commit.c

   1 #include "cache.h"
   2 #include "tag.h"
   3 #include "commit.h"
   4 #include "repository.h"
   5 #include "object-store.h"
   6 #include "pkt-line.h"
   7 #include "utf8.h"
   8 #include "diff.h"
   9 #include "revision.h"
  10 #include "notes.h"
  11 #include "alloc.h"
  12 #include "gpg-interface.h"
  13 #include "mergesort.h"
  14 #include "commit-slab.h"
  15 #include "prio-queue.h"
  16 #include "sha1-lookup.h"
  17 #include "wt-status.h"
  18
  19 static struct commit_extra_header *read_commit_extra_header_lines(const char *buf, size_t len, const char **);
  20
  21 int save_commit_buffer = 1;
  22
  23 const char *commit_type = "commit";
  24
  25 struct commit *lookup_commit_reference_gently(const struct object_id *oid,
  26                                               int quiet)
  27 {
  28         struct object *obj = deref_tag(parse_object(oid), NULL, 0);
  29
  30         if (!obj)
  31                 return NULL;
  32         return object_as_type(obj, OBJ_COMMIT, quiet);
  33 }
  34
  35 struct commit *lookup_commit_reference(const struct object_id *oid)
  36 {
  37         return lookup_commit_reference_gently(oid, 0);
  38 }
  39
  40 struct commit *lookup_commit_or_die(const struct object_id *oid, const char *ref_name)
  41 {
  42         struct commit *c = lookup_commit_reference(oid);
  43         if (!c)
  44                 die(_("could not parse %s"), ref_name);
  45         if (oidcmp(oid, &c->object.oid)) {
  46                 warning(_("%s %s is not a commit!"),
  47                         ref_name, oid_to_hex(oid));
  48         }
  49         return c;
  50 }
  51
  52 struct commit *lookup_commit(const struct object_id *oid)
  53 {
  54         struct object *obj = lookup_object(oid->hash);
  55         if (!obj)
  56                 return create_object(the_repository, oid->hash,
  57                                      alloc_commit_node(the_repository));
  58         return object_as_type(obj, OBJ_COMMIT, 0);
  59 }
  60
  61 struct commit *lookup_commit_reference_by_name(const char *name)
  62 {
  63         struct object_id oid;
  64         struct commit *commit;
  65
  66         if (get_oid_committish(name, &oid))
  67                 return NULL;
  68         commit = lookup_commit_reference(&oid);
  69         if (parse_commit(commit))
  70                 return NULL;
  71         return commit;
  72 }
  73
  74 static timestamp_t parse_commit_date(const char *buf, const char *tail)
  75 {
  76         const char *dateptr;
  77
  78         if (buf + 6 >= tail)
  79                 return 0;
  80         if (memcmp(buf, "author", 6))
  81                 return 0;
  82         while (buf < tail && *buf++ != '\n')
  83                 /* nada */;
  84         if (buf + 9 >= tail)
  85                 return 0;
  86         if (memcmp(buf, "committer", 9))
  87                 return 0;
  88         while (buf < tail && *buf++ != '>')
  89                 /* nada */;
  90         if (buf >= tail)
  91                 return 0;
  92         dateptr = buf;
  93         while (buf < tail && *buf++ != '\n')
  94                 /* nada */;
  95         if (buf >= tail)
  96                 return 0;
  97         /* dateptr < buf && buf[-1] == '\n', so parsing will stop at buf-1 */
  98         return parse_timestamp(dateptr, NULL, 10);
  99 }
 100
 101 static const unsigned char *commit_graft_sha1_access(size_t index, void *table)
 102 {
 103         struct commit_graft **commit_graft_table = table;
 104         return commit_graft_table[index]->oid.hash;
 105 }
 106
 107 #define commit_graft_pos(r, s) commit_graft_pos_##r(s)
 108 static int commit_graft_pos_the_repository(const unsigned char *sha1)
 109 {
 110         return sha1_pos(sha1, the_repository->parsed_objects->grafts,
 111                         the_repository->parsed_objects->grafts_nr,
 112                         commit_graft_sha1_access);
 113 }
 114
 115 int register_commit_graft_the_repository(struct commit_graft *graft, int ignore_dups)
 116 {
 117         int pos = commit_graft_pos(the_repository, graft->oid.hash);
 118
 119         if (0 <= pos) {
 120                 if (ignore_dups)
 121                         free(graft);
 122                 else {
 123                         free(the_repository->parsed_objects->grafts[pos]);
 124                         the_repository->parsed_objects->grafts[pos] = graft;
 125                 }
 126                 return 1;
 127         }
 128         pos = -pos - 1;
 129         ALLOC_GROW(the_repository->parsed_objects->grafts,
 130                    the_repository->parsed_objects->grafts_nr + 1,
 131                    the_repository->parsed_objects->grafts_alloc);
 132         the_repository->parsed_objects->grafts_nr++;
 133         if (pos < the_repository->parsed_objects->grafts_nr)
 134                 memmove(the_repository->parsed_objects->grafts + pos + 1,
 135                         the_repository->parsed_objects->grafts + pos,
 136                         (the_repository->parsed_objects->grafts_nr - pos - 1) *
 137                         sizeof(*the_repository->parsed_objects->grafts));
 138         the_repository->parsed_objects->grafts[pos] = graft;
 139         return 0;
 140 }
 141
 142 struct commit_graft *read_graft_line(struct strbuf *line)
 143 {
 144         /* The format is just "Commit Parent1 Parent2 ...\n" */
 145         int i, phase;
 146         const char *tail = NULL;
 147         struct commit_graft *graft = NULL;
 148         struct object_id dummy_oid, *oid;
 149
 150         strbuf_rtrim(line);
 151         if (!line->len || line->buf[0] == '#')
 152                 return NULL;
 153         /*
 154          * phase 0 verifies line, counts hashes in line and allocates graft
 155          * phase 1 fills graft
 156          */
 157         for (phase = 0; phase < 2; phase++) {
 158                 oid = graft ? &graft->oid : &dummy_oid;
 159                 if (parse_oid_hex(line->buf, oid, &tail))
 160                         goto bad_graft_data;
 161                 for (i = 0; *tail != '\0'; i++) {
 162                         oid = graft ? &graft->parent[i] : &dummy_oid;
 163                         if (!isspace(*tail++) || parse_oid_hex(tail, oid, &tail))
 164                                 goto bad_graft_data;
 165                 }
 166                 if (!graft) {
 167                         graft = xmalloc(st_add(sizeof(*graft),
 168                                                st_mult(sizeof(struct object_id), i)));
 169                         graft->nr_parent = i;
 170                 }
 171         }
 172         return graft;
 173
 174 bad_graft_data:
 175         error("bad graft data: %s", line->buf);
 176         assert(!graft);
 177         return NULL;
 178 }
 179
 180 static int read_graft_file(const char *graft_file)
 181 {
 182         FILE *fp = fopen_or_warn(graft_file, "r");
 183         struct strbuf buf = STRBUF_INIT;
 184         if (!fp)
 185                 return -1;
 186         while (!strbuf_getwholeline(&buf, fp, '\n')) {
