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_oid_committish(name, &oid))
  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                 MOVE_ARRAY(commit_graft + pos + 1, commit_graft + pos,
 130                            commit_graft_nr - pos - 1);
 131         commit_graft[pos] = graft;
 132         return 0;
 133 }
 134
 135 struct commit_graft *read_graft_line(struct strbuf *line)
 136 {
 137         /* The format is just "Commit Parent1 Parent2 ...\n" */
 138         int i, phase;
 139         const char *tail = NULL;
 140         struct commit_graft *graft = NULL;
 141         struct object_id dummy_oid, *oid;
 142
 143         strbuf_rtrim(line);
 144         if (!line->len || line->buf[0] == '#')
 145                 return NULL;
 146         /*
 147          * phase 0 verifies line, counts hashes in line and allocates graft
 148          * phase 1 fills graft
 149          */
 150         for (phase = 0; phase < 2; phase++) {
 151                 oid = graft ? &graft->oid : &dummy_oid;
 152                 if (parse_oid_hex(line->buf, oid, &tail))
 153                         goto bad_graft_data;
 154                 for (i = 0; *tail != '\0'; i++) {
 155                         oid = graft ? &graft->parent[i] : &dummy_oid;
 156                         if (!isspace(*tail++) || parse_oid_hex(tail, oid, &tail))
 157                                 goto bad_graft_data;
 158                 }
 159                 if (!graft) {
 160                         graft = xmalloc(st_add(sizeof(*graft),
 161                                                st_mult(sizeof(struct object_id), i)));
 162                         graft->nr_parent = i;
 163                 }
 164         }
 165         return graft;
 166
 167 bad_graft_data:
 168         error("bad graft data: %s", line->buf);
 169         assert(!graft);
 170         return NULL;
 171 }
 172
 173 static int read_graft_file(const char *graft_file)
 174 {
 175         FILE *fp = fopen_or_warn(graft_file, "r");
 176         struct strbuf buf = STRBUF_INIT;
 177         if (!fp)
 178                 return -1;
 179         while (!strbuf_getwholeline(&buf, fp, '\n')) {
 180                 /* The format is just "Commit Parent1 Parent2 ...\n" */
 181                 struct commit_graft *graft = read_graft_line(&buf);
 182                 if (!graft)
 183                         continue;
 184                 if (register_commit_graft(graft, 1))
 185                         error("duplicate graft data: %s", buf.buf);
 186         }
 187         fclose(fp);
 188         strbuf_release(&buf);
 189         return 0;
 190 }
 191
 192 static void prepare_commit_graft(void)
 193 {
 194         static int commit_graft_prepared;
 195         char *graft_file;
 196
 197         if (commit_graft_prepared)
 198                 return;
 199         graft_file = get_graft_file();
 200         read_graft_file(graft_file);
 201         /* make sure shallows are read */
 202         is_repository_shallow();
 203         commit_graft_prepared = 1;
 204 }
 205
 206 struct commit_graft *lookup_commit_graft(const struct object_id *oid)
 207 {
 208         int pos;
 209         prepare_commit_graft();
 210         pos = commit_graft_pos(oid->hash);
 211         if (pos < 0)
 212                 return NULL;
 213         return commit_graft[pos];
 214 }
 215
 216 int for_each_commit_graft(each_commit_graft_fn fn, void *cb_data)
 217 {
 218         int i, ret;
 219         for (i = ret = 0; i < commit_graft_nr && !ret; i++)
 220                 ret = fn(commit_graft[i], cb_data);
 221         return ret;
 222 }
 223
 224 int unregister_shallow(const struct object_id *oid)
 225 {
 226         int pos = commit_graft_pos(oid->hash);
 227         if (pos < 0)
 228                 return -1;
 229         if (pos + 1 < commit_graft_nr)
 230                 MOVE_ARRAY(commit_graft + pos, commit_graft + pos + 1,
 231                            commit_graft_nr - pos - 1);
 232         commit_graft_nr--;
 233         return 0;
 234 }
 235
 236 struct commit_buffer {
 237         void *buffer;
 238         unsigned long size;
 239 };
 240 define_commit_slab(buffer_slab, struct commit_buffer);
 241 static struct buffer_slab buffer_slab = COMMIT_SLAB_INIT(1, buffer_slab);
 242
 243 void set_commit_buffer(struct commit *commit, void *buffer, unsigned long size)
 244 {
 245         struct commit_buffer *v = buffer_slab_at(&buffer_slab, commit);
 246         v->buffer = buffer;
 247         v->size = size;
 248 }
 249
 250 const void *get_cached_commit_buffer(const struct commit *commit, unsigned long *sizep)
 251 {
 252         struct commit_buffer *v = buffer_slab_peek(&buffer_slab, commit);
 253         if (!v) {
 254                 if (sizep)
 255                         *sizep = 0;
 256                 return NULL;
 257         }
 258         if (sizep)
 259                 *sizep = v->size;
 260         return v->buffer;
 261 }
 262
 263 const void *get_commit_buffer(const struct commit *commit, unsigned long *sizep)
 264 {
 265         const void *ret = get_cached_commit_buffer(commit, sizep);
 266         if (!ret) {
 267                 enum object_type type;
 268                 unsigned long size;
 269                 ret = read_sha1_file(commit->object.oid.hash, &type, &size);
 270                 if (!ret)
 271                         die("cannot read commit object %s",
 272                             oid_to_hex(&commit->object.oid));
 273                 if (type != OBJ_COMMIT)
 274                         die("expected commit for %s, got %s",
 275                             oid_to_hex(&commit->object.oid), type_name(type));
 276                 if (sizep)
 277                         *sizep = size;
 278         }
 279         return ret;
 280 }
 281
 282 void unuse_commit_buffer(const struct commit *commit, const void *buffer)
 283 {
 284         struct commit_buffer *v = buffer_slab_peek(&buffer_slab, commit);
 285         if (!(v && v->buffer == buffer))
 286                 free((void *)buffer);
 287 }
 288
 289 void free_commit_buffer(struct commit *commit)
 290 {
 291         struct commit_buffer *v = buffer_slab_peek(&buffer_slab, commit);
 292         if (v) {
 293                 FREE_AND_NULL(v->buffer);
 294                 v->size = 0;
 295         }
 296 }
 297
 298 const void *detach_commit_buffer(struct commit *commit, unsigned long *sizep)
 299 {
 300         struct commit_buffer *v = buffer_slab_peek(&buffer_slab, commit);
 301         void *ret;
 302
 303         if (!v) {
 304                 if (sizep)
 305                         *sizep = 0;
 306                 return NULL;
 307         }
 308         ret = v->buffer;
 309         if (sizep)
 310                 *sizep = v->size;
 311
 312         v->buffer = NULL;
 313         v->size = 0;
 314         return ret;
 315 }
 316
 317 int parse_commit_buffer(struct commit *item, const void *buffer, unsigned long size)
 318 {
 319         const char *tail = buffer;
 320         const char *bufptr = buffer;
 321         struct object_id parent;
 322         struct commit_list **pptr;
 323         struct commit_graft *graft;
 324         const int tree_entry_len = GIT_SHA1_HEXSZ + 5;
 325         const int parent_entry_len = GIT_SHA1_HEXSZ + 7;
 326
 327         if (item->object.parsed)
 328                 return 0;
 329         item->object.parsed = 1;
 330         tail += size;
 331         if (tail <= bufptr + tree_entry_len + 1 || memcmp(bufptr, "tree ", 5) ||
 332                         bufptr[tree_entry_len] != '\n')
 333                 return error("bogus commit object %s", oid_to_hex(&item->object.oid));
 334         if (get_sha1_hex(bufptr + 5, parent.hash) < 0)
 335                 return error("bad tree pointer in commit %s",
 336                              oid_to_hex(&item->object.oid));
 337         item->tree = lookup_tree(&parent);
 338         bufptr += tree_entry_len + 1; /* "tree " + "hex sha1" + "\n" */
 339         pptr = &item->parents;
 340
 341         graft = lookup_commit_graft(&item->object.oid);
 342         while (bufptr + parent_entry_len < tail && !memcmp(bufptr, "parent ", 7)) {
 343                 struct commit *new_parent;
 344
 345                 if (tail <= bufptr + parent_entry_len + 1 ||
 346                     get_sha1_hex(bufptr + 7, parent.hash) ||
 347                     bufptr[parent_entry_len] != '\n')
 348                         return error("bad parents in commit %s", oid_to_hex(&item->object.oid));
 349                 bufptr += parent_entry_len + 1;
 350                 /*
 351                  * The clone is shallow if nr_parent < 0, and we must
 352                  * not traverse its real parents even when we unhide them.
