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 #include "wt-status.h"
  15
  16 static struct commit_extra_header *read_commit_extra_header_lines(const char *buf, size_t len, const char **);
  17
  18 int save_commit_buffer = 1;
  19
  20 const char *commit_type = "commit";
  21
  22 struct commit *lookup_commit_reference_gently(const struct object_id *oid,
  23                                               int quiet)
  24 {
  25         struct object *obj = deref_tag(parse_object(oid), NULL, 0);
  26
  27         if (!obj)
  28                 return NULL;
  29         return object_as_type(obj, OBJ_COMMIT, quiet);
  30 }
  31
  32 struct commit *lookup_commit_reference(const struct object_id *oid)
  33 {
  34         return lookup_commit_reference_gently(oid, 0);
  35 }
  36
  37 struct commit *lookup_commit_or_die(const struct object_id *oid, const char *ref_name)
  38 {
  39         struct commit *c = lookup_commit_reference(oid);
  40         if (!c)
  41                 die(_("could not parse %s"), ref_name);
  42         if (oidcmp(oid, &c->object.oid)) {
  43                 warning(_("%s %s is not a commit!"),
  44                         ref_name, oid_to_hex(oid));
  45         }
  46         return c;
  47 }
  48
  49 struct commit *lookup_commit(const struct object_id *oid)
  50 {
  51         struct object *obj = lookup_object(oid->hash);
  52         if (!obj)
  53                 return create_object(oid->hash, alloc_commit_node());
  54         return object_as_type(obj, OBJ_COMMIT, 0);
  55 }
  56
  57 struct commit *lookup_commit_reference_by_name(const char *name)
  58 {
  59         struct object_id oid;
  60         struct commit *commit;
  61
  62         if (get_sha1_committish(name, oid.hash))
  63                 return NULL;
  64         commit = lookup_commit_reference(&oid);
  65         if (parse_commit(commit))
  66                 return NULL;
  67         return commit;
  68 }
  69
  70 static timestamp_t parse_commit_date(const char *buf, const char *tail)
  71 {
  72         const char *dateptr;
  73
  74         if (buf + 6 >= tail)
  75                 return 0;
  76         if (memcmp(buf, "author", 6))
  77                 return 0;
  78         while (buf < tail && *buf++ != '\n')
  79                 /* nada */;
  80         if (buf + 9 >= tail)
  81                 return 0;
  82         if (memcmp(buf, "committer", 9))
  83                 return 0;
  84         while (buf < tail && *buf++ != '>')
  85                 /* nada */;
  86         if (buf >= tail)
  87                 return 0;
  88         dateptr = buf;
  89         while (buf < tail && *buf++ != '\n')
  90                 /* nada */;
  91         if (buf >= tail)
  92                 return 0;
  93         /* dateptr < buf && buf[-1] == '\n', so parsing will stop at buf-1 */
  94         return parse_timestamp(dateptr, NULL, 10);
  95 }
  96
  97 static struct commit_graft **commit_graft;
  98 static int commit_graft_alloc, commit_graft_nr;
  99
 100 static const unsigned char *commit_graft_sha1_access(size_t index, void *table)
 101 {
 102         struct commit_graft **commit_graft_table = table;
 103         return commit_graft_table[index]->oid.hash;
 104 }
 105
 106 static int commit_graft_pos(const unsigned char *sha1)
 107 {
 108         return sha1_pos(sha1, commit_graft, commit_graft_nr,
 109                         commit_graft_sha1_access);
 110 }
 111
 112 int register_commit_graft(struct commit_graft *graft, int ignore_dups)
 113 {
 114         int pos = commit_graft_pos(graft->oid.hash);
 115
 116         if (0 <= pos) {
 117                 if (ignore_dups)
 118                         free(graft);
 119                 else {
 120                         free(commit_graft[pos]);
 121                         commit_graft[pos] = graft;
 122                 }
 123                 return 1;
 124         }
 125         pos = -pos - 1;
 126         ALLOC_GROW(commit_graft, commit_graft_nr + 1, commit_graft_alloc);
 127         commit_graft_nr++;
 128         if (pos < commit_graft_nr)
 129                 memmove(commit_graft + pos + 1,
 130                         commit_graft + pos,
 131                         (commit_graft_nr - pos - 1) *
 132                         sizeof(*commit_graft));
 133         commit_graft[pos] = graft;
 134         return 0;
 135 }
 136
 137 struct commit_graft *read_graft_line(char *buf, int len)
 138 {
 139         /* The format is just "Commit Parent1 Parent2 ...\n" */
 140         int i;
 141         struct commit_graft *graft = NULL;
 142         const int entry_size = GIT_SHA1_HEXSZ + 1;
 143
 144         while (len && isspace(buf[len-1]))
 145                 buf[--len] = '\0';
 146         if (buf[0] == '#' || buf[0] == '\0')
 147                 return NULL;
 148         if ((len + 1) % entry_size)
 149                 goto bad_graft_data;
 150         i = (len + 1) / entry_size - 1;
 151         graft = xmalloc(st_add(sizeof(*graft), st_mult(GIT_SHA1_RAWSZ, i)));
 152         graft->nr_parent = i;
 153         if (get_oid_hex(buf, &graft->oid))
 154                 goto bad_graft_data;
 155         for (i = GIT_SHA1_HEXSZ; i < len; i += entry_size) {
 156                 if (buf[i] != ' ')
 157                         goto bad_graft_data;
 158                 if (get_sha1_hex(buf + i + 1, graft->parent[i/entry_size].hash))
 159                         goto bad_graft_data;
 160         }
 161         return graft;
 162
 163 bad_graft_data:
 164         error("bad graft data: %s", buf);
 165         free(graft);
 166         return NULL;
 167 }
 168
 169 static int read_graft_file(const char *graft_file)
 170 {
 171         FILE *fp = fopen(graft_file, "r");
 172         struct strbuf buf = STRBUF_INIT;
 173         if (!fp)
 174                 return -1;
 175         while (!strbuf_getwholeline(&buf, fp, '\n')) {
 176                 /* The format is just "Commit Parent1 Parent2 ...\n" */
 177                 struct commit_graft *graft = read_graft_line(buf.buf, buf.len);
 178                 if (!graft)
 179                         continue;
 180                 if (register_commit_graft(graft, 1))
 181                         error("duplicate graft data: %s", buf.buf);
 182         }
 183         fclose(fp);
 184         strbuf_release(&buf);
 185         return 0;
 186 }
 187
 188 static void prepare_commit_graft(void)
 189 {
 190         static int commit_graft_prepared;
 191         char *graft_file;
 192
 193         if (commit_graft_prepared)
 194                 return;
 195         graft_file = get_graft_file();
 196         read_graft_file(graft_file);
 197         /* make sure shallows are read */
 198         is_repository_shallow();
 199         commit_graft_prepared = 1;
 200 }
 201
 202 struct commit_graft *lookup_commit_graft(const unsigned char *sha1)
 203 {
 204         int pos;
 205         prepare_commit_graft();
 206         pos = commit_graft_pos(sha1);
 207         if (pos < 0)
 208                 return NULL;
 209         return commit_graft[pos];
 210 }
 211
 212 int for_each_commit_graft(each_commit_graft_fn fn, void *cb_data)
 213 {
 214         int i, ret;
 215         for (i = ret = 0; i < commit_graft_nr && !ret; i++)
 216                 ret = fn(commit_graft[i], cb_data);
 217         return ret;
 218 }
 219
 220 int unregister_shallow(const struct object_id *oid)
 221 {
 222         int pos = commit_graft_pos(oid->hash);
 223         if (pos < 0)
 224                 return -1;
 225         if (pos + 1 < commit_graft_nr)
 226                 memmove(commit_graft + pos, commit_graft + pos + 1,
