commit.c

   1 #include "cache.h"
   2 #include "tag.h"
   3 #include "commit.h"
   4 #include "pkt-line.h"
   5 #include "utf8.h"
   6 #include "diff.h"
   7 #include "revision.h"
   8 #include "notes.h"
   9 #include "gpg-interface.h"
  10 #include "mergesort.h"
  11 #include "commit-slab.h"
  12 #include "prio-queue.h"
  13 #include "sha1-lookup.h"
  14
  15 static struct commit_extra_header *read_commit_extra_header_lines(const char *buf, size_t len, const char **);
  16
  17 int save_commit_buffer = 1;
  18
  19 const char *commit_type = "commit";
  20
  21 struct commit *lookup_commit_reference_gently(const unsigned char *sha1,
  22                                               int quiet)
  23 {
  24         struct object *obj = deref_tag(parse_object(sha1), NULL, 0);
  25
  26         if (!obj)
  27                 return NULL;
  28         return object_as_type(obj, OBJ_COMMIT, quiet);
  29 }
  30
  31 struct commit *lookup_commit_reference(const unsigned char *sha1)
  32 {
  33         return lookup_commit_reference_gently(sha1, 0);
  34 }
  35
  36 struct commit *lookup_commit_or_die(const unsigned char *sha1, const char *ref_name)
  37 {
  38         struct commit *c = lookup_commit_reference(sha1);
  39         if (!c)
  40                 die(_("could not parse %s"), ref_name);
  41         if (hashcmp(sha1, c->object.sha1)) {
  42                 warning(_("%s %s is not a commit!"),
  43                         ref_name, sha1_to_hex(sha1));
  44         }
  45         return c;
  46 }
  47
  48 struct commit *lookup_commit(const unsigned char *sha1)
  49 {
  50         struct object *obj = lookup_object(sha1);
  51         if (!obj)
  52                 return create_object(sha1, alloc_commit_node());
  53         return object_as_type(obj, OBJ_COMMIT, 0);
  54 }
  55
  56 struct commit *lookup_commit_reference_by_name(const char *name)
  57 {
  58         unsigned char sha1[20];
  59         struct commit *commit;
  60
  61         if (get_sha1_committish(name, sha1))
  62                 return NULL;
  63         commit = lookup_commit_reference(sha1);
  64         if (parse_commit(commit))
  65                 return NULL;
  66         return commit;
  67 }
  68
  69 static unsigned long parse_commit_date(const char *buf, const char *tail)
  70 {
  71         const char *dateptr;
  72
  73         if (buf + 6 >= tail)
  74                 return 0;
  75         if (memcmp(buf, "author", 6))
  76                 return 0;
  77         while (buf < tail && *buf++ != '\n')
  78                 /* nada */;
  79         if (buf + 9 >= tail)
  80                 return 0;
  81         if (memcmp(buf, "committer", 9))
  82                 return 0;
  83         while (buf < tail && *buf++ != '>')
  84                 /* nada */;
  85         if (buf >= tail)
  86                 return 0;
  87         dateptr = buf;
  88         while (buf < tail && *buf++ != '\n')
  89                 /* nada */;
  90         if (buf >= tail)
  91                 return 0;
  92         /* dateptr < buf && buf[-1] == '\n', so strtoul will stop at buf-1 */
  93         return strtoul(dateptr, NULL, 10);
  94 }
  95
  96 static struct commit_graft **commit_graft;
  97 static int commit_graft_alloc, commit_graft_nr;
  98
  99 static const unsigned char *commit_graft_sha1_access(size_t index, void *table)
 100 {
 101         struct commit_graft **commit_graft_table = table;
 102         return commit_graft_table[index]->sha1;
 103 }
 104
 105 static int commit_graft_pos(const unsigned char *sha1)
 106 {
 107         return sha1_pos(sha1, commit_graft, commit_graft_nr,
 108                         commit_graft_sha1_access);
 109 }
 110
 111 int register_commit_graft(struct commit_graft *graft, int ignore_dups)
 112 {
 113         int pos = commit_graft_pos(graft->sha1);
 114
 115         if (0 <= pos) {
 116                 if (ignore_dups)
 117                         free(graft);
 118                 else {
 119                         free(commit_graft[pos]);
 120                         commit_graft[pos] = graft;
 121                 }
 122                 return 1;
 123         }
 124         pos = -pos - 1;
 125         ALLOC_GROW(commit_graft, commit_graft_nr + 1, commit_graft_alloc);
 126         commit_graft_nr++;
 127         if (pos < commit_graft_nr)
 128                 memmove(commit_graft + pos + 1,
 129                         commit_graft + pos,
 130                         (commit_graft_nr - pos - 1) *
 131                         sizeof(*commit_graft));
 132         commit_graft[pos] = graft;
 133         return 0;
 134 }
 135
 136 struct commit_graft *read_graft_line(char *buf, int len)
 137 {
 138         /* The format is just "Commit Parent1 Parent2 ...\n" */
 139         int i;
 140         struct commit_graft *graft = NULL;
 141
 142         while (len && isspace(buf[len-1]))
 143                 buf[--len] = '\0';
 144         if (buf[0] == '#' || buf[0] == '\0')
 145                 return NULL;
 146         if ((len + 1) % 41)
 147                 goto bad_graft_data;
 148         i = (len + 1) / 41 - 1;
 149         graft = xmalloc(sizeof(*graft) + 20 * i);
 150         graft->nr_parent = i;
 151         if (get_sha1_hex(buf, graft->sha1))
 152                 goto bad_graft_data;
 153         for (i = 40; i < len; i += 41) {
 154                 if (buf[i] != ' ')
 155                         goto bad_graft_data;
 156                 if (get_sha1_hex(buf + i + 1, graft->parent[i/41]))
 157                         goto bad_graft_data;
 158         }
 159         return graft;
 160
 161 bad_graft_data:
 162         error("bad graft data: %s", buf);
 163         free(graft);
 164         return NULL;
 165 }
 166
 167 static int read_graft_file(const char *graft_file)
 168 {
 169         FILE *fp = fopen(graft_file, "r");
 170         struct strbuf buf = STRBUF_INIT;
 171         if (!fp)
 172                 return -1;
 173         while (!strbuf_getwholeline(&buf, fp, '\n')) {
 174                 /* The format is just "Commit Parent1 Parent2 ...\n" */
 175                 struct commit_graft *graft = read_graft_line(buf.buf, buf.len);
 176                 if (!graft)
 177                         continue;
 178                 if (register_commit_graft(graft, 1))
 179                         error("duplicate graft data: %s", buf.buf);
 180         }
 181         fclose(fp);
 182         strbuf_release(&buf);
 183         return 0;
 184 }
 185
 186 static void prepare_commit_graft(void)
 187 {
 188         static int commit_graft_prepared;
 189         char *graft_file;
 190
 191         if (commit_graft_prepared)
 192                 return;
 193         graft_file = get_graft_file();
 194         read_graft_file(graft_file);
 195         /* make sure shallows are read */
 196         is_repository_shallow();
 197         commit_graft_prepared = 1;
 198 }
 199
 200 struct commit_graft *lookup_commit_graft(const unsigned char *sha1)
 201 {
 202         int pos;
 203         prepare_commit_graft();
 204         pos = commit_graft_pos(sha1);
 205         if (pos < 0)
 206                 return NULL;
 207         return commit_graft[pos];
 208 }
 209
 210 int for_each_commit_graft(each_commit_graft_fn fn, void *cb_data)
 211 {
 212         int i, ret;
 213         for (i = ret = 0; i < commit_graft_nr && !ret; i++)
 214                 ret = fn(commit_graft[i], cb_data);
 215         return ret;
 216 }
 217
 218 int unregister_shallow(const unsigned char *sha1)
