commit.c

   1 #include "cache.h"
   2 #include "tag.h"
   3 #include "commit.h"
   4 #include "pkt-line.h"
   5 #include "utf8.h"
   6 #include "diff.h"
   7 #include "revision.h"
   8 #include "notes.h"
   9 #include "gpg-interface.h"
  10 #include "mergesort.h"
  11 #include "commit-slab.h"
  12 #include "prio-queue.h"
  13 #include "sha1-lookup.h"
  14 #include "wt-status.h"
  15
  16 static struct commit_extra_header *read_commit_extra_header_lines(const char *buf, size_t len, const char **);
  17
  18 int save_commit_buffer = 1;
  19
  20 const char *commit_type = "commit";
  21
  22 struct commit *lookup_commit_reference_gently(const struct object_id *oid,
  23                                               int quiet)
  24 {
  25         struct object *obj = deref_tag(parse_object(oid), NULL, 0);
  26
  27         if (!obj)
  28                 return NULL;
  29         return object_as_type(obj, OBJ_COMMIT, quiet);
  30 }
  31
  32 struct commit *lookup_commit_reference(const struct object_id *oid)
  33 {
  34         return lookup_commit_reference_gently(oid, 0);
  35 }
  36
  37 struct commit *lookup_commit_or_die(const struct object_id *oid, const char *ref_name)
  38 {
  39         struct commit *c = lookup_commit_reference(oid);
  40         if (!c)
  41                 die(_("could not parse %s"), ref_name);
  42         if (oidcmp(oid, &c->object.oid)) {
  43                 warning(_("%s %s is not a commit!"),
  44                         ref_name, oid_to_hex(oid));
  45         }
  46         return c;
  47 }
  48
  49 struct commit *lookup_commit(const struct object_id *oid)
  50 {
  51         struct object *obj = lookup_object(oid->hash);
  52         if (!obj)
  53                 return create_object(oid->hash, alloc_commit_node());
  54         return object_as_type(obj, OBJ_COMMIT, 0);
  55 }
  56
  57 struct commit *lookup_commit_reference_by_name(const char *name)
  58 {
  59         struct object_id oid;
  60         struct commit *commit;
  61
  62         if (get_oid_committish(name, &oid))
  63                 return NULL;
  64         commit = lookup_commit_reference(&oid);
  65         if (parse_commit(commit))
  66                 return NULL;
  67         return commit;
  68 }
  69
  70 static timestamp_t parse_commit_date(const char *buf, const char *tail)
  71 {
  72         const char *dateptr;
  73
  74         if (buf + 6 >= tail)
  75                 return 0;
  76         if (memcmp(buf, "author", 6))
  77                 return 0;
  78         while (buf < tail && *buf++ != '\n')
  79                 /* nada */;
  80         if (buf + 9 >= tail)
  81                 return 0;
  82         if (memcmp(buf, "committer", 9))
  83                 return 0;
  84         while (buf < tail && *buf++ != '>')
  85                 /* nada */;
  86         if (buf >= tail)
  87                 return 0;
  88         dateptr = buf;
  89         while (buf < tail && *buf++ != '\n')
  90                 /* nada */;
  91         if (buf >= tail)
  92                 return 0;
  93         /* dateptr < buf && buf[-1] == '\n', so parsing will stop at buf-1 */
  94         return parse_timestamp(dateptr, NULL, 10);
  95 }
  96
  97 static struct commit_graft **commit_graft;
  98 static int commit_graft_alloc, commit_graft_nr;
  99
 100 static const unsigned char *commit_graft_sha1_access(size_t index, void *table)
 101 {
 102         struct commit_graft **commit_graft_table = table;
 103         return commit_graft_table[index]->oid.hash;
 104 }
 105
 106 static int commit_graft_pos(const unsigned char *sha1)
 107 {
 108         return sha1_pos(sha1, commit_graft, commit_graft_nr,
 109                         commit_graft_sha1_access);
 110 }
 111
 112 int register_commit_graft(struct commit_graft *graft, int ignore_dups)
 113 {
 114         int pos = commit_graft_pos(graft->oid.hash);
 115
 116         if (0 <= pos) {
 117                 if (ignore_dups)
 118                         free(graft);
 119                 else {
 120                         free(commit_graft[pos]);
 121                         commit_graft[pos] = graft;
 122                 }
 123                 return 1;
 124         }
 125         pos = -pos - 1;
 126         ALLOC_GROW(commit_graft, commit_graft_nr + 1, commit_graft_alloc);
 127         commit_graft_nr++;
 128         if (pos < commit_graft_nr)
 129                 memmove(commit_graft + pos + 1,
 130                         commit_graft + pos,
 131                         (commit_graft_nr - pos - 1) *
 132                         sizeof(*commit_graft));
 133         commit_graft[pos] = graft;
 134         return 0;
 135 }
 136
 137 struct commit_graft *read_graft_line(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_sha1_file(commit->object.oid.hash, &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), typename(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 const void *detach_commit_buffer(struct commit *commit, unsigned long *sizep)
 301 {
 302         struct commit_buffer *v = buffer_slab_peek(&buffer_slab, commit);
 303         void *ret;
 304
 305         if (!v) {
 306                 if (sizep)
 307                         *sizep = 0;
 308                 return NULL;
 309         }
 310         ret = v->buffer;
 311         if (sizep)
 312                 *sizep = v->size;
 313
 314         v->buffer = NULL;
 315         v->size = 0;
 316         return ret;
 317 }
 318
 319 int parse_commit_buffer(struct commit *item, const void *buffer, unsigned long size)
 320 {
 321         const char *tail = buffer;
 322         const char *bufptr = buffer;
 323         struct object_id parent;
 324         struct commit_list **pptr;
 325         struct commit_graft *graft;
 326         const int tree_entry_len = GIT_SHA1_HEXSZ + 5;
 327         const int parent_entry_len = GIT_SHA1_HEXSZ + 7;
 328
 329         if (item->object.parsed)
 330                 return 0;
 331         item->object.parsed = 1;
 332         tail += size;
 333         if (tail <= bufptr + tree_entry_len + 1 || memcmp(bufptr, "tree ", 5) ||
 334                         bufptr[tree_entry_len] != '\n')
 335                 return error("bogus commit object %s", oid_to_hex(&item->object.oid));
 336         if (get_sha1_hex(bufptr + 5, parent.hash) < 0)
 337                 return error("bad tree pointer in commit %s",
 338                              oid_to_hex(&item->object.oid));
 339         item->tree = lookup_tree(&parent);
 340         bufptr += tree_entry_len + 1; /* "tree " + "hex sha1" + "\n" */
 341         pptr = &item->parents;
 342
 343         graft = lookup_commit_graft(&item->object.oid);
 344         while (bufptr + parent_entry_len < tail && !memcmp(bufptr, "parent ", 7)) {
 345                 struct commit *new_parent;
 346
 347                 if (tail <= bufptr + parent_entry_len + 1 ||
 348                     get_sha1_hex(bufptr + 7, parent.hash) ||
 349                     bufptr[parent_entry_len] != '\n')
 350                         return error("bad parents in commit %s", oid_to_hex(&item->object.oid));
 351                 bufptr += parent_entry_len + 1;
 352                 /*
 353                  * The clone is shallow if nr_parent < 0, and we must
 354                  * not traverse its real parents even when we unhide them.