 227                                 sizeof(struct commit_graft *)
 228                                 * (commit_graft_nr - pos - 1));
 229         commit_graft_nr--;
 230         return 0;
 231 }
 232
 233 struct commit_buffer {
 234         void *buffer;
 235         unsigned long size;
 236 };
 237 define_commit_slab(buffer_slab, struct commit_buffer);
 238 static struct buffer_slab buffer_slab = COMMIT_SLAB_INIT(1, buffer_slab);
 239
 240 void set_commit_buffer(struct commit *commit, void *buffer, unsigned long size)
 241 {
 242         struct commit_buffer *v = buffer_slab_at(&buffer_slab, commit);
 243         v->buffer = buffer;
 244         v->size = size;
 245 }
 246
 247 const void *get_cached_commit_buffer(const struct commit *commit, unsigned long *sizep)
 248 {
 249         struct commit_buffer *v = buffer_slab_peek(&buffer_slab, commit);
 250         if (!v) {
 251                 if (sizep)
 252                         *sizep = 0;
 253                 return NULL;
 254         }
 255         if (sizep)
 256                 *sizep = v->size;
 257         return v->buffer;
 258 }
 259
 260 const void *get_commit_buffer(const struct commit *commit, unsigned long *sizep)
 261 {
 262         const void *ret = get_cached_commit_buffer(commit, sizep);
 263         if (!ret) {
 264                 enum object_type type;
 265                 unsigned long size;
 266                 ret = read_sha1_file(commit->object.oid.hash, &type, &size);
 267                 if (!ret)
 268                         die("cannot read commit object %s",
 269                             oid_to_hex(&commit->object.oid));
 270                 if (type != OBJ_COMMIT)
 271                         die("expected commit for %s, got %s",
 272                             oid_to_hex(&commit->object.oid), typename(type));
 273                 if (sizep)
 274                         *sizep = size;
 275         }
 276         return ret;
 277 }
 278
 279 void unuse_commit_buffer(const struct commit *commit, const void *buffer)
 280 {
 281         struct commit_buffer *v = buffer_slab_peek(&buffer_slab, commit);
 282         if (!(v && v->buffer == buffer))
 283                 free((void *)buffer);
 284 }
 285
 286 void free_commit_buffer(struct commit *commit)
 287 {
 288         struct commit_buffer *v = buffer_slab_peek(&buffer_slab, commit);
 289         if (v) {
 290                 free(v->buffer);
 291                 v->buffer = NULL;
 292                 v->size = 0;
 293         }
 294 }
 295
 296 const void *detach_commit_buffer(struct commit *commit, unsigned long *sizep)
 297 {
 298         struct commit_buffer *v = buffer_slab_peek(&buffer_slab, commit);
 299         void *ret;
 300
 301         if (!v) {
 302                 if (sizep)
 303                         *sizep = 0;
 304                 return NULL;
 305         }
 306         ret = v->buffer;
 307         if (sizep)
 308                 *sizep = v->size;
 309
 310         v->buffer = NULL;
 311         v->size = 0;
 312         return ret;
 313 }
 314
 315 int parse_commit_buffer(struct commit *item, const void *buffer, unsigned long size)
 316 {
 317         const char *tail = buffer;
 318         const char *bufptr = buffer;
 319         struct object_id parent;
 320         struct commit_list **pptr;
 321         struct commit_graft *graft;
 322         const int tree_entry_len = GIT_SHA1_HEXSZ + 5;
 323         const int parent_entry_len = GIT_SHA1_HEXSZ + 7;
 324
 325         if (item->object.parsed)
 326                 return 0;
 327         item->object.parsed = 1;
 328         tail += size;
 329         if (tail <= bufptr + tree_entry_len + 1 || memcmp(bufptr, "tree ", 5) ||
 330                         bufptr[tree_entry_len] != '\n')
 331                 return error("bogus commit object %s", oid_to_hex(&item->object.oid));
 332         if (get_sha1_hex(bufptr + 5, parent.hash) < 0)
 333                 return error("bad tree pointer in commit %s",
 334                              oid_to_hex(&item->object.oid));
 335         item->tree = lookup_tree(&parent);
 336         bufptr += tree_entry_len + 1; /* "tree " + "hex sha1" + "\n" */
 337         pptr = &item->parents;
 338
 339         graft = lookup_commit_graft(item->object.oid.hash);
 340         while (bufptr + parent_entry_len < tail && !memcmp(bufptr, "parent ", 7)) {
 341                 struct commit *new_parent;
 342
 343                 if (tail <= bufptr + parent_entry_len + 1 ||
 344                     get_sha1_hex(bufptr + 7, parent.hash) ||
 345                     bufptr[parent_entry_len] != '\n')
 346                         return error("bad parents in commit %s", oid_to_hex(&item->object.oid));
 347                 bufptr += parent_entry_len + 1;
 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_gently(struct commit *item, int quiet_on_missing)
 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.oid.hash, &type, &size);
 385         if (!buffer)
 386                 return quiet_on_missing ? -1 :
 387                         error("Could not read %s",
 388                              oid_to_hex(&item->object.oid));
 389         if (type != OBJ_COMMIT) {
 390                 free(buffer);
 391                 return error("Object %s not a commit",
 392                              oid_to_hex(&item->object.oid));
 393         }
 394         ret = parse_commit_buffer(item, buffer, size);
 395         if (save_commit_buffer && !ret) {
 396                 set_commit_buffer(item, buffer, size);
 397                 return 0;
 398         }
 399         free(buffer);
 400         return ret;
 401 }
 402
 403 void parse_commit_or_die(struct commit *item)
 404 {
 405         if (parse_commit(item))
 406                 die("unable to parse commit %s",
 407                     item ? oid_to_hex(&item->object.oid) : "(null)");
 408 }
 409
 410 int find_commit_subject(const char *commit_buffer, const char **subject)
 411 {
 412         const char *eol;
 413         const char *p = commit_buffer;
 414
 415         while (*p && (*p != '\n' || p[1] != '\n'))
 416                 p++;
 417         if (*p) {
 418                 p = skip_blank_lines(p + 2);
 419                 eol = strchrnul(p, '\n');
 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                 pop_commit(&list);
 460 }
 461
 462 struct commit_list * commit_list_insert_by_date(struct commit *item, struct commit_list **list)
 463 {
 464         struct commit_list **pp = list;
 465         struct commit_list *p;
 466         while ((p = *pp) != NULL) {
 467                 if (p->item->date < item->date) {
 468                         break;
 469                 }
 470                 pp = &p->next;
 471         }
 472         return commit_list_insert(item, pp);
 473 }
 474
 475 static int commit_list_compare_by_date(const void *a, const void *b)
 476 {
 477         timestamp_t a_date = ((const struct commit_list *)a)->item->date;
 478         timestamp_t b_date = ((const struct commit_list *)b)->item->date;
 479         if (a_date < b_date)
 480                 return 1;
 481         if (a_date > b_date)
 482                 return -1;
 483         return 0;
 484 }
 485
 486 static void *commit_list_get_next(const void *a)
 487 {
 488         return ((const struct commit_list *)a)->next;
 489 }
 490
 491 static void commit_list_set_next(void *a, void *next)
 492 {
 493         ((struct commit_list *)a)->next = next;
 494 }
 495
 496 void commit_list_sort_by_date(struct commit_list **list)
 497 {
 498         *list = llist_mergesort(*list, commit_list_get_next, commit_list_set_next,
