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                 clear_commit_marks_1(&list, *commit, mark);
 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 struct commit *pop_commit(struct commit_list **stack)
 563 {
 564         struct commit_list *top = *stack;
 565         struct commit *item = top ? top->item : NULL;
 566
 567         if (top) {
 568                 *stack = top->next;
 569                 free(top);
 570         }
 571         return item;
 572 }
 573
 574 /*
 575  * Topological sort support
 576  */
 577
 578 /* count number of children that have not been emitted */
 579 define_commit_slab(indegree_slab, int);
 580
 581 /* record author-date for each commit object */
 582 define_commit_slab(author_date_slab, unsigned long);
 583
 584 static void record_author_date(struct author_date_slab *author_date,
 585                                struct commit *commit)
 586 {
 587         const char *buffer = get_commit_buffer(commit, NULL);
 588         struct ident_split ident;
 589         const char *ident_line;
 590         size_t ident_len;
 591         char *date_end;
 592         timestamp_t date;
 593
 594         ident_line = find_commit_header(buffer, "author", &ident_len);
 595         if (!ident_line)
 596                 goto fail_exit; /* no author line */
 597         if (split_ident_line(&ident, ident_line, ident_len) ||
 598             !ident.date_begin || !ident.date_end)
 599                 goto fail_exit; /* malformed "author" line */
 600
 601         date = parse_timestamp(ident.date_begin, &date_end, 10);
 602         if (date_end != ident.date_end)
 603                 goto fail_exit; /* malformed date */
 604         *(author_date_slab_at(author_date, commit)) = date;
 605
 606 fail_exit:
 607         unuse_commit_buffer(commit, buffer);
 608 }
 609
 610 static int compare_commits_by_author_date(const void *a_, const void *b_,
 611                                           void *cb_data)
 612 {
 613         const struct commit *a = a_, *b = b_;
 614         struct author_date_slab *author_date = cb_data;
 615         timestamp_t a_date = *(author_date_slab_at(author_date, a));
 616         timestamp_t b_date = *(author_date_slab_at(author_date, b));
 617
 618         /* newer commits with larger date first */
 619         if (a_date < b_date)
 620                 return 1;
 621         else if (a_date > b_date)
 622                 return -1;
 623         return 0;
 624 }
 625
 626 int compare_commits_by_commit_date(const void *a_, const void *b_, void *unused)
 627 {
 628         const struct commit *a = a_, *b = b_;
 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 /*
 638  * Performs an in-place topological sort on the list supplied.
 639  */
 640 void sort_in_topological_order(struct commit_list **list, enum rev_sort_order sort_order)
 641 {
 642         struct commit_list *next, *orig = *list;
 643         struct commit_list **pptr;
 644         struct indegree_slab indegree;
 645         struct prio_queue queue;
 646         struct commit *commit;
 647         struct author_date_slab author_date;
 648
 649         if (!orig)
 650                 return;
 651         *list = NULL;
 652
 653         init_indegree_slab(&indegree);
 654         memset(&queue, '\0', sizeof(queue));
 655
 656         switch (sort_order) {
 657         default: /* REV_SORT_IN_GRAPH_ORDER */
 658                 queue.compare = NULL;
 659                 break;
 660         case REV_SORT_BY_COMMIT_DATE:
 661                 queue.compare = compare_commits_by_commit_date;
 662                 break;
 663         case REV_SORT_BY_AUTHOR_DATE:
 664                 init_author_date_slab(&author_date);
 665                 queue.compare = compare_commits_by_author_date;
 666                 queue.cb_data = &author_date;
 667                 break;
 668         }
 669
 670         /* Mark them and clear the indegree */
 671         for (next = orig; next; next = next->next) {
 672                 struct commit *commit = next->item;
 673                 *(indegree_slab_at(&indegree, commit)) = 1;
 674                 /* also record the author dates, if needed */
 675                 if (sort_order == REV_SORT_BY_AUTHOR_DATE)
 676                         record_author_date(&author_date, commit);
 677         }
 678
 679         /* update the indegree */
 680         for (next = orig; next; next = next->next) {
 681                 struct commit_list *parents = next->item->parents;
 682                 while (parents) {
 683                         struct commit *parent = parents->item;
 684                         int *pi = indegree_slab_at(&indegree, parent);
 685
 686                         if (*pi)
 687                                 (*pi)++;
 688                         parents = parents->next;
 689                 }
 690         }
 691
 692         /*
 693          * find the tips
 694          *
 695          * tips are nodes not reachable from any other node in the list
 696          *
 697          * the tips serve as a starting set for the work queue.