 187                 /* The format is just "Commit Parent1 Parent2 ...\n" */
 188                 struct commit_graft *graft = read_graft_line(&buf);
 189                 if (!graft)
 190                         continue;
 191                 if (register_commit_graft(the_repository, graft, 1))
 192                         error("duplicate graft data: %s", buf.buf);
 193         }
 194         fclose(fp);
 195         strbuf_release(&buf);
 196         return 0;
 197 }
 198
 199 static void prepare_commit_graft(void)
 200 {
 201         static int commit_graft_prepared;
 202         char *graft_file;
 203
 204         if (commit_graft_prepared)
 205                 return;
 206         graft_file = get_graft_file();
 207         read_graft_file(graft_file);
 208         /* make sure shallows are read */
 209         is_repository_shallow();
 210         commit_graft_prepared = 1;
 211 }
 212
 213 struct commit_graft *lookup_commit_graft(const struct object_id *oid)
 214 {
 215         int pos;
 216         prepare_commit_graft();
 217         pos = commit_graft_pos(the_repository, oid->hash);
 218         if (pos < 0)
 219                 return NULL;
 220         return the_repository->parsed_objects->grafts[pos];
 221 }
 222
 223 int for_each_commit_graft(each_commit_graft_fn fn, void *cb_data)
 224 {
 225         int i, ret;
 226         for (i = ret = 0; i < the_repository->parsed_objects->grafts_nr && !ret; i++)
 227                 ret = fn(the_repository->parsed_objects->grafts[i], cb_data);
 228         return ret;
 229 }
 230
 231 int unregister_shallow(const struct object_id *oid)
 232 {
 233         int pos = commit_graft_pos(the_repository, oid->hash);
 234         if (pos < 0)
 235                 return -1;
 236         if (pos + 1 < the_repository->parsed_objects->grafts_nr)
 237                 MOVE_ARRAY(the_repository->parsed_objects->grafts + pos,
 238                            the_repository->parsed_objects->grafts + pos + 1,
 239                            the_repository->parsed_objects->grafts_nr - pos - 1);
 240         the_repository->parsed_objects->grafts_nr--;
 241         return 0;
 242 }
 243
 244 struct commit_buffer {
 245         void *buffer;
 246         unsigned long size;
 247 };
 248 define_commit_slab(buffer_slab, struct commit_buffer);
 249 static struct buffer_slab buffer_slab = COMMIT_SLAB_INIT(1, buffer_slab);
 250
 251 void set_commit_buffer(struct commit *commit, void *buffer, unsigned long size)
 252 {
 253         struct commit_buffer *v = buffer_slab_at(&buffer_slab, commit);
 254         v->buffer = buffer;
 255         v->size = size;
 256 }
 257
 258 const void *get_cached_commit_buffer(const struct commit *commit, unsigned long *sizep)
 259 {
 260         struct commit_buffer *v = buffer_slab_peek(&buffer_slab, commit);
 261         if (!v) {
 262                 if (sizep)
 263                         *sizep = 0;
 264                 return NULL;
 265         }
 266         if (sizep)
 267                 *sizep = v->size;
 268         return v->buffer;
 269 }
 270
 271 const void *get_commit_buffer(const struct commit *commit, unsigned long *sizep)
 272 {
 273         const void *ret = get_cached_commit_buffer(commit, sizep);
 274         if (!ret) {
 275                 enum object_type type;
 276                 unsigned long size;
 277                 ret = read_object_file(&commit->object.oid, &type, &size);
 278                 if (!ret)
 279                         die("cannot read commit object %s",
 280                             oid_to_hex(&commit->object.oid));
 281                 if (type != OBJ_COMMIT)
 282                         die("expected commit for %s, got %s",
 283                             oid_to_hex(&commit->object.oid), type_name(type));
 284                 if (sizep)
 285                         *sizep = size;
 286         }
 287         return ret;
 288 }
 289
 290 void unuse_commit_buffer(const struct commit *commit, const void *buffer)
 291 {
 292         struct commit_buffer *v = buffer_slab_peek(&buffer_slab, commit);
 293         if (!(v && v->buffer == buffer))
 294                 free((void *)buffer);
 295 }
 296
 297 void free_commit_buffer(struct commit *commit)
 298 {
 299         struct commit_buffer *v = buffer_slab_peek(&buffer_slab, commit);
 300         if (v) {
 301                 FREE_AND_NULL(v->buffer);
 302                 v->size = 0;
 303         }
 304 }
 305
 306 void release_commit_memory(struct commit *c)
 307 {
 308         c->tree = NULL;
 309         c->index = 0;
 310         free_commit_buffer(c);
 311         free_commit_list(c->parents);
 312         /* TODO: what about commit->util? */
 313
 314         c->object.parsed = 0;
 315 }
 316
 317 const void *detach_commit_buffer(struct commit *commit, unsigned long *sizep)
 318 {
 319         struct commit_buffer *v = buffer_slab_peek(&buffer_slab, commit);
 320         void *ret;
 321
 322         if (!v) {
 323                 if (sizep)
 324                         *sizep = 0;
 325                 return NULL;
 326         }
 327         ret = v->buffer;
 328         if (sizep)
 329                 *sizep = v->size;
 330
 331         v->buffer = NULL;
 332         v->size = 0;
 333         return ret;
 334 }
 335
 336 int parse_commit_buffer(struct commit *item, const void *buffer, unsigned long size)
 337 {
 338         const char *tail = buffer;
 339         const char *bufptr = buffer;
 340         struct object_id parent;
 341         struct commit_list **pptr;
 342         struct commit_graft *graft;
 343         const int tree_entry_len = GIT_SHA1_HEXSZ + 5;
 344         const int parent_entry_len = GIT_SHA1_HEXSZ + 7;
 345
 346         if (item->object.parsed)
 347                 return 0;
 348         item->object.parsed = 1;
 349         tail += size;
 350         if (tail <= bufptr + tree_entry_len + 1 || memcmp(bufptr, "tree ", 5) ||
 351                         bufptr[tree_entry_len] != '\n')
 352                 return error("bogus commit object %s", oid_to_hex(&item->object.oid));
 353         if (get_sha1_hex(bufptr + 5, parent.hash) < 0)
 354                 return error("bad tree pointer in commit %s",
 355                              oid_to_hex(&item->object.oid));
 356         item->tree = lookup_tree(&parent);
 357         bufptr += tree_entry_len + 1; /* "tree " + "hex sha1" + "\n" */
 358         pptr = &item->parents;
 359
 360         graft = lookup_commit_graft(&item->object.oid);
 361         while (bufptr + parent_entry_len < tail && !memcmp(bufptr, "parent ", 7)) {
 362                 struct commit *new_parent;
 363
 364                 if (tail <= bufptr + parent_entry_len + 1 ||
 365                     get_sha1_hex(bufptr + 7, parent.hash) ||
 366                     bufptr[parent_entry_len] != '\n')
 367                         return error("bad parents in commit %s", oid_to_hex(&item->object.oid));
 368                 bufptr += parent_entry_len + 1;
 369                 /*
 370                  * The clone is shallow if nr_parent < 0, and we must
 371                  * not traverse its real parents even when we unhide them.