 499                                 commit_list_compare_by_date);
 500 }
 501
 502 struct commit *pop_most_recent_commit(struct commit_list **list,
 503                                       unsigned int mark)
 504 {
 505         struct commit *ret = pop_commit(list);
 506         struct commit_list *parents = ret->parents;
 507
 508         while (parents) {
 509                 struct commit *commit = parents->item;
 510                 if (!parse_commit(commit) && !(commit->object.flags & mark)) {
 511                         commit->object.flags |= mark;
 512                         commit_list_insert_by_date(commit, list);
 513                 }
 514                 parents = parents->next;
 515         }
 516         return ret;
 517 }
 518
 519 static void clear_commit_marks_1(struct commit_list **plist,
 520                                  struct commit *commit, unsigned int mark)
 521 {
 522         while (commit) {
 523                 struct commit_list *parents;
 524
 525                 if (!(mark & commit->object.flags))
 526                         return;
 527
 528                 commit->object.flags &= ~mark;
 529
 530                 parents = commit->parents;
 531                 if (!parents)
 532                         return;
 533
 534                 while ((parents = parents->next))
 535                         commit_list_insert(parents->item, plist);
 536
 537                 commit = commit->parents->item;
 538         }
 539 }
 540
 541 void clear_commit_marks_many(int nr, struct commit **commit, unsigned int mark)
 542 {
 543         struct commit_list *list = NULL;
 544
 545         while (nr--) {
 546                 commit_list_insert(*commit, &list);
 547                 commit++;
 548         }
 549         while (list)
 550                 clear_commit_marks_1(&list, pop_commit(&list), mark);
 551 }
 552
 553 void clear_commit_marks(struct commit *commit, unsigned int mark)
 554 {
 555         clear_commit_marks_many(1, &commit, mark);
 556 }
 557
 558 void clear_commit_marks_for_object_array(struct object_array *a, unsigned mark)
 559 {
 560         struct object *object;
 561         struct commit *commit;
 562         unsigned int i;
 563
 564         for (i = 0; i < a->nr; i++) {
 565                 object = a->objects[i].item;
 566                 commit = lookup_commit_reference_gently(&object->oid, 1);
 567                 if (commit)
 568                         clear_commit_marks(commit, mark);
 569         }
 570 }
 571
 572 struct commit *pop_commit(struct commit_list **stack)
 573 {
 574         struct commit_list *top = *stack;
 575         struct commit *item = top ? top->item : NULL;
 576
 577         if (top) {
 578                 *stack = top->next;
 579                 free(top);
 580         }
 581         return item;
 582 }
 583
 584 /*
 585  * Topological sort support
 586  */
 587
 588 /* count number of children that have not been emitted */
 589 define_commit_slab(indegree_slab, int);
 590
 591 /* record author-date for each commit object */
 592 define_commit_slab(author_date_slab, unsigned long);
 593
 594 static void record_author_date(struct author_date_slab *author_date,
 595                                struct commit *commit)
 596 {
 597         const char *buffer = get_commit_buffer(commit, NULL);
 598         struct ident_split ident;
 599         const char *ident_line;
 600         size_t ident_len;
 601         char *date_end;
 602         timestamp_t date;
 603
 604         ident_line = find_commit_header(buffer, "author", &ident_len);
 605         if (!ident_line)
 606                 goto fail_exit; /* no author line */
 607         if (split_ident_line(&ident, ident_line, ident_len) ||
 608             !ident.date_begin || !ident.date_end)
 609                 goto fail_exit; /* malformed "author" line */
 610
 611         date = parse_timestamp(ident.date_begin, &date_end, 10);
 612         if (date_end != ident.date_end)
 613                 goto fail_exit; /* malformed date */
 614         *(author_date_slab_at(author_date, commit)) = date;
 615
 616 fail_exit:
 617         unuse_commit_buffer(commit, buffer);
 618 }
 619
 620 static int compare_commits_by_author_date(const void *a_, const void *b_,
 621                                           void *cb_data)
 622 {
 623         const struct commit *a = a_, *b = b_;
 624         struct author_date_slab *author_date = cb_data;
 625         timestamp_t a_date = *(author_date_slab_at(author_date, a));
 626         timestamp_t b_date = *(author_date_slab_at(author_date, b));
 627
 628         /* newer commits with larger date first */
 629         if (a_date < b_date)
 630                 return 1;
 631         else if (a_date > b_date)
 632                 return -1;
 633         return 0;
 634 }
 635
 636 int compare_commits_by_commit_date(const void *a_, const void *b_, void *unused)
 637 {
 638         const struct commit *a = a_, *b = b_;
 639         /* newer commits with larger date first */
 640         if (a->date < b->date)
 641                 return 1;
 642         else if (a->date > b->date)
 643                 return -1;
 644         return 0;
 645 }
 646
 647 /*
 648  * Performs an in-place topological sort on the list supplied.
 649  */
 650 void sort_in_topological_order(struct commit_list **list, enum rev_sort_order sort_order)
 651 {
 652         struct commit_list *next, *orig = *list;
 653         struct commit_list **pptr;
 654         struct indegree_slab indegree;
 655         struct prio_queue queue;
 656         struct commit *commit;
 657         struct author_date_slab author_date;
 658
 659         if (!orig)
 660                 return;
 661         *list = NULL;
 662
 663         init_indegree_slab(&indegree);
 664         memset(&queue, '\0', sizeof(queue));
 665
 666         switch (sort_order) {
 667         default: /* REV_SORT_IN_GRAPH_ORDER */
 668                 queue.compare = NULL;
 669                 break;
 670         case REV_SORT_BY_COMMIT_DATE:
 671                 queue.compare = compare_commits_by_commit_date;
 672                 break;
 673         case REV_SORT_BY_AUTHOR_DATE:
 674                 init_author_date_slab(&author_date);
 675                 queue.compare = compare_commits_by_author_date;
 676                 queue.cb_data = &author_date;
 677                 break;
 678         }
 679
 680         /* Mark them and clear the indegree */
 681         for (next = orig; next; next = next->next) {
 682                 struct commit *commit = next->item;
 683                 *(indegree_slab_at(&indegree, commit)) = 1;
 684                 /* also record the author dates, if needed */
 685                 if (sort_order == REV_SORT_BY_AUTHOR_DATE)
 686                         record_author_date(&author_date, commit);
 687         }
 688
 689         /* update the indegree */
 690         for (next = orig; next; next = next->next) {
 691                 struct commit_list *parents = next->item->parents;
 692                 while (parents) {
 693                         struct commit *parent = parents->item;
 694                         int *pi = indegree_slab_at(&indegree, parent);
 695
 696                         if (*pi)
 697                                 (*pi)++;
 698                         parents = parents->next;
 699                 }
 700         }
 701
 702         /*
 703          * find the tips
 704          *
 705          * tips are nodes not reachable from any other node in the list
 706          *
 707          * the tips serve as a starting set for the work queue.