 355                  */
 356                 if (graft && (graft->nr_parent < 0 || grafts_replace_parents))
 357                         continue;
 358                 new_parent = lookup_commit(&parent);
 359                 if (new_parent)
 360                         pptr = &commit_list_insert(new_parent, pptr)->next;
 361         }
 362         if (graft) {
 363                 int i;
 364                 struct commit *new_parent;
 365                 for (i = 0; i < graft->nr_parent; i++) {
 366                         new_parent = lookup_commit(&graft->parent[i]);
 367                         if (!new_parent)
 368                                 continue;
 369                         pptr = &commit_list_insert(new_parent, pptr)->next;
 370                 }
 371         }
 372         item->date = parse_commit_date(bufptr, tail);
 373
 374         return 0;
 375 }
 376
 377 int parse_commit_gently(struct commit *item, int quiet_on_missing)
 378 {
 379         enum object_type type;
 380         void *buffer;
 381         unsigned long size;
 382         int ret;
 383
 384         if (!item)
 385                 return -1;
 386         if (item->object.parsed)
 387                 return 0;
 388         buffer = read_sha1_file(item->object.oid.hash, &type, &size);
 389         if (!buffer)
 390                 return quiet_on_missing ? -1 :
 391                         error("Could not read %s",
 392                              oid_to_hex(&item->object.oid));
 393         if (type != OBJ_COMMIT) {
 394                 free(buffer);
 395                 return error("Object %s not a commit",
 396                              oid_to_hex(&item->object.oid));
 397         }
 398         ret = parse_commit_buffer(item, buffer, size);
 399         if (save_commit_buffer && !ret) {
 400                 set_commit_buffer(item, buffer, size);
 401                 return 0;
 402         }
 403         free(buffer);
 404         return ret;
 405 }
 406
 407 void parse_commit_or_die(struct commit *item)
 408 {
 409         if (parse_commit(item))
 410                 die("unable to parse commit %s",
 411                     item ? oid_to_hex(&item->object.oid) : "(null)");
 412 }
 413
 414 int find_commit_subject(const char *commit_buffer, const char **subject)
 415 {
 416         const char *eol;
 417         const char *p = commit_buffer;
 418
 419         while (*p && (*p != '\n' || p[1] != '\n'))
 420                 p++;
 421         if (*p) {
 422                 p = skip_blank_lines(p + 2);
 423                 eol = strchrnul(p, '\n');
 424         } else
 425                 eol = p;
 426
 427         *subject = p;
 428
 429         return eol - p;
 430 }
 431
 432 struct commit_list *commit_list_insert(struct commit *item, struct commit_list **list_p)
 433 {
 434         struct commit_list *new_list = xmalloc(sizeof(struct commit_list));
 435         new_list->item = item;
 436         new_list->next = *list_p;
 437         *list_p = new_list;
 438         return new_list;
 439 }
 440
 441 unsigned commit_list_count(const struct commit_list *l)
 442 {
 443         unsigned c = 0;
 444         for (; l; l = l->next )
 445                 c++;
 446         return c;
 447 }
 448
 449 struct commit_list *copy_commit_list(struct commit_list *list)
 450 {
 451         struct commit_list *head = NULL;
 452         struct commit_list **pp = &head;
 453         while (list) {
 454                 pp = commit_list_append(list->item, pp);
 455                 list = list->next;
 456         }
 457         return head;
 458 }
 459
 460 void free_commit_list(struct commit_list *list)
 461 {
 462         while (list)
 463                 pop_commit(&list);
 464 }
 465
 466 struct commit_list * commit_list_insert_by_date(struct commit *item, struct commit_list **list)
 467 {
 468         struct commit_list **pp = list;
 469         struct commit_list *p;
 470         while ((p = *pp) != NULL) {
 471                 if (p->item->date < item->date) {
 472                         break;
 473                 }
 474                 pp = &p->next;
 475         }
 476         return commit_list_insert(item, pp);
 477 }
 478
 479 static int commit_list_compare_by_date(const void *a, const void *b)
 480 {
 481         timestamp_t a_date = ((const struct commit_list *)a)->item->date;
 482         timestamp_t b_date = ((const struct commit_list *)b)->item->date;
 483         if (a_date < b_date)
 484                 return 1;
 485         if (a_date > b_date)
 486                 return -1;
 487         return 0;
 488 }
 489
 490 static void *commit_list_get_next(const void *a)
 491 {
 492         return ((const struct commit_list *)a)->next;
 493 }
 494
 495 static void commit_list_set_next(void *a, void *next)
 496 {
 497         ((struct commit_list *)a)->next = next;
 498 }
 499
 500 void commit_list_sort_by_date(struct commit_list **list)
 501 {
 502         *list = llist_mergesort(*list, commit_list_get_next, commit_list_set_next,
 503                                 commit_list_compare_by_date);
 504 }
 505
 506 struct commit *pop_most_recent_commit(struct commit_list **list,
 507                                       unsigned int mark)
 508 {
 509         struct commit *ret = pop_commit(list);
 510         struct commit_list *parents = ret->parents;
 511
 512         while (parents) {
 513                 struct commit *commit = parents->item;
 514                 if (!parse_commit(commit) && !(commit->object.flags & mark)) {
 515                         commit->object.flags |= mark;
 516                         commit_list_insert_by_date(commit, list);
 517                 }
 518                 parents = parents->next;
 519         }
 520         return ret;
 521 }
 522
 523 static void clear_commit_marks_1(struct commit_list **plist,
 524                                  struct commit *commit, unsigned int mark)
 525 {
 526         while (commit) {
 527                 struct commit_list *parents;
 528
 529                 if (!(mark & commit->object.flags))
 530                         return;
 531
 532                 commit->object.flags &= ~mark;
 533
 534                 parents = commit->parents;
 535                 if (!parents)
 536                         return;
 537
 538                 while ((parents = parents->next))
 539                         commit_list_insert(parents->item, plist);
 540
 541                 commit = commit->parents->item;
 542         }
 543 }
 544
 545 void clear_commit_marks_many(int nr, struct commit **commit, unsigned int mark)
 546 {
 547         struct commit_list *list = NULL;
 548
 549         while (nr--) {
 550                 commit_list_insert(*commit, &list);
 551                 commit++;
 552         }
 553         while (list)
 554                 clear_commit_marks_1(&list, pop_commit(&list), mark);
 555 }
 556
 557 void clear_commit_marks(struct commit *commit, unsigned int mark)
 558 {
 559         clear_commit_marks_many(1, &commit, mark);
 560 }