 219 {
 220         int pos = commit_graft_pos(sha1);
 221         if (pos < 0)
 222                 return -1;
 223         if (pos + 1 < commit_graft_nr)
 224                 memmove(commit_graft + pos, commit_graft + pos + 1,
 225                                 sizeof(struct commit_graft *)
 226                                 * (commit_graft_nr - pos - 1));
 227         commit_graft_nr--;
 228         return 0;
 229 }
 230
 231 struct commit_buffer {
 232         void *buffer;
 233         unsigned long size;
 234 };
 235 define_commit_slab(buffer_slab, struct commit_buffer);
 236 static struct buffer_slab buffer_slab = COMMIT_SLAB_INIT(1, buffer_slab);
 237
 238 void set_commit_buffer(struct commit *commit, void *buffer, unsigned long size)
 239 {
 240         struct commit_buffer *v = buffer_slab_at(&buffer_slab, commit);
 241         v->buffer = buffer;
 242         v->size = size;
 243 }
 244
 245 const void *get_cached_commit_buffer(const struct commit *commit, unsigned long *sizep)
 246 {
 247         struct commit_buffer *v = buffer_slab_at(&buffer_slab, commit);
 248         if (sizep)
 249                 *sizep = v->size;
 250         return v->buffer;
 251 }
 252
 253 const void *get_commit_buffer(const struct commit *commit, unsigned long *sizep)
 254 {
 255         const void *ret = get_cached_commit_buffer(commit, sizep);
 256         if (!ret) {
 257                 enum object_type type;
 258                 unsigned long size;
 259                 ret = read_sha1_file(commit->object.sha1, &type, &size);
 260                 if (!ret)
 261                         die("cannot read commit object %s",
 262                             sha1_to_hex(commit->object.sha1));
 263                 if (type != OBJ_COMMIT)
 264                         die("expected commit for %s, got %s",
 265                             sha1_to_hex(commit->object.sha1), typename(type));
 266                 if (sizep)
 267                         *sizep = size;
 268         }
 269         return ret;
 270 }
 271
 272 void unuse_commit_buffer(const struct commit *commit, const void *buffer)
 273 {
 274         struct commit_buffer *v = buffer_slab_at(&buffer_slab, commit);
 275         if (v->buffer != buffer)
 276                 free((void *)buffer);
 277 }
 278
 279 void free_commit_buffer(struct commit *commit)
 280 {
 281         struct commit_buffer *v = buffer_slab_at(&buffer_slab, commit);
 282         free(v->buffer);
 283         v->buffer = NULL;
 284         v->size = 0;
 285 }
 286
 287 const void *detach_commit_buffer(struct commit *commit, unsigned long *sizep)
 288 {
 289         struct commit_buffer *v = buffer_slab_at(&buffer_slab, commit);
 290         void *ret;
 291
 292         ret = v->buffer;
 293         if (sizep)
 294                 *sizep = v->size;
 295
 296         v->buffer = NULL;
 297         v->size = 0;
 298         return ret;
 299 }
 300
 301 int parse_commit_buffer(struct commit *item, const void *buffer, unsigned long size)
 302 {
 303         const char *tail = buffer;
 304         const char *bufptr = buffer;
 305         unsigned char parent[20];
 306         struct commit_list **pptr;
 307         struct commit_graft *graft;
 308
 309         if (item->object.parsed)
 310                 return 0;
 311         item->object.parsed = 1;
 312         tail += size;
 313         if (tail <= bufptr + 46 || memcmp(bufptr, "tree ", 5) || bufptr[45] != '\n')
 314                 return error("bogus commit object %s", sha1_to_hex(item->object.sha1));
 315         if (get_sha1_hex(bufptr + 5, parent) < 0)
 316                 return error("bad tree pointer in commit %s",
 317                              sha1_to_hex(item->object.sha1));
 318         item->tree = lookup_tree(parent);
 319         bufptr += 46; /* "tree " + "hex sha1" + "\n" */
 320         pptr = &item->parents;
 321
 322         graft = lookup_commit_graft(item->object.sha1);
 323         while (bufptr + 48 < tail && !memcmp(bufptr, "parent ", 7)) {
 324                 struct commit *new_parent;
 325
 326                 if (tail <= bufptr + 48 ||
 327                     get_sha1_hex(bufptr + 7, parent) ||
 328                     bufptr[47] != '\n')
 329                         return error("bad parents in commit %s", sha1_to_hex(item->object.sha1));
 330                 bufptr += 48;
 331                 /*
 332                  * The clone is shallow if nr_parent < 0, and we must
 333                  * not traverse its real parents even when we unhide them.
 334                  */
 335                 if (graft && (graft->nr_parent < 0 || grafts_replace_parents))
 336                         continue;
 337                 new_parent = lookup_commit(parent);
 338                 if (new_parent)
 339                         pptr = &commit_list_insert(new_parent, pptr)->next;
 340         }
 341         if (graft) {
 342                 int i;
 343                 struct commit *new_parent;
 344                 for (i = 0; i < graft->nr_parent; i++) {
 345                         new_parent = lookup_commit(graft->parent[i]);
 346                         if (!new_parent)
 347                                 continue;
 348                         pptr = &commit_list_insert(new_parent, pptr)->next;
 349                 }
 350         }
 351         item->date = parse_commit_date(bufptr, tail);
 352
 353         return 0;
 354 }
 355
 356 int parse_commit(struct commit *item)
 357 {
 358         enum object_type type;
 359         void *buffer;
 360         unsigned long size;
 361         int ret;
 362
 363         if (!item)
 364                 return -1;
 365         if (item->object.parsed)
 366                 return 0;
 367         buffer = read_sha1_file(item->object.sha1, &type, &size);
 368         if (!buffer)
 369                 return error("Could not read %s",
 370                              sha1_to_hex(item->object.sha1));
 371         if (type != OBJ_COMMIT) {
 372                 free(buffer);
 373                 return error("Object %s not a commit",
 374                              sha1_to_hex(item->object.sha1));
 375         }
 376         ret = parse_commit_buffer(item, buffer, size);
 377         if (save_commit_buffer && !ret) {
 378                 set_commit_buffer(item, buffer, size);
 379                 return 0;
 380         }
 381         free(buffer);
 382         return ret;
 383 }
 384
 385 void parse_commit_or_die(struct commit *item)
 386 {
 387         if (parse_commit(item))
 388                 die("unable to parse commit %s",
 389                     item ? sha1_to_hex(item->object.sha1) : "(null)");
 390 }
 391
 392 int find_commit_subject(const char *commit_buffer, const char **subject)
 393 {
 394         const char *eol;
 395         const char *p = commit_buffer;
 396
 397         while (*p && (*p != '\n' || p[1] != '\n'))
 398                 p++;
 399         if (*p) {
 400                 p += 2;
 401                 for (eol = p; *eol && *eol != '\n'; eol++)
 402                         ; /* do nothing */
 403         } else
 404                 eol = p;
 405
 406         *subject = p;
 407
 408         return eol - p;
 409 }
 410
 411 struct commit_list *commit_list_insert(struct commit *item, struct commit_list **list_p)
 412 {
 413         struct commit_list *new_list = xmalloc(sizeof(struct commit_list));
 414         new_list->item = item;
 415         new_list->next = *list_p;
 416         *list_p = new_list;