 353                  */
 354                 if (graft && (graft->nr_parent < 0 || grafts_replace_parents))
 355                         continue;
 356                 new_parent = lookup_commit(&parent);
 357                 if (new_parent)
 358                         pptr = &commit_list_insert(new_parent, pptr)->next;
 359         }
 360         if (graft) {
 361                 int i;
 362                 struct commit *new_parent;
 363                 for (i = 0; i < graft->nr_parent; i++) {
 364                         new_parent = lookup_commit(&graft->parent[i]);
 365                         if (!new_parent)
 366                                 continue;
 367                         pptr = &commit_list_insert(new_parent, pptr)->next;
 368                 }
 369         }
 370         item->date = parse_commit_date(bufptr, tail);
 371
 372         return 0;
 373 }
 374
 375 int parse_commit_gently(struct commit *item, int quiet_on_missing)
 376 {
 377         enum object_type type;
 378         void *buffer;
 379         unsigned long size;
 380         int ret;
 381
 382         if (!item)
 383                 return -1;
 384         if (item->object.parsed)
 385                 return 0;
 386         buffer = read_sha1_file(item->object.oid.hash, &type, &size);
 387         if (!buffer)
 388                 return quiet_on_missing ? -1 :
 389                         error("Could not read %s",
 390                              oid_to_hex(&item->object.oid));
 391         if (type != OBJ_COMMIT) {
 392                 free(buffer);
 393                 return error("Object %s not a commit",
 394                              oid_to_hex(&item->object.oid));
 395         }
 396         ret = parse_commit_buffer(item, buffer, size);
 397         if (save_commit_buffer && !ret) {
 398                 set_commit_buffer(item, buffer, size);
 399                 return 0;
 400         }
 401         free(buffer);
 402         return ret;
 403 }
 404
 405 void parse_commit_or_die(struct commit *item)
 406 {
 407         if (parse_commit(item))
 408                 die("unable to parse commit %s",
 409                     item ? oid_to_hex(&item->object.oid) : "(null)");
 410 }
 411
 412 int find_commit_subject(const char *commit_buffer, const char **subject)
 413 {
 414         const char *eol;
 415         const char *p = commit_buffer;
 416
 417         while (*p && (*p != '\n' || p[1] != '\n'))
 418                 p++;
 419         if (*p) {
 420                 p = skip_blank_lines(p + 2);
 421                 eol = strchrnul(p, '\n');
 422         } else
 423                 eol = p;
 424
 425         *subject = p;
 426
 427         return eol - p;
 428 }
 429
 430 struct commit_list *commit_list_insert(struct commit *item, struct commit_list **list_p)
 431 {
 432         struct commit_list *new_list = xmalloc(sizeof(struct commit_list));
 433         new_list->item = item;
 434         new_list->next = *list_p;
 435         *list_p = new_list;
 436         return new_list;
 437 }
 438
 439 unsigned commit_list_count(const struct commit_list *l)
 440 {
 441         unsigned c = 0;
 442         for (; l; l = l->next )
 443                 c++;
 444         return c;
 445 }
 446
 447 struct commit_list *copy_commit_list(struct commit_list *list)
 448 {
 449         struct commit_list *head = NULL;
 450         struct commit_list **pp = &head;
 451         while (list) {
 452                 pp = commit_list_append(list->item, pp);
 453                 list = list->next;
 454         }
 455         return head;
 456 }
 457
 458 void free_commit_list(struct commit_list *list)
 459 {
 460         while (list)
 461                 pop_commit(&list);
 462 }
 463
 464 struct commit_list * commit_list_insert_by_date(struct commit *item, struct commit_list **list)
 465 {
 466         struct commit_list **pp = list;
 467         struct commit_list *p;
 468         while ((p = *pp) != NULL) {
 469                 if (p->item->date < item->date) {
 470                         break;
 471                 }
 472                 pp = &p->next;
 473         }
 474         return commit_list_insert(item, pp);
 475 }
 476
 477 static int commit_list_compare_by_date(const void *a, const void *b)
 478 {
 479         timestamp_t a_date = ((const struct commit_list *)a)->item->date;
 480         timestamp_t b_date = ((const struct commit_list *)b)->item->date;
 481         if (a_date < b_date)
 482                 return 1;
 483         if (a_date > b_date)
 484                 return -1;
 485         return 0;
 486 }
 487
 488 static void *commit_list_get_next(const void *a)
 489 {
 490         return ((const struct commit_list *)a)->next;
 491 }
 492
 493 static void commit_list_set_next(void *a, void *next)
 494 {
 495         ((struct commit_list *)a)->next = next;
 496 }
 497
 498 void commit_list_sort_by_date(struct commit_list **list)
 499 {
 500         *list = llist_mergesort(*list, commit_list_get_next, commit_list_set_next,
 501                                 commit_list_compare_by_date);
 502 }
 503
 504 struct commit *pop_most_recent_commit(struct commit_list **list,
 505                                       unsigned int mark)
 506 {
 507         struct commit *ret = pop_commit(list);
 508         struct commit_list *parents = ret->parents;
 509
 510         while (parents) {
 511                 struct commit *commit = parents->item;
 512                 if (!parse_commit(commit) && !(commit->object.flags & mark)) {
 513                         commit->object.flags |= mark;
 514                         commit_list_insert_by_date(commit, list);
 515                 }
 516                 parents = parents->next;
 517         }
 518         return ret;
 519 }
 520
 521 static void clear_commit_marks_1(struct commit_list **plist,
 522                                  struct commit *commit, unsigned int mark)
 523 {
 524         while (commit) {
 525                 struct commit_list *parents;
 526
 527                 if (!(mark & commit->object.flags))
 528                         return;
 529
 530                 commit->object.flags &= ~mark;
 531
 532                 parents = commit->parents;
 533                 if (!parents)
 534                         return;
 535