 708          */
 709         for (next = orig; next; next = next->next) {
 710                 struct commit *commit = next->item;
 711
 712                 if (*(indegree_slab_at(&indegree, commit)) == 1)
 713                         prio_queue_put(&queue, commit);
 714         }
 715
 716         /*
 717          * This is unfortunate; the initial tips need to be shown
 718          * in the order given from the revision traversal machinery.
 719          */
 720         if (sort_order == REV_SORT_IN_GRAPH_ORDER)
 721                 prio_queue_reverse(&queue);
 722
 723         /* We no longer need the commit list */
 724         free_commit_list(orig);
 725
 726         pptr = list;
 727         *list = NULL;
 728         while ((commit = prio_queue_get(&queue)) != NULL) {
 729                 struct commit_list *parents;
 730
 731                 for (parents = commit->parents; parents ; parents = parents->next) {
 732                         struct commit *parent = parents->item;
 733                         int *pi = indegree_slab_at(&indegree, parent);
 734
 735                         if (!*pi)
 736                                 continue;
 737
 738                         /*
 739                          * parents are only enqueued for emission
 740                          * when all their children have been emitted thereby
 741                          * guaranteeing topological order.
 742                          */
 743                         if (--(*pi) == 1)
 744                                 prio_queue_put(&queue, parent);
 745                 }
 746                 /*
 747                  * all children of commit have already been
 748                  * emitted. we can emit it now.
 749                  */
 750                 *(indegree_slab_at(&indegree, commit)) = 0;
 751
 752                 pptr = &commit_list_insert(commit, pptr)->next;
 753         }
 754
 755         clear_indegree_slab(&indegree);
 756         clear_prio_queue(&queue);
 757         if (sort_order == REV_SORT_BY_AUTHOR_DATE)
 758                 clear_author_date_slab(&author_date);
 759 }
 760
 761 /* merge-base stuff */
 762
 763 /* Remember to update object flag allocation in object.h */
 764 #define PARENT1         (1u<<16)
 765 #define PARENT2         (1u<<17)
 766 #define STALE           (1u<<18)
 767 #define RESULT          (1u<<19)
 768
 769 static const unsigned all_flags = (PARENT1 | PARENT2 | STALE | RESULT);
 770
 771 static int queue_has_nonstale(struct prio_queue *queue)
 772 {
 773         int i;
 774         for (i = 0; i < queue->nr; i++) {
 775                 struct commit *commit = queue->array[i].data;
 776                 if (!(commit->object.flags & STALE))
 777                         return 1;
 778         }
 779         return 0;
 780 }
 781
 782 /* all input commits in one and twos[] must have been parsed! */
 783 static struct commit_list *paint_down_to_common(struct commit *one, int n, struct commit **twos)
 784 {
 785         struct prio_queue queue = { compare_commits_by_commit_date };
 786         struct commit_list *result = NULL;
 787         int i;
 788
 789         one->object.flags |= PARENT1;
 790         if (!n) {
 791                 commit_list_append(one, &result);
 792                 return result;
 793         }
 794         prio_queue_put(&queue, one);
 795
 796         for (i = 0; i < n; i++) {
 797                 twos[i]->object.flags |= PARENT2;
 798                 prio_queue_put(&queue, twos[i]);
 799         }
 800
 801         while (queue_has_nonstale(&queue)) {
 802                 struct commit *commit = prio_queue_get(&queue);
 803                 struct commit_list *parents;
 804                 int flags;
 805
 806                 flags = commit->object.flags & (PARENT1 | PARENT2 | STALE);
 807                 if (flags == (PARENT1 | PARENT2)) {
 808                         if (!(commit->object.flags & RESULT)) {
 809                                 commit->object.flags |= RESULT;
 810                                 commit_list_insert_by_date(commit, &result);
 811                         }
 812                         /* Mark parents of a found merge stale */
 813                         flags |= STALE;
 814                 }
 815                 parents = commit->parents;
 816                 while (parents) {
 817                         struct commit *p = parents->item;
 818                         parents = parents->next;
 819                         if ((p->object.flags & flags) == flags)
 820                                 continue;
 821                         if (parse_commit(p))
 822                                 return NULL;
 823                         p->object.flags |= flags;
 824                         prio_queue_put(&queue, p);
 825                 }
 826         }
 827
 828         clear_prio_queue(&queue);
 829         return result;
 830 }
 831
 832 static struct commit_list *merge_bases_many(struct commit *one, int n, struct commit **twos)
 833 {
 834         struct commit_list *list = NULL;
 835         struct commit_list *result = NULL;
 836         int i;
 837
 838         for (i = 0; i < n; i++) {
 839                 if (one == twos[i])
 840                         /*
 841                          * We do not mark this even with RESULT so we do not
 842                          * have to clean it up.
 843                          */
 844                         return commit_list_insert(one, &result);
 845         }
 846
 847         if (parse_commit(one))
 848                 return NULL;
 849         for (i = 0; i < n; i++) {
 850                 if (parse_commit(twos[i]))
 851                         return NULL;
 852         }
 853
 854         list = paint_down_to_common(one, n, twos);
 855
 856         while (list) {
 857                 struct commit *commit = pop_commit(&list);
 858                 if (!(commit->object.flags & STALE))
 859                         commit_list_insert_by_date(commit, &result);
 860         }
 861         return result;
 862 }
 863
 864 struct commit_list *get_octopus_merge_bases(struct commit_list *in)
 865 {
 866         struct commit_list *i, *j, *k, *ret = NULL;
 867
 868         if (!in)
 869                 return ret;
 870
 871         commit_list_insert(in->item, &ret);
 872
 873         for (i = in->next; i; i = i->next) {
 874                 struct commit_list *new = NULL, *end = NULL;
 875
 876                 for (j = ret; j; j = j->next) {
 877                         struct commit_list *bases;
 878                         bases = get_merge_bases(i->item, j->item);
 879                         if (!new)
 880                                 new = bases;
 881                         else
 882                                 end->next = bases;
 883                         for (k = bases; k; k = k->next)
 884                                 end = k;
 885                 }
 886                 ret = new;
 887         }
 888         return ret;
 889 }
 890
 891 static int remove_redundant(struct commit **array, int cnt)
 892 {
 893         /*
 894          * Some commit in the array may be an ancestor of
 895          * another commit.  Move such commit to the end of
 896          * the array, and return the number of commits that
 897          * are independent from each other.