 561
 562 void clear_commit_marks_for_object_array(struct object_array *a, unsigned mark)
 563 {
 564         struct object *object;
 565         struct commit *commit;
 566         unsigned int i;
 567
 568         for (i = 0; i < a->nr; i++) {
 569                 object = a->objects[i].item;
 570                 commit = lookup_commit_reference_gently(&object->oid, 1);
 571                 if (commit)
 572                         clear_commit_marks(commit, mark);
 573         }
 574 }
 575
 576 struct commit *pop_commit(struct commit_list **stack)
 577 {
 578         struct commit_list *top = *stack;
 579         struct commit *item = top ? top->item : NULL;
 580
 581         if (top) {
 582                 *stack = top->next;
 583                 free(top);
 584         }
 585         return item;
 586 }
 587
 588 /*
 589  * Topological sort support
 590  */
 591
 592 /* count number of children that have not been emitted */
 593 define_commit_slab(indegree_slab, int);
 594
 595 /* record author-date for each commit object */
 596 define_commit_slab(author_date_slab, unsigned long);
 597
 598 static void record_author_date(struct author_date_slab *author_date,
 599                                struct commit *commit)
 600 {
 601         const char *buffer = get_commit_buffer(commit, NULL);
 602         struct ident_split ident;
 603         const char *ident_line;
 604         size_t ident_len;
 605         char *date_end;
 606         timestamp_t date;
 607
 608         ident_line = find_commit_header(buffer, "author", &ident_len);
 609         if (!ident_line)
 610                 goto fail_exit; /* no author line */
 611         if (split_ident_line(&ident, ident_line, ident_len) ||
 612             !ident.date_begin || !ident.date_end)
 613                 goto fail_exit; /* malformed "author" line */
 614
 615         date = parse_timestamp(ident.date_begin, &date_end, 10);
 616         if (date_end != ident.date_end)
 617                 goto fail_exit; /* malformed date */
 618         *(author_date_slab_at(author_date, commit)) = date;
 619
 620 fail_exit:
 621         unuse_commit_buffer(commit, buffer);
 622 }
 623
 624 static int compare_commits_by_author_date(const void *a_, const void *b_,
 625                                           void *cb_data)
 626 {
 627         const struct commit *a = a_, *b = b_;
 628         struct author_date_slab *author_date = cb_data;
 629         timestamp_t a_date = *(author_date_slab_at(author_date, a));
 630         timestamp_t b_date = *(author_date_slab_at(author_date, b));
 631
 632         /* newer commits with larger date first */
 633         if (a_date < b_date)
 634                 return 1;
 635         else if (a_date > b_date)
 636                 return -1;
 637         return 0;
 638 }
 639
 640 int compare_commits_by_commit_date(const void *a_, const void *b_, void *unused)
 641 {
 642         const struct commit *a = a_, *b = b_;
 643         /* newer commits with larger date first */
 644         if (a->date < b->date)
 645                 return 1;
 646         else if (a->date > b->date)
 647                 return -1;
 648         return 0;
 649 }
 650
 651 /*
 652  * Performs an in-place topological sort on the list supplied.
 653  */
 654 void sort_in_topological_order(struct commit_list **list, enum rev_sort_order sort_order)
 655 {
 656         struct commit_list *next, *orig = *list;
 657         struct commit_list **pptr;
 658         struct indegree_slab indegree;
 659         struct prio_queue queue;
 660         struct commit *commit;
 661         struct author_date_slab author_date;
 662
 663         if (!orig)
 664                 return;
 665         *list = NULL;
 666
 667         init_indegree_slab(&indegree);
 668         memset(&queue, '\0', sizeof(queue));
 669
 670         switch (sort_order) {
 671         default: /* REV_SORT_IN_GRAPH_ORDER */
 672                 queue.compare = NULL;
 673                 break;
 674         case REV_SORT_BY_COMMIT_DATE:
 675                 queue.compare = compare_commits_by_commit_date;
 676                 break;
 677         case REV_SORT_BY_AUTHOR_DATE:
 678                 init_author_date_slab(&author_date);
 679                 queue.compare = compare_commits_by_author_date;
 680                 queue.cb_data = &author_date;
 681                 break;
 682         }
 683
 684         /* Mark them and clear the indegree */
 685         for (next = orig; next; next = next->next) {
 686                 struct commit *commit = next->item;
 687                 *(indegree_slab_at(&indegree, commit)) = 1;
 688                 /* also record the author dates, if needed */
 689                 if (sort_order == REV_SORT_BY_AUTHOR_DATE)
 690                         record_author_date(&author_date, commit);
 691         }
 692
 693         /* update the indegree */
 694         for (next = orig; next; next = next->next) {
 695                 struct commit_list *parents = next->item->parents;
 696                 while (parents) {
 697                         struct commit *parent = parents->item;
 698                         int *pi = indegree_slab_at(&indegree, parent);
 699
 700                         if (*pi)
 701                                 (*pi)++;
 702                         parents = parents->next;
 703                 }
 704         }
 705
 706         /*
 707          * find the tips
 708          *
 709          * tips are nodes not reachable from any other node in the list
 710          *
 711          * the tips serve as a starting set for the work queue.
 712          */
 713         for (next = orig; next; next = next->next) {
 714                 struct commit *commit = next->item;
 715
 716                 if (*(indegree_slab_at(&indegree, commit)) == 1)
 717                         prio_queue_put(&queue, commit);
 718         }
 719
 720         /*
 721          * This is unfortunate; the initial tips need to be shown
 722          * in the order given from the revision traversal machinery.
 723          */
 724         if (sort_order == REV_SORT_IN_GRAPH_ORDER)
 725                 prio_queue_reverse(&queue);
 726
 727         /* We no longer need the commit list */
 728         free_commit_list(orig);
 729
 730         pptr = list;
 731         *list = NULL;
 732         while ((commit = prio_queue_get(&queue)) != NULL) {
 733                 struct commit_list *parents;
 734
 735                 for (parents = commit->parents; parents ; parents = parents->next) {
 736                         struct commit *parent = parents->item;
 737                         int *pi = indegree_slab_at(&indegree, parent);
 738
 739                         if (!*pi)
 740                                 continue;
 741
 742                         /*
 743                          * parents are only enqueued for emission
 744                          * when all their children have been emitted thereby
 745                          * guaranteeing topological order.