 372                  */
 373                 if (graft && (graft->nr_parent < 0 || grafts_replace_parents))
 374                         continue;
 375                 new_parent = lookup_commit(&parent);
 376                 if (new_parent)
 377                         pptr = &commit_list_insert(new_parent, pptr)->next;
 378         }
 379         if (graft) {
 380                 int i;
 381                 struct commit *new_parent;
 382                 for (i = 0; i < graft->nr_parent; i++) {
 383                         new_parent = lookup_commit(&graft->parent[i]);
 384                         if (!new_parent)
 385                                 continue;
 386                         pptr = &commit_list_insert(new_parent, pptr)->next;
 387                 }
 388         }
 389         item->date = parse_commit_date(bufptr, tail);
 390
 391         return 0;
 392 }
 393
 394 int parse_commit_gently(struct commit *item, int quiet_on_missing)
 395 {
 396         enum object_type type;
 397         void *buffer;
 398         unsigned long size;
 399         int ret;
 400
 401         if (!item)
 402                 return -1;
 403         if (item->object.parsed)
 404                 return 0;
 405         buffer = read_object_file(&item->object.oid, &type, &size);
 406         if (!buffer)
 407                 return quiet_on_missing ? -1 :
 408                         error("Could not read %s",
 409                              oid_to_hex(&item->object.oid));
 410         if (type != OBJ_COMMIT) {
 411                 free(buffer);
 412                 return error("Object %s not a commit",
 413                              oid_to_hex(&item->object.oid));
 414         }
 415         ret = parse_commit_buffer(item, buffer, size);
 416         if (save_commit_buffer && !ret) {
 417                 set_commit_buffer(item, buffer, size);
 418                 return 0;
 419         }
 420         free(buffer);
 421         return ret;
 422 }
 423
 424 void parse_commit_or_die(struct commit *item)
 425 {
 426         if (parse_commit(item))
 427                 die("unable to parse commit %s",
 428                     item ? oid_to_hex(&item->object.oid) : "(null)");
 429 }
 430
 431 int find_commit_subject(const char *commit_buffer, const char **subject)
 432 {
 433         const char *eol;
 434         const char *p = commit_buffer;
 435
 436         while (*p && (*p != '\n' || p[1] != '\n'))
 437                 p++;
 438         if (*p) {
 439                 p = skip_blank_lines(p + 2);
 440                 eol = strchrnul(p, '\n');
 441         } else
 442                 eol = p;
 443
 444         *subject = p;
 445
 446         return eol - p;
 447 }
 448
 449 struct commit_list *commit_list_insert(struct commit *item, struct commit_list **list_p)
 450 {
 451         struct commit_list *new_list = xmalloc(sizeof(struct commit_list));
 452         new_list->item = item;
 453         new_list->next = *list_p;
 454         *list_p = new_list;
 455         return new_list;
 456 }
 457
 458 unsigned commit_list_count(const struct commit_list *l)
 459 {
 460         unsigned c = 0;
 461         for (; l; l = l->next )
 462                 c++;
 463         return c;
 464 }
 465
 466 struct commit_list *copy_commit_list(struct commit_list *list)
 467 {
 468         struct commit_list *head = NULL;
 469         struct commit_list **pp = &head;
 470         while (list) {
 471                 pp = commit_list_append(list->item, pp);
 472                 list = list->next;
 473         }
 474         return head;
 475 }
 476
 477 void free_commit_list(struct commit_list *list)
 478 {
 479         while (list)
 480                 pop_commit(&list);
 481 }
 482
 483 struct commit_list * commit_list_insert_by_date(struct commit *item, struct commit_list **list)
 484 {
 485         struct commit_list **pp = list;
 486         struct commit_list *p;
 487         while ((p = *pp) != NULL) {
 488                 if (p->item->date < item->date) {
 489                         break;
 490                 }
 491                 pp = &p->next;
 492         }
 493         return commit_list_insert(item, pp);
 494 }
 495
 496 static int commit_list_compare_by_date(const void *a, const void *b)
 497 {
 498         timestamp_t a_date = ((const struct commit_list *)a)->item->date;
 499         timestamp_t b_date = ((const struct commit_list *)b)->item->date;
 500         if (a_date < b_date)
 501                 return 1;
 502         if (a_date > b_date)
 503                 return -1;
 504         return 0;
 505 }
 506
 507 static void *commit_list_get_next(const void *a)
 508 {
 509         return ((const struct commit_list *)a)->next;
 510 }
 511
 512 static void commit_list_set_next(void *a, void *next)
 513 {
 514         ((struct commit_list *)a)->next = next;
 515 }
 516
 517 void commit_list_sort_by_date(struct commit_list **list)
 518 {
 519         *list = llist_mergesort(*list, commit_list_get_next, commit_list_set_next,
 520                                 commit_list_compare_by_date);
 521 }
 522
 523 struct commit *pop_most_recent_commit(struct commit_list **list,
 524                                       unsigned int mark)
 525 {
 526         struct commit *ret = pop_commit(list);
 527         struct commit_list *parents = ret->parents;
 528
 529         while (parents) {
 530                 struct commit *commit = parents->item;
 531                 if (!parse_commit(commit) && !(commit->object.flags & mark)) {
 532                         commit->object.flags |= mark;
 533                         commit_list_insert_by_date(commit, list);
 534                 }
 535                 parents = parents->next;
 536         }
 537         return ret;
 538 }
 539
 540 static void clear_commit_marks_1(struct commit_list **plist,
 541                                  struct commit *commit, unsigned int mark)
 542 {
 543         while (commit) {
 544                 struct commit_list *parents;
 545
 546                 if (!(mark & commit->object.flags))
 547                         return;
 548
 549                 commit->object.flags &= ~mark;
 550
 551                 parents = commit->parents;
 552                 if (!parents)
 553                         return;
 554
 555                 while ((parents = parents->next))
 556                         commit_list_insert(parents->item, plist);
 557
 558                 commit = commit->parents->item;
 559         }
 560 }
 561
 562 void clear_commit_marks_many(int nr, struct commit **commit, unsigned int mark)
 563 {
 564         struct commit_list *list = NULL;
 565
 566         while (nr--) {
 567                 clear_commit_marks_1(&list, *commit, mark);
 568                 commit++;
 569         }
 570         while (list)
 571                 clear_commit_marks_1(&list, pop_commit(&list), mark);
 572 }
 573
 574 void clear_commit_marks(struct commit *commit, unsigned int mark)
 575 {
 576         clear_commit_marks_many(1, &commit, mark);
 577 }
 578
 579 struct commit *pop_commit(struct commit_list **stack)
 580 {
 581         struct commit_list *top = *stack;
 582         struct commit *item = top ? top->item : NULL;
 583
 584         if (top) {
 585                 *stack = top->next;
 586                 free(top);
 587         }
 588         return item;
 589 }
 590
 591 /*
 592  * Topological sort support
 593  */
 594
 595 /* count number of children that have not been emitted */
 596 define_commit_slab(indegree_slab, int);
 597
 598 /* record author-date for each commit object */
 599 define_commit_slab(author_date_slab, unsigned long);
 600
 601 static void record_author_date(struct author_date_slab *author_date,
 602                                struct commit *commit)
 603 {
 604         const char *buffer = get_commit_buffer(commit, NULL);
 605         struct ident_split ident;
 606         const char *ident_line;
 607         size_t ident_len;
 608         char *date_end;
 609         timestamp_t date;
 610
 611         ident_line = find_commit_header(buffer, "author", &ident_len);
 612         if (!ident_line)
 613                 goto fail_exit; /* no author line */
 614         if (split_ident_line(&ident, ident_line, ident_len) ||
 615             !ident.date_begin || !ident.date_end)
 616                 goto fail_exit; /* malformed "author" line */
 617
 618         date = parse_timestamp(ident.date_begin, &date_end, 10);
 619         if (date_end != ident.date_end)
 620                 goto fail_exit; /* malformed date */
 621         *(author_date_slab_at(author_date, commit)) = date;
 622
 623 fail_exit:
 624         unuse_commit_buffer(commit, buffer);
 625 }
 626
 627 static int compare_commits_by_author_date(const void *a_, const void *b_,
 628                                           void *cb_data)
 629 {
 630         const struct commit *a = a_, *b = b_;
 631         struct author_date_slab *author_date = cb_data;
 632         timestamp_t a_date = *(author_date_slab_at(author_date, a));
 633         timestamp_t b_date = *(author_date_slab_at(author_date, b));
 634
 635         /* newer commits with larger date first */
 636         if (a_date < b_date)
 637                 return 1;
 638         else if (a_date > b_date)
 639                 return -1;
 640         return 0;
 641 }
 642
 643 int compare_commits_by_commit_date(const void *a_, const void *b_, void *unused)
 644 {
 645         const struct commit *a = a_, *b = b_;
 646         /* newer commits with larger date first */
 647         if (a->date < b->date)
 648                 return 1;
 649         else if (a->date > b->date)
 650                 return -1;
 651         return 0;
 652 }
 653
 654 /*
 655  * Performs an in-place topological sort on the list supplied.