 698          */
 699         for (next = orig; next; next = next->next) {
 700                 struct commit *commit = next->item;
 701
 702                 if (*(indegree_slab_at(&indegree, commit)) == 1)
 703                         prio_queue_put(&queue, commit);
 704         }
 705
 706         /*
 707          * This is unfortunate; the initial tips need to be shown
 708          * in the order given from the revision traversal machinery.
 709          */
 710         if (sort_order == REV_SORT_IN_GRAPH_ORDER)
 711                 prio_queue_reverse(&queue);
 712
 713         /* We no longer need the commit list */
 714         free_commit_list(orig);
 715
 716         pptr = list;
 717         *list = NULL;
 718         while ((commit = prio_queue_get(&queue)) != NULL) {
 719                 struct commit_list *parents;
 720
 721                 for (parents = commit->parents; parents ; parents = parents->next) {
 722                         struct commit *parent = parents->item;
 723                         int *pi = indegree_slab_at(&indegree, parent);
 724
 725                         if (!*pi)
 726                                 continue;
 727
 728                         /*
 729                          * parents are only enqueued for emission
 730                          * when all their children have been emitted thereby
 731                          * guaranteeing topological order.
 732                          */
 733                         if (--(*pi) == 1)
 734                                 prio_queue_put(&queue, parent);
 735                 }
 736                 /*
 737                  * all children of commit have already been
 738                  * emitted. we can emit it now.
 739                  */
 740                 *(indegree_slab_at(&indegree, commit)) = 0;
 741
 742                 pptr = &commit_list_insert(commit, pptr)->next;
 743         }
 744
 745         clear_indegree_slab(&indegree);
 746         clear_prio_queue(&queue);
 747         if (sort_order == REV_SORT_BY_AUTHOR_DATE)
 748                 clear_author_date_slab(&author_date);
 749 }
 750
 751 /* merge-base stuff */
 752
 753 /* Remember to update object flag allocation in object.h */
 754 #define PARENT1         (1u<<16)
 755 #define PARENT2         (1u<<17)
 756 #define STALE           (1u<<18)
 757 #define RESULT          (1u<<19)
 758
 759 static const unsigned all_flags = (PARENT1 | PARENT2 | STALE | RESULT);
 760
 761 static int queue_has_nonstale(struct prio_queue *queue)
 762 {
 763         int i;
 764         for (i = 0; i < queue->nr; i++) {
 765                 struct commit *commit = queue->array[i].data;
 766                 if (!(commit->object.flags & STALE))
 767                         return 1;
 768         }
 769         return 0;
 770 }
 771
 772 /* all input commits in one and twos[] must have been parsed! */
 773 static struct commit_list *paint_down_to_common(struct commit *one, int n, struct commit **twos)
 774 {
 775         struct prio_queue queue = { compare_commits_by_commit_date };
 776         struct commit_list *result = NULL;
 777         int i;
 778
 779         one->object.flags |= PARENT1;
 780         if (!n) {
 781                 commit_list_append(one, &result);
 782                 return result;
 783         }
 784         prio_queue_put(&queue, one);
 785
 786         for (i = 0; i < n; i++) {
 787                 twos[i]->object.flags |= PARENT2;
 788                 prio_queue_put(&queue, twos[i]);
 789         }
 790
 791         while (queue_has_nonstale(&queue)) {
 792                 struct commit *commit = prio_queue_get(&queue);