 536                 while ((parents = parents->next))
 537                         commit_list_insert(parents->item, plist);
 538
 539                 commit = commit->parents->item;
 540         }
 541 }
 542
 543 void clear_commit_marks_many(int nr, struct commit **commit, unsigned int mark)
 544 {
 545         struct commit_list *list = NULL;
 546
 547         while (nr--) {
 548                 clear_commit_marks_1(&list, *commit, mark);
 549                 commit++;
 550         }
 551         while (list)
 552                 clear_commit_marks_1(&list, pop_commit(&list), mark);
 553 }
 554
 555 void clear_commit_marks(struct commit *commit, unsigned int mark)
 556 {
 557         clear_commit_marks_many(1, &commit, mark);
 558 }
 559
 560 struct commit *pop_commit(struct commit_list **stack)
 561 {
 562         struct commit_list *top = *stack;
 563         struct commit *item = top ? top->item : NULL;
 564
 565         if (top) {
 566                 *stack = top->next;
 567                 free(top);
 568         }
 569         return item;
 570 }
 571
 572 /*
 573  * Topological sort support
 574  */
 575
 576 /* count number of children that have not been emitted */
 577 define_commit_slab(indegree_slab, int);
 578
 579 /* record author-date for each commit object */
 580 define_commit_slab(author_date_slab, unsigned long);
 581
 582 static void record_author_date(struct author_date_slab *author_date,
 583                                struct commit *commit)
 584 {
 585         const char *buffer = get_commit_buffer(commit, NULL);
 586         struct ident_split ident;
 587         const char *ident_line;
 588         size_t ident_len;
 589         char *date_end;
 590         timestamp_t date;
 591
 592         ident_line = find_commit_header(buffer, "author", &ident_len);
 593         if (!ident_line)
 594                 goto fail_exit; /* no author line */
 595         if (split_ident_line(&ident, ident_line, ident_len) ||
 596             !ident.date_begin || !ident.date_end)
 597                 goto fail_exit; /* malformed "author" line */
 598
 599         date = parse_timestamp(ident.date_begin, &date_end, 10);
 600         if (date_end != ident.date_end)
 601                 goto fail_exit; /* malformed date */
 602         *(author_date_slab_at(author_date, commit)) = date;
 603
 604 fail_exit:
 605         unuse_commit_buffer(commit, buffer);
 606 }
 607
 608 static int compare_commits_by_author_date(const void *a_, const void *b_,
 609                                           void *cb_data)
 610 {
 611         const struct commit *a = a_, *b = b_;
 612         struct author_date_slab *author_date = cb_data;
 613         timestamp_t a_date = *(author_date_slab_at(author_date, a));
 614         timestamp_t b_date = *(author_date_slab_at(author_date, b));
 615
 616         /* newer commits with larger date first */
 617         if (a_date < b_date)
 618                 return 1;
 619         else if (a_date > b_date)
 620                 return -1;
 621         return 0;
 622 }
 623
 624 int compare_commits_by_commit_date(const void *a_, const void *b_, void *unused)
 625 {
 626         const struct commit *a = a_, *b = b_;
 627         /* newer commits with larger date first */
 628         if (a->date < b->date)
 629                 return 1;
 630         else if (a->date > b->date)
 631                 return -1;
 632         return 0;
 633 }
 634
 635 /*
 636  * Performs an in-place topological sort on the list supplied.
 637  */
 638 void sort_in_topological_order(struct commit_list **list, enum rev_sort_order sort_order)
 639 {
 640         struct commit_list *next, *orig = *list;
 641         struct commit_list **pptr;
 642         struct indegree_slab indegree;
 643         struct prio_queue queue;
 644         struct commit *commit;
 645         struct author_date_slab author_date;
 646
 647         if (!orig)
 648                 return;
 649         *list = NULL;
 650
 651         init_indegree_slab(&indegree);
 652         memset(&queue, '\0', sizeof(queue));
 653
 654         switch (sort_order) {
 655         default: /* REV_SORT_IN_GRAPH_ORDER */
 656                 queue.compare = NULL;
 657                 break;
 658         case REV_SORT_BY_COMMIT_DATE:
 659                 queue.compare = compare_commits_by_commit_date;
 660                 break;
 661         case REV_SORT_BY_AUTHOR_DATE:
 662                 init_author_date_slab(&author_date);
 663                 queue.compare = compare_commits_by_author_date;
 664                 queue.cb_data = &author_date;
 665                 break;
 666         }
 667
 668         /* Mark them and clear the indegree */
 669         for (next = orig; next; next = next->next) {
 670                 struct commit *commit = next->item;
 671                 *(indegree_slab_at(&indegree, commit)) = 1;
 672                 /* also record the author dates, if needed */
 673                 if (sort_order == REV_SORT_BY_AUTHOR_DATE)
 674                         record_author_date(&author_date, commit);
 675         }
 676
 677         /* update the indegree */
 678         for (next = orig; next; next = next->next) {
 679                 struct commit_list *parents = next->item->parents;
 680                 while (parents) {
 681                         struct commit *parent = parents->item;
 682                         int *pi = indegree_slab_at(&indegree, parent);
 683
 684                         if (*pi)
 685                                 (*pi)++;
 686                         parents = parents->next;
 687                 }
 688         }
 689
 690         /*
 691          * find the tips
 692          *
 693          * tips are nodes not reachable from any other node in the list
 694          *
 695          * the tips serve as a starting set for the work queue.
 696          */
 697         for (next = orig; next; next = next->next) {
 698                 struct commit *commit = next->item;
 699
 700                 if (*(indegree_slab_at(&indegree, commit)) == 1)
 701                         prio_queue_put(&queue, commit);
 702         }
 703
 704         /*
 705          * This is unfortunate; the initial tips need to be shown
 706          * in the order given from the revision traversal machinery.