 656  */
 657 void sort_in_topological_order(struct commit_list **list, enum rev_sort_order sort_order)
 658 {
 659         struct commit_list *next, *orig = *list;
 660         struct commit_list **pptr;
 661         struct indegree_slab indegree;
 662         struct prio_queue queue;
 663         struct commit *commit;
 664         struct author_date_slab author_date;
 665
 666         if (!orig)
 667                 return;
 668         *list = NULL;
 669
 670         init_indegree_slab(&indegree);
 671         memset(&queue, '\0', sizeof(queue));
 672
 673         switch (sort_order) {
 674         default: /* REV_SORT_IN_GRAPH_ORDER */
 675                 queue.compare = NULL;
 676                 break;
 677         case REV_SORT_BY_COMMIT_DATE:
 678                 queue.compare = compare_commits_by_commit_date;
 679                 break;
 680         case REV_SORT_BY_AUTHOR_DATE:
 681                 init_author_date_slab(&author_date);
 682                 queue.compare = compare_commits_by_author_date;
 683                 queue.cb_data = &author_date;
 684                 break;
 685         }
 686
 687         /* Mark them and clear the indegree */
 688         for (next = orig; next; next = next->next) {
 689                 struct commit *commit = next->item;
 690                 *(indegree_slab_at(&indegree, commit)) = 1;
 691                 /* also record the author dates, if needed */
 692                 if (sort_order == REV_SORT_BY_AUTHOR_DATE)
 693                         record_author_date(&author_date, commit);
 694         }
 695
 696         /* update the indegree */
 697         for (next = orig; next; next = next->next) {
 698                 struct commit_list *parents = next->item->parents;
 699                 while (parents) {
 700                         struct commit *parent = parents->item;
 701                         int *pi = indegree_slab_at(&indegree, parent);
 702
 703                         if (*pi)
 704                                 (*pi)++;
 705                         parents = parents->next;
 706                 }
 707         }
 708
 709         /*
 710          * find the tips
 711          *
 712          * tips are nodes not reachable from any other node in the list
 713          *
 714          * the tips serve as a starting set for the work queue.
 715          */
 716         for (next = orig; next; next = next->next) {
 717                 struct commit *commit = next->item;
 718
 719                 if (*(indegree_slab_at(&indegree, commit)) == 1)
 720                         prio_queue_put(&queue, commit);
 721         }
 722
 723         /*
 724          * This is unfortunate; the initial tips need to be shown
 725          * in the order given from the revision traversal machinery.
 726          */
 727         if (sort_order == REV_SORT_IN_GRAPH_ORDER)
 728                 prio_queue_reverse(&queue);
 729
 730         /* We no longer need the commit list */
 731         free_commit_list(orig);
 732
 733         pptr = list;
 734         *list = NULL;
 735         while ((commit = prio_queue_get(&queue)) != NULL) {
 736                 struct commit_list *parents;
 737
 738                 for (parents = commit->parents; parents ; parents = parents->next) {
 739                         struct commit *parent = parents->item;
 740                         int *pi = indegree_slab_at(&indegree, parent);
 741
 742                         if (!*pi)
 743                                 continue;
 744
 745                         /*
 746                          * parents are only enqueued for emission
 747                          * when all their children have been emitted thereby
 748                          * guaranteeing topological order.
 749                          */
 750                         if (--(*pi) == 1)
 751                                 prio_queue_put(&queue, parent);
 752                 }
 753                 /*
 754                  * all children of commit have already been
 755                  * emitted. we can emit it now.
 756                  */
 757                 *(indegree_slab_at(&indegree, commit)) = 0;
 758
 759                 pptr = &commit_list_insert(commit, pptr)->next;
 760         }
 761
 762         clear_indegree_slab(&indegree);
 763         clear_prio_queue(&queue);
 764         if (sort_order == REV_SORT_BY_AUTHOR_DATE)
 765                 clear_author_date_slab(&author_date);
 766 }
 767
 768 /* merge-base stuff */
 769
 770 /* Remember to update object flag allocation in object.h */
 771 #define PARENT1         (1u<<16)
 772 #define PARENT2         (1u<<17)
 773 #define STALE           (1u<<18)
 774 #define RESULT          (1u<<19)
 775
 776 static const unsigned all_flags = (PARENT1 | PARENT2 | STALE | RESULT);
 777
 778 static int queue_has_nonstale(struct prio_queue *queue)
 779 {
 780         int i;
 781         for (i = 0; i < queue->nr; i++) {
 782                 struct commit *commit = queue->array[i].data;
 783                 if (!(commit->object.flags & STALE))
 784                         return 1;
 785         }
 786         return 0;
 787 }
 788
 789 /* all input commits in one and twos[] must have been parsed! */
 790 static struct commit_list *paint_down_to_common(struct commit *one, int n, struct commit **twos)
 791 {
 792         struct prio_queue queue = { compare_commits_by_commit_date };
 793         struct commit_list *result = NULL;
 794         int i;
 795
 796         one->object.flags |= PARENT1;
 797         if (!n) {
 798                 commit_list_append(one, &result);
 799                 return result;
 800         }
 801         prio_queue_put(&queue, one);
 802
 803         for (i = 0; i < n; i++) {
 804                 twos[i]->object.flags |= PARENT2;
 805                 prio_queue_put(&queue, twos[i]);
 806         }
 807
 808         while (queue_has_nonstale(&queue)) {
 809                 struct commit *commit = prio_queue_get(&queue);
 810                 struct commit_list *parents;
 811                 int flags;
 812
 813                 flags = commit->object.flags & (PARENT1 | PARENT2 | STALE);
 814                 if (flags == (PARENT1 | PARENT2)) {
 815                         if (!(commit->object.flags & RESULT)) {
 816                                 commit->object.flags |= RESULT;
 817                                 commit_list_insert_by_date(commit, &result);
 818                         }
 819                         /* Mark parents of a found merge stale */
 820                         flags |= STALE;
 821                 }
 822                 parents = commit->parents;
 823                 while (parents) {
 824                         struct commit *p = parents->item;
 825                         parents = parents->next;
 826                         if ((p->object.flags & flags) == flags)
 827                                 continue;
 828                         if (parse_commit(p))
 829                                 return NULL;
 830                         p->object.flags |= flags;
 831                         prio_queue_put(&queue, p);
 832                 }
 833         }
 834
 835         clear_prio_queue(&queue);
 836         return result;
 837 }
 838
 839 static struct commit_list *merge_bases_many(struct commit *one, int n, struct commit **twos)
 840 {
 841         struct commit_list *list = NULL;
 842         struct commit_list *result = NULL;
 843         int i;
 844
 845         for (i = 0; i < n; i++) {
 846                 if (one == twos[i])
 847                         /*
 848                          * We do not mark this even with RESULT so we do not
 849                          * have to clean it up.