 898          */
 899         struct commit **work;
 900         unsigned char *redundant;
 901         int *filled_index;
 902         int i, j, filled;
 903
 904         work = xcalloc(cnt, sizeof(*work));
 905         redundant = xcalloc(cnt, 1);
 906         ALLOC_ARRAY(filled_index, cnt - 1);
 907
 908         for (i = 0; i < cnt; i++)
 909                 parse_commit(array[i]);
 910         for (i = 0; i < cnt; i++) {
 911                 struct commit_list *common;
 912
 913                 if (redundant[i])
 914                         continue;
 915                 for (j = filled = 0; j < cnt; j++) {
 916                         if (i == j || redundant[j])
 917                                 continue;
 918                         filled_index[filled] = j;
 919                         work[filled++] = array[j];
 920                 }
 921                 common = paint_down_to_common(array[i], filled, work);
 922                 if (array[i]->object.flags & PARENT2)
 923                         redundant[i] = 1;
 924                 for (j = 0; j < filled; j++)
 925                         if (work[j]->object.flags & PARENT1)
 926                                 redundant[filled_index[j]] = 1;
 927                 clear_commit_marks(array[i], all_flags);
 928                 for (j = 0; j < filled; j++)
 929                         clear_commit_marks(work[j], all_flags);
 930                 free_commit_list(common);
 931         }
 932
 933         /* Now collect the result */
 934         COPY_ARRAY(work, array, cnt);
 935         for (i = filled = 0; i < cnt; i++)
 936                 if (!redundant[i])
 937                         array[filled++] = work[i];
 938         for (j = filled, i = 0; i < cnt; i++)
 939                 if (redundant[i])
 940                         array[j++] = work[i];
 941         free(work);
 942         free(redundant);
 943         free(filled_index);
 944         return filled;
 945 }
 946
 947 static struct commit_list *get_merge_bases_many_0(struct commit *one,
 948                                                   int n,
 949                                                   struct commit **twos,
 950                                                   int cleanup)
 951 {
 952         struct commit_list *list;
 953         struct commit **rslt;
 954         struct commit_list *result;
 955         int cnt, i;
 956
 957         result = merge_bases_many(one, n, twos);
 958         for (i = 0; i < n; i++) {
 959                 if (one == twos[i])
 960                         return result;
 961         }
 962         if (!result || !result->next) {
 963                 if (cleanup) {
 964                         clear_commit_marks(one, all_flags);
 965                         clear_commit_marks_many(n, twos, all_flags);
 966                 }
 967                 return result;
 968         }
 969
 970         /* There are more than one */
 971         cnt = commit_list_count(result);
 972         rslt = xcalloc(cnt, sizeof(*rslt));
 973         for (list = result, i = 0; list; list = list->next)
 974                 rslt[i++] = list->item;
 975         free_commit_list(result);
 976
 977         clear_commit_marks(one, all_flags);
 978         clear_commit_marks_many(n, twos, all_flags);
 979
 980         cnt = remove_redundant(rslt, cnt);
 981         result = NULL;
 982         for (i = 0; i < cnt; i++)
 983                 commit_list_insert_by_date(rslt[i], &result);
 984         free(rslt);
 985         return result;
 986 }
 987
 988 struct commit_list *get_merge_bases_many(struct commit *one,
 989                                          int n,
 990                                          struct commit **twos)
 991 {
 992         return get_merge_bases_many_0(one, n, twos, 1);
 993 }
 994
 995 struct commit_list *get_merge_bases_many_dirty(struct commit *one,
 996                                                int n,
 997                                                struct commit **twos)
 998 {
 999         return get_merge_bases_many_0(one, n, twos, 0);
1000 }
1001
1002 struct commit_list *get_merge_bases(struct commit *one, struct commit *two)
1003 {
1004         return get_merge_bases_many_0(one, 1, &two, 1);
1005 }
1006
1007 /*
1008  * Is "commit" a descendant of one of the elements on the "with_commit" list?
1009  */
1010 int is_descendant_of(struct commit *commit, struct commit_list *with_commit)
1011 {
1012         if (!with_commit)
1013                 return 1;
1014         while (with_commit) {
1015                 struct commit *other;
1016
1017                 other = with_commit->item;
1018                 with_commit = with_commit->next;
1019                 if (in_merge_bases(other, commit))
1020                         return 1;
1021         }
1022         return 0;
1023 }
1024
1025 /*
1026  * Is "commit" an ancestor of one of the "references"?
1027  */
1028 int in_merge_bases_many(struct commit *commit, int nr_reference, struct commit **reference)
1029 {
1030         struct commit_list *bases;
1031         int ret = 0, i;
1032
1033         if (parse_commit(commit))
1034                 return ret;
1035         for (i = 0; i < nr_reference; i++)
1036                 if (parse_commit(reference[i]))
1037                         return ret;
1038
1039         bases = paint_down_to_common(commit, nr_reference, reference);
1040         if (commit->object.flags & PARENT2)
1041                 ret = 1;
1042         clear_commit_marks(commit, all_flags);
1043         clear_commit_marks_many(nr_reference, reference, all_flags);
1044         free_commit_list(bases);
1045         return ret;
1046 }
1047
1048 /*
1049  * Is "commit" an ancestor of (i.e. reachable from) the "reference"?