 417         return new_list;
 418 }
 419
 420 unsigned commit_list_count(const struct commit_list *l)
 421 {
 422         unsigned c = 0;
 423         for (; l; l = l->next )
 424                 c++;
 425         return c;
 426 }
 427
 428 struct commit_list *copy_commit_list(struct commit_list *list)
 429 {
 430         struct commit_list *head = NULL;
 431         struct commit_list **pp = &head;
 432         while (list) {
 433                 pp = commit_list_append(list->item, pp);
 434                 list = list->next;
 435         }
 436         return head;
 437 }
 438
 439 void free_commit_list(struct commit_list *list)
 440 {
 441         while (list) {
 442                 struct commit_list *temp = list;
 443                 list = temp->next;
 444                 free(temp);
 445         }
 446 }
 447
 448 struct commit_list * commit_list_insert_by_date(struct commit *item, struct commit_list **list)
 449 {
 450         struct commit_list **pp = list;
 451         struct commit_list *p;
 452         while ((p = *pp) != NULL) {
 453                 if (p->item->date < item->date) {
 454                         break;
 455                 }
 456                 pp = &p->next;
 457         }
 458         return commit_list_insert(item, pp);
 459 }
 460
 461 static int commit_list_compare_by_date(const void *a, const void *b)
 462 {
 463         unsigned long a_date = ((const struct commit_list *)a)->item->date;
 464         unsigned long b_date = ((const struct commit_list *)b)->item->date;
 465         if (a_date < b_date)
 466                 return 1;
 467         if (a_date > b_date)
 468                 return -1;
 469         return 0;
 470 }
 471
 472 static void *commit_list_get_next(const void *a)
 473 {
 474         return ((const struct commit_list *)a)->next;
 475 }
 476
 477 static void commit_list_set_next(void *a, void *next)
 478 {
 479         ((struct commit_list *)a)->next = next;
 480 }
 481
 482 void commit_list_sort_by_date(struct commit_list **list)
 483 {
 484         *list = llist_mergesort(*list, commit_list_get_next, commit_list_set_next,
 485                                 commit_list_compare_by_date);
 486 }
 487
 488 struct commit *pop_most_recent_commit(struct commit_list **list,
 489                                       unsigned int mark)
 490 {
 491         struct commit *ret = (*list)->item;
 492         struct commit_list *parents = ret->parents;
 493         struct commit_list *old = *list;
 494
 495         *list = (*list)->next;
 496         free(old);
 497
 498         while (parents) {
 499                 struct commit *commit = parents->item;
 500                 if (!parse_commit(commit) && !(commit->object.flags & mark)) {
 501                         commit->object.flags |= mark;
 502                         commit_list_insert_by_date(commit, list);
 503                 }
 504                 parents = parents->next;
 505         }
 506         return ret;
 507 }
 508
 509 static void clear_commit_marks_1(struct commit_list **plist,
 510                                  struct commit *commit, unsigned int mark)
 511 {
 512         while (commit) {
 513                 struct commit_list *parents;
 514
 515                 if (!(mark & commit->object.flags))
 516                         return;
 517
 518                 commit->object.flags &= ~mark;
 519
 520                 parents = commit->parents;
 521                 if (!parents)
 522                         return;
 523
 524                 while ((parents = parents->next))
 525                         commit_list_insert(parents->item, plist);
 526
 527                 commit = commit->parents->item;
 528         }
 529 }
 530
 531 void clear_commit_marks_many(int nr, struct commit **commit, unsigned int mark)
 532 {
 533         struct commit_list *list = NULL;
 534
 535         while (nr--) {
 536                 commit_list_insert(*commit, &list);
 537                 commit++;
 538         }
 539         while (list)
 540                 clear_commit_marks_1(&list, pop_commit(&list), mark);
 541 }
 542
 543 void clear_commit_marks(struct commit *commit, unsigned int mark)
 544 {
 545         clear_commit_marks_many(1, &commit, mark);
 546 }
 547
 548 void clear_commit_marks_for_object_array(struct object_array *a, unsigned mark)
 549 {
 550         struct object *object;
 551         struct commit *commit;
 552         unsigned int i;
 553
 554         for (i = 0; i < a->nr; i++) {
 555                 object = a->objects[i].item;
 556                 commit = lookup_commit_reference_gently(object->sha1, 1);
 557                 if (commit)
 558                         clear_commit_marks(commit, mark);
 559         }
 560 }
 561
 562 struct commit *pop_commit(struct commit_list **stack)
 563 {
 564         struct commit_list *top = *stack;
 565         struct commit *item = top ? top->item : NULL;
 566
 567         if (top) {
 568                 *stack = top->next;
 569                 free(top);
 570         }
 571         return item;
 572 }
 573
 574 /*
 575  * Topological sort support
 576  */
 577
 578 /* count number of children that have not been emitted */
 579 define_commit_slab(indegree_slab, int);
 580
 581 /* record author-date for each commit object */
 582 define_commit_slab(author_date_slab, unsigned long);
 583
 584 static void record_author_date(struct author_date_slab *author_date,
 585                                struct commit *commit)
 586 {
 587         const char *buffer = get_commit_buffer(commit, NULL);
 588         struct ident_split ident;
 589         const char *ident_line;
 590         size_t ident_len;
 591         char *date_end;
 592         unsigned long date;
 593
 594         ident_line = find_commit_header(buffer, "author", &ident_len);
 595         if (!ident_line)
 596                 goto fail_exit; /* no author line */
 597         if (split_ident_line(&ident, ident_line, ident_len) ||
 598             !ident.date_begin || !ident.date_end)
 599                 goto fail_exit; /* malformed "author" line */
 600
 601         date = strtoul(ident.date_begin, &date_end, 10);
 602         if (date_end != ident.date_end)
 603                 goto fail_exit; /* malformed date */
 604         *(author_date_slab_at(author_date, commit)) = date;
 605
 606 fail_exit:
 607         unuse_commit_buffer(commit, buffer);
 608 }
 609
 610 static int compare_commits_by_author_date(const void *a_, const void *b_,
 611                                           void *cb_data)
 612 {
 613         const struct commit *a = a_, *b = b_;
 614         struct author_date_slab *author_date = cb_data;
 615         unsigned long a_date = *(author_date_slab_at(author_date, a));
 616         unsigned long b_date = *(author_date_slab_at(author_date, b));
 617
 618         /* newer commits with larger date first */
 619         if (a_date < b_date)
 620                 return 1;
 621         else if (a_date > b_date)
 622                 return -1;
 623         return 0;
 624 }
 625
 626 int compare_commits_by_commit_date(const void *a_, const void *b_, void *unused)
 627 {
 628         const struct commit *a = a_, *b = b_;
 629         /* newer commits with larger date first */
 630         if (a->date < b->date)
 631                 return 1;
 632         else if (a->date > b->date)
 633                 return -1;
 634         return 0;
 635 }
 636
 637 /*
 638  * Performs an in-place topological sort on the list supplied.