 850                          */
 851                         return commit_list_insert(one, &result);
 852         }
 853
 854         if (parse_commit(one))
 855                 return NULL;
 856         for (i = 0; i < n; i++) {
 857                 if (parse_commit(twos[i]))
 858                         return NULL;
 859         }
 860
 861         list = paint_down_to_common(one, n, twos);
 862
 863         while (list) {
 864                 struct commit *commit = pop_commit(&list);
 865                 if (!(commit->object.flags & STALE))
 866                         commit_list_insert_by_date(commit, &result);
 867         }
 868         return result;
 869 }
 870
 871 struct commit_list *get_octopus_merge_bases(struct commit_list *in)
 872 {
 873         struct commit_list *i, *j, *k, *ret = NULL;
 874
 875         if (!in)
 876                 return ret;
 877
 878         commit_list_insert(in->item, &ret);
 879
 880         for (i = in->next; i; i = i->next) {
 881                 struct commit_list *new_commits = NULL, *end = NULL;
 882
 883                 for (j = ret; j; j = j->next) {
 884                         struct commit_list *bases;
 885                         bases = get_merge_bases(i->item, j->item);
 886                         if (!new_commits)
 887                                 new_commits = bases;
 888                         else
 889                                 end->next = bases;
 890                         for (k = bases; k; k = k->next)
 891                                 end = k;
 892                 }
 893                 ret = new_commits;
 894         }
 895         return ret;
 896 }
 897
 898 static int remove_redundant(struct commit **array, int cnt)
 899 {
 900         /*
 901          * Some commit in the array may be an ancestor of
 902          * another commit.  Move such commit to the end of
 903          * the array, and return the number of commits that
 904          * are independent from each other.
 905          */
 906         struct commit **work;
 907         unsigned char *redundant;
 908         int *filled_index;
 909         int i, j, filled;
 910
 911         work = xcalloc(cnt, sizeof(*work));
 912         redundant = xcalloc(cnt, 1);
 913         ALLOC_ARRAY(filled_index, cnt - 1);
 914
 915         for (i = 0; i < cnt; i++)
 916                 parse_commit(array[i]);
 917         for (i = 0; i < cnt; i++) {
 918                 struct commit_list *common;
 919
 920                 if (redundant[i])
 921                         continue;
 922                 for (j = filled = 0; j < cnt; j++) {
 923                         if (i == j || redundant[j])
 924                                 continue;
 925                         filled_index[filled] = j;
 926                         work[filled++] = array[j];
 927                 }
 928                 common = paint_down_to_common(array[i], filled, work);
 929                 if (array[i]->object.flags & PARENT2)
 930                         redundant[i] = 1;
 931                 for (j = 0; j < filled; j++)
 932                         if (work[j]->object.flags & PARENT1)
 933                                 redundant[filled_index[j]] = 1;
 934                 clear_commit_marks(array[i], all_flags);
 935                 clear_commit_marks_many(filled, work, all_flags);
 936                 free_commit_list(common);
 937         }
 938
 939         /* Now collect the result */
 940         COPY_ARRAY(work, array, cnt);
 941         for (i = filled = 0; i < cnt; i++)
 942                 if (!redundant[i])
 943                         array[filled++] = work[i];
 944         for (j = filled, i = 0; i < cnt; i++)
 945                 if (redundant[i])
 946                         array[j++] = work[i];
 947         free(work);
 948         free(redundant);
 949         free(filled_index);
 950         return filled;
 951 }
 952
 953 static struct commit_list *get_merge_bases_many_0(struct commit *one,
 954                                                   int n,
 955                                                   struct commit **twos,
 956                                                   int cleanup)
 957 {
 958         struct commit_list *list;
 959         struct commit **rslt;
 960         struct commit_list *result;
 961         int cnt, i;
 962
 963         result = merge_bases_many(one, n, twos);
 964         for (i = 0; i < n; i++) {
 965                 if (one == twos[i])
 966                         return result;
 967         }
 968         if (!result || !result->next) {
 969                 if (cleanup) {
 970                         clear_commit_marks(one, all_flags);
 971                         clear_commit_marks_many(n, twos, all_flags);
 972                 }
 973                 return result;
 974         }
 975
 976         /* There are more than one */
 977         cnt = commit_list_count(result);
 978         rslt = xcalloc(cnt, sizeof(*rslt));
 979         for (list = result, i = 0; list; list = list->next)
 980                 rslt[i++] = list->item;
 981         free_commit_list(result);
 982
 983         clear_commit_marks(one, all_flags);
 984         clear_commit_marks_many(n, twos, all_flags);
 985
 986         cnt = remove_redundant(rslt, cnt);
 987         result = NULL;
 988         for (i = 0; i < cnt; i++)
 989                 commit_list_insert_by_date(rslt[i], &result);
 990         free(rslt);
 991         return result;
 992 }
 993
 994 struct commit_list *get_merge_bases_many(struct commit *one,
 995                                          int n,
 996                                          struct commit **twos)
 997 {
 998         return get_merge_bases_many_0(one, n, twos, 1);
 999 }
1000
1001 struct commit_list *get_merge_bases_many_dirty(struct commit *one,
1002                                                int n,
1003                                                struct commit **twos)
1004 {
1005         return get_merge_bases_many_0(one, n, twos, 0);
1006 }
1007
1008 struct commit_list *get_merge_bases(struct commit *one, struct commit *two)
1009 {
1010         return get_merge_bases_many_0(one, 1, &two, 1);
1011 }
1012
1013 /*
1014  * Is "commit" a descendant of one of the elements on the "with_commit" list?
1015  */
1016 int is_descendant_of(struct commit *commit, struct commit_list *with_commit)
1017 {
1018         if (!with_commit)
1019                 return 1;
1020         while (with_commit) {
1021                 struct commit *other;
1022
1023                 other = with_commit->item;
1024                 with_commit = with_commit->next;
1025                 if (in_merge_bases(other, commit))
1026                         return 1;
1027         }
1028         return 0;
1029 }
1030
1031 /*
1032  * Is "commit" an ancestor of one of the "references"?
1033  */
1034 int in_merge_bases_many(struct commit *commit, int nr_reference, struct commit **reference)
1035 {
1036         struct commit_list *bases;
1037         int ret = 0, i;
1038
1039         if (parse_commit(commit))
1040                 return ret;
1041         for (i = 0; i < nr_reference; i++)
1042                 if (parse_commit(reference[i]))
1043                         return ret;
1044
1045         bases = paint_down_to_common(commit, nr_reference, reference);
1046         if (commit->object.flags & PARENT2)
1047                 ret = 1;
1048         clear_commit_marks(commit, all_flags);
1049         clear_commit_marks_many(nr_reference, reference, all_flags);
1050         free_commit_list(bases);
1051         return ret;
1052 }
1053
1054 /*
1055  * Is "commit" an ancestor of (i.e. reachable from) the "reference"?