 746                          */
 747                         if (--(*pi) == 1)
 748                                 prio_queue_put(&queue, parent);
 749                 }
 750                 /*
 751                  * all children of commit have already been
 752                  * emitted. we can emit it now.
 753                  */
 754                 *(indegree_slab_at(&indegree, commit)) = 0;
 755
 756                 pptr = &commit_list_insert(commit, pptr)->next;
 757         }
 758
 759         clear_indegree_slab(&indegree);
 760         clear_prio_queue(&queue);
 761         if (sort_order == REV_SORT_BY_AUTHOR_DATE)
 762                 clear_author_date_slab(&author_date);
 763 }
 764
 765 /* merge-base stuff */
 766
 767 /* Remember to update object flag allocation in object.h */
 768 #define PARENT1         (1u<<16)
 769 #define PARENT2         (1u<<17)
 770 #define STALE           (1u<<18)
 771 #define RESULT          (1u<<19)
 772
 773 static const unsigned all_flags = (PARENT1 | PARENT2 | STALE | RESULT);
 774
 775 static int queue_has_nonstale(struct prio_queue *queue)
 776 {
 777         int i;
 778         for (i = 0; i < queue->nr; i++) {
 779                 struct commit *commit = queue->array[i].data;
 780                 if (!(commit->object.flags & STALE))
 781                         return 1;
 782         }
 783         return 0;
 784 }
 785
 786 /* all input commits in one and twos[] must have been parsed! */
 787 static struct commit_list *paint_down_to_common(struct commit *one, int n, struct commit **twos)
 788 {
 789         struct prio_queue queue = { compare_commits_by_commit_date };
 790         struct commit_list *result = NULL;
 791         int i;
 792
 793         one->object.flags |= PARENT1;
 794         if (!n) {
 795                 commit_list_append(one, &result);
 796                 return result;
 797         }
 798         prio_queue_put(&queue, one);
 799
 800         for (i = 0; i < n; i++) {
 801                 twos[i]->object.flags |= PARENT2;
 802                 prio_queue_put(&queue, twos[i]);
 803         }
 804
 805         while (queue_has_nonstale(&queue)) {
 806                 struct commit *commit = prio_queue_get(&queue);
 807                 struct commit_list *parents;
 808                 int flags;
 809
 810                 flags = commit->object.flags & (PARENT1 | PARENT2 | STALE);
 811                 if (flags == (PARENT1 | PARENT2)) {
 812                         if (!(commit->object.flags & RESULT)) {
 813                                 commit->object.flags |= RESULT;
 814                                 commit_list_insert_by_date(commit, &result);
 815                         }
 816                         /* Mark parents of a found merge stale */
 817                         flags |= STALE;
 818                 }
 819                 parents = commit->parents;
 820                 while (parents) {
 821                         struct commit *p = parents->item;
 822                         parents = parents->next;
 823                         if ((p->object.flags & flags) == flags)
 824                                 continue;
 825                         if (parse_commit(p))
 826                                 return NULL;
 827                         p->object.flags |= flags;
 828                         prio_queue_put(&queue, p);
 829                 }
 830         }
 831
 832         clear_prio_queue(&queue);
 833         return result;
 834 }
 835
 836 static struct commit_list *merge_bases_many(struct commit *one, int n, struct commit **twos)
 837 {
 838         struct commit_list *list = NULL;
 839         struct commit_list *result = NULL;
 840         int i;
 841
 842         for (i = 0; i < n; i++) {
 843                 if (one == twos[i])
 844                         /*
 845                          * We do not mark this even with RESULT so we do not
 846                          * have to clean it up.
 847                          */
 848                         return commit_list_insert(one, &result);
 849         }
 850
 851         if (parse_commit(one))
 852                 return NULL;
 853         for (i = 0; i < n; i++) {
 854                 if (parse_commit(twos[i]))
 855                         return NULL;
 856         }
 857
 858         list = paint_down_to_common(one, n, twos);
 859
 860         while (list) {
 861                 struct commit *commit = pop_commit(&list);
 862                 if (!(commit->object.flags & STALE))
 863                         commit_list_insert_by_date(commit, &result);
 864         }
 865         return result;
 866 }
 867
 868 struct commit_list *get_octopus_merge_bases(struct commit_list *in)
 869 {
 870         struct commit_list *i, *j, *k, *ret = NULL;
 871
 872         if (!in)
 873                 return ret;
 874
 875         commit_list_insert(in->item, &ret);
 876
 877         for (i = in->next; i; i = i->next) {
 878                 struct commit_list *new = NULL, *end = NULL;
 879
 880                 for (j = ret; j; j = j->next) {
 881                         struct commit_list *bases;
 882                         bases = get_merge_bases(i->item, j->item);
 883                         if (!new)
 884                                 new = bases;
 885                         else
 886                                 end->next = bases;
 887                         for (k = bases; k; k = k->next)
 888                                 end = k;
 889                 }
 890                 ret = new;
 891         }
 892         return ret;
 893 }
 894
 895 static int remove_redundant(struct commit **array, int cnt)
 896 {
 897         /*
 898          * Some commit in the array may be an ancestor of
 899          * another commit.  Move such commit to the end of
 900          * the array, and return the number of commits that
 901          * are independent from each other.
