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