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