1050  */
1051 int in_merge_bases(struct commit *commit, struct commit *reference)
1052 {
1053         return in_merge_bases_many(commit, 1, &reference);
1054 }
1055
1056 struct commit_list *reduce_heads(struct commit_list *heads)
1057 {
1058         struct commit_list *p;
1059         struct commit_list *result = NULL, **tail = &result;
1060         struct commit **array;
1061         int num_head, i;
1062
1063         if (!heads)
1064                 return NULL;
1065
1066         /* Uniquify */
1067         for (p = heads; p; p = p->next)
1068                 p->item->object.flags &= ~STALE;
1069         for (p = heads, num_head = 0; p; p = p->next) {
1070                 if (p->item->object.flags & STALE)
1071                         continue;
1072                 p->item->object.flags |= STALE;
1073                 num_head++;
1074         }
1075         array = xcalloc(num_head, sizeof(*array));
1076         for (p = heads, i = 0; p; p = p->next) {
1077                 if (p->item->object.flags & STALE) {
1078                         array[i++] = p->item;
1079                         p->item->object.flags &= ~STALE;
1080                 }
1081         }
1082         num_head = remove_redundant(array, num_head);
1083         for (i = 0; i < num_head; i++)
1084                 tail = &commit_list_insert(array[i], tail)->next;
1085         return result;
1086 }
1087
1088 static const char gpg_sig_header[] = "gpgsig";
1089 static const int gpg_sig_header_len = sizeof(gpg_sig_header) - 1;
1090
1091 static int do_sign_commit(struct strbuf *buf, const char *keyid)
1092 {
1093         struct strbuf sig = STRBUF_INIT;
1094         int inspos, copypos;
1095         const char *eoh;
1096
1097         /* find the end of the header */
1098         eoh = strstr(buf->buf, "\n\n");
1099         if (!eoh)
1100                 inspos = buf->len;
1101         else
1102                 inspos = eoh - buf->buf + 1;
1103
1104         if (!keyid || !*keyid)
1105                 keyid = get_signing_key();
1106         if (sign_buffer(buf, &sig, keyid)) {
1107                 strbuf_release(&sig);
1108                 return -1;
1109         }
1110
1111         for (copypos = 0; sig.buf[copypos]; ) {
1112                 const char *bol = sig.buf + copypos;
1113                 const char *eol = strchrnul(bol, '\n');
1114                 int len = (eol - bol) + !!*eol;
1115
1116                 if (!copypos) {
1117                         strbuf_insert(buf, inspos, gpg_sig_header, gpg_sig_header_len);
1118                         inspos += gpg_sig_header_len;
1119                 }
1120                 strbuf_insert(buf, inspos++, " ", 1);
1121                 strbuf_insert(buf, inspos, bol, len);
1122                 inspos += len;
1123                 copypos += len;
1124         }
1125         strbuf_release(&sig);
1126         return 0;
1127 }
1128
1129 int parse_signed_commit(const struct commit *commit,
1130                         struct strbuf *payload, struct strbuf *signature)
1131 {
1132
1133         unsigned long size;
1134         const char *buffer = get_commit_buffer(commit, &size);
1135         int in_signature, saw_signature = -1;
1136         const char *line, *tail;
1137
1138         line = buffer;
1139         tail = buffer + size;
1140         in_signature = 0;
1141         saw_signature = 0;
1142         while (line < tail) {
1143                 const char *sig = NULL;
1144                 const char *next = memchr(line, '\n', tail - line);
1145
1146                 next = next ? next + 1 : tail;
1147                 if (in_signature && line[0] == ' ')
1148                         sig = line + 1;
1149                 else if (starts_with(line, gpg_sig_header) &&
1150                          line[gpg_sig_header_len] == ' ')
1151                         sig = line + gpg_sig_header_len + 1;
1152                 if (sig) {
1153                         strbuf_add(signature, sig, next - sig);
1154                         saw_signature = 1;
1155                         in_signature = 1;
1156                 } else {
1157                         if (*line == '\n')
1158                                 /* dump the whole remainder of the buffer */
1159                                 next = tail;
1160                         strbuf_add(payload, line, next - line);
1161                         in_signature = 0;
1162                 }
1163                 line = next;
1164         }
1165         unuse_commit_buffer(commit, buffer);
1166         return saw_signature;
1167 }
1168
1169 int remove_signature(struct strbuf *buf)
1170 {
1171         const char *line = buf->buf;
1172         const char *tail = buf->buf + buf->len;
1173         int in_signature = 0;
1174         const char *sig_start = NULL;
1175         const char *sig_end = NULL;
1176
1177         while (line < tail) {
1178                 const char *next = memchr(line, '\n', tail - line);
1179                 next = next ? next + 1 : tail;
1180
1181                 if (in_signature && line[0] == ' ')
1182                         sig_end = next;
1183                 else if (starts_with(line, gpg_sig_header) &&
1184                          line[gpg_sig_header_len] == ' ') {
1185                         sig_start = line;
1186                         sig_end = next;
1187                         in_signature = 1;
1188                 } else {
1189                         if (*line == '\n')
1190                                 /* dump the whole remainder of the buffer */
1191                                 next = tail;
1192                         in_signature = 0;
1193                 }
1194                 line = next;
1195         }
1196
1197         if (sig_start)
1198                 strbuf_remove(buf, sig_start - buf->buf, sig_end - sig_start);
1199
1200         return sig_start != NULL;
1201 }
1202
1203 static void handle_signed_tag(struct commit *parent, struct commit_extra_header ***tail)
1204 {
1205         struct merge_remote_desc *desc;
1206         struct commit_extra_header *mergetag;
1207         char *buf;
1208         unsigned long size, len;
1209         enum object_type type;
1210
1211         desc = merge_remote_util(parent);
1212         if (!desc || !desc->obj)
1213                 return;
1214         buf = read_sha1_file(desc->obj->oid.hash, &type, &size);
1215         if (!buf || type != OBJ_TAG)
1216                 goto free_return;
1217         len = parse_signature(buf, size);
1218         if (size == len)
1219                 goto free_return;
1220         /*
1221          * We could verify this signature and either omit the tag when
1222          * it does not validate, but the integrator may not have the
1223          * public key of the signer of the tag he is merging, while a
1224          * later auditor may have it while auditing, so let's not run
1225          * verify-signed-buffer here for now...
1226          *
1227          * if (verify_signed_buffer(buf, len, buf + len, size - len, ...))
1228          *      warn("warning: signed tag unverified.");
1229          */
1230         mergetag = xcalloc(1, sizeof(*mergetag));
1231         mergetag->key = xstrdup("mergetag");
1232         mergetag->value = buf;
1233         mergetag->len = size;
1234
1235         **tail = mergetag;
1236         *tail = &mergetag->next;
1237         return;
1238
1239 free_return:
1240         free(buf);
1241 }
1242
1243 int check_commit_signature(const struct commit *commit, struct signature_check *sigc)
1244 {
1245         struct strbuf payload = STRBUF_INIT;
1246         struct strbuf signature = STRBUF_INIT;
1247         int ret = 1;
1248
1249         sigc->result = 'N';
1250
1251         if (parse_signed_commit(commit, &payload, &signature) <= 0)
1252                 goto out;
1253         ret = check_signature(payload.buf, payload.len, signature.buf,
1254                 signature.len, sigc);
1255
1256  out:
1257         strbuf_release(&payload);
1258         strbuf_release(&signature);
1259
1260         return ret;
1261 }
1262
1263
1264
1265 void append_merge_tag_headers(struct commit_list *parents,
1266                               struct commit_extra_header ***tail)
1267 {
1268         while (parents) {
1269                 struct commit *parent = parents->item;
1270                 handle_signed_tag(parent, tail);
1271                 parents = parents->next;