 639  */
 640 void sort_in_topological_order(struct commit_list **list, enum rev_sort_order sort_order)
 641 {
 642         struct commit_list *next, *orig = *list;
 643         struct commit_list **pptr;
 644         struct indegree_slab indegree;
 645         struct prio_queue queue;
 646         struct commit *commit;
 647         struct author_date_slab author_date;
 648
 649         if (!orig)
 650                 return;
 651         *list = NULL;
 652
 653         init_indegree_slab(&indegree);
 654         memset(&queue, '\0', sizeof(queue));
 655
 656         switch (sort_order) {
 657         default: /* REV_SORT_IN_GRAPH_ORDER */
 658                 queue.compare = NULL;
 659                 break;
 660         case REV_SORT_BY_COMMIT_DATE:
 661                 queue.compare = compare_commits_by_commit_date;
 662                 break;
 663         case REV_SORT_BY_AUTHOR_DATE:
 664                 init_author_date_slab(&author_date);
 665                 queue.compare = compare_commits_by_author_date;
 666                 queue.cb_data = &author_date;
 667                 break;
 668         }
 669
 670         /* Mark them and clear the indegree */
 671         for (next = orig; next; next = next->next) {
 672                 struct commit *commit = next->item;
 673                 *(indegree_slab_at(&indegree, commit)) = 1;
 674                 /* also record the author dates, if needed */
 675                 if (sort_order == REV_SORT_BY_AUTHOR_DATE)
 676                         record_author_date(&author_date, commit);
 677         }
 678
 679         /* update the indegree */
 680         for (next = orig; next; next = next->next) {
 681                 struct commit_list *parents = next->item->parents;
 682                 while (parents) {
 683                         struct commit *parent = parents->item;
 684                         int *pi = indegree_slab_at(&indegree, parent);
 685
 686                         if (*pi)
 687                                 (*pi)++;
 688                         parents = parents->next;
 689                 }
 690         }
 691
 692         /*
 693          * find the tips
 694          *
 695          * tips are nodes not reachable from any other node in the list
 696          *
 697          * the tips serve as a starting set for the work queue.
 698          */
 699         for (next = orig; next; next = next->next) {
 700                 struct commit *commit = next->item;
 701
 702                 if (*(indegree_slab_at(&indegree, commit)) == 1)
 703                         prio_queue_put(&queue, commit);
 704         }
 705
 706         /*
 707          * This is unfortunate; the initial tips need to be shown
 708          * in the order given from the revision traversal machinery.
 709          */
 710         if (sort_order == REV_SORT_IN_GRAPH_ORDER)
 711                 prio_queue_reverse(&queue);
 712
 713         /* We no longer need the commit list */
 714         free_commit_list(orig);
 715
 716         pptr = list;
 717         *list = NULL;
 718         while ((commit = prio_queue_get(&queue)) != NULL) {
 719                 struct commit_list *parents;
 720
 721                 for (parents = commit->parents; parents ; parents = parents->next) {
 722                         struct commit *parent = parents->item;
 723                         int *pi = indegree_slab_at(&indegree, parent);
 724
 725                         if (!*pi)
 726                                 continue;
 727
 728                         /*
 729                          * parents are only enqueued for emission
 730                          * when all their children have been emitted thereby
 731                          * guaranteeing topological order.
 732                          */
 733                         if (--(*pi) == 1)
 734                                 prio_queue_put(&queue, parent);
 735                 }
 736                 /*
 737                  * all children of commit have already been
 738                  * emitted. we can emit it now.
 739                  */
 740                 *(indegree_slab_at(&indegree, commit)) = 0;
 741
 742                 pptr = &commit_list_insert(commit, pptr)->next;
 743         }
 744
 745         clear_indegree_slab(&indegree);
 746         clear_prio_queue(&queue);
 747         if (sort_order == REV_SORT_BY_AUTHOR_DATE)
 748                 clear_author_date_slab(&author_date);
 749 }
 750
 751 /* merge-base stuff */
 752
 753 /* Remember to update object flag allocation in object.h */
 754 #define PARENT1         (1u<<16)
 755 #define PARENT2         (1u<<17)
 756 #define STALE           (1u<<18)
 757 #define RESULT          (1u<<19)
 758
 759 static const unsigned all_flags = (PARENT1 | PARENT2 | STALE | RESULT);
 760
 761 static int queue_has_nonstale(struct prio_queue *queue)
 762 {
 763         int i;
 764         for (i = 0; i < queue->nr; i++) {
 765                 struct commit *commit = queue->array[i].data;
 766                 if (!(commit->object.flags & STALE))
 767                         return 1;
 768         }
 769         return 0;
 770 }
 771
 772 /* all input commits in one and twos[] must have been parsed! */
 773 static struct commit_list *paint_down_to_common(struct commit *one, int n, struct commit **twos)
 774 {
 775         struct prio_queue queue = { compare_commits_by_commit_date };
 776         struct commit_list *result = NULL;
 777         int i;
 778
 779         one->object.flags |= PARENT1;
 780         if (!n) {
 781                 commit_list_append(one, &result);
 782                 return result;
 783         }
 784         prio_queue_put(&queue, one);
 785
 786         for (i = 0; i < n; i++) {
 787                 twos[i]->object.flags |= PARENT2;
 788                 prio_queue_put(&queue, twos[i]);
 789         }
 790
 791         while (queue_has_nonstale(&queue)) {
 792                 struct commit *commit = prio_queue_get(&queue);
 793                 struct commit_list *parents;
 794                 int flags;
 795
 796                 flags = commit->object.flags & (PARENT1 | PARENT2 | STALE);
 797                 if (flags == (PARENT1 | PARENT2)) {
 798                         if (!(commit->object.flags & RESULT)) {
 799                                 commit->object.flags |= RESULT;
 800                                 commit_list_insert_by_date(commit, &result);
 801                         }
 802                         /* Mark parents of a found merge stale */
 803                         flags |= STALE;
 804                 }
 805                 parents = commit->parents;
 806                 while (parents) {
 807                         struct commit *p = parents->item;
 808                         parents = parents->next;
 809                         if ((p->object.flags & flags) == flags)
 810                                 continue;
 811                         if (parse_commit(p))
 812                                 return NULL;
 813                         p->object.flags |= flags;
 814                         prio_queue_put(&queue, p);
 815                 }
 816         }
 817
 818         clear_prio_queue(&queue);
 819         return result;
 820 }
 821
 822 static struct commit_list *merge_bases_many(struct commit *one, int n, struct commit **twos)
 823 {
 824         struct commit_list *list = NULL;
 825         struct commit_list *result = NULL;
 826         int i;
 827
 828         for (i = 0; i < n; i++) {
 829                 if (one == twos[i])
 830                         /*
 831                          * We do not mark this even with RESULT so we do not
 832                          * have to clean it up.
 833                          */
 834                         return commit_list_insert(one, &result);
 835         }
 836
 837         if (parse_commit(one))
 838                 return NULL;
 839         for (i = 0; i < n; i++) {
 840                 if (parse_commit(twos[i]))
 841                         return NULL;
 842         }
 843
 844         list = paint_down_to_common(one, n, twos);
 845
 846         while (list) {
 847                 struct commit_list *next = list->next;
 848                 if (!(list->item->object.flags & STALE))
 849                         commit_list_insert_by_date(list->item, &result);
 850                 free(list);
 851                 list = next;
 852         }
 853         return result;
 854 }
 855
 856 struct commit_list *get_octopus_merge_bases(struct commit_list *in)
 857 {
 858         struct commit_list *i, *j, *k, *ret = NULL;
 859
 860         if (!in)
 861                 return ret;
 862
 863         commit_list_insert(in->item, &ret);
 864
 865         for (i = in->next; i; i = i->next) {
 866                 struct commit_list *new = NULL, *end = NULL;
 867
 868                 for (j = ret; j; j = j->next) {
 869                         struct commit_list *bases;
 870                         bases = get_merge_bases(i->item, j->item);
 871                         if (!new)
 872                                 new = bases;
 873                         else
 874                                 end->next = bases;
 875                         for (k = bases; k; k = k->next)
 876                                 end = k;
 877                 }
 878                 ret = new;
 879         }
 880         return ret;
 881 }
 882
 883 static int remove_redundant(struct commit **array, int cnt)
 884 {
 885         /*
 886          * Some commit in the array may be an ancestor of
 887          * another commit.  Move such commit to the end of
 888          * the array, and return the number of commits that
 889          * are independent from each other.
