length. */
1463                 if (codepoint < min_val || codepoint > max_val)
1464                         return bad_offset;
1465                 /* Surrogates are only for UTF-16 and cannot be encoded in UTF-8. */
1466                 if ((codepoint & 0x1ff800) == 0xd800)
1467                         return bad_offset;
1468                 /* U+xxFFFE and U+xxFFFF are guaranteed non-characters. */
1469                 if ((codepoint & 0xfffe) == 0xfffe)
1470                         return bad_offset;
1471                 /* So are anything in the range U+FDD0..U+FDEF. */
1472                 if (codepoint >= 0xfdd0 && codepoint <= 0xfdef)
1473                         return bad_offset;
1474         }
1475         return -1;
1476 }
1477
1478 /*
1479  * This verifies that the buffer is in proper utf8 format.
1480  *
1481  * If it isn't, it assumes any non-utf8 characters are Latin1,
1482  * and does the conversion.
1483  */
1484 static int verify_utf8(struct strbuf *buf)
1485 {
1486         int ok = 1;
1487         long pos = 0;
1488
1489         for (;;) {
1490                 int bad;
1491                 unsigned char c;
1492                 unsigned char replace[2];
1493
1494                 bad = find_invalid_utf8(buf->buf + pos, buf->len - pos);
1495                 if (bad < 0)
1496                         return ok;
1497                 pos += bad;
1498                 ok = 0;
1499                 c = buf->buf[pos];
1500                 strbuf_remove(buf, pos, 1);
1501
1502                 /* We know 'c' must be in the range 128-255 */
1503                 replace[0] = 0xc0 + (c >> 6);
1504                 replace[1] = 0x80 + (c & 0x3f);
1505                 strbuf_insert(buf, pos, replace, 2);
1506                 pos += 2;
1507         }
1508 }
1509
1510 static const char commit_utf8_warn[] =
1511 "Warning: commit message did not conform to UTF-8.\n"
1512 "You may want to amend it after fixing the message, or set the config\n"
1513 "variable i18n.commitencoding to the encoding your project uses.\n";
1514
1515 int commit_tree_extended(const char *msg, size_t msg_len,
1516                          const unsigned char *tree,
1517                          struct commit_list *parents, unsigned char *ret,
1518                          const char *author, const char *sign_commit,
1519                          struct commit_extra_header *extra)
1520 {
1521         int result;
1522         int encoding_is_utf8;
1523         struct strbuf buffer;
1524
1525         assert_sha1_type(tree, OBJ_TREE);
1526
1527         if (memchr(msg, '\0', msg_len))
1528                 return error("a NUL byte in commit log message not allowed.");
1529
1530         /* Not having i18n.commitencoding is the same as having utf-8 */
1531         encoding_is_utf8 = is_encoding_utf8(git_commit_encoding);
1532
1533         strbuf_init(&buffer, 8192); /* should avoid reallocs for the headers */
1534         strbuf_addf(&buffer, "tree %s\n", sha1_to_hex(tree));
1535
1536         /*
1537          * NOTE! This ordering means that the same exact tree merged with a
1538          * different order of parents will be a _different_ changeset even
1539          * if everything else stays the same.
1540          */
1541         while (parents) {
1542                 struct commit_list *next = parents->next;
1543                 struct commit *parent = parents->item;
1544
1545                 strbuf_addf(&buffer, "parent %s\n",
1546                             sha1_to_hex(parent->object.sha1));
1547                 free(parents);
1548                 parents = next;
1549         }
1550
1551         /* Person/date information */
1552         if (!author)
1553                 author = git_author_info(IDENT_STRICT);
1554         strbuf_addf(&buffer, "author %s\n", author);
1555         strbuf_addf(&buffer, "committer %s\n", git_committer_info(IDENT_STRICT));
1556         if (!encoding_is_utf8)
1557                 strbuf_addf(&buffer, "encoding %s\n", git_commit_encoding);
1558
1559         while (extra) {
1560                 add_extra_header(&buffer, extra);
1561                 extra = extra->next;
1562         }
1563         strbuf_addch(&buffer, '\n');
1564
1565         /* And add the comment */
1566         strbuf_add(&buffer, msg, msg_len);
1567
1568         /* And check the encoding */
1569         if (encoding_is_utf8 && !verify_utf8(&buffer))
1570                 fprintf(stderr, commit_utf8_warn);
1571
1572         if (sign_commit && do_sign_commit(&buffer, sign_commit))
1573                 return -1;
1574
1575         result = write_sha1_file(buffer.buf, buffer.len, commit_type, ret);
1576         strbuf_release(&buffer);
1577         return result;
1578 }
1579
1580 struct commit *get_merge_parent(const char *name)
1581 {
1582         struct object *obj;
1583         struct commit *commit;
1584         unsigned char sha1[20];
1585         if (get_sha1(name, sha1))
1586                 return NULL;
1587         obj = parse_object(sha1);
1588         commit = (struct commit *)peel_to_type(name, 0, obj, OBJ_COMMIT);
1589         if (commit && !commit->util) {
1590                 struct merge_remote_desc *desc;
1591                 desc = xmalloc(sizeof(*desc));
1592                 desc->obj = obj;
1593                 desc->name = strdup(name);
1594                 commit->util = desc;
1595         }
1596         return commit;
1597 }
1598
1599 /*
1600  * Append a commit to the end of the commit_list.