1272         }
1273 }
1274
1275 static void add_extra_header(struct strbuf *buffer,
1276                              struct commit_extra_header *extra)
1277 {
1278         strbuf_addstr(buffer, extra->key);
1279         if (extra->len)
1280                 strbuf_add_lines(buffer, " ", extra->value, extra->len);
1281         else
1282                 strbuf_addch(buffer, '\n');
1283 }
1284
1285 struct commit_extra_header *read_commit_extra_headers(struct commit *commit,
1286                                                       const char **exclude)
1287 {
1288         struct commit_extra_header *extra = NULL;
1289         unsigned long size;
1290         const char *buffer = get_commit_buffer(commit, &size);
1291         extra = read_commit_extra_header_lines(buffer, size, exclude);
1292         unuse_commit_buffer(commit, buffer);
1293         return extra;
1294 }
1295
1296 void for_each_mergetag(each_mergetag_fn fn, struct commit *commit, void *data)
1297 {
1298         struct commit_extra_header *extra, *to_free;
1299
1300         to_free = read_commit_extra_headers(commit, NULL);
1301         for (extra = to_free; extra; extra = extra->next) {
1302                 if (strcmp(extra->key, "mergetag"))
1303                         continue; /* not a merge tag */
1304                 fn(commit, extra, data);
1305         }
1306         free_commit_extra_headers(to_free);
1307 }
1308
1309 static inline int standard_header_field(const char *field, size_t len)
1310 {
1311         return ((len == 4 && !memcmp(field, "tree", 4)) ||
1312                 (len == 6 && !memcmp(field, "parent", 6)) ||
1313                 (len == 6 && !memcmp(field, "author", 6)) ||
1314                 (len == 9 && !memcmp(field, "committer", 9)) ||
1315                 (len == 8 && !memcmp(field, "encoding", 8)));
1316 }
1317
1318 static int excluded_header_field(const char *field, size_t len, const char **exclude)
1319 {
1320         if (!exclude)
1321                 return 0;
1322
1323         while (*exclude) {
1324                 size_t xlen = strlen(*exclude);
1325                 if (len == xlen && !memcmp(field, *exclude, xlen))
1326                         return 1;
1327                 exclude++;
1328         }
1329         return 0;
1330 }
1331
1332 static struct commit_extra_header *read_commit_extra_header_lines(
1333         const char *buffer, size_t size,
1334         const char **exclude)
1335 {
1336         struct commit_extra_header *extra = NULL, **tail = &extra, *it = NULL;
1337         const char *line, *next, *eof, *eob;
1338         struct strbuf buf = STRBUF_INIT;
1339
1340         for (line = buffer, eob = line + size;
1341              line < eob && *line != '\n';
1342              line = next) {
1343                 next = memchr(line, '\n', eob - line);
1344                 next = next ? next + 1 : eob;
1345                 if (*line == ' ') {
1346                         /* continuation */
1347                         if (it)
1348                                 strbuf_add(&buf, line + 1, next - (line + 1));
1349                         continue;
1350                 }
1351                 if (it)
1352                         it->value = strbuf_detach(&buf, &it->len);
1353                 strbuf_reset(&buf);
1354                 it = NULL;
1355
1356                 eof = memchr(line, ' ', next - line);
1357                 if (!eof)
1358                         eof = next;
1359                 else if (standard_header_field(line, eof - line) ||
1360                          excluded_header_field(line, eof - line, exclude))
1361                         continue;
1362
1363                 it = xcalloc(1, sizeof(*it));
1364                 it->key = xmemdupz(line, eof-line);
1365                 *tail = it;
1366                 tail = &it->next;
1367                 if (eof + 1 < next)
1368                         strbuf_add(&buf, eof + 1, next - (eof + 1));
1369         }
1370         if (it)
1371                 it->value = strbuf_detach(&buf, &it->len);
1372         return extra;
1373 }
1374
1375 void free_commit_extra_headers(struct commit_extra_header *extra)
1376 {
1377         while (extra) {
1378                 struct commit_extra_header *next = extra->next;
1379                 free(extra->key);
1380                 free(extra->value);
1381                 free(extra);
1382                 extra = next;
1383         }
1384 }
1385
1386 int commit_tree(const char *msg, size_t msg_len,
1387                 const unsigned char *tree,
1388                 struct commit_list *parents, unsigned char *ret,
1389                 const char *author, const char *sign_commit)
1390 {
1391         struct commit_extra_header *extra = NULL, **tail = &extra;
1392         int result;
1393
1394         append_merge_tag_headers(parents, &tail);
1395         result = commit_tree_extended(msg, msg_len, tree, parents, ret,
1396                                       author, sign_commit, extra);
1397         free_commit_extra_headers(extra);
1398         return result;
1399 }
1400
1401 static int find_invalid_utf8(const char *buf, int len)
1402 {
1403         int offset = 0;
1404         static const unsigned int max_codepoint[] = {
1405                 0x7f, 0x7ff, 0xffff, 0x10ffff
1406         };
1407
1408         while (len) {
1409                 unsigned char c = *buf++;
1410                 int bytes, bad_offset;
1411                 unsigned int codepoint;
1412                 unsigned int min_val, max_val;
1413
1414                 len--;
1415                 offset++;
1416
1417                 /* Simple US-ASCII? No worries. */
1418                 if (c < 0x80)
1419                         continue;
1420
1421                 bad_offset = offset-1;
1422
1423                 /*
1424                  * Count how many more high bits set: that's how
1425                  * many more bytes this sequence should have.
1426                  */
1427                 bytes = 0;
1428                 while (c & 0x40) {
1429                         c <<= 1;
1430                         bytes++;
1431                 }
1432
1433                 /*
1434                  * Must be between 1 and 3 more bytes.  Longer sequences result in
1435                  * codepoints beyond U+10FFFF, which are guaranteed never to exist.
1436                  */
1437                 if (bytes < 1 || 3 < bytes)
1438                         return bad_offset;
1439
1440                 /* Do we *have* that many bytes? */
1441                 if (len < bytes)
1442                         return bad_offset;
1443
1444                 /*
1445                  * Place the encoded bits at the bottom of the value and compute the
1446                  * valid range.
1447                  */
1448                 codepoint = (c & 0x7f) >> bytes;
1449                 min_val = max_codepoint[bytes-1] + 1;
1450                 max_val = max_codepoint[bytes];
1451
1452                 offset += bytes;
1453                 len -= bytes;
1454
1455                 /* And verify that they are good continuation bytes */
1456                 do {
1457                         codepoint <<= 6;
1458                         codepoint |= *buf & 0x3f;
1459                         if ((*buf++ & 0xc0) != 0x80)
1460                                 return bad_offset;
1461                 } while (--bytes);
1462
1463                 /* Reject codepoints that are out of range for the sequence length. */
1464                 if (codepoint < min_val || codepoint > max_val)
1465                         return bad_offset;
1466                 /* Surrogates are only for UTF-16 and cannot be encoded in UTF-8. */
1467                 if ((codepoint & 0x1ff800) == 0xd800)
1468                         return bad_offset;
1469                 /* U+xxFFFE and U+xxFFFF are guaranteed non-characters. */
1470                 if ((codepoint & 0xfffe) == 0xfffe)
1471                         return bad_offset;
1472                 /* So are anything in the range U+FDD0..U+FDEF. */
1473                 if (codepoint >= 0xfdd0 && codepoint <= 0xfdef)
1474                         return bad_offset;
1475         }
1476         return -1;
1477 }
1478
1479 /*
1480  * This verifies that the buffer is in proper utf8 format.
1481  *
1482  * If it isn't, it assumes any non-utf8 characters are Latin1,
1483  * and does the conversion.