 902          */
 903         struct commit **work;
 904         unsigned char *redundant;
 905         int *filled_index;
 906         int i, j, filled;
 907
 908         work = xcalloc(cnt, sizeof(*work));
 909         redundant = xcalloc(cnt, 1);
 910         ALLOC_ARRAY(filled_index, cnt - 1);
 911
 912         for (i = 0; i < cnt; i++)
 913                 parse_commit(array[i]);
 914         for (i = 0; i < cnt; i++) {
 915                 struct commit_list *common;
 916
 917                 if (redundant[i])
 918                         continue;
 919                 for (j = filled = 0; j < cnt; j++) {
 920                         if (i == j || redundant[j])
 921                                 continue;
 922                         filled_index[filled] = j;
 923                         work[filled++] = array[j];
 924                 }
 925                 common = paint_down_to_common(array[i], filled, work);
 926                 if (array[i]->object.flags & PARENT2)
 927                         redundant[i] = 1;
 928                 for (j = 0; j < filled; j++)
 929                         if (work[j]->object.flags & PARENT1)
 930                                 redundant[filled_index[j]] = 1;
 931                 clear_commit_marks(array[i], all_flags);
 932                 for (j = 0; j < filled; j++)
 933                         clear_commit_marks(work[j], all_flags);
 934                 free_commit_list(common);
 935         }
 936
 937         /* Now collect the result */
 938         COPY_ARRAY(work, array, cnt);
 939         for (i = filled = 0; i < cnt; i++)
 940                 if (!redundant[i])
 941                         array[filled++] = work[i];
 942         for (j = filled, i = 0; i < cnt; i++)
 943                 if (redundant[i])
 944                         array[j++] = work[i];
 945         free(work);
 946         free(redundant);
 947         free(filled_index);
 948         return filled;
 949 }
 950
 951 static struct commit_list *get_merge_bases_many_0(struct commit *one,
 952                                                   int n,
 953                                                   struct commit **twos,
 954                                                   int cleanup)
 955 {
 956         struct commit_list *list;
 957         struct commit **rslt;
 958         struct commit_list *result;
 959         int cnt, i;
 960
 961         result = merge_bases_many(one, n, twos);
 962         for (i = 0; i < n; i++) {
 963                 if (one == twos[i])
 964                         return result;
 965         }
 966         if (!result || !result->next) {
 967                 if (cleanup) {
 968                         clear_commit_marks(one, all_flags);
 969                         clear_commit_marks_many(n, twos, all_flags);
 970                 }
 971                 return result;
 972         }
 973
 974         /* There are more than one */
 975         cnt = commit_list_count(result);
 976         rslt = xcalloc(cnt, sizeof(*rslt));
 977         for (list = result, i = 0; list; list = list->next)
 978                 rslt[i++] = list->item;
 979         free_commit_list(result);
 980
 981         clear_commit_marks(one, all_flags);
 982         clear_commit_marks_many(n, twos, all_flags);
 983
 984         cnt = remove_redundant(rslt, cnt);
 985         result = NULL;
 986         for (i = 0; i < cnt; i++)
 987                 commit_list_insert_by_date(rslt[i], &result);
 988         free(rslt);
 989         return result;
 990 }
 991
 992 struct commit_list *get_merge_bases_many(struct commit *one,
 993                                          int n,
 994                                          struct commit **twos)
 995 {
 996         return get_merge_bases_many_0(one, n, twos, 1);
 997 }
 998
 999 struct commit_list *get_merge_bases_many_dirty(struct commit *one,
1000                                                int n,
1001                                                struct commit **twos)
1002 {
1003         return get_merge_bases_many_0(one, n, twos, 0);
1004 }
1005
1006 struct commit_list *get_merge_bases(struct commit *one, struct commit *two)
1007 {
1008         return get_merge_bases_many_0(one, 1, &two, 1);
1009 }
1010
1011 /*
1012  * Is "commit" a descendant of one of the elements on the "with_commit" list?
1013  */
1014 int is_descendant_of(struct commit *commit, struct commit_list *with_commit)
1015 {
1016         if (!with_commit)
1017                 return 1;
1018         while (with_commit) {
1019                 struct commit *other;
1020
1021                 other = with_commit->item;
1022                 with_commit = with_commit->next;
1023                 if (in_merge_bases(other, commit))
1024                         return 1;
1025         }
1026         return 0;
1027 }
1028
1029 /*
1030  * Is "commit" an ancestor of one of the "references"?
1031  */
1032 int in_merge_bases_many(struct commit *commit, int nr_reference, struct commit **reference)
1033 {
1034         struct commit_list *bases;
1035         int ret = 0, i;
1036
1037         if (parse_commit(commit))
1038                 return ret;
1039         for (i = 0; i < nr_reference; i++)
1040                 if (parse_commit(reference[i]))
1041                         return ret;
1042
1043         bases = paint_down_to_common(commit, nr_reference, reference);
1044         if (commit->object.flags & PARENT2)
1045                 ret = 1;
1046         clear_commit_marks(commit, all_flags);
1047         clear_commit_marks_many(nr_reference, reference, all_flags);
1048         free_commit_list(bases);
1049         return ret;
1050 }
1051
1052 /*
1053  * Is "commit" an ancestor of (i.e. reachable from) the "reference"?
1054  */
1055 int in_merge_bases(struct commit *commit, struct commit *reference)
1056 {
1057         return in_merge_bases_many(commit, 1, &reference);
1058 }
1059
1060 struct commit_list *reduce_heads(struct commit_list *heads)
1061 {
1062         struct commit_list *p;
1063         struct commit_list *result = NULL, **tail = &result;
1064         struct commit **array;
1065         int num_head, i;
1066
1067         if (!heads)
1068                 return NULL;
1069
1070         /* Uniquify */
1071         for (p = heads; p; p = p->next)
1072                 p->item->object.flags &= ~STALE;
1073         for (p = heads, num_head = 0; p; p = p->next) {
1074                 if (p->item->object.flags & STALE)
1075                         continue;
1076                 p->item->object.flags |= STALE;
1077                 num_head++;
1078         }
1079         array = xcalloc(num_head, sizeof(*array));
1080         for (p = heads, i = 0; p; p = p->next) {
1081                 if (p->item->object.flags & STALE) {
1082                         array[i++] = p->item;
1083                         p->item->object.flags &= ~STALE;
1084                 }
1085         }
1086         num_head = remove_redundant(array, num_head);
1087         for (i = 0; i < num_head; i++)
1088                 tail = &commit_list_insert(array[i], tail)->next;
1089         free(array);
1090         return result;
1091 }
1092
1093 static const char gpg_sig_header[] = "gpgsig";
1094 static const int gpg_sig_header_len = sizeof(gpg_sig_header) - 1;
1095
1096 static int do_sign_commit(struct strbuf *buf, const char *keyid)
1097 {
1098         struct strbuf sig = STRBUF_INIT;
1099         int inspos, copypos;
1100         const char *eoh;
1101
1102         /* find the end of the header */