 890          */
 891         struct commit **work;
 892         unsigned char *redundant;
 893         int *filled_index;
 894         int i, j, filled;
 895
 896         work = xcalloc(cnt, sizeof(*work));
 897         redundant = xcalloc(cnt, 1);
 898         filled_index = xmalloc(sizeof(*filled_index) * (cnt - 1));
 899
 900         for (i = 0; i < cnt; i++)
 901                 parse_commit(array[i]);
 902         for (i = 0; i < cnt; i++) {
 903                 struct commit_list *common;
 904
 905                 if (redundant[i])
 906                         continue;
 907                 for (j = filled = 0; j < cnt; j++) {
 908                         if (i == j || redundant[j])
 909                                 continue;
 910                         filled_index[filled] = j;
 911                         work[filled++] = array[j];
 912                 }
 913                 common = paint_down_to_common(array[i], filled, work);
 914                 if (array[i]->object.flags & PARENT2)
 915                         redundant[i] = 1;
 916                 for (j = 0; j < filled; j++)
 917                         if (work[j]->object.flags & PARENT1)
 918                                 redundant[filled_index[j]] = 1;
 919                 clear_commit_marks(array[i], all_flags);
 920                 for (j = 0; j < filled; j++)
 921                         clear_commit_marks(work[j], all_flags);
 922                 free_commit_list(common);
 923         }
 924
 925         /* Now collect the result */
 926         memcpy(work, array, sizeof(*array) * cnt);
 927         for (i = filled = 0; i < cnt; i++)
 928                 if (!redundant[i])
 929                         array[filled++] = work[i];
 930         for (j = filled, i = 0; i < cnt; i++)
 931                 if (redundant[i])
 932                         array[j++] = work[i];
 933         free(work);
 934         free(redundant);
 935         free(filled_index);
 936         return filled;
 937 }
 938
 939 static struct commit_list *get_merge_bases_many_0(struct commit *one,
 940                                                   int n,
 941                                                   struct commit **twos,
 942                                                   int cleanup)
 943 {
 944         struct commit_list *list;
 945         struct commit **rslt;
 946         struct commit_list *result;
 947         int cnt, i;
 948
 949         result = merge_bases_many(one, n, twos);
 950         for (i = 0; i < n; i++) {
 951                 if (one == twos[i])
 952                         return result;
 953         }
 954         if (!result || !result->next) {
 955                 if (cleanup) {
 956                         clear_commit_marks(one, all_flags);
 957                         clear_commit_marks_many(n, twos, all_flags);
 958                 }
 959                 return result;
 960         }
 961
 962         /* There are more than one */
 963         cnt = commit_list_count(result);
 964         rslt = xcalloc(cnt, sizeof(*rslt));
 965         for (list = result, i = 0; list; list = list->next)
 966                 rslt[i++] = list->item;
 967         free_commit_list(result);
 968
 969         clear_commit_marks(one, all_flags);
 970         clear_commit_marks_many(n, twos, all_flags);
 971
 972         cnt = remove_redundant(rslt, cnt);
 973         result = NULL;
 974         for (i = 0; i < cnt; i++)
 975                 commit_list_insert_by_date(rslt[i], &result);
 976         free(rslt);
 977         return result;
 978 }
 979
 980 struct commit_list *get_merge_bases_many(struct commit *one,
 981                                          int n,
 982                                          struct commit **twos)
 983 {
 984         return get_merge_bases_many_0(one, n, twos, 1);
 985 }
 986
 987 struct commit_list *get_merge_bases_many_dirty(struct commit *one,
 988                                                int n,
 989                                                struct commit **twos)
 990 {
 991         return get_merge_bases_many_0(one, n, twos, 0);
 992 }
 993
 994 struct commit_list *get_merge_bases(struct commit *one, struct commit *two)
 995 {
 996         return get_merge_bases_many_0(one, 1, &two, 1);
 997 }
 998
 999 /*
1000  * Is "commit" a descendant of one of the elements on the "with_commit" list?
1001  */
1002 int is_descendant_of(struct commit *commit, struct commit_list *with_commit)
1003 {
1004         if (!with_commit)
1005                 return 1;
1006         while (with_commit) {
1007                 struct commit *other;
1008
1009                 other = with_commit->item;
1010                 with_commit = with_commit->next;
1011                 if (in_merge_bases(other, commit))
1012                         return 1;
1013         }
1014         return 0;
1015 }
1016
1017 /*
1018  * Is "commit" an ancestor of one of the "references"?
1019  */
1020 int in_merge_bases_many(struct commit *commit, int nr_reference, struct commit **reference)
1021 {
1022         struct commit_list *bases;
1023         int ret = 0, i;
1024
1025         if (parse_commit(commit))
1026                 return ret;
1027         for (i = 0; i < nr_reference; i++)
1028                 if (parse_commit(reference[i]))
1029                         return ret;
1030
1031         bases = paint_down_to_common(commit, nr_reference, reference);
1032         if (commit->object.flags & PARENT2)
1033                 ret = 1;
1034         clear_commit_marks(commit, all_flags);
1035         clear_commit_marks_many(nr_reference, reference, all_flags);
1036         free_commit_list(bases);
1037         return ret;
1038 }
1039
1040 /*
1041  * Is "commit" an ancestor of (i.e. reachable from) the "reference"?