1056  */
1057 int in_merge_bases(struct commit *commit, struct commit *reference)
1058 {
1059         return in_merge_bases_many(commit, 1, &reference);
1060 }
1061
1062 struct commit_list *reduce_heads(struct commit_list *heads)
1063 {
1064         struct commit_list *p;
1065         struct commit_list *result = NULL, **tail = &result;
1066         struct commit **array;
1067         int num_head, i;
1068
1069         if (!heads)
1070                 return NULL;
1071
1072         /* Uniquify */
1073         for (p = heads; p; p = p->next)
1074                 p->item->object.flags &= ~STALE;
1075         for (p = heads, num_head = 0; p; p = p->next) {
1076                 if (p->item->object.flags & STALE)
1077                         continue;
1078                 p->item->object.flags |= STALE;
1079                 num_head++;
1080         }
1081         array = xcalloc(num_head, sizeof(*array));
1082         for (p = heads, i = 0; p; p = p->next) {
1083                 if (p->item->object.flags & STALE) {
1084                         array[i++] = p->item;
1085                         p->item->object.flags &= ~STALE;
1086                 }
1087         }
1088         num_head = remove_redundant(array, num_head);
1089         for (i = 0; i < num_head; i++)
1090                 tail = &commit_list_insert(array[i], tail)->next;
1091         free(array);
1092         return result;
1093 }
1094
1095 void reduce_heads_replace(struct commit_list **heads)
1096 {
1097         struct commit_list *result = reduce_heads(*heads);
1098         free_commit_list(*heads);
1099         *heads = result;
1100 }
1101
1102 static const char gpg_sig_header[] = "gpgsig";
1103 static const int gpg_sig_header_len = sizeof(gpg_sig_header) - 1;
1104
1105 static int do_sign_commit(struct strbuf *buf, const char *keyid)
1106 {
1107         struct strbuf sig = STRBUF_INIT;
1108         int inspos, copypos;
1109         const char *eoh;
1110
1111         /* find the end of the header */
1112         eoh = strstr(buf->buf, "\n\n");
1113         if (!eoh)
1114                 inspos = buf->len;
1115         else
1116                 inspos = eoh - buf->buf + 1;
1117
1118         if (!keyid || !*keyid)
1119                 keyid = get_signing_key();
1120         if (sign_buffer(buf, &sig, keyid)) {
1121                 strbuf_release(&sig);
1122                 return -1;
1123         }
1124
1125         for (copypos = 0; sig.buf[copypos]; ) {
1126                 const char *bol = sig.buf + copypos;
1127                 const char *eol = strchrnul(bol, '\n');
1128                 int len = (eol - bol) + !!*eol;
1129
1130                 if (!copypos) {
1131                         strbuf_insert(buf, inspos, gpg_sig_header, gpg_sig_header_len);
1132                         inspos += gpg_sig_header_len;
1133                 }
1134                 strbuf_insert(buf, inspos++, " ", 1);
1135                 strbuf_insert(buf, inspos, bol, len);
1136                 inspos += len;
1137                 copypos += len;
1138         }
1139         strbuf_release(&sig);
1140         return 0;
1141 }
1142
1143 int parse_signed_commit(const struct commit *commit,
1144                         struct strbuf *payload, struct strbuf *signature)
1145 {
1146
1147         unsigned long size;
1148         const char *buffer = get_commit_buffer(commit, &size);
1149         int in_signature, saw_signature = -1;
1150         const char *line, *tail;
1151
1152         line = buffer;
1153         tail = buffer + size;
1154         in_signature = 0;
1155         saw_signature = 0;
1156         while (line < tail) {
1157                 const char *sig = NULL;
1158                 const char *next = memchr(line, '\n', tail - line);
1159
1160                 next = next ? next + 1 : tail;
1161                 if (in_signature && line[0] == ' ')
1162                         sig = line + 1;
1163                 else if (starts_with(line, gpg_sig_header) &&
1164                          line[gpg_sig_header_len] == ' ')
1165                         sig = line + gpg_sig_header_len + 1;
1166                 if (sig) {
1167                         strbuf_add(signature, sig, next - sig);
1168                         saw_signature = 1;
1169                         in_signature = 1;
1170                 } else {
1171                         if (*line == '\n')
1172                                 /* dump the whole remainder of the buffer */
1173                                 next = tail;
1174                         strbuf_add(payload, line, next - line);
1175                         in_signature = 0;
1176                 }
1177                 line = next;
1178         }
1179         unuse_commit_buffer(commit, buffer);
1180         return saw_signature;
1181 }
1182
1183 int remove_signature(struct strbuf *buf)
1184 {
1185         const char *line = buf->buf;
1186         const char *tail = buf->buf + buf->len;
1187         int in_signature = 0;
1188         const char *sig_start = NULL;
1189         const char *sig_end = NULL;
1190
1191         while (line < tail) {
1192                 const char *next = memchr(line, '\n', tail - line);
1193                 next = next ? next + 1 : tail;
1194
1195                 if (in_signature && line[0] == ' ')
1196                         sig_end = next;
1197                 else if (starts_with(line, gpg_sig_header) &&
1198                          line[gpg_sig_header_len] == ' ') {
1199                         sig_start = line;
1200                         sig_end = next;
1201                         in_signature = 1;
1202                 } else {
1203                         if (*line == '\n')
1204                                 /* dump the whole remainder of the buffer */
1205                                 next = tail;
1206                         in_signature = 0;
1207                 }
1208                 line = next;
1209         }
1210
1211         if (sig_start)
1212                 strbuf_remove(buf, sig_start - buf->buf, sig_end - sig_start);
1213
1214         return sig_start != NULL;
1215 }
1216
1217 static void handle_signed_tag(struct commit *parent, struct commit_extra_header ***tail)
1218 {
1219         struct merge_remote_desc *desc;
1220         struct commit_extra_header *mergetag;
1221         char *buf;
1222         unsigned long size, len;
1223         enum object_type type;
1224
1225         desc = merge_remote_util(parent);
1226         if (!desc || !desc->obj)
1227                 return;
1228         buf = read_object_file(&desc->obj->oid, &type, &size);
1229         if (!buf || type != OBJ_TAG)
1230                 goto free_return;
1231         len = parse_signature(buf, size);
1232         if (size == len)
1233                 goto free_return;
1234         /*
1235          * We could verify this signature and either omit the tag when
1236          * it does not validate, but the integrator may not have the
1237          * public key of the signer of the tag he is merging, while a
1238          * later auditor may have it while auditing, so let's not run
1239          * verify-signed-buffer here for now...
1240          *
1241          * if (verify_signed_buffer(buf, len, buf + len, size - len, ...))