 793                 struct commit_list *parents;
 794                 int flags;
 795
 796                 flags = commit->object.flags & (PARENT1 | PARENT2 | STALE);
 797                 if (flags == (PARENT1 | PARENT2)) {
 798                         if (!(commit->object.flags & RESULT)) {
 799                                 commit->object.flags |= RESULT;
 800                                 commit_list_insert_by_date(commit, &result);
 801                         }
 802                         /* Mark parents of a found merge stale */
 803                         flags |= STALE;
 804                 }
 805                 parents = commit->parents;
 806                 while (parents) {
 807                         struct commit *p = parents->item;
 808                         parents = parents->next;
 809                         if ((p->object.flags & flags) == flags)
 810                                 continue;
 811                         if (parse_commit(p))
 812                                 return NULL;
 813                         p->object.flags |= flags;
 814                         prio_queue_put(&queue, p);
 815                 }
 816         }
 817
 818         clear_prio_queue(&queue);
 819         return result;
 820 }
 821
 822 static struct commit_list *merge_bases_many(struct commit *one, int n, struct commit **twos)
 823 {
 824         struct commit_list *list = NULL;
 825         struct commit_list *result = NULL;
 826         int i;
 827
 828         for (i = 0; i < n; i++) {
 829                 if (one == twos[i])
 830                         /*
 831                          * We do not mark this even with RESULT so we do not
 832                          * have to clean it up.
 833                          */
 834                         return commit_list_insert(one, &result);
 835         }
 836
 837         if (parse_commit(one))
 838                 return NULL;
 839         for (i = 0; i < n; i++) {
 840                 if (parse_commit(twos[i]))
 841                         return NULL;
 842         }
 843
 844         list = paint_down_to_common(one, n, twos);
 845
 846         while (list) {
 847                 struct commit *commit = pop_commit(&list);
 848                 if (!(commit->object.flags & STALE))
 849                         commit_list_insert_by_date(commit, &result);
 850         }
 851         return result;
 852 }
 853
 854 struct commit_list *get_octopus_merge_bases(struct commit_list *in)
 855 {
 856         struct commit_list *i, *j, *k, *ret = NULL;
 857
 858         if (!in)
 859                 return ret;
 860
 861         commit_list_insert(in->item, &ret);
 862
 863         for (i = in->next; i; i = i->next) {
 864                 struct commit_list *new = NULL, *end = NULL;
 865
 866                 for (j = ret; j; j = j->next) {
 867                         struct commit_list *bases;
 868                         bases = get_merge_bases(i->item, j->item);
 869                         if (!new)
 870                                 new = bases;
 871                         else
 872                                 end->next = bases;
 873                         for (k = bases; k; k = k->next)
 874                                 end = k;
 875                 }
 876                 ret = new;
 877         }
 878         return ret;
 879 }
 880
 881 static int remove_redundant(struct commit **array, int cnt)
 882 {
 883         /*
 884          * Some commit in the array may be an ancestor of
 885          * another commit.  Move such commit to the end of
 886          * the array, and return the number of commits that
 887          * are independent from each other.