 707          */
 708         if (sort_order == REV_SORT_IN_GRAPH_ORDER)
 709                 prio_queue_reverse(&queue);
 710
 711         /* We no longer need the commit list */
 712         free_commit_list(orig);
 713
 714         pptr = list;
 715         *list = NULL;
 716         while ((commit = prio_queue_get(&queue)) != NULL) {
 717                 struct commit_list *parents;
 718
 719                 for (parents = commit->parents; parents ; parents = parents->next) {
 720                         struct commit *parent = parents->item;
 721                         int *pi = indegree_slab_at(&indegree, parent);
 722
 723                         if (!*pi)
 724                                 continue;
 725
 726                         /*
 727                          * parents are only enqueued for emission
 728                          * when all their children have been emitted thereby
 729                          * guaranteeing topological order.
 730                          */
 731                         if (--(*pi) == 1)
 732                                 prio_queue_put(&queue, parent);
 733                 }
 734                 /*
 735                  * all children of commit have already been
 736                  * emitted. we can emit it now.
 737                  */
 738                 *(indegree_slab_at(&indegree, commit)) = 0;
 739
 740                 pptr = &commit_list_insert(commit, pptr)->next;
 741         }
 742
 743         clear_indegree_slab(&indegree);
 744         clear_prio_queue(&queue);
 745         if (sort_order == REV_SORT_BY_AUTHOR_DATE)
 746                 clear_author_date_slab(&author_date);
 747 }
 748
 749 /* merge-base stuff */
 750
 751 /* Remember to update object flag allocation in object.h */
 752 #define PARENT1         (1u<<16)
 753 #define PARENT2         (1u<<17)
 754 #define STALE           (1u<<18)
 755 #define RESULT          (1u<<19)
 756
 757 static const unsigned all_flags = (PARENT1 | PARENT2 | STALE | RESULT);
 758
 759 static int queue_has_nonstale(struct prio_queue *queue)
 760 {
 761         int i;
 762         for (i = 0; i < queue->nr; i++) {
 763                 struct commit *commit = queue->array[i].data;
 764                 if (!(commit->object.flags & STALE))
 765                         return 1;
 766         }
 767         return 0;
 768 }
 769
 770 /* all input commits in one and twos[] must have been parsed! */
 771 static struct commit_list *paint_down_to_common(struct commit *one, int n, struct commit **twos)
 772 {
 773         struct prio_queue queue = { compare_commits_by_commit_date };
 774         struct commit_list *result = NULL;
 775         int i;
 776
 777         one->object.flags |= PARENT1;
 778         if (!n) {
 779                 commit_list_append(one, &result);
 780                 return result;
 781         }
 782         prio_queue_put(&queue, one);
 783
 784         for (i = 0; i < n; i++) {
 785                 twos[i]->object.flags |= PARENT2;
 786                 prio_queue_put(&queue, twos[i]);
 787         }
 788
 789         while (queue_has_nonstale(&queue)) {
 790                 struct commit *commit = prio_queue_get(&queue);
 791                 struct commit_list *parents;
 792                 int flags;
 793
 794                 flags = commit->object.flags & (PARENT1 | PARENT2 | STALE);
 795                 if (flags == (PARENT1 | PARENT2)) {
 796                         if (!(commit->object.flags & RESULT)) {
 797                                 commit->object.flags |= RESULT;
 798                                 commit_list_insert_by_date(commit, &result);
 799                         }
 800                         /* Mark parents of a found merge stale */
 801                         flags |= STALE;
 802                 }
 803                 parents = commit->parents;
 804                 while (parents) {
 805                         struct commit *p = parents->item;
 806                         parents = parents->next;
 807                         if ((p->object.flags & flags) == flags)
 808                                 continue;
 809                         if (parse_commit(p))
 810                                 return NULL;
 811                         p->object.flags |= flags;
 812                         prio_queue_put(&queue, p);
 813                 }
 814         }
 815
 816         clear_prio_queue(&queue);
 817         return result;
 818 }
 819
 820 static struct commit_list *merge_bases_many(struct commit *one, int n, struct commit **twos)
 821 {
 822         struct commit_list *list = NULL;
 823         struct commit_list *result = NULL;
 824         int i;
 825
 826         for (i = 0; i < n; i++) {
 827                 if (one == twos[i])
 828                         /*
 829                          * We do not mark this even with RESULT so we do not
 830                          * have to clean it up.
 831                          */
 832                         return commit_list_insert(one, &result);
 833         }
 834
 835         if (parse_commit(one))
 836                 return NULL;
 837         for (i = 0; i < n; i++) {
 838                 if (parse_commit(twos[i]))
 839                         return NULL;
 840         }
 841
 842         list = paint_down_to_common(one, n, twos);
 843
 844         while (list) {
 845                 struct commit *commit = pop_commit(&list);
 846                 if (!(commit->object.flags & STALE))
 847                         commit_list_insert_by_date(commit, &result);
 848         }
 849         return result;
 850 }
 851
 852 struct commit_list *get_octopus_merge_bases(struct commit_list *in)
 853 {
 854         struct commit_list *i, *j, *k, *ret = NULL;
 855
 856         if (!in)
 857                 return ret;
 858
 859         commit_list_insert(in->item, &ret);
 860
 861         for (i = in->next; i; i = i->next) {
 862                 struct commit_list *new_commits = NULL, *end = NULL;
 863
 864                 for (j = ret; j; j = j->next) {
 865                         struct commit_list *bases;
 866                         bases = get_merge_bases(i->item, j->item);
 867                         if (!new_commits)
 868                                 new_commits = bases;
 869                         else
 870                                 end->next = bases;
 871                         for (k = bases; k; k = k->next)
 872                                 end = k;
 873                 }
 874                 ret = new_commits;
 875         }
 876         return ret;
 877 }
 878
 879 static int remove_redundant(struct commit **array, int cnt)
 880 {
 881         /*
 882          * Some commit in the array may be an ancestor of
 883          * another commit.  Move such commit to the end of
 884          * the array, and return the number of commits that
 885          * are independent from each other.