1242          *      warn("warning: signed tag unverified.");
1243          */
1244         mergetag = xcalloc(1, sizeof(*mergetag));
1245         mergetag->key = xstrdup("mergetag");
1246         mergetag->value = buf;
1247         mergetag->len = size;
1248
1249         **tail = mergetag;
1250         *tail = &mergetag->next;
1251         return;
1252
1253 free_return:
1254         free(buf);
1255 }
1256
1257 int check_commit_signature(const struct commit *commit, struct signature_check *sigc)
1258 {
1259         struct strbuf payload = STRBUF_INIT;
1260         struct strbuf signature = STRBUF_INIT;
1261         int ret = 1;
1262
1263         sigc->result = 'N';
1264
1265         if (parse_signed_commit(commit, &payload, &signature) <= 0)
1266                 goto out;
1267         ret = check_signature(payload.buf, payload.len, signature.buf,
1268                 signature.len, sigc);
1269
1270  out:
1271         strbuf_release(&payload);
1272         strbuf_release(&signature);
1273
1274         return ret;
1275 }
1276
1277
1278
1279 void append_merge_tag_headers(struct commit_list *parents,
1280                               struct commit_extra_header ***tail)
1281 {
1282         while (parents) {
1283                 struct commit *parent = parents->item;
1284                 handle_signed_tag(parent, tail);
1285                 parents = parents->next;
1286         }
1287 }
1288
1289 static void add_extra_header(struct strbuf *buffer,
1290                              struct commit_extra_header *extra)
1291 {
1292         strbuf_addstr(buffer, extra->key);
1293         if (extra->len)
1294                 strbuf_add_lines(buffer, " ", extra->value, extra->len);
1295         else
1296                 strbuf_addch(buffer, '\n');
1297 }
1298
1299 struct commit_extra_header *read_commit_extra_headers(struct commit *commit,
1300                                                       const char **exclude)
1301 {
1302         struct commit_extra_header *extra = NULL;
1303         unsigned long size;
1304         const char *buffer = get_commit_buffer(commit, &size);
1305         extra = read_commit_extra_header_lines(buffer, size, exclude);
1306         unuse_commit_buffer(commit, buffer);
1307         return extra;
1308 }
1309
1310 void for_each_mergetag(each_mergetag_fn fn, struct commit *commit, void *data)
1311 {
1312         struct commit_extra_header *extra, *to_free;
1313
1314         to_free = read_commit_extra_headers(commit, NULL);
1315         for (extra = to_free; extra; extra = extra->next) {
1316                 if (strcmp(extra->key, "mergetag"))
1317                         continue; /* not a merge tag */
1318                 fn(commit, extra, data);
1319         }
1320         free_commit_extra_headers(to_free);
1321 }
1322
1323 static inline int standard_header_field(const char *field, size_t len)
1324 {
1325         return ((len == 4 && !memcmp(field, "tree", 4)) ||
1326                 (len == 6 && !memcmp(field, "parent", 6)) ||
1327                 (len == 6 && !memcmp(field, "author", 6)) ||
1328                 (len == 9 && !memcmp(field, "committer", 9)) ||
1329                 (len == 8 && !memcmp(field, "encoding", 8)));
1330 }
1331
1332 static int excluded_header_field(const char *field, size_t len, const char **exclude)
1333 {
1334         if (!exclude)
1335                 return 0;
1336
1337         while (*exclude) {
1338                 size_t xlen = strlen(*exclude);
1339                 if (len == xlen && !memcmp(field, *exclude, xlen))
1340                         return 1;
1341                 exclude++;
1342         }
1343         return 0;
1344 }
1345
1346 static struct commit_extra_header *read_commit_extra_header_lines(
1347         const char *buffer, size_t size,
1348         const char **exclude)
1349 {
1350         struct commit_extra_header *extra = NULL, **tail = &extra, *it = NULL;
1351         const char *line, *next, *eof, *eob;
1352         struct strbuf buf = STRBUF_INIT;
1353
1354         for (line = buffer, eob = line + size;
1355              line < eob && *line != '\n';
1356              line = next) {
1357                 next = memchr(line, '\n', eob - line);
1358                 next = next ? next + 1 : eob;
1359                 if (*line == ' ') {
1360                         /* continuation */
1361                         if (it)
1362                                 strbuf_add(&buf, line + 1, next - (line + 1));
1363                         continue;
1364                 }
1365                 if (it)
1366                         it->value = strbuf_detach(&buf, &it->len);
1367                 strbuf_reset(&buf);
1368                 it = NULL;
1369
1370                 eof = memchr(line, ' ', next - line);
1371                 if (!eof)
1372                         eof = next;
1373                 else if (standard_header_field(line, eof - line) ||
1374                          excluded_header_field(line, eof - line, exclude))
1375                         continue;
1376
1377                 it = xcalloc(1, sizeof(*it));
1378                 it->key = xmemdupz(line, eof-line);
1379                 *tail = it;
1380                 tail = &it->next;
1381                 if (eof + 1 < next)
1382                         strbuf_add(&buf, eof + 1, next - (eof + 1));
1383         }
1384         if (it)
1385                 it->value = strbuf_detach(&buf, &it->len);
1386         return extra;
1387 }
1388
1389 void free_commit_extra_headers(struct commit_extra_header *extra)
1390 {
1391         while (extra) {
1392                 struct commit_extra_header *next = extra->next;
1393                 free(extra->key);
1394                 free(extra->value);
1395                 free(extra);
1396                 extra = next;
1397         }
1398 }
1399
1400 int commit_tree(const char *msg, size_t msg_len, const struct object_id *tree,
1401                 struct commit_list *parents, struct object_id *ret,
1402                 const char *author, const char *sign_commit)
1403 {
1404         struct commit_extra_header *extra = NULL, **tail = &extra;
1405         int result;
1406
1407         append_merge_tag_headers(parents, &tail);
1408         result = commit_tree_extended(msg, msg_len, tree, parents, ret,
1409                                       author, sign_commit, extra);
1410         free_commit_extra_headers(extra);
1411         return result;
1412 }
1413
1414 static int find_invalid_utf8(const char *buf, int len)
1415 {
1416         int offset = 0;
1417         static const unsigned int max_codepoint[] = {
1418                 0x7f, 0x7ff, 0xffff, 0x10ffff
1419         };
1420
1421         while (len) {
1422                 unsigned char c = *buf++;
1423                 int bytes, bad_offset;
1424                 unsigned int codepoint;
1425                 unsigned int min_val, max_val;
1426
1427                 len--;
1428                 offset++;
1429
1430                 /* Simple US-ASCII? No worries. */
1431                 if (c < 0x80)
1432                         continue;
1433
1434                 bad_offset = offset-1;
1435
1436                 /*
1437                  * Count how many more high bits set: that's how
1438                  * many more bytes this sequence should have.
1439                  */
1440                 bytes = 0;
1441                 while (c & 0x40) {
1442                         c <<= 1;
1443                         bytes++;
1444                 }
1445
1446                 /*
1447                  * Must be between 1 and 3 more bytes.  Longer sequences result in
1448                  * codepoints beyond U+10FFFF, which are guaranteed never to exist.
1449                  */
1450                 if (bytes < 1 || 3 < bytes)
1451                         return bad_offset;
1452
1453                 /* Do we *have* that many bytes? */
1454                 if (len < bytes)
1455                         return bad_offset;
1456
1457                 /*
1458                  * Place the encoded bits at the bottom of the value and compute the
1459                  * valid range.
1460                  */
1461                 codepoint = (c & 0x7f) >> bytes;
1462                 min_val = max_codepoint[bytes-1] + 1;
1463                 max_val = max_codepoint[bytes];
1464
1465                 offset += bytes;
1466                 len -= bytes;
1467
1468                 /* And verify that they are good continuation bytes */
1469                 do {
1470                         codepoint <<= 6;
1471                         codepoint |= *buf & 0x3f;
1472                         if ((*buf++ & 0xc0) != 0x80)
1473                                 return bad_offset;
1474                 } while (--bytes);
1475
1476                 /* Reject codepoints that are out of range for the sequence length. */
1477                 if (codepoint < min_val || codepoint > max_val)
1478                         return bad_offset;
1479                 /* Surrogates are only for UTF-16 and cannot be encoded in UTF-8. */
1480                 if ((codepoint & 0x1ff800) == 0xd800)
1481                         return bad_offset;
1482                 /* U+xxFFFE and U+xxFFFF are guaranteed non-characters. */
1483                 if ((codepoint & 0xfffe) == 0xfffe)
1484                         return bad_offset;
1485                 /* So are anything in the range U+FDD0..U+FDEF. */
1486                 if (codepoint >= 0xfdd0 && codepoint <= 0xfdef)
1487                         return bad_offset;
1488         }
1489         return -1;
1490 }
1491
1492 /*
1493  * This verifies that the buffer is in proper utf8 format.
1494  *
1495  * If it isn't, it assumes any non-utf8 characters are Latin1,
1496  * and does the conversion.