1042  */
1043 int in_merge_bases(struct commit *commit, struct commit *reference)
1044 {
1045         return in_merge_bases_many(commit, 1, &reference);
1046 }
1047
1048 struct commit_list *reduce_heads(struct commit_list *heads)
1049 {
1050         struct commit_list *p;
1051         struct commit_list *result = NULL, **tail = &result;
1052         struct commit **array;
1053         int num_head, i;
1054
1055         if (!heads)
1056                 return NULL;
1057
1058         /* Uniquify */
1059         for (p = heads; p; p = p->next)
1060                 p->item->object.flags &= ~STALE;
1061         for (p = heads, num_head = 0; p; p = p->next) {
1062                 if (p->item->object.flags & STALE)
1063                         continue;
1064                 p->item->object.flags |= STALE;
1065                 num_head++;
1066         }
1067         array = xcalloc(num_head, sizeof(*array));
1068         for (p = heads, i = 0; p; p = p->next) {
1069                 if (p->item->object.flags & STALE) {
1070                         array[i++] = p->item;
1071                         p->item->object.flags &= ~STALE;
1072                 }
1073         }
1074         num_head = remove_redundant(array, num_head);
1075         for (i = 0; i < num_head; i++)
1076                 tail = &commit_list_insert(array[i], tail)->next;
1077         return result;
1078 }
1079
1080 static const char gpg_sig_header[] = "gpgsig";
1081 static const int gpg_sig_header_len = sizeof(gpg_sig_header) - 1;
1082
1083 static int do_sign_commit(struct strbuf *buf, const char *keyid)
1084 {
1085         struct strbuf sig = STRBUF_INIT;
1086         int inspos, copypos;
1087
1088         /* find the end of the header */
1089         inspos = strstr(buf->buf, "\n\n") - buf->buf + 1;
1090
1091         if (!keyid || !*keyid)
1092                 keyid = get_signing_key();
1093         if (sign_buffer(buf, &sig, keyid)) {
1094                 strbuf_release(&sig);
1095                 return -1;
1096         }
1097
1098         for (copypos = 0; sig.buf[copypos]; ) {
1099                 const char *bol = sig.buf + copypos;
1100                 const char *eol = strchrnul(bol, '\n');
1101                 int len = (eol - bol) + !!*eol;
1102
1103                 if (!copypos) {
1104                         strbuf_insert(buf, inspos, gpg_sig_header, gpg_sig_header_len);
1105                         inspos += gpg_sig_header_len;
1106                 }
1107                 strbuf_insert(buf, inspos++, " ", 1);
1108                 strbuf_insert(buf, inspos, bol, len);
1109                 inspos += len;
1110                 copypos += len;
1111         }
1112         strbuf_release(&sig);
1113         return 0;
1114 }
1115
1116 int parse_signed_commit(const struct commit *commit,
1117                         struct strbuf *payload, struct strbuf *signature)
1118 {
1119
1120         unsigned long size;
1121         const char *buffer = get_commit_buffer(commit, &size);
1122         int in_signature, saw_signature = -1;
1123         const char *line, *tail;
1124
1125         line = buffer;
1126         tail = buffer + size;
1127         in_signature = 0;
1128         saw_signature = 0;
1129         while (line < tail) {
1130                 const char *sig = NULL;
1131                 const char *next = memchr(line, '\n', tail - line);
1132
1133                 next = next ? next + 1 : tail;
1134                 if (in_signature && line[0] == ' ')
1135                         sig = line + 1;
1136                 else if (starts_with(line, gpg_sig_header) &&
1137                          line[gpg_sig_header_len] == ' ')
1138                         sig = line + gpg_sig_header_len + 1;
1139                 if (sig) {
1140                         strbuf_add(signature, sig, next - sig);
1141                         saw_signature = 1;
1142                         in_signature = 1;
1143                 } else {
1144                         if (*line == '\n')
1145                                 /* dump the whole remainder of the buffer */
1146                                 next = tail;
1147                         strbuf_add(payload, line, next - line);
1148                         in_signature = 0;
1149                 }
1150                 line = next;
1151         }
1152         unuse_commit_buffer(commit, buffer);
1153         return saw_signature;
1154 }
1155
1156 int remove_signature(struct strbuf *buf)
1157 {
1158         const char *line = buf->buf;
1159         const char *tail = buf->buf + buf->len;
1160         int in_signature = 0;
1161         const char *sig_start = NULL;
1162         const char *sig_end = NULL;
1163
1164         while (line < tail) {
1165                 const char *next = memchr(line, '\n', tail - line);
1166                 next = next ? next + 1 : tail;
1167
1168                 if (in_signature && line[0] == ' ')
1169                         sig_end = next;
1170                 else if (starts_with(line, gpg_sig_header) &&
1171                          line[gpg_sig_header_len] == ' ') {
1172                         sig_start = line;
1173                         sig_end = next;
1174                         in_signature = 1;
1175                 } else {
1176                         if (*line == '\n')
1177                                 /* dump the whole remainder of the buffer */
1178                                 next = tail;
1179                         in_signature = 0;
1180                 }
1181                 line = next;
1182         }
1183
1184         if (sig_start)
1185                 strbuf_remove(buf, sig_start - buf->buf, sig_end - sig_start);
1186
1187         return sig_start != NULL;
1188 }
1189
1190 static void handle_signed_tag(struct commit *parent, struct commit_extra_header ***tail)
1191 {
1192         struct merge_remote_desc *desc;
1193         struct commit_extra_header *mergetag;
1194         char *buf;
1195         unsigned long size, len;
1196         enum object_type type;
1197
1198         desc = merge_remote_util(parent);
1199         if (!desc || !desc->obj)
1200                 return;
1201         buf = read_sha1_file(desc->obj->sha1, &type, &size);
1202         if (!buf || type != OBJ_TAG)
1203                 goto free_return;
1204         len = parse_signature(buf, size);
1205         if (size == len)
1206                 goto free_return;
1207         /*
1208          * We could verify this signature and either omit the tag when
1209          * it does not validate, but the integrator may not have the
1210          * public key of the signer of the tag he is merging, while a
1211          * later auditor may have it while auditing, so let's not run
1212          * verify-signed-buffer here for now...
1213          *
1214          * if (verify_signed_buffer(buf, len, buf + len, size - len, ...))
1215          *      warn("warning: signed tag unverified.");
1216          */
1217         mergetag = xcalloc(1, sizeof(*mergetag));
1218         mergetag->key = xstrdup("mergetag");
1219         mergetag->value = buf;
1220         mergetag->len = size;
1221
1222         **tail = mergetag;
1223         *tail = &mergetag->next;
1224         return;
1225
1226 free_return:
1227         free(buf);
1228 }
1229
1230 void check_commit_signature(const struct commit *commit, struct signature_check *sigc)
1231 {
1232         struct strbuf payload = STRBUF_INIT;
1233         struct strbuf signature = STRBUF_INIT;
1234         struct strbuf gpg_output = STRBUF_INIT;
1235         struct strbuf gpg_status = STRBUF_INIT;
1236         int status;
1237
1238         sigc->result = 'N';
1239
1240         if (parse_signed_commit(commit, &payload, &signature) <= 0)
1241                 goto out;
1242         status = verify_signed_buffer(payload.buf, payload.len,
1243                                       signature.buf, signature.len,
1244                                       &gpg_output, &gpg_status);
1245         if (status && !gpg_output.len)
1246                 goto out;
1247         sigc->payload = strbuf_detach(&payload, NULL);
1248         sigc->gpg_output = strbuf_detach(&gpg_output, NULL);
1249         sigc->gpg_status = strbuf_detach(&gpg_status, NULL);
1250         parse_gpg_output(sigc);
1251
1252  out:
1253         strbuf_release(&gpg_status);
1254         strbuf_release(&gpg_output);
1255         strbuf_release(&payload);
1256         strbuf_release(&signature);
1257 }
1258
1259
1260
1261 void append_merge_tag_headers(struct commit_list *parents,
1262                               struct commit_extra_header ***tail)
1263 {