 888          */
 889         struct commit **work;
 890         unsigned char *redundant;
 891         int *filled_index;
 892         int i, j, filled;
 893
 894         work = xcalloc(cnt, sizeof(*work));
 895         redundant = xcalloc(cnt, 1);
 896         ALLOC_ARRAY(filled_index, cnt - 1);
 897
 898         for (i = 0; i < cnt; i++)
 899                 parse_commit(array[i]);
 900         for (i = 0; i < cnt; i++) {
 901                 struct commit_list *common;
 902
 903                 if (redundant[i])
 904                         continue;
 905                 for (j = filled = 0; j < cnt; j++) {
 906                         if (i == j || redundant[j])
 907                                 continue;
 908                         filled_index[filled] = j;
 909                         work[filled++] = array[j];
 910                 }
 911                 common = paint_down_to_common(array[i], filled, work);
 912                 if (array[i]->object.flags & PARENT2)
 913                         redundant[i] = 1;
 914                 for (j = 0; j < filled; j++)
 915                         if (work[j]->object.flags & PARENT1)
 916                                 redundant[filled_index[j]] = 1;
 917                 clear_commit_marks(array[i], all_flags);
 918                 clear_commit_marks_many(filled, work, all_flags);
 919                 free_commit_list(common);
 920         }
 921
 922         /* Now collect the result */
 923         COPY_ARRAY(work, array, cnt);
 924         for (i = filled = 0; i < cnt; i++)
 925                 if (!redundant[i])
 926                         array[filled++] = work[i];
 927         for (j = filled, i = 0; i < cnt; i++)
 928                 if (redundant[i])
 929                         array[j++] = work[i];
 930         free(work);
 931         free(redundant);
 932         free(filled_index);
 933         return filled;
 934 }
 935
 936 static struct commit_list *get_merge_bases_many_0(struct commit *one,
 937                                                   int n,
 938                                                   struct commit **twos,
 939                                                   int cleanup)
 940 {
 941         struct commit_list *list;
 942         struct commit **rslt;
 943         struct commit_list *result;
 944         int cnt, i;
 945
 946         result = merge_bases_many(one, n, twos);
 947         for (i = 0; i < n; i++) {
 948                 if (one == twos[i])
 949                         return result;
 950         }
 951         if (!result || !result->next) {
 952                 if (cleanup) {
 953                         clear_commit_marks(one, all_flags);
 954                         clear_commit_marks_many(n, twos, all_flags);
 955                 }
 956                 return result;
 957         }
 958
 959         /* There are more than one */
 960         cnt = commit_list_count(result);
 961         rslt = xcalloc(cnt, sizeof(*rslt));
 962         for (list = result, i = 0; list; list = list->next)
 963                 rslt[i++] = list->item;
 964         free_commit_list(result);
 965
 966         clear_commit_marks(one, all_flags);
 967         clear_commit_marks_many(n, twos, all_flags);
 968
 969         cnt = remove_redundant(rslt, cnt);
 970         result = NULL;
 971         for (i = 0; i < cnt; i++)
 972                 commit_list_insert_by_date(rslt[i], &result);
 973         free(rslt);
 974         return result;
 975 }
 976
 977 struct commit_list *get_merge_bases_many(struct commit *one,
 978                                          int n,
 979                                          struct commit **twos)
 980 {
 981         return get_merge_bases_many_0(one, n, twos, 1);
 982 }
 983
 984 struct commit_list *get_merge_bases_many_dirty(struct commit *one,
 985                                                int n,
 986                                                struct commit **twos)
 987 {
 988         return get_merge_bases_many_0(one, n, twos, 0);
 989 }
 990
 991 struct commit_list *get_merge_bases(struct commit *one, struct commit *two)
 992 {
 993         return get_merge_bases_many_0(one, 1, &two, 1);
 994 }
 995
 996 /*
 997  * Is "commit" a descendant of one of the elements on the "with_commit" list?
 998  */
 999 int is_descendant_of(struct commit *commit, struct commit_list *with_commit)
1000 {
1001         if (!with_commit)
1002                 return 1;
1003         while (with_commit) {
1004                 struct commit *other;
1005
1006                 other = with_commit->item;
1007                 with_commit = with_commit->next;
1008                 if (in_merge_bases(other, commit))
1009                         return 1;
1010         }
1011         return 0;
1012 }
1013
1014 /*
1015  * Is "commit" an ancestor of one of the "references"?