1103         eoh = strstr(buf->buf, "\n\n");
1104         if (!eoh)
1105                 inspos = buf->len;
1106         else
1107                 inspos = eoh - buf->buf + 1;
1108
1109         if (!keyid || !*keyid)
1110                 keyid = get_signing_key();
1111         if (sign_buffer(buf, &sig, keyid)) {
1112                 strbuf_release(&sig);
1113                 return -1;
1114         }
1115
1116         for (copypos = 0; sig.buf[copypos]; ) {
1117                 const char *bol = sig.buf + copypos;
1118                 const char *eol = strchrnul(bol, '\n');
1119                 int len = (eol - bol) + !!*eol;
1120
1121                 if (!copypos) {
1122                         strbuf_insert(buf, inspos, gpg_sig_header, gpg_sig_header_len);
1123                         inspos += gpg_sig_header_len;
1124                 }
1125                 strbuf_insert(buf, inspos++, " ", 1);
1126                 strbuf_insert(buf, inspos, bol, len);
1127                 inspos += len;
1128                 copypos += len;
1129         }
1130         strbuf_release(&sig);
1131         return 0;
1132 }
1133
1134 int parse_signed_commit(const struct commit *commit,
1135                         struct strbuf *payload, struct strbuf *signature)
1136 {
1137
1138         unsigned long size;
1139         const char *buffer = get_commit_buffer(commit, &size);
1140         int in_signature, saw_signature = -1;
1141         const char *line, *tail;
1142
1143         line = buffer;
1144         tail = buffer + size;
1145         in_signature = 0;
1146         saw_signature = 0;
1147         while (line < tail) {
1148                 const char *sig = NULL;
1149                 const char *next = memchr(line, '\n', tail - line);
1150
1151                 next = next ? next + 1 : tail;
1152                 if (in_signature && line[0] == ' ')
1153                         sig = line + 1;
1154                 else if (starts_with(line, gpg_sig_header) &&
1155                          line[gpg_sig_header_len] == ' ')
1156                         sig = line + gpg_sig_header_len + 1;
1157                 if (sig) {
1158                         strbuf_add(signature, sig, next - sig);
1159                         saw_signature = 1;
1160                         in_signature = 1;
1161                 } else {
1162                         if (*line == '\n')
1163                                 /* dump the whole remainder of the buffer */
1164                                 next = tail;
1165                         strbuf_add(payload, line, next - line);
1166                         in_signature = 0;
1167                 }
1168                 line = next;
1169         }
1170         unuse_commit_buffer(commit, buffer);
1171         return saw_signature;
1172 }
1173
1174 int remove_signature(struct strbuf *buf)
1175 {
1176         const char *line = buf->buf;
1177         const char *tail = buf->buf + buf->len;
1178         int in_signature = 0;
1179         const char *sig_start = NULL;
1180         const char *sig_end = NULL;
1181
1182         while (line < tail) {
1183                 const char *next = memchr(line, '\n', tail - line);
1184                 next = next ? next + 1 : tail;
1185
1186                 if (in_signature && line[0] == ' ')
1187                         sig_end = next;
1188                 else if (starts_with(line, gpg_sig_header) &&
1189                          line[gpg_sig_header_len] == ' ') {
1190                         sig_start = line;
1191                         sig_end = next;
1192                         in_signature = 1;
1193                 } else {
1194                         if (*line == '\n')
1195                                 /* dump the whole remainder of the buffer */
1196                                 next = tail;
1197                         in_signature = 0;
1198                 }
1199                 line = next;
1200         }
1201
1202         if (sig_start)
1203                 strbuf_remove(buf, sig_start - buf->buf, sig_end - sig_start);
1204
1205         return sig_start != NULL;
1206 }
1207
1208 static void handle_signed_tag(struct commit *parent, struct commit_extra_header ***tail)
1209 {
1210         struct merge_remote_desc *desc;
1211         struct commit_extra_header *mergetag;
1212         char *buf;
1213         unsigned long size, len;
1214         enum object_type type;
1215
1216         desc = merge_remote_util(parent);
1217         if (!desc || !desc->obj)
1218                 return;
1219         buf = read_sha1_file(desc->obj->oid.hash, &type, &size);
1220         if (!buf || type != OBJ_TAG)
1221                 goto free_return;
1222         len = parse_signature(buf, size);
1223         if (size == len)
1224                 goto free_return;
1225         /*
1226          * We could verify this signature and either omit the tag when
1227          * it does not validate, but the integrator may not have the
1228          * public key of the signer of the tag he is merging, while a
1229          * later auditor may have it while auditing, so let's not run
1230          * verify-signed-buffer here for now...
1231          *
1232          * if (verify_signed_buffer(buf, len, buf + len, size - len, ...))
1233          *      warn("warning: signed tag unverified.");
1234          */
1235         mergetag = xcalloc(1, sizeof(*mergetag));
1236         mergetag->key = xstrdup("mergetag");
1237         mergetag->value = buf;
1238         mergetag->len = size;
1239
1240         **tail = mergetag;
1241         *tail = &mergetag->next;
1242         return;
1243
1244 free_return:
1245         free(buf);
1246 }
1247
1248 int check_commit_signature(const struct commit *commit, struct signature_check *sigc)
1249 {
1250         struct strbuf payload = STRBUF_INIT;
1251         struct strbuf signature = STRBUF_INIT;
1252         int ret = 1;
1253
1254         sigc->result = 'N';
1255
1256         if (parse_signed_commit(commit, &payload, &signature) <= 0)
1257                 goto out;
1258         ret = check_signature(payload.buf, payload.len, signature.buf,
1259                 signature.len, sigc);
1260
1261  out:
1262         strbuf_release(&payload);
1263         strbuf_release(&signature);
1264
1265         return ret;
1266 }
1267
1268
1269
1270 void append_merge_tag_headers(struct commit_list *parents,
1271                               struct commit_extra_header ***tail)
1272 {
1273         while (parents) {
1274                 struct commit *parent = parents->item;
1275                 handle_signed_tag(parent, tail);
1276                 parents = parents->next;
1277         }
1278 }
1279
1280 static void add_extra_header(struct strbuf *buffer,
1281                              struct commit_extra_header *extra)
1282 {
1283         strbuf_addstr(buffer, extra->key);
1284         if (extra->len)
1285                 strbuf_add_lines(buffer, " ", extra->value, extra->len);
1286         else
1287                 strbuf_addch(buffer, '\n');
1288 }
1289
1290 struct commit_extra_header *read_commit_extra_headers(struct commit *commit,
1291                                                       const char **exclude)
1292 {
1293         struct commit_extra_header *extra = NULL;
1294         unsigned long size;
1295         const char *buffer = get_commit_buffer(commit, &size);
1296         extra = read_commit_extra_header_lines(buffer, size, exclude);
1297         unuse_commit_buffer(commit, buffer);
1298         return extra;