 886          */
 887         struct commit **work;
 888         unsigned char *redundant;
 889         int *filled_index;
 890         int i, j, filled;
 891
 892         work = xcalloc(cnt, sizeof(*work));
 893         redundant = xcalloc(cnt, 1);
 894         ALLOC_ARRAY(filled_index, cnt - 1);
 895
 896         for (i = 0; i < cnt; i++)
 897                 parse_commit(array[i]);
 898         for (i = 0; i < cnt; i++) {
 899                 struct commit_list *common;
 900
 901                 if (redundant[i])
 902                         continue;
 903                 for (j = filled = 0; j < cnt; j++) {
 904                         if (i == j || redundant[j])
 905                                 continue;
 906                         filled_index[filled] = j;
 907                         work[filled++] = array[j];
 908                 }
 909                 common = paint_down_to_common(array[i], filled, work);
 910                 if (array[i]->object.flags & PARENT2)
 911                         redundant[i] = 1;
 912                 for (j = 0; j < filled; j++)
 913                         if (work[j]->object.flags & PARENT1)
 914                                 redundant[filled_index[j]] = 1;
 915                 clear_commit_marks(array[i], all_flags);
 916                 clear_commit_marks_many(filled, work, all_flags);
 917                 free_commit_list(common);
 918         }
 919
 920         /* Now collect the result */
 921         COPY_ARRAY(work, array, cnt);
 922         for (i = filled = 0; i < cnt; i++)
 923                 if (!redundant[i])
 924                         array[filled++] = work[i];
 925         for (j = filled, i = 0; i < cnt; i++)
 926                 if (redundant[i])
 927                         array[j++] = work[i];
 928         free(work);
 929         free(redundant);
 930         free(filled_index);
 931         return filled;
 932 }
 933
 934 static struct commit_list *get_merge_bases_many_0(struct commit *one,
 935                                                   int n,
 936                                                   struct commit **twos,
 937                                                   int cleanup)
 938 {
 939         struct commit_list *list;
 940         struct commit **rslt;
 941         struct commit_list *result;
 942         int cnt, i;
 943
 944         result = merge_bases_many(one, n, twos);
 945         for (i = 0; i < n; i++) {
 946                 if (one == twos[i])
 947                         return result;
 948         }
 949         if (!result || !result->next) {
 950                 if (cleanup) {
 951                         clear_commit_marks(one, all_flags);
 952                         clear_commit_marks_many(n, twos, all_flags);
 953                 }
 954                 return result;
 955         }
 956
 957         /* There are more than one */
 958         cnt = commit_list_count(result);
 959         rslt = xcalloc(cnt, sizeof(*rslt));
 960         for (list = result, i = 0; list; list = list->next)
 961                 rslt[i++] = list->item;
 962         free_commit_list(result);
 963
 964         clear_commit_marks(one, all_flags);
 965         clear_commit_marks_many(n, twos, all_flags);
 966
 967         cnt = remove_redundant(rslt, cnt);
 968         result = NULL;
 969         for (i = 0; i < cnt; i++)
 970                 commit_list_insert_by_date(rslt[i], &result);
 971         free(rslt);
 972         return result;
 973 }
 974
 975 struct commit_list *get_merge_bases_many(struct commit *one,
 976                                          int n,
 977                                          struct commit **twos)
 978 {
 979         return get_merge_bases_many_0(one, n, twos, 1);
 980 }
 981
 982 struct commit_list *get_merge_bases_many_dirty(struct commit *one,
 983                                                int n,
 984                                                struct commit **twos)
 985 {
 986         return get_merge_bases_many_0(one, n, twos, 0);
 987 }
 988
 989 struct commit_list *get_merge_bases(struct commit *one, struct commit *two)
 990 {
 991         return get_merge_bases_many_0(one, 1, &two, 1);
 992 }
 993
 994 /*
 995  * Is "commit" a descendant of one of the elements on the "with_commit" list?
 996  */
 997 int is_descendant_of(struct commit *commit, struct commit_list *with_commit)
 998 {
 999         if (!with_commit)
1000                 return 1;
1001         while (with_commit) {
1002                 struct commit *other;
1003
1004                 other = with_commit->item;
1005                 with_commit = with_commit->next;
1006                 if (in_merge_bases(other, commit))
1007                         return 1;
1008         }
1009         return 0;
1010 }
1011
1012 /*
1013  * Is "commit" an ancestor of one of the "references"?
1014  */
1015 int in_merge_bases_many(struct commit *commit, int nr_reference, struct commit **reference)
1016 {
1017         struct commit_list *bases;
1018         int ret = 0, i;
1019
1020         if (parse_commit(commit))
1021                 return ret;
1022         for (i = 0; i < nr_reference; i++)
1023                 if (parse_commit(reference[i]))
1024                         return ret;
1025
1026         bases = paint_down_to_common(commit, nr_reference, reference);
1027         if (commit->object.flags & PARENT2)
1028                 ret = 1;
1029         clear_commit_marks(commit, all_flags);
1030         clear_commit_marks_many(nr_reference, reference, all_flags);
1031         free_commit_list(bases);
1032         return ret;
1033 }
1034
1035 /*
1036  * Is "commit" an ancestor of (i.e. reachable from) the "reference"?
1037  */
1038 int in_merge_bases(struct commit *commit, struct commit *reference)
1039 {
1040         return in_merge_bases_many(commit, 1, &reference);
1041 }
1042
1043 struct commit_list *reduce_heads(struct commit_list *heads)
1044 {
1045         struct commit_list *p;
1046         struct commit_list *result = NULL, **tail = &result;
1047         struct commit **array;
1048         int num_head, i;
1049
1050         if (!heads)
1051                 return NULL;
1052
1053         /* Uniquify */
1054         for (p = heads; p; p = p->next)
1055                 p->item->object.flags &= ~STALE;
1056         for (p = heads, num_head = 0; p; p = p->next) {
1057                 if (p->item->object.flags & STALE)
1058                         continue;
1059                 p->item->object.flags |= STALE;
1060                 num_head++;
1061         }
1062         array = xcalloc(num_head, sizeof(*array));
1063         for (p = heads, i = 0; p; p = p->next) {
1064                 if (p->item->object.flags & STALE) {
1065                         array[i++] = p->item;
1066                         p->item->object.flags &= ~STALE;
1067                 }
1068         }
1069         num_head = remove_redundant(array, num_head);
1070         for (i = 0; i < num_head; i++)
1071                 tail = &commit_list_insert(array[i], tail)->next;
1072         free(array);
1073         return result;
1074 }
1075
1076 void reduce_heads_replace(struct commit_list **heads)
1077 {
1078         struct commit_list *result = reduce_heads(*heads);
1079         free_commit_list(*heads);
1080         *heads = result;
1081 }
1082
1083 static const char gpg_sig_header[] = "gpgsig";
1084 static const int gpg_sig_header_len = sizeof(gpg_sig_header) - 1;
1085
1086 static int do_sign_commit(struct strbuf *buf, const char *keyid)
1087 {
1088         struct strbuf sig = STRBUF_INIT;