1016  */
1017 int in_merge_bases_many(struct commit *commit, int nr_reference, struct commit **reference)
1018 {
1019         struct commit_list *bases;
1020         int ret = 0, i;
1021
1022         if (parse_commit(commit))
1023                 return ret;
1024         for (i = 0; i < nr_reference; i++)
1025                 if (parse_commit(reference[i]))
1026                         return ret;
1027
1028         bases = paint_down_to_common(commit, nr_reference, reference);
1029         if (commit->object.flags & PARENT2)
1030                 ret = 1;
1031         clear_commit_marks(commit, all_flags);
1032         clear_commit_marks_many(nr_reference, reference, all_flags);
1033         free_commit_list(bases);
1034         return ret;
1035 }
1036
1037 /*
1038  * Is "commit" an ancestor of (i.e. reachable from) the "reference"?
1039  */
1040 int in_merge_bases(struct commit *commit, struct commit *reference)
1041 {
1042         return in_merge_bases_many(commit, 1, &reference);
1043 }
1044
1045 struct commit_list *reduce_heads(struct commit_list *heads)
1046 {
1047         struct commit_list *p;
1048         struct commit_list *result = NULL, **tail = &result;
1049         struct commit **array;
1050         int num_head, i;
1051
1052         if (!heads)
1053                 return NULL;
1054
1055         /* Uniquify */
1056         for (p = heads; p; p = p->next)
1057                 p->item->object.flags &= ~STALE;
1058         for (p = heads, num_head = 0; p; p = p->next) {
1059                 if (p->item->object.flags & STALE)
1060                         continue;
1061                 p->item->object.flags |= STALE;
1062                 num_head++;
1063         }
1064         array = xcalloc(num_head, sizeof(*array));
1065         for (p = heads, i = 0; p; p = p->next) {
1066                 if (p->item->object.flags & STALE) {
1067                         array[i++] = p->item;
1068                         p->item->object.flags &= ~STALE;
1069                 }
1070         }
1071         num_head = remove_redundant(array, num_head);
1072         for (i = 0; i < num_head; i++)
1073                 tail = &commit_list_insert(array[i], tail)->next;
1074         free(array);
1075         return result;
1076 }
1077
1078 void reduce_heads_replace(struct commit_list **heads)
1079 {
1080         struct commit_list *result = reduce_heads(*heads);
1081         free_commit_list(*heads);
1082         *heads = result;
1083 }
1084
1085 static const char gpg_sig_header[] = "gpgsig";
1086 static const int gpg_sig_header_len = sizeof(gpg_sig_header) - 1;
1087
1088 static int do_sign_commit(struct strbuf *buf, const char *keyid)
1089 {
1090         struct strbuf sig = STRBUF_INIT;
1091         int inspos, copypos;
1092         const char *eoh;
1093
1094         /* find the end of the header */
1095         eoh = strstr(buf->buf, "\n\n");
1096         if (!eoh)
1097                 inspos = buf->len;
1098         else
1099                 inspos = eoh - buf->buf + 1;
1100
1101         if (!keyid || !*keyid)
1102                 keyid = get_signing_key();
1103         if (sign_buffer(buf, &sig, keyid)) {
1104                 strbuf_release(&sig);
1105                 return -1;
1106         }
1107
1108         for (copypos = 0; sig.buf[copypos]; ) {
1109                 const char *bol = sig.buf + copypos;
1110                 const char *eol = strchrnul(bol, '\n');
1111                 int len = (eol - bol) + !!*eol;
1112
1113                 if (!copypos) {
1114                         strbuf_insert(buf, inspos, gpg_sig_header, gpg_sig_header_len);
1115                         inspos += gpg_sig_header_len;
1116                 }
1117                 strbuf_insert(buf, inspos++, " ", 1);
1118                 strbuf_insert(buf, inspos, bol, len);
1119                 inspos += len;
1120                 copypos += len;
1121         }
1122         strbuf_release(&sig);
1123         return 0;
1124 }
1125
1126 int parse_signed_commit(const struct commit *commit,