1299 }
1300
1301 void for_each_mergetag(each_mergetag_fn fn, struct commit *commit, void *data)
1302 {
1303         struct commit_extra_header *extra, *to_free;
1304
1305         to_free = read_commit_extra_headers(commit, NULL);
1306         for (extra = to_free; extra; extra = extra->next) {
1307                 if (strcmp(extra->key, "mergetag"))
1308                         continue; /* not a merge tag */
1309                 fn(commit, extra, data);
1310         }
1311         free_commit_extra_headers(to_free);
1312 }
1313
1314 static inline int standard_header_field(const char *field, size_t len)
1315 {
1316         return ((len == 4 && !memcmp(field, "tree", 4)) ||
1317                 (len == 6 && !memcmp(field, "parent", 6)) ||
1318                 (len == 6 && !memcmp(field, "author", 6)) ||
1319                 (len == 9 && !memcmp(field, "committer", 9)) ||
1320                 (len == 8 && !memcmp(field, "encoding", 8)));
1321 }
1322
1323 static int excluded_header_field(const char *field, size_t len, const char **exclude)
1324 {
1325         if (!exclude)
1326                 return 0;
1327
1328         while (*exclude) {
1329                 size_t xlen = strlen(*exclude);
1330                 if (len == xlen && !memcmp(field, *exclude, xlen))
1331                         return 1;
1332                 exclude++;
1333         }
1334         return 0;
1335 }
1336
1337 static struct commit_extra_header *read_commit_extra_header_lines(
1338         const char *buffer, size_t size,
1339         const char **exclude)
1340 {
1341         struct commit_extra_header *extra = NULL, **tail = &extra, *it = NULL;
1342         const char *line, *next, *eof, *eob;
1343         struct strbuf buf = STRBUF_INIT;
1344
1345         for (line = buffer, eob = line + size;
1346              line < eob && *line != '\n';
1347              line = next) {
1348                 next = memchr(line, '\n', eob - line);
1349                 next = next ? next + 1 : eob;
1350                 if (*line == ' ') {
1351                         /* continuation */
1352                         if (it)
1353                                 strbuf_add(&buf, line + 1, next - (line + 1));
1354                         continue;
1355                 }
1356                 if (it)
1357                         it->value = strbuf_detach(&buf, &it->len);
1358                 strbuf_reset(&buf);
1359                 it = NULL;
1360
1361                 eof = memchr(line, ' ', next - line);
1362                 if (!eof)
1363                         eof = next;
1364                 else if (standard_header_field(line, eof - line) ||
1365                          excluded_header_field(line, eof - line, exclude))
1366                         continue;
1367
1368                 it = xcalloc(1, sizeof(*it));
1369                 it->key = xmemdupz(line, eof-line);
1370                 *tail = it;
1371                 tail = &it->next;
1372                 if (eof + 1 < next)
1373                         strbuf_add(&buf, eof + 1, next - (eof + 1));
1374         }
1375         if (it)
1376                 it->value = strbuf_detach(&buf, &it->len);
1377         return extra;
1378 }
1379
1380 void free_commit_extra_headers(struct commit_extra_header *extra)
1381 {
1382         while (extra) {
1383                 struct commit_extra_header *next = extra->next;
1384                 free(extra->key);
1385                 free(extra->value);
1386                 free(extra);
1387                 extra = next;
1388         }
1389 }
1390
1391 int commit_tree(const char *msg, size_t msg_len,
1392                 const unsigned char *tree,
1393                 struct commit_list *parents, unsigned char *ret,
1394                 const char *author, const char *sign_commit)
1395 {
1396         struct commit_extra_header *extra = NULL, **tail = &extra;
1397         int result;
1398
1399         append_merge_tag_headers(parents, &tail);
1400         result = commit_tree_extended(msg, msg_len, tree, parents, ret,
1401                                       author, sign_commit, extra);
1402         free_commit_extra_headers(extra);
1403         return result;
1404 }
1405
1406 static int find_invalid_utf8(const char *buf, int len)
1407 {
1408         int offset = 0;
1409         static const unsigned int max_codepoint[] = {
1410                 0x7f, 0x7ff, 0xffff, 0x10ffff
1411         };
1412
1413         while (len) {
1414                 unsigned char c = *buf++;
1415                 int bytes, bad_offset;
1416                 unsigned int codepoint;
1417                 unsigned int min_val, max_val;
1418
1419                 len--;
1420                 offset++;
1421
1422                 /* Simple US-ASCII? No worries. */
1423                 if (c < 0x80)
1424                         continue;
1425
1426                 bad_offset = offset-1;
1427
1428                 /*
1429                  * Count how many more high bits set: that's how
1430                  * many more bytes this sequence should have.
1431                  */
1432                 bytes = 0;
1433                 while (c & 0x40) {
1434                         c <<= 1;
1435                         bytes++;
1436                 }
1437
1438                 /*
1439                  * Must be between 1 and 3 more bytes.  Longer sequences result in
1440                  * codepoints beyond U+10FFFF, which are guaranteed never to exist.
1441                  */
1442                 if (bytes < 1 || 3 < bytes)
1443                         return bad_offset;
1444
1445                 /* Do we *have* that many bytes? */
1446                 if (len < bytes)
1447                         return bad_offset;
1448
1449                 /*
1450                  * Place the encoded bits at the bottom of the value and compute the
1451                  * valid range.
1452                  */
1453                 codepoint = (c & 0x7f) >> bytes;
1454                 min_val = max_codepoint[bytes-1] + 1;
1455                 max_val = max_codepoint[bytes];
1456
1457                 offset += bytes;
1458                 len -= bytes;
1459
1460                 /* And verify that they are good continuation bytes */
1461                 do {
1462                         codepoint <<= 6;
1463                         codepoint |= *buf & 0x3f;
1464                         if ((*buf++ & 0xc0) != 0x80)
1465                                 return bad_offset;
1466                 } while (--bytes);
1467
1468                 /* Reject codepoints that are out of range for the sequence length. */
1469                 if (codepoint < min_val || codepoint > max_val)
1470                         return bad_offset;
1471                 /* Surrogates are only for UTF-16 and cannot be encoded in UTF-8. */
1472                 if ((codepoint & 0x1ff800) == 0xd800)
1473                         return bad_offset;
1474                 /* U+xxFFFE and U+xxFFFF are guaranteed non-characters. */
1475                 if ((codepoint & 0xfffe) == 0xfffe)
1476                         return bad_offset;
1477                 /* So are anything in the range U+FDD0..U+FDEF. */
1478                 if (codepoint >= 0xfdd0 && codepoint <= 0xfdef)
1479                         return bad_offset;
1480         }
1481         return -1;
1482 }
1483
1484 /*
1485  * This verifies that the buffer is in proper utf8 format.
1486  *
1487  * If it isn't, it assumes any non-utf8 characters are Latin1,
1488  * and does the conversion.