1089         int inspos, copypos;
1090         const char *eoh;
1091
1092         /* find the end of the header */
1093         eoh = strstr(buf->buf, "\n\n");
1094         if (!eoh)
1095                 inspos = buf->len;
1096         else
1097                 inspos = eoh - buf->buf + 1;
1098
1099         if (!keyid || !*keyid)
1100                 keyid = get_signing_key();
1101         if (sign_buffer(buf, &sig, keyid)) {
1102                 strbuf_release(&sig);
1103                 return -1;
1104         }
1105
1106         for (copypos = 0; sig.buf[copypos]; ) {
1107                 const char *bol = sig.buf + copypos;
1108                 const char *eol = strchrnul(bol, '\n');
1109                 int len = (eol - bol) + !!*eol;
1110
1111                 if (!copypos) {
1112                         strbuf_insert(buf, inspos, gpg_sig_header, gpg_sig_header_len);
1113                         inspos += gpg_sig_header_len;
1114                 }
1115                 strbuf_insert(buf, inspos++, " ", 1);
1116                 strbuf_insert(buf, inspos, bol, len);
1117                 inspos += len;
1118                 copypos += len;
1119         }
1120         strbuf_release(&sig);
1121         return 0;
1122 }
1123
1124 int parse_signed_commit(const struct commit *commit,
1125                         struct strbuf *payload, struct strbuf *signature)
1126 {
1127
1128         unsigned long size;
1129         const char *buffer = get_commit_buffer(commit, &size);
1130         int in_signature, saw_signature = -1;
1131         const char *line, *tail;
1132
1133         line = buffer;
1134         tail = buffer + size;
1135         in_signature = 0;
1136         saw_signature = 0;
1137         while (line < tail) {
1138                 const char *sig = NULL;
1139                 const char *next = memchr(line, '\n', tail - line);
1140
1141                 next = next ? next + 1 : tail;
1142                 if (in_signature && line[0] == ' ')
1143                         sig = line + 1;
1144                 else if (starts_with(line, gpg_sig_header) &&
1145                          line[gpg_sig_header_len] == ' ')
1146                         sig = line + gpg_sig_header_len + 1;
1147                 if (sig) {
1148                         strbuf_add(signature, sig, next - sig);
1149                         saw_signature = 1;
1150                         in_signature = 1;
1151                 } else {
1152                         if (*line == '\n')
1153                                 /* dump the whole remainder of the buffer */
1154                                 next = tail;
1155                         strbuf_add(payload, line, next - line);
1156                         in_signature = 0;
1157                 }
1158                 line = next;
1159         }
1160         unuse_commit_buffer(commit, buffer);
1161         return saw_signature;
1162 }
1163
1164 int remove_signature(struct strbuf *buf)
1165 {
1166         const char *line = buf->buf;
1167         const char *tail = buf->buf + buf->len;
1168         int in_signature = 0;
1169         const char *sig_start = NULL;
1170         const char *sig_end = NULL;
1171
1172         while (line < tail) {
1173                 const char *next = memchr(line, '\n', tail - line);
1174                 next = next ? next + 1 : tail;
1175
1176                 if (in_signature && line[0] == ' ')
1177                         sig_end = next;
1178                 else if (starts_with(line, gpg_sig_header) &&
1179                          line[gpg_sig_header_len] == ' ') {
1180                         sig_start = line;
1181                         sig_end = next;
1182                         in_signature = 1;
1183                 } else {
1184                         if (*line == '\n')
1185                                 /* dump the whole remainder of the buffer */
1186                                 next = tail;
1187                         in_signature = 0;
1188                 }
1189                 line = next;
1190         }
1191
1192         if (sig_start)
1193                 strbuf_remove(buf, sig_start - buf->buf, sig_end - sig_start);
1194
1195         return sig_start != NULL;
1196 }
1197
1198 static void handle_signed_tag(struct commit *parent, struct commit_extra_header ***tail)
1199 {
1200         struct merge_remote_desc *desc;
1201         struct commit_extra_header *mergetag;
1202         char *buf;
1203         unsigned long size, len;
1204         enum object_type type;
1205
1206         desc = merge_remote_util(parent);
1207         if (!desc || !desc->obj)
1208                 return;
1209         buf = read_sha1_file(desc->obj->oid.hash, &type, &size);
1210         if (!buf || type != OBJ_TAG)
1211                 goto free_return;
1212         len = parse_signature(buf, size);
1213         if (size == len)
1214                 goto free_return;
1215         /*
1216          * We could verify this signature and either omit the tag when
1217          * it does not validate, but the integrator may not have the
1218          * public key of the signer of the tag he is merging, while a
1219          * later auditor may have it while auditing, so let's not run
1220          * verify-signed-buffer here for now...
1221          *
1222          * if (verify_signed_buffer(buf, len, buf + len, size - len, ...))
1223          *      warn("warning: signed tag unverified.");
1224          */
1225         mergetag = xcalloc(1, sizeof(*mergetag));
1226         mergetag->key = xstrdup("mergetag");
1227         mergetag->value = buf;
1228         mergetag->len = size;
1229
1230         **tail = mergetag;
1231         *tail = &mergetag->next;
1232         return;
1233
1234 free_return:
1235         free(buf);
1236 }
1237
1238 int check_commit_signature(const struct commit *commit, struct signature_check *sigc)
1239 {
1240         struct strbuf payload = STRBUF_INIT;
1241         struct strbuf signature = STRBUF_INIT;
1242         int ret = 1;
1243
1244         sigc->result = 'N';
1245
1246         if (parse_signed_commit(commit, &payload, &signature) <= 0)
1247                 goto out;
1248         ret = check_signature(payload.buf, payload.len, signature.buf,
1249                 signature.len, sigc);
1250
1251  out:
1252         strbuf_release(&payload);
1253         strbuf_release(&signature);
1254
1255         return ret;
1256 }
1257
1258
1259
1260 void append_merge_tag_headers(struct commit_list *parents,
1261                               struct commit_extra_header ***tail)
1262 {
1263         while (parents) {
1264                 struct commit *parent = parents->item;
1265                 handle_signed_tag(parent, tail);
1266                 parents = parents->next;
1267         }
1268 }
1269
1270 static void add_extra_header(struct strbuf *buffer,
1271                              struct commit_extra_header *extra)
1272 {
1273         strbuf_addstr(buffer, extra->key);
1274         if (extra->len)
1275                 strbuf_add_lines(buffer, " ", extra->value, extra->len);
1276         else
1277                 strbuf_addch(buffer, '\n');
1278 }
1279
1280 struct commit_extra_header *read_commit_extra_headers(struct commit *commit,
1281                                                       const char **exclude)
1282 {
1283         struct commit_extra_header *extra = NULL;
1284         unsigned long size;
1285         const char *buffer = get_commit_buffer(commit, &size);
1286         extra = read_commit_extra_header_lines(buffer, size, exclude);