1484  */
1485 static int verify_utf8(struct strbuf *buf)
1486 {
1487         int ok = 1;
1488         long pos = 0;
1489
1490         for (;;) {
1491                 int bad;
1492                 unsigned char c;
1493                 unsigned char replace[2];
1494
1495                 bad = find_invalid_utf8(buf->buf + pos, buf->len - pos);
1496                 if (bad < 0)
1497                         return ok;
1498                 pos += bad;
1499                 ok = 0;
1500                 c = buf->buf[pos];
1501                 strbuf_remove(buf, pos, 1);
1502
1503                 /* We know 'c' must be in the range 128-255 */
1504                 replace[0] = 0xc0 + (c >> 6);
1505                 replace[1] = 0x80 + (c & 0x3f);
1506                 strbuf_insert(buf, pos, replace, 2);
1507                 pos += 2;
1508         }
1509 }
1510
1511 static const char commit_utf8_warn[] =
1512 N_("Warning: commit message did not conform to UTF-8.\n"
1513    "You may want to amend it after fixing the message, or set the config\n"
1514    "variable i18n.commitencoding to the encoding your project uses.\n");
1515
1516 int commit_tree_extended(const char *msg, size_t msg_len,
1517                          const unsigned char *tree,
1518                          struct commit_list *parents, unsigned char *ret,
1519                          const char *author, const char *sign_commit,
1520                          struct commit_extra_header *extra)
1521 {
1522         int result;
1523         int encoding_is_utf8;
1524         struct strbuf buffer;
1525
1526         assert_sha1_type(tree, OBJ_TREE);
1527
1528         if (memchr(msg, '\0', msg_len))
1529                 return error("a NUL byte in commit log message not allowed.");
1530
1531         /* Not having i18n.commitencoding is the same as having utf-8 */
1532         encoding_is_utf8 = is_encoding_utf8(git_commit_encoding);
1533
1534         strbuf_init(&buffer, 8192); /* should avoid reallocs for the headers */
1535         strbuf_addf(&buffer, "tree %s\n", sha1_to_hex(tree));
1536
1537         /*
1538          * NOTE! This ordering means that the same exact tree merged with a
1539          * different order of parents will be a _different_ changeset even
1540          * if everything else stays the same.
1541          */
1542         while (parents) {
1543                 struct commit *parent = pop_commit(&parents);
1544                 strbuf_addf(&buffer, "parent %s\n",
1545                             oid_to_hex(&parent->object.oid));
1546         }
1547
1548         /* Person/date information */
1549         if (!author)
1550                 author = git_author_info(IDENT_STRICT);
1551         strbuf_addf(&buffer, "author %s\n", author);
1552         strbuf_addf(&buffer, "committer %s\n", git_committer_info(IDENT_STRICT));
1553         if (!encoding_is_utf8)
1554                 strbuf_addf(&buffer, "encoding %s\n", git_commit_encoding);
1555
1556         while (extra) {
1557                 add_extra_header(&buffer, extra);
1558                 extra = extra->next;
1559         }
1560         strbuf_addch(&buffer, '\n');
1561
1562         /* And add the comment */
1563         strbuf_add(&buffer, msg, msg_len);
1564
1565         /* And check the encoding */
1566         if (encoding_is_utf8 && !verify_utf8(&buffer))
1567                 fprintf(stderr, _(commit_utf8_warn));
1568
1569         if (sign_commit && do_sign_commit(&buffer, sign_commit))
1570                 return -1;
1571
1572         result = write_sha1_file(buffer.buf, buffer.len, commit_type, ret);
1573         strbuf_release(&buffer);
1574         return result;
1575 }
1576
1577 void set_merge_remote_desc(struct commit *commit,
1578                            const char *name, struct object *obj)
1579 {
1580         struct merge_remote_desc *desc;
1581         FLEX_ALLOC_STR(desc, name, name);
1582         desc->obj = obj;
1583         commit->util = desc;
1584 }
1585
1586 struct commit *get_merge_parent(const char *name)
1587 {
1588         struct object *obj;
1589         struct commit *commit;
1590         struct object_id oid;
1591         if (get_sha1(name, oid.hash))
1592                 return NULL;
1593         obj = parse_object(&oid);
1594         commit = (struct commit *)peel_to_type(name, 0, obj, OBJ_COMMIT);
1595         if (commit && !commit->util)
1596                 set_merge_remote_desc(commit, name, obj);
1597         return commit;
1598 }
1599
1600 /*
1601  * Append a commit to the end of the commit_list.
1602  *
1603  * next starts by pointing to the variable that holds the head of an
1604  * empty commit_list, and is updated to point to the "next" field of
1605  * the last item on the list as new commits are appended.
1606  *
1607  * Usage example:
1608  *
1609  *     struct commit_list *list;
1610  *     struct commit_list **next = &list;
1611  *
1612  *     next = commit_list_append(c1, next);
1613  *     next = commit_list_append(c2, next);
1614  *     assert(commit_list_count(list) == 2);
1615  *     return list;
1616  */
1617 struct commit_list **commit_list_append(struct commit *commit,
1618                                         struct commit_list **next)
1619 {
1620         struct commit_list *new = xmalloc(sizeof(struct commit_list));
1621         new->item = commit;
1622         *next = new;
1623         new->next = NULL;
1624         return &new->next;
1625 }
1626
1627 const char *find_commit_header(const char *msg, const char *key, size_t *out_len)
1628 {
1629         int key_len = strlen(key);
1630         const char *line = msg;
1631
1632         while (line) {
1633                 const char *eol = strchrnul(line, '\n');
1634
1635                 if (line == eol)
1636                         return NULL;
1637
1638                 if (eol - line > key_len &&
1639                     !strncmp(line, key, key_len) &&
1640                     line[key_len] == ' ') {
1641                         *out_len = eol - line - key_len - 1;
1642                         return line + key_len + 1;
1643                 }
1644                 line = *eol ? eol + 1 : NULL;
1645         }
1646         return NULL;
1647 }
1648
1649 /*
1650  * Inspect the given string and determine the true "end" of the log message, in
1651  * order to find where to put a new Signed-off-by: line.  Ignored are
1652  * trailing comment lines and blank lines.  To support "git commit -s
1653  * --amend" on an existing commit, we also ignore "Conflicts:".  To
1654  * support "git commit -v", we truncate at cut lines.
1655  *
1656  * Returns the number of bytes from the tail to ignore, to be fed as
1657  * the second parameter to append_signoff().
1658  */
1659 int ignore_non_trailer(const char *buf, size_t len)
1660 {
1661         int boc = 0;
1662         int bol = 0;
1663         int in_old_conflicts_block = 0;
1664         size_t cutoff = wt_status_locate_end(buf, len);
1665
1666         while (bol < cutoff) {
1667                 const char *next_line = memchr(buf + bol, '\n', len - bol);
1668
1669                 if (!next_line)
1670                         next_line = buf + len;
1671                 else
1672                         next_line++;
1673
1674                 if (buf[bol] == comment_line_char || buf[bol] == '\n') {
1675                         /* is this the first of the run of comments? */
1676                         if (!boc)
1677                                 boc = bol;
1678                         /* otherwise, it is just continuing */
1679                 } else if (starts_with(buf + bol, "Conflicts:\n")) {
1680                         in_old_conflicts_block = 1;
1681                         if (!boc)
1682                                 boc = bol;
1683                 } else if (in_old_conflicts_block && buf[bol] == '\t') {
1684                         ; /* a pathname in the conflicts block */
1685                 } else if (boc) {
1686                         /* the previous was not trailing comment */
1687                         boc = 0;
1688                         in_old_conflicts_block = 0;
1689                 }
1690                 bol = next_line - buf;
1691         }
1692         return boc ? len - boc : len - cutoff;
1693 }