1264         while (parents) {
1265                 struct commit *parent = parents->item;
1266                 handle_signed_tag(parent, tail);
1267                 parents = parents->next;
1268         }
1269 }
1270
1271 static void add_extra_header(struct strbuf *buffer,
1272                              struct commit_extra_header *extra)
1273 {
1274         strbuf_addstr(buffer, extra->key);
1275         if (extra->len)
1276                 strbuf_add_lines(buffer, " ", extra->value, extra->len);
1277         else
1278                 strbuf_addch(buffer, '\n');
1279 }
1280
1281 struct commit_extra_header *read_commit_extra_headers(struct commit *commit,
1282                                                       const char **exclude)
1283 {
1284         struct commit_extra_header *extra = NULL;
1285         unsigned long size;
1286         const char *buffer = get_commit_buffer(commit, &size);
1287         extra = read_commit_extra_header_lines(buffer, size, exclude);
1288         unuse_commit_buffer(commit, buffer);
1289         return extra;
1290 }
1291
1292 void for_each_mergetag(each_mergetag_fn fn, struct commit *commit, void *data)
1293 {
1294         struct commit_extra_header *extra, *to_free;
1295
1296         to_free = read_commit_extra_headers(commit, NULL);
1297         for (extra = to_free; extra; extra = extra->next) {
1298                 if (strcmp(extra->key, "mergetag"))
1299                         continue; /* not a merge tag */
1300                 fn(commit, extra, data);
1301         }
1302         free_commit_extra_headers(to_free);
1303 }
1304
1305 static inline int standard_header_field(const char *field, size_t len)
1306 {
1307         return ((len == 4 && !memcmp(field, "tree ", 5)) ||
1308                 (len == 6 && !memcmp(field, "parent ", 7)) ||
1309                 (len == 6 && !memcmp(field, "author ", 7)) ||
1310                 (len == 9 && !memcmp(field, "committer ", 10)) ||
1311                 (len == 8 && !memcmp(field, "encoding ", 9)));
1312 }
1313
1314 static int excluded_header_field(const char *field, size_t len, const char **exclude)
1315 {
1316         if (!exclude)
1317                 return 0;
1318
1319         while (*exclude) {
1320                 size_t xlen = strlen(*exclude);
1321                 if (len == xlen &&
1322                     !memcmp(field, *exclude, xlen) && field[xlen] == ' ')
1323                         return 1;
1324                 exclude++;
1325         }
1326         return 0;
1327 }
1328
1329 static struct commit_extra_header *read_commit_extra_header_lines(
1330         const char *buffer, size_t size,
1331         const char **exclude)
1332 {
1333         struct commit_extra_header *extra = NULL, **tail = &extra, *it = NULL;
1334         const char *line, *next, *eof, *eob;
1335         struct strbuf buf = STRBUF_INIT;
1336
1337         for (line = buffer, eob = line + size;
1338              line < eob && *line != '\n';
1339              line = next) {
1340                 next = memchr(line, '\n', eob - line);
1341                 next = next ? next + 1 : eob;
1342                 if (*line == ' ') {
1343                         /* continuation */
1344                         if (it)
1345                                 strbuf_add(&buf, line + 1, next - (line + 1));
1346                         continue;
1347                 }
1348                 if (it)
1349                         it->value = strbuf_detach(&buf, &it->len);
1350                 strbuf_reset(&buf);
1351                 it = NULL;
1352
1353                 eof = strchr(line, ' ');
1354                 if (next <= eof)
1355                         eof = next;
1356
1357                 if (standard_header_field(line, eof - line) ||
1358                     excluded_header_field(line, eof - line, exclude))
1359                         continue;
1360
1361                 it = xcalloc(1, sizeof(*it));
1362                 it->key = xmemdupz(line, eof-line);
1363                 *tail = it;
1364                 tail = &it->next;
1365                 if (eof + 1 < next)
1366                         strbuf_add(&buf, eof + 1, next - (eof + 1));
1367         }
1368         if (it)
1369                 it->value = strbuf_detach(&buf, &it->len);
1370         return extra;
1371 }
1372
1373 void free_commit_extra_headers(struct commit_extra_header *extra)
1374 {
1375         while (extra) {
1376                 struct commit_extra_header *next = extra->next;
1377                 free(extra->key);
1378                 free(extra->value);
1379                 free(extra);
1380                 extra = next;
1381         }
1382 }
1383
1384 int commit_tree(const char *msg, size_t msg_len,
1385                 const unsigned char *tree,
1386                 struct commit_list *parents, unsigned char *ret,
1387                 const char *author, const char *sign_commit)
1388 {
1389         struct commit_extra_header *extra = NULL, **tail = &extra;
1390         int result;
1391
1392         append_merge_tag_headers(parents, &tail);
1393         result = commit_tree_extended(msg, msg_len, tree, parents, ret,
1394                                       author, sign_commit, extra);
1395         free_commit_extra_headers(extra);
1396         return result;
1397 }
1398
1399 static int find_invalid_utf8(const char *buf, int len)
1400 {
1401         int offset = 0;
1402         static const unsigned int max_codepoint[] = {
1403                 0x7f, 0x7ff, 0xffff, 0x10ffff
1404         };
1405
1406         while (len) {
1407                 unsigned char c = *buf++;
1408                 int bytes, bad_offset;
1409                 unsigned int codepoint;
1410                 unsigned int min_val, max_val;
1411
1412                 len--;
1413                 offset++;
1414
1415                 /* Simple US-ASCII? No worries. */
1416                 if (c < 0x80)
1417                         continue;
1418
1419                 bad_offset = offset-1;
1420
1421                 /*
1422                  * Count how many more high bits set: that's how
1423                  * many more bytes this sequence should have.
1424                  */
1425                 bytes = 0;
1426                 while (c & 0x40) {
1427                         c <<= 1;
1428                         bytes++;
1429                 }
1430
1431                 /*
1432                  * Must be between 1 and 3 more bytes.  Longer sequences result in
1433                  * codepoints beyond U+10FFFF, which are guaranteed never to exist.
1434                  */
1435                 if (bytes < 1 || 3 < bytes)
1436                         return bad_offset;
1437
1438                 /* Do we *have* that many bytes? */
1439                 if (len < bytes)
1440                         return bad_offset;
1441
1442                 /*
1443                  * Place the encoded bits at the bottom of the value and compute the
1444                  * valid range.
1445                  */
1446                 codepoint = (c & 0x7f) >> bytes;
1447                 min_val = max_codepoint[bytes-1] + 1;
1448                 max_val = max_codepoint[bytes];
1449
1450                 offset += bytes;
1451                 len -= bytes;
1452
1453                 /* And verify that they are good continuation bytes */
1454                 do {
1455                         codepoint <<= 6;
1456                         codepoint |= *buf & 0x3f;
1457                         if ((*buf++ & 0xc0) != 0x80)
1458                                 return bad_offset;
1459                 } while (--bytes);
1460
1461                 /* Reject codepoints that are out of range for the sequence length. */
1462                 if (codepoint < min_val || codepoint > max_val)
1463                         return bad_offset;
1464                 /* Surrogates are only for UTF-16 and cannot be encoded in UTF-8. */
1465                 if ((codepoint & 0x1ff800) == 0xd800)
1466                         return bad_offset;
1467                 /* U+xxFFFE and U+xxFFFF are guaranteed non-characters. */
1468                 if ((codepoint & 0xfffe) == 0xfffe)
1469                         return bad_offset;
1470                 /* So are anything in the range U+FDD0..U+FDEF. */
1471                 if (codepoint >= 0xfdd0 && codepoint <= 0xfdef)
1472                         return bad_offset;
1473         }
1474         return -1;
1475 }
1476
1477 /*
1478  * This verifies that the buffer is in proper utf8 format.
1479  *
1480  * If it isn't, it assumes any non-utf8 characters are Latin1,
1481  * and does the conversion.