1127                         struct strbuf *payload, struct strbuf *signature)
1128 {
1129
1130         unsigned long size;
1131         const char *buffer = get_commit_buffer(commit, &size);
1132         int in_signature, saw_signature = -1;
1133         const char *line, *tail;
1134
1135         line = buffer;
1136         tail = buffer + size;
1137         in_signature = 0;
1138         saw_signature = 0;
1139         while (line < tail) {
1140                 const char *sig = NULL;
1141                 const char *next = memchr(line, '\n', tail - line);
1142
1143                 next = next ? next + 1 : tail;
1144                 if (in_signature && line[0] == ' ')
1145                         sig = line + 1;
1146                 else if (starts_with(line, gpg_sig_header) &&
1147                          line[gpg_sig_header_len] == ' ')
1148                         sig = line + gpg_sig_header_len + 1;
1149                 if (sig) {
1150                         strbuf_add(signature, sig, next - sig);
1151                         saw_signature = 1;
1152                         in_signature = 1;
1153                 } else {
1154                         if (*line == '\n')
1155                                 /* dump the whole remainder of the buffer */
1156                                 next = tail;
1157                         strbuf_add(payload, line, next - line);
1158                         in_signature = 0;
1159                 }
1160                 line = next;
1161         }
1162         unuse_commit_buffer(commit, buffer);
1163         return saw_signature;
1164 }
1165
1166 int remove_signature(struct strbuf *buf)
1167 {
1168         const char *line = buf->buf;
1169         const char *tail = buf->buf + buf->len;
1170         int in_signature = 0;
1171         const char *sig_start = NULL;
1172         const char *sig_end = NULL;
1173
1174         while (line < tail) {
1175                 const char *next = memchr(line, '\n', tail - line);
1176                 next = next ? next + 1 : tail;
1177
1178                 if (in_signature && line[0] == ' ')
1179                         sig_end = next;
1180                 else if (starts_with(line, gpg_sig_header) &&
1181                          line[gpg_sig_header_len] == ' ') {
1182                         sig_start = line;
1183                         sig_end = next;
1184                         in_signature = 1;
1185                 } else {
1186                         if (*line == '\n')
1187                                 /* dump the whole remainder of the buffer */
1188                                 next = tail;
1189                         in_signature = 0;
1190                 }
1191                 line = next;
1192         }
1193
1194         if (sig_start)
1195                 strbuf_remove(buf, sig_start - buf->buf, sig_end - sig_start);
1196
1197         return sig_start != NULL;
1198 }
1199
1200 static void handle_signed_tag(struct commit *parent, struct commit_extra_header ***tail)
1201 {
1202         struct merge_remote_desc *desc;
1203         struct commit_extra_header *mergetag;
1204         char *buf;
1205         unsigned long size, len;
1206         enum object_type type;
1207
1208         desc = merge_remote_util(parent);
1209         if (!desc || !desc->obj)
1210                 return;
1211         buf = read_sha1_file(desc->obj->oid.hash, &type, &size);
1212         if (!buf || type != OBJ_TAG)
1213                 goto free_return;
1214         len = parse_signature(buf, size);
1215         if (size == len)
1216                 goto free_return;
1217         /*
1218          * We could verify this signature and either omit the tag when
1219          * it does not validate, but the integrator may not have the
1220          * public key of the signer of the tag he is merging, while a
1221          * later auditor may have it while auditing, so let's not run
1222          * verify-signed-buffer here for now...
1223          *
1224          * if (verify_signed_buffer(buf, len, buf + len, size - len, ...))