1601  *
1602  * next starts by pointing to the variable that holds the head of an
1603  * empty commit_list, and is updated to point to the "next" field of
1604  * the last item on the list as new commits are appended.
1605  *
1606  * Usage example:
1607  *
1608  *     struct commit_list *list;
1609  *     struct commit_list **next = &list;
1610  *
1611  *     next = commit_list_append(c1, next);
1612  *     next = commit_list_append(c2, next);
1613  *     assert(commit_list_count(list) == 2);
1614  *     return list;
1615  */
1616 struct commit_list **commit_list_append(struct commit *commit,
1617                                         struct commit_list **next)
1618 {
1619         struct commit_list *new = xmalloc(sizeof(struct commit_list));
1620         new->item = commit;
1621         *next = new;
1622         new->next = NULL;
1623         return &new->next;
1624 }
1625
1626 void print_commit_list(struct commit_list *list,
1627                        const char *format_cur,
1628                        const char *format_last)
1629 {
1630         for ( ; list; list = list->next) {
1631                 const char *format = list->next ? format_cur : format_last;
1632                 printf(format, sha1_to_hex(list->item->object.sha1));
1633         }
1634 }
1635
1636 const char *find_commit_header(const char *msg, const char *key, size_t *out_len)
1637 {
1638         int key_len = strlen(key);
1639         const char *line = msg;
1640
1641         while (line) {
1642                 const char *eol = strchrnul(line, '\n');
1643
1644                 if (line == eol)
1645                         return NULL;
1646
1647                 if (eol - line > key_len &&
1648                     !strncmp(line, key, key_len) &&
1649                     line[key_len] == ' ') {
1650                         *out_len = eol - line - key_len - 1;
1651                         return line + key_len + 1;
1652                 }
1653                 line = *eol ? eol + 1 : NULL;
1654         }
1655         return NULL;
1656 }
1657
1658 /*
1659  * Inspect sb and determine the true "end" of the log message, in
1660  * order to find where to put a new Signed-off-by: line.  Ignored are
1661  * trailing comment lines and blank lines, and also the traditional
1662  * "Conflicts:" block that is not commented out, so that we can use
1663  * "git commit -s --amend" on an existing commit that forgot to remove
1664  * it.
1665  *
1666  * Returns the number of bytes from the tail to ignore, to be fed as
1667  * the second parameter to append_signoff().
1668  */
1669 int ignore_non_trailer(struct strbuf *sb)
1670 {
1671         int boc = 0;
1672         int bol = 0;
1673         int in_old_conflicts_block = 0;
1674
1675         while (bol < sb->len) {
1676                 char *next_line;
1677
1678                 if (!(next_line = memchr(sb->buf + bol, '\n', sb->len - bol)))
1679                         next_line = sb->buf + sb->len;
1680                 else
1681                         next_line++;
1682
1683                 if (sb->buf[bol] == comment_line_char || sb->buf[bol] == '\n') {
1684                         /* is this the first of the run of comments? */
1685                         if (!boc)
1686                                 boc = bol;
1687                         /* otherwise, it is just continuing */
1688                 } else if (starts_with(sb->buf + bol, "Conflicts:\n")) {
1689                         in_old_conflicts_block = 1;
1690                         if (!boc)
1691                                 boc = bol;
1692                 } else if (in_old_conflicts_block && sb->buf[bol] == '\t') {
1693                         ; /* a pathname in the conflicts block */
1694                 } else if (boc) {
1695                         /* the previous was not trailing comment */
1696                         boc = 0;
1697                         in_old_conflicts_block = 0;
1698                 }
1699                 bol = next_line - sb->buf;
1700         }
1701         return boc ? sb->len - boc : 0;
1702 }