1489  */
1490 static int verify_utf8(struct strbuf *buf)
1491 {
1492         int ok = 1;
1493         long pos = 0;
1494
1495         for (;;) {
1496                 int bad;
1497                 unsigned char c;
1498                 unsigned char replace[2];
1499
1500                 bad = find_invalid_utf8(buf->buf + pos, buf->len - pos);
1501                 if (bad < 0)
1502                         return ok;
1503                 pos += bad;
1504                 ok = 0;
1505                 c = buf->buf[pos];
1506                 strbuf_remove(buf, pos, 1);
1507
1508                 /* We know 'c' must be in the range 128-255 */
1509                 replace[0] = 0xc0 + (c >> 6);
1510                 replace[1] = 0x80 + (c & 0x3f);
1511                 strbuf_insert(buf, pos, replace, 2);
1512                 pos += 2;
1513         }
1514 }
1515
1516 static const char commit_utf8_warn[] =
1517 N_("Warning: commit message did not conform to UTF-8.\n"
1518    "You may want to amend it after fixing the message, or set the config\n"
1519    "variable i18n.commitencoding to the encoding your project uses.\n");
1520
1521 int commit_tree_extended(const char *msg, size_t msg_len,
1522                          const unsigned char *tree,
1523                          struct commit_list *parents, unsigned char *ret,
1524                          const char *author, const char *sign_commit,
1525                          struct commit_extra_header *extra)
1526 {
1527         int result;
1528         int encoding_is_utf8;
1529         struct strbuf buffer;
1530
1531         assert_sha1_type(tree, OBJ_TREE);
1532
1533         if (memchr(msg, '\0', msg_len))
1534                 return error("a NUL byte in commit log message not allowed.");
1535
1536         /* Not having i18n.commitencoding is the same as having utf-8 */
1537         encoding_is_utf8 = is_encoding_utf8(git_commit_encoding);
1538
1539         strbuf_init(&buffer, 8192); /* should avoid reallocs for the headers */
1540         strbuf_addf(&buffer, "tree %s\n", sha1_to_hex(tree));
1541
1542         /*
1543          * NOTE! This ordering means that the same exact tree merged with a
1544          * different order of parents will be a _different_ changeset even
1545          * if everything else stays the same.
1546          */
1547         while (parents) {
1548                 struct commit *parent = pop_commit(&parents);
1549                 strbuf_addf(&buffer, "parent %s\n",
1550                             oid_to_hex(&parent->object.oid));
1551         }
1552
1553         /* Person/date information */
1554         if (!author)
1555                 author = git_author_info(IDENT_STRICT);
1556         strbuf_addf(&buffer, "author %s\n", author);
1557         strbuf_addf(&buffer, "committer %s\n", git_committer_info(IDENT_STRICT));
1558         if (!encoding_is_utf8)
1559                 strbuf_addf(&buffer, "encoding %s\n", git_commit_encoding);
1560
1561         while (extra) {
1562                 add_extra_header(&buffer, extra);
1563                 extra = extra->next;
1564         }
1565         strbuf_addch(&buffer, '\n');
1566
1567         /* And add the comment */
1568         strbuf_add(&buffer, msg, msg_len);
1569
1570         /* And check the encoding */
1571         if (encoding_is_utf8 && !verify_utf8(&buffer))
1572                 fprintf(stderr, _(commit_utf8_warn));
1573
1574         if (sign_commit && do_sign_commit(&buffer, sign_commit)) {
1575                 result = -1;
1576                 goto out;
1577         }
1578
1579         result = write_sha1_file(buffer.buf, buffer.len, commit_type, ret);
1580 out:
1581         strbuf_release(&buffer);
1582         return result;
1583 }
1584
1585 void set_merge_remote_desc(struct commit *commit,
1586                            const char *name, struct object *obj)
1587 {
1588         struct merge_remote_desc *desc;
1589         FLEX_ALLOC_STR(desc, name, name);
1590         desc->obj = obj;
1591         commit->util = desc;
1592 }
1593
1594 struct commit *get_merge_parent(const char *name)
1595 {
1596         struct object *obj;
1597         struct commit *commit;
1598         struct object_id oid;
1599         if (get_oid(name, &oid))
1600                 return NULL;
1601         obj = parse_object(&oid);
1602         commit = (struct commit *)peel_to_type(name, 0, obj, OBJ_COMMIT);
1603         if (commit && !commit->util)
1604                 set_merge_remote_desc(commit, name, obj);
1605         return commit;
1606 }
1607
1608 /*
1609  * Append a commit to the end of the commit_list.
1610  *
1611  * next starts by pointing to the variable that holds the head of an
1612  * empty commit_list, and is updated to point to the "next" field of
1613  * the last item on the list as new commits are appended.
1614  *
1615  * Usage example:
1616  *
1617  *     struct commit_list *list;
1618  *     struct commit_list **next = &list;
1619  *
1620  *     next = commit_list_append(c1, next);
1621  *     next = commit_list_append(c2, next);
1622  *     assert(commit_list_count(list) == 2);
1623  *     return list;
1624  */
1625 struct commit_list **commit_list_append(struct commit *commit,
1626                                         struct commit_list **next)
1627 {
1628         struct commit_list *new = xmalloc(sizeof(struct commit_list));
1629         new->item = commit;
1630         *next = new;
1631         new->next = NULL;
1632         return &new->next;
1633 }
1634
1635 const char *find_commit_header(const char *msg, const char *key, size_t *out_len)
1636 {
1637         int key_len = strlen(key);
1638         const char *line = msg;
1639
1640         while (line) {
1641                 const char *eol = strchrnul(line, '\n');
1642
1643                 if (line == eol)
1644                         return NULL;
1645
1646                 if (eol - line > key_len &&
1647                     !strncmp(line, key, key_len) &&
1648                     line[key_len] == ' ') {
1649                         *out_len = eol - line - key_len - 1;
1650                         return line + key_len + 1;
1651                 }
1652                 line = *eol ? eol + 1 : NULL;
1653         }
1654         return NULL;
1655 }
1656
1657 /*
1658  * Inspect the given string and determine the true "end" of the log message, in
1659  * order to find where to put a new Signed-off-by: line.  Ignored are
1660  * trailing comment lines and blank lines.  To support "git commit -s
1661  * --amend" on an existing commit, we also ignore "Conflicts:".  To
1662  * support "git commit -v", we truncate at cut lines.
1663  *
1664  * Returns the number of bytes from the tail to ignore, to be fed as
1665  * the second parameter to append_signoff().
1666  */
1667 int ignore_non_trailer(const char *buf, size_t len)
1668 {
1669         int boc = 0;
1670         int bol = 0;
1671         int in_old_conflicts_block = 0;
1672         size_t cutoff = wt_status_locate_end(buf, len);
1673
1674         while (bol < cutoff) {
1675                 const char *next_line = memchr(buf + bol, '\n', len - bol);
1676
1677                 if (!next_line)
1678                         next_line = buf + len;
1679                 else
1680                         next_line++;
1681
1682                 if (buf[bol] == comment_line_char || buf[bol] == '\n') {
1683                         /* is this the first of the run of comments? */
1684                         if (!boc)
1685                                 boc = bol;
1686                         /* otherwise, it is just continuing */
1687                 } else if (starts_with(buf + bol, "Conflicts:\n")) {
1688                         in_old_conflicts_block = 1;
1689                         if (!boc)
1690                                 boc = bol;
1691                 } else if (in_old_conflicts_block && buf[bol] == '\t') {
1692                         ; /* a pathname in the conflicts block */
1693                 } else if (boc) {
1694                         /* the previous was not trailing comment */
1695                         boc = 0;
1696                         in_old_conflicts_block = 0;
1697                 }
1698                 bol = next_line - buf;
1699         }
1700         return boc ? len - boc : len - cutoff;
1701 }