1482  */
1483 static int verify_utf8(struct strbuf *buf)
1484 {
1485         int ok = 1;
1486         long pos = 0;
1487
1488         for (;;) {
1489                 int bad;
1490                 unsigned char c;
1491                 unsigned char replace[2];
1492
1493                 bad = find_invalid_utf8(buf->buf + pos, buf->len - pos);
1494                 if (bad < 0)
1495                         return ok;
1496                 pos += bad;
1497                 ok = 0;
1498                 c = buf->buf[pos];
1499                 strbuf_remove(buf, pos, 1);
1500
1501                 /* We know 'c' must be in the range 128-255 */
1502                 replace[0] = 0xc0 + (c >> 6);
1503                 replace[1] = 0x80 + (c & 0x3f);
1504                 strbuf_insert(buf, pos, replace, 2);
1505                 pos += 2;
1506         }
1507 }
1508
1509 static const char commit_utf8_warn[] =
1510 "Warning: commit message did not conform to UTF-8.\n"
1511 "You may want to amend it after fixing the message, or set the config\n"
1512 "variable i18n.commitencoding to the encoding your project uses.\n";
1513
1514 int commit_tree_extended(const char *msg, size_t msg_len,
1515                          const unsigned char *tree,
1516                          struct commit_list *parents, unsigned char *ret,
1517                          const char *author, const char *sign_commit,
1518                          struct commit_extra_header *extra)
1519 {
1520         int result;
1521         int encoding_is_utf8;
1522         struct strbuf buffer;
1523
1524         assert_sha1_type(tree, OBJ_TREE);
1525
1526         if (memchr(msg, '\0', msg_len))
1527                 return error("a NUL byte in commit log message not allowed.");
1528
1529         /* Not having i18n.commitencoding is the same as having utf-8 */
1530         encoding_is_utf8 = is_encoding_utf8(git_commit_encoding);
1531
1532         strbuf_init(&buffer, 8192); /* should avoid reallocs for the headers */
1533         strbuf_addf(&buffer, "tree %s\n", sha1_to_hex(tree));
1534
1535         /*
1536          * NOTE! This ordering means that the same exact tree merged with a
1537          * different order of parents will be a _different_ changeset even
1538          * if everything else stays the same.
1539          */
1540         while (parents) {
1541                 struct commit_list *next = parents->next;
1542                 struct commit *parent = parents->item;
1543
1544                 strbuf_addf(&buffer, "parent %s\n",
1545                             sha1_to_hex(parent->object.sha1));
1546                 free(parents);
1547                 parents = next;
1548         }
1549
1550         /* Person/date information */
1551         if (!author)
1552                 author = git_author_info(IDENT_STRICT);
1553         strbuf_addf(&buffer, "author %s\n", author);
1554         strbuf_addf(&buffer, "committer %s\n", git_committer_info(IDENT_STRICT));
1555         if (!encoding_is_utf8)
1556                 strbuf_addf(&buffer, "encoding %s\n", git_commit_encoding);
1557
1558         while (extra) {
1559                 add_extra_header(&buffer, extra);
1560                 extra = extra->next;
1561         }
1562         strbuf_addch(&buffer, '\n');
1563
1564         /* And add the comment */
1565         strbuf_add(&buffer, msg, msg_len);
1566
1567         /* And check the encoding */
1568         if (encoding_is_utf8 && !verify_utf8(&buffer))
1569                 fprintf(stderr, commit_utf8_warn);
1570
1571         if (sign_commit && do_sign_commit(&buffer, sign_commit))
1572                 return -1;
1573
1574         result = write_sha1_file(buffer.buf, buffer.len, commit_type, ret);
1575         strbuf_release(&buffer);
1576         return result;
1577 }
1578
1579 struct commit *get_merge_parent(const char *name)
1580 {
1581         struct object *obj;
1582         struct commit *commit;
1583         unsigned char sha1[20];
1584         if (get_sha1(name, sha1))
1585                 return NULL;
1586         obj = parse_object(sha1);
1587         commit = (struct commit *)peel_to_type(name, 0, obj, OBJ_COMMIT);
1588         if (commit && !commit->util) {
1589                 struct merge_remote_desc *desc;
1590                 desc = xmalloc(sizeof(*desc));
1591                 desc->obj = obj;
1592                 desc->name = strdup(name);
1593                 commit->util = desc;
1594         }
1595         return commit;
1596 }
1597
1598 /*
1599  * Append a commit to the end of the commit_list.
1600  *
1601  * next starts by pointing to the variable that holds the head of an
1602  * empty commit_list, and is updated to point to the "next" field of
1603  * the last item on the list as new commits are appended.
1604  *
1605  * Usage example:
1606  *
1607  *     struct commit_list *list;
1608  *     struct commit_list **next = &list;
1609  *
1610  *     next = commit_list_append(c1, next);
1611  *     next = commit_list_append(c2, next);
1612  *     assert(commit_list_count(list) == 2);
1613  *     return list;
1614  */
1615 struct commit_list **commit_list_append(struct commit *commit,
1616                                         struct commit_list **next)
1617 {
1618         struct commit_list *new = xmalloc(sizeof(struct commit_list));
1619         new->item = commit;
1620         *next = new;
1621         new->next = NULL;
1622         return &new->next;
1623 }
1624
1625 void print_commit_list(struct commit_list *list,
1626                        const char *format_cur,
1627                        const char *format_last)
1628 {
1629         for ( ; list; list = list->next) {
1630                 const char *format = list->next ? format_cur : format_last;
1631                 printf(format, sha1_to_hex(list->item->object.sha1));
1632         }
1633 }
1634
1635 const char *find_commit_header(const char *msg, const char *key, size_t *out_len)
1636 {
1637         int key_len = strlen(key);
1638         const char *line = msg;
1639
1640         while (line) {
1641                 const char *eol = strchrnul(line, '\n');
1642
1643                 if (line == eol)
1644                         return NULL;
1645
1646                 if (eol - line > key_len &&
1647                     !strncmp(line, key, key_len) &&
1648                     line[key_len] == ' ') {
1649                         *out_len = eol - line - key_len - 1;
1650                         return line + key_len + 1;
1651                 }
1652                 line = *eol ? eol + 1 : NULL;
1653         }
1654         return NULL;
1655 }
1656
1657 /*
1658  * Inspect sb and determine the true "end" of the log message, in
1659  * order to find where to put a new Signed-off-by: line.  Ignored are
1660  * trailing comment lines and blank lines, and also the traditional
1661  * "Conflicts:" block that is not commented out, so that we can use
1662  * "git commit -s --amend" on an existing commit that forgot to remove
1663  * it.
1664  *
1665  * Returns the number of bytes from the tail to ignore, to be fed as
1666  * the second parameter to append_signoff().
1667  */
1668 int ignore_non_trailer(struct strbuf *sb)
1669 {
1670         int boc = 0;
1671         int bol = 0;
1672         int in_old_conflicts_block = 0;
1673
1674         while (bol < sb->len) {
1675                 char *next_line;
1676
1677                 if (!(next_line = memchr(sb->buf + bol, '\n', sb->len - bol)))
1678                         next_line = sb->buf + sb->len;
1679                 else
1680                         next_line++;
1681
1682                 if (sb->buf[bol] == comment_line_char || sb->buf[bol] == '\n') {
1683                         /* is this the first of the run of comments? */
1684                         if (!boc)
1685                                 boc = bol;
1686                         /* otherwise, it is just continuing */
1687                 } else if (starts_with(sb->buf + bol, "Conflicts:\n")) {
1688                         in_old_conflicts_block = 1;
1689                         if (!boc)
1690                                 boc = bol;
1691                 } else if (in_old_conflicts_block && sb->buf[bol] == '\t') {
1692                         ; /* a pathname in the conflicts block */
1693                 } else if (boc) {
1694                         /* the previous was not trailing comment */
1695                         boc = 0;
1696                         in_old_conflicts_block = 0;
1697                 }
1698                 bol = next_line - sb->buf;
1699         }
1700         return boc ? sb->len - boc : 0;
1701 }