1287         unuse_commit_buffer(commit, buffer);
1288         return extra;
1289 }
1290
1291 void for_each_mergetag(each_mergetag_fn fn, struct commit *commit, void *data)
1292 {
1293         struct commit_extra_header *extra, *to_free;
1294
1295         to_free = read_commit_extra_headers(commit, NULL);
1296         for (extra = to_free; extra; extra = extra->next) {
1297                 if (strcmp(extra->key, "mergetag"))
1298                         continue; /* not a merge tag */
1299                 fn(commit, extra, data);
1300         }
1301         free_commit_extra_headers(to_free);
1302 }
1303
1304 static inline int standard_header_field(const char *field, size_t len)
1305 {
1306         return ((len == 4 && !memcmp(field, "tree", 4)) ||
1307                 (len == 6 && !memcmp(field, "parent", 6)) ||
1308                 (len == 6 && !memcmp(field, "author", 6)) ||
1309                 (len == 9 && !memcmp(field, "committer", 9)) ||
1310                 (len == 8 && !memcmp(field, "encoding", 8)));
1311 }
1312
1313 static int excluded_header_field(const char *field, size_t len, const char **exclude)
1314 {
1315         if (!exclude)
1316                 return 0;
1317
1318         while (*exclude) {
1319                 size_t xlen = strlen(*exclude);
1320                 if (len == xlen && !memcmp(field, *exclude, xlen))
1321                         return 1;
1322                 exclude++;
1323         }
1324         return 0;
1325 }
1326
1327 static struct commit_extra_header *read_commit_extra_header_lines(
1328         const char *buffer, size_t size,
1329         const char **exclude)
1330 {
1331         struct commit_extra_header *extra = NULL, **tail = &extra, *it = NULL;
1332         const char *line, *next, *eof, *eob;
1333         struct strbuf buf = STRBUF_INIT;
1334
1335         for (line = buffer, eob = line + size;
1336              line < eob && *line != '\n';
1337              line = next) {
1338                 next = memchr(line, '\n', eob - line);
1339                 next = next ? next + 1 : eob;
1340                 if (*line == ' ') {
1341                         /* continuation */
1342                         if (it)
1343                                 strbuf_add(&buf, line + 1, next - (line + 1));
1344                         continue;
1345                 }
1346                 if (it)
1347                         it->value = strbuf_detach(&buf, &it->len);
1348                 strbuf_reset(&buf);
1349                 it = NULL;
1350
1351                 eof = memchr(line, ' ', next - line);
1352                 if (!eof)
1353                         eof = next;
1354                 else if (standard_header_field(line, eof - line) ||
1355                          excluded_header_field(line, eof - line, exclude))
1356                         continue;
1357
1358                 it = xcalloc(1, sizeof(*it));
1359                 it->key = xmemdupz(line, eof-line);
1360                 *tail = it;
1361                 tail = &it->next;
1362                 if (eof + 1 < next)
1363                         strbuf_add(&buf, eof + 1, next - (eof + 1));
1364         }
1365         if (it)
1366                 it->value = strbuf_detach(&buf, &it->len);
1367         return extra;
1368 }
1369
1370 void free_commit_extra_headers(struct commit_extra_header *extra)
1371 {
1372         while (extra) {
1373                 struct commit_extra_header *next = extra->next;
1374                 free(extra->key);
1375                 free(extra->value);
1376                 free(extra);
1377                 extra = next;
1378         }
1379 }
1380
1381 int commit_tree(const char *msg, size_t msg_len, const struct object_id *tree,
1382                 struct commit_list *parents, struct object_id *ret,
1383                 const char *author, const char *sign_commit)
1384 {
1385         struct commit_extra_header *extra = NULL, **tail = &extra;
1386         int result;
1387
1388         append_merge_tag_headers(parents, &tail);
1389         result = commit_tree_extended(msg, msg_len, tree, parents, ret,
1390                                       author, sign_commit, extra);
1391         free_commit_extra_headers(extra);
1392         return result;
1393 }
1394
1395 static int find_invalid_utf8(const char *buf, int len)
1396 {
1397         int offset = 0;
1398         static const unsigned int max_codepoint[] = {
1399                 0x7f, 0x7ff, 0xffff, 0x10ffff
1400         };
1401
1402         while (len) {
1403                 unsigned char c = *buf++;
1404                 int bytes, bad_offset;
1405                 unsigned int codepoint;
1406                 unsigned int min_val, max_val;
1407
1408                 len--;
1409                 offset++;
1410
1411                 /* Simple US-ASCII? No worries. */
1412                 if (c < 0x80)
1413                         continue;
1414
1415                 bad_offset = offset-1;
1416
1417                 /*
1418                  * Count how many more high bits set: that's how
1419                  * many more bytes this sequence should have.
1420                  */
1421                 bytes = 0;
1422                 while (c & 0x40) {
1423                         c <<= 1;
1424                         bytes++;
1425                 }
1426
1427                 /*
1428                  * Must be between 1 and 3 more bytes.  Longer sequences result in
1429                  * codepoints beyond U+10FFFF, which are guaranteed never to exist.
1430                  */
1431                 if (bytes < 1 || 3 < bytes)
1432                         return bad_offset;
1433
1434                 /* Do we *have* that many bytes? */
1435                 if (len < bytes)
1436                         return bad_offset;
1437
1438                 /*
1439                  * Place the encoded bits at the bottom of the value and compute the
1440                  * valid range.
1441                  */
1442                 codepoint = (c & 0x7f) >> bytes;
1443                 min_val = max_codepoint[bytes-1] + 1;
1444                 max_val = max_codepoint[bytes];
1445
1446                 offset += bytes;
1447                 len -= bytes;
1448
1449                 /* And verify that they are good continuation bytes */
1450                 do {
1451                         codepoint <<= 6;
1452                         codepoint |= *buf & 0x3f;
1453                         if ((*buf++ & 0xc0) != 0x80)
1454                                 return bad_offset;
1455                 } while (--bytes);
1456
1457                 /* Reject codepoints that are out of range for the sequence length. */
1458                 if (codepoint < min_val || codepoint > max_val)
1459                         return bad_offset;
1460                 /* Surrogates are only for UTF-16 and cannot be encoded in UTF-8. */
1461                 if ((codepoint & 0x1ff800) == 0xd800)
1462                         return bad_offset;
1463                 /* U+xxFFFE and U+xxFFFF are guaranteed non-characters. */
1464                 if ((codepoint & 0xfffe) == 0xfffe)
1465                         return bad_offset;
1466                 /* So are anything in the range U+FDD0..U+FDEF. */
1467                 if (codepoint >= 0xfdd0 && codepoint <= 0xfdef)
1468                         return bad_offset;
1469         }
1470         return -1;
1471 }
1472
1473 /*
1474  * This verifies that the buffer is in proper utf8 format.
1475  *
1476  * If it isn't, it assumes any non-utf8 characters are Latin1,
1477  * and does the conversion.