1225          *      warn("warning: signed tag unverified.");
1226          */
1227         mergetag = xcalloc(1, sizeof(*mergetag));
1228         mergetag->key = xstrdup("mergetag");
1229         mergetag->value = buf;
1230         mergetag->len = size;
1231
1232         **tail = mergetag;
1233         *tail = &mergetag->next;
1234         return;
1235
1236 free_return:
1237         free(buf);
1238 }
1239
1240 int check_commit_signature(const struct commit *commit, struct signature_check *sigc)
1241 {
1242         struct strbuf payload = STRBUF_INIT;
1243         struct strbuf signature = STRBUF_INIT;
1244         int ret = 1;
1245
1246         sigc->result = 'N';
1247
1248         if (parse_signed_commit(commit, &payload, &signature) <= 0)
1249                 goto out;
1250         ret = check_signature(payload.buf, payload.len, signature.buf,
1251                 signature.len, sigc);
1252
1253  out:
1254         strbuf_release(&payload);
1255         strbuf_release(&signature);
1256
1257         return ret;
1258 }
1259
1260
1261
1262 void append_merge_tag_headers(struct commit_list *parents,
1263                               struct commit_extra_header ***tail)
1264 {
1265         while (parents) {
1266                 struct commit *parent = parents->item;
1267                 handle_signed_tag(parent, tail);
1268                 parents = parents->next;
1269         }
1270 }
1271
1272 static void add_extra_header(struct strbuf *buffer,
1273                              struct commit_extra_header *extra)
1274 {
1275         strbuf_addstr(buffer, extra->key);
1276         if (extra->len)
1277                 strbuf_add_lines(buffer, " ", extra->value, extra->len);
1278         else
1279                 strbuf_addch(buffer, '\n');
1280 }
1281
1282 struct commit_extra_header *read_commit_extra_headers(struct commit *commit,
1283                                                       const char **exclude)
1284 {
1285         struct commit_extra_header *extra = NULL;
1286         unsigned long size;
1287         const char *buffer = get_commit_buffer(commit, &size);
1288         extra = read_commit_extra_header_lines(buffer, size, exclude);
1289         unuse_commit_buffer(commit, buffer);
1290         return extra;
1291 }
1292
1293 void for_each_mergetag(each_mergetag_fn fn, struct commit *commit, void *data)
1294 {
1295         struct commit_extra_header *extra, *to_free;
1296
1297         to_free = read_commit_extra_headers(commit, NULL);
1298         for (extra = to_free; extra; extra = extra->next) {
1299                 if (strcmp(extra->key, "mergetag"))
1300                         continue; /* not a merge tag */
1301                 fn(commit, extra, data);
1302         }
1303         free_commit_extra_headers(to_free);
1304 }
1305
1306 static inline int standard_header_field(const char *field, size_t len)
1307 {
1308         return ((len == 4 && !memcmp(field, "tree", 4)) ||
1309                 (len == 6 && !memcmp(field, "parent", 6)) ||
1310                 (len == 6 && !memcmp(field, "author", 6)) ||
1311                 (len == 9 && !memcmp(field, "committer", 9)) ||
1312                 (len == 8 && !memcmp(field, "encoding", 8)));
1313 }
1314
1315 static int excluded_header_field(const char *field, size_t len, const char **exclude)
1316 {
1317         if (!exclude)
1318                 return 0;
1319
1320         while (*exclude) {
1321                 size_t xlen = strlen(*exclude);
1322                 if (len == xlen && !memcmp(field, *exclude, xlen))
1323                         return 1;
1324                 exclude++;
1325         }
1326         return 0;
1327 }
1328
1329 static struct commit_extra_header *read_commit_extra_header_lines(
1330         const char *buffer, size_t size,
1331         const char **exclude)
1332 {
1333         struct commit_extra_header *extra = NULL, **tail = &extra, *it = NULL;
1334         const char *line, *next, *eof, *eob;
1335         struct strbuf buf = STRBUF_INIT;
1336
1337         for (line = buffer, eob = line + size;
1338              line < eob && *line != '\n';
1339              line = next) {
1340                 next = memchr(line, '\n', eob - line);
1341                 next = next ? next + 1 : eob;
1342                 if (*line == ' ') {
1343                         /* continuation */
1344                         if (it)
