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