1497  */
1498 static int verify_utf8(struct strbuf *buf)
1499 {
1500         int ok = 1;
1501         long pos = 0;
1502
1503         for (;;) {
1504                 int bad;
1505                 unsigned char c;
1506                 unsigned char replace[2];
1507
1508                 bad = find_invalid_utf8(buf->buf + pos, buf->len - pos);
1509                 if (bad < 0)
1510                         return ok;
1511                 pos += bad;
1512                 ok = 0;
1513                 c = buf->buf[pos];
1514                 strbuf_remove(buf, pos, 1);
1515
1516                 /* We know 'c' must be in the range 128-255 */
1517                 replace[0] = 0xc0 + (c >> 6);
1518                 replace[1] = 0x80 + (c & 0x3f);
1519                 strbuf_insert(buf, pos, replace, 2);
1520                 pos += 2;
1521         }
1522 }
1523
1524 static const char commit_utf8_warn[] =
1525 N_("Warning: commit message did not conform to UTF-8.\n"
1526    "You may want to amend it after fixing the message, or set the config\n"
1527    "variable i18n.commitencoding to the encoding your project uses.\n");
1528
1529 int commit_tree_extended(const char *msg, size_t msg_len,
1530                          const struct object_id *tree,
1531                          struct commit_list *parents, struct object_id *ret,
1532                          const char *author, const char *sign_commit,
1533                          struct commit_extra_header *extra)
1534 {
1535         int result;
1536         int encoding_is_utf8;
1537         struct strbuf buffer;
1538
1539         assert_oid_type(tree, OBJ_TREE);
1540
1541         if (memchr(msg, '\0', msg_len))
1542                 return error("a NUL byte in commit log message not allowed.");
1543
1544         /* Not having i18n.commitencoding is the same as having utf-8 */
1545         encoding_is_utf8 = is_encoding_utf8(git_commit_encoding);
1546
1547         strbuf_init(&buffer, 8192); /* should avoid reallocs for the headers */
1548         strbuf_addf(&buffer, "tree %s\n", oid_to_hex(tree));
1549
1550         /*
1551          * NOTE! This ordering means that the same exact tree merged with a
1552          * different order of parents will be a _different_ changeset even
1553          * if everything else stays the same.
1554          */
1555         while (parents) {
1556                 struct commit *parent = pop_commit(&parents);
1557                 strbuf_addf(&buffer, "parent %s\n",
1558                             oid_to_hex(&parent->object.oid));
1559         }
1560
1561         /* Person/date information */
1562         if (!author)
1563                 author = git_author_info(IDENT_STRICT);
1564         strbuf_addf(&buffer, "author %s\n", author);
1565         strbuf_addf(&buffer, "committer %s\n", git_committer_info(IDENT_STRICT));
1566         if (!encoding_is_utf8)
1567                 strbuf_addf(&buffer, "encoding %s\n", git_commit_encoding);
1568
1569         while (extra) {
1570                 add_extra_header(&buffer, extra);
1571                 extra = extra->next;
1572         }
1573         strbuf_addch(&buffer, '\n');
1574
1575         /* And add the comment */
1576         strbuf_add(&buffer, msg, msg_len);
1577
1578         /* And check the encoding */
1579         if (encoding_is_utf8 && !verify_utf8(&buffer))
1580                 fprintf(stderr, _(commit_utf8_warn));
1581
1582         if (sign_commit && do_sign_commit(&buffer, sign_commit)) {
1583                 result = -1;
1584                 goto out;
1585         }
1586
1587         result = write_object_file(buffer.buf, buffer.len, commit_type, ret);
1588 out:
1589         strbuf_release(&buffer);
1590         return result;
1591 }
1592
1593 void set_merge_remote_desc(struct commit *commit,
1594                            const char *name, struct object *obj)
1595 {
1596         struct merge_remote_desc *desc;
1597         FLEX_ALLOC_STR(desc, name, name);
1598         desc->obj = obj;
1599         commit->util = desc;
1600 }
1601
1602 struct commit *get_merge_parent(const char *name)
1603 {
1604         struct object *obj;
1605         struct commit *commit;
1606         struct object_id oid;
1607         if (get_oid(name, &oid))
1608                 return NULL;
1609         obj = parse_object(&oid);
1610         commit = (struct commit *)peel_to_type(name, 0, obj, OBJ_COMMIT);
1611         if (commit && !commit->util)
1612                 set_merge_remote_desc(commit, name, obj);
1613         return commit;
1614 }
1615
1616 /*
1617  * Append a commit to the end of the commit_list.
1618  *
1619  * next starts by pointing to the variable that holds the head of an
1620  * empty commit_list, and is updated to point to the "next" field of
1621  * the last item on the list as new commits are appended.
1622  *
1623  * Usage example:
1624  *
1625  *     struct commit_list *list;
1626  *     struct commit_list **next = &list;
1627  *
1628  *     next = commit_list_append(c1, next);
1629  *     next = commit_list_append(c2, next);
1630  *     assert(commit_list_count(list) == 2);
1631  *     return list;
1632  */
1633 struct commit_list **commit_list_append(struct commit *commit,
1634                                         struct commit_list **next)
1635 {
1636         struct commit_list *new_commit = xmalloc(sizeof(struct commit_list));
1637         new_commit->item = commit;
1638         *next = new_commit;
1639         new_commit->next = NULL;
1640         return &new_commit->next;
1641 }
1642
1643 const char *find_commit_header(const char *msg, const char *key, size_t *out_len)
1644 {
1645         int key_len = strlen(key);
1646         const char *line = msg;
1647
1648         while (line) {
1649                 const char *eol = strchrnul(line, '\n');
1650
1651                 if (line == eol)
1652                         return NULL;
1653
1654                 if (eol - line > key_len &&
1655                     !strncmp(line, key, key_len) &&
1656                     line[key_len] == ' ') {
1657                         *out_len = eol - line - key_len - 1;
1658                         return line + key_len + 1;
1659                 }
1660                 line = *eol ? eol + 1 : NULL;
1661         }
1662         return NULL;
1663 }
1664
1665 /*
1666  * Inspect the given string and determine the true "end" of the log message, in
1667  * order to find where to put a new Signed-off-by: line.  Ignored are
1668  * trailing comment lines and blank lines.  To support "git commit -s
1669  * --amend" on an existing commit, we also ignore "Conflicts:".  To
1670  * support "git commit -v", we truncate at cut lines.
1671  *
1672  * Returns the number of bytes from the tail to ignore, to be fed as
1673  * the second parameter to append_signoff().
1674  */
1675 int ignore_non_trailer(const char *buf, size_t len)
1676 {
1677         int boc = 0;
1678         int bol = 0;
1679         int in_old_conflicts_block = 0;
1680         size_t cutoff = wt_status_locate_end(buf, len);
1681
1682         while (bol < cutoff) {
1683                 const char *next_line = memchr(buf + bol, '\n', len - bol);
1684
1685                 if (!next_line)
1686                         next_line = buf + len;
1687                 else
1688                         next_line++;
1689
1690                 if (buf[bol] == comment_line_char || buf[bol] == '\n') {
1691                         /* is this the first of the run of comments? */
1692                         if (!boc)
1693                                 boc = bol;
1694                         /* otherwise, it is just continuing */
1695                 } else if (starts_with(buf + bol, "Conflicts:\n")) {
1696                         in_old_conflicts_block = 1;
1697                         if (!boc)
1698                                 boc = bol;
1699                 } else if (in_old_conflicts_block && buf[bol] == '\t') {
1700                         ; /* a pathname in the conflicts block */
1701                 } else if (boc) {
1702                         /* the previous was not trailing comment */
1703                         boc = 0;
1704                         in_old_conflicts_block = 0;
1705                 }
1706                 bol = next_line - buf;
1707         }
1708         return boc ? len - boc : len - cutoff;
1709 }