1478  */
1479 static int verify_utf8(struct strbuf *buf)
1480 {
1481         int ok = 1;
1482         long pos = 0;
1483
1484         for (;;) {
1485                 int bad;
1486                 unsigned char c;
1487                 unsigned char replace[2];
1488
1489                 bad = find_invalid_utf8(buf->buf + pos, buf->len - pos);
1490                 if (bad < 0)
1491                         return ok;
1492                 pos += bad;
1493                 ok = 0;
1494                 c = buf->buf[pos];
1495                 strbuf_remove(buf, pos, 1);
1496
1497                 /* We know 'c' must be in the range 128-255 */
1498                 replace[0] = 0xc0 + (c >> 6);
1499                 replace[1] = 0x80 + (c & 0x3f);
1500                 strbuf_insert(buf, pos, replace, 2);
1501                 pos += 2;
1502         }
1503 }
1504
1505 static const char commit_utf8_warn[] =
1506 N_("Warning: commit message did not conform to UTF-8.\n"
1507    "You may want to amend it after fixing the message, or set the config\n"
1508    "variable i18n.commitencoding to the encoding your project uses.\n");
1509
1510 int commit_tree_extended(const char *msg, size_t msg_len,
1511                          const struct object_id *tree,
1512                          struct commit_list *parents, struct object_id *ret,
1513                          const char *author, const char *sign_commit,
1514                          struct commit_extra_header *extra)
1515 {
1516         int result;
1517         int encoding_is_utf8;
1518         struct strbuf buffer;
1519
1520         assert_sha1_type(tree->hash, OBJ_TREE);
1521
1522         if (memchr(msg, '\0', msg_len))
1523                 return error("a NUL byte in commit log message not allowed.");
1524
1525         /* Not having i18n.commitencoding is the same as having utf-8 */
1526         encoding_is_utf8 = is_encoding_utf8(git_commit_encoding);
1527
1528         strbuf_init(&buffer, 8192); /* should avoid reallocs for the headers */
1529         strbuf_addf(&buffer, "tree %s\n", oid_to_hex(tree));
1530
1531         /*
1532          * NOTE! This ordering means that the same exact tree merged with a
1533          * different order of parents will be a _different_ changeset even
1534          * if everything else stays the same.
1535          */
1536         while (parents) {
1537                 struct commit *parent = pop_commit(&parents);
1538                 strbuf_addf(&buffer, "parent %s\n",
1539                             oid_to_hex(&parent->object.oid));
1540         }
1541
1542         /* Person/date information */
1543         if (!author)
1544                 author = git_author_info(IDENT_STRICT);
1545         strbuf_addf(&buffer, "author %s\n", author);
1546         strbuf_addf(&buffer, "committer %s\n", git_committer_info(IDENT_STRICT));
1547         if (!encoding_is_utf8)
1548                 strbuf_addf(&buffer, "encoding %s\n", git_commit_encoding);
1549
1550         while (extra) {
1551                 add_extra_header(&buffer, extra);
1552                 extra = extra->next;
1553         }
1554         strbuf_addch(&buffer, '\n');
1555
1556         /* And add the comment */
1557         strbuf_add(&buffer, msg, msg_len);
1558
1559         /* And check the encoding */
1560         if (encoding_is_utf8 && !verify_utf8(&buffer))
1561                 fprintf(stderr, _(commit_utf8_warn));
1562
1563         if (sign_commit && do_sign_commit(&buffer, sign_commit)) {
1564                 result = -1;
1565                 goto out;
1566         }
1567
1568         result = write_object_file(buffer.buf, buffer.len, commit_type, ret);
1569 out:
1570         strbuf_release(&buffer);
1571         return result;
1572 }
1573
1574 void set_merge_remote_desc(struct commit *commit,
1575                            const char *name, struct object *obj)
1576 {
1577         struct merge_remote_desc *desc;
1578         FLEX_ALLOC_STR(desc, name, name);
1579         desc->obj = obj;
1580         commit->util = desc;
1581 }
1582
1583 struct commit *get_merge_parent(const char *name)
1584 {
1585         struct object *obj;
1586         struct commit *commit;
1587         struct object_id oid;
1588         if (get_oid(name, &oid))
1589                 return NULL;
1590         obj = parse_object(&oid);
1591         commit = (struct commit *)peel_to_type(name, 0, obj, OBJ_COMMIT);
1592         if (commit && !commit->util)
1593                 set_merge_remote_desc(commit, name, obj);
1594         return commit;
1595 }
1596
1597 /*
1598  * Append a commit to the end of the commit_list.
1599  *
1600  * next starts by pointing to the variable that holds the head of an
1601  * empty commit_list, and is updated to point to the "next" field of
1602  * the last item on the list as new commits are appended.
1603  *
1604  * Usage example:
1605  *
1606  *     struct commit_list *list;
1607  *     struct commit_list **next = &list;
1608  *
1609  *     next = commit_list_append(c1, next);
1610  *     next = commit_list_append(c2, next);
1611  *     assert(commit_list_count(list) == 2);
1612  *     return list;
1613  */
1614 struct commit_list **commit_list_append(struct commit *commit,
1615                                         struct commit_list **next)
1616 {
1617         struct commit_list *new_commit = xmalloc(sizeof(struct commit_list));
1618         new_commit->item = commit;
1619         *next = new_commit;
1620         new_commit->next = NULL;
1621         return &new_commit->next;
1622 }
1623
1624 const char *find_commit_header(const char *msg, const char *key, size_t *out_len)
1625 {
1626         int key_len = strlen(key);
1627         const char *line = msg;
1628
1629         while (line) {
1630                 const char *eol = strchrnul(line, '\n');
1631
1632                 if (line == eol)
1633                         return NULL;
1634
1635                 if (eol - line > key_len &&
1636                     !strncmp(line, key, key_len) &&
1637                     line[key_len] == ' ') {
1638                         *out_len = eol - line - key_len - 1;
1639                         return line + key_len + 1;
1640                 }
1641                 line = *eol ? eol + 1 : NULL;
1642         }
1643         return NULL;
1644 }
1645
1646 /*
1647  * Inspect the given string and determine the true "end" of the log message, in
1648  * order to find where to put a new Signed-off-by: line.  Ignored are
1649  * trailing comment lines and blank lines.  To support "git commit -s
1650  * --amend" on an existing commit, we also ignore "Conflicts:".  To
1651  * support "git commit -v", we truncate at cut lines.
1652  *
1653  * Returns the number of bytes from the tail to ignore, to be fed as
1654  * the second parameter to append_signoff().
1655  */
1656 int ignore_non_trailer(const char *buf, size_t len)
1657 {
1658         int boc = 0;
1659         int bol = 0;
1660         int in_old_conflicts_block = 0;
1661         size_t cutoff = wt_status_locate_end(buf, len);
1662
1663         while (bol < cutoff) {
1664                 const char *next_line = memchr(buf + bol, '\n', len - bol);
1665
1666                 if (!next_line)
1667                         next_line = buf + len;
1668                 else
1669                         next_line++;
1670
1671                 if (buf[bol] == comment_line_char || buf[bol] == '\n') {
1672                         /* is this the first of the run of comments? */
1673                         if (!boc)
1674                                 boc = bol;
1675                         /* otherwise, it is just continuing */
1676                 } else if (starts_with(buf + bol, "Conflicts:\n")) {
1677                         in_old_conflicts_block = 1;
1678                         if (!boc)
1679                                 boc = bol;
1680                 } else if (in_old_conflicts_block && buf[bol] == '\t') {
1681                         ; /* a pathname in the conflicts block */
1682                 } else if (boc) {
1683                         /* the previous was not trailing comment */
1684                         boc = 0;
1685                         in_old_conflicts_block = 0;
1686                 }
1687                 bol = next_line - buf;
1688         }
1689         return boc ? len - boc : len - cutoff;
1690 }