1345                                 strbuf_add(&buf, line + 1, next - (line + 1));
1346                         continue;
1347                 }
1348                 if (it)
1349                         it->value = strbuf_detach(&buf, &it->len);
1350                 strbuf_reset(&buf);
1351                 it = NULL;
1352
1353                 eof = memchr(line, ' ', next - line);
1354                 if (!eof)
1355                         eof = next;
1356                 else if (standard_header_field(line, eof - line) ||
1357                          excluded_header_field(line, eof - line, exclude))
1358                         continue;
1359
1360                 it = xcalloc(1, sizeof(*it));
1361                 it->key = xmemdupz(line, eof-line);
1362                 *tail = it;
1363                 tail = &it->next;
1364                 if (eof + 1 < next)
1365                         strbuf_add(&buf, eof + 1, next - (eof + 1));
1366         }
1367         if (it)
1368                 it->value = strbuf_detach(&buf, &it->len);
1369         return extra;
1370 }
1371
1372 void free_commit_extra_headers(struct commit_extra_header *extra)
1373 {
1374         while (extra) {
1375                 struct commit_extra_header *next = extra->next;
1376                 free(extra->key);
1377                 free(extra->value);
1378                 free(extra);
1379                 extra = next;
1380         }
1381 }
1382
1383 int commit_tree(const char *msg, size_t msg_len,
1384                 const unsigned char *tree,
1385                 struct commit_list *parents, unsigned char *ret,
1386                 const char *author, const char *sign_commit)
1387 {
1388         struct commit_extra_header *extra = NULL, **tail = &extra;
1389         int result;
1390
1391         append_merge_tag_headers(parents, &tail);
1392         result = commit_tree_extended(msg, msg_len, tree, parents, ret,
1393                                       author, sign_commit, extra);
1394         free_commit_extra_headers(extra);
1395         return result;
1396 }
1397
1398 static int find_invalid_utf8(const char *buf, int len)
1399 {
1400         int offset = 0;
1401         static const unsigned int max_codepoint[] = {
1402                 0x7f, 0x7ff, 0xffff, 0x10ffff
1403         };
1404
1405         while (len) {
1406                 unsigned char c = *buf++;
1407                 int bytes, bad_offset;
1408                 unsigned int codepoint;
1409                 unsigned int min_val, max_val;
1410
1411                 len--;
1412                 offset++;
1413
1414                 /* Simple US-ASCII? No worries. */
1415                 if (c < 0x80)
1416                         continue;
1417
1418                 bad_offset = offset-1;
1419
1420                 /*
1421                  * Count how many more high bits set: that's how
1422                  * many more bytes this sequence should have.
1423                  */
1424                 bytes = 0;
1425                 while (c & 0x40) {
1426                         c <<= 1;
1427                         bytes++;
1428                 }
1429
1430                 /*
1431                  * Must be between 1 and 3 more bytes.  Longer sequences result in
1432                  * codepoints beyond U+10FFFF, which are guaranteed never to exist.
1433                  */
1434                 if (bytes < 1 || 3 < bytes)
1435                         return bad_offset;
1436
1437                 /* Do we *have* that many bytes? */
1438                 if (len < bytes)
1439                         return bad_offset;
1440
1441                 /*
1442                  * Place the encoded bits at the bottom of the value and compute the
1443                  * valid range.
1444                  */
1445                 codepoint = (c & 0x7f) >> bytes;
1446                 min_val = max_codepoint[bytes-1] + 1;
1447                 max_val = max_codepoint[bytes];
1448
1449                 offset += bytes;
1450                 len -= bytes;
1451
1452                 /* And verify that they are good continuation bytes */
1453                 do {
1454                         codepoint <<= 6;
1455                         codepoint |= *buf & 0x3f;
1456                         if ((*buf++ & 0xc0) != 0x80)
1457                                 return bad_offset;
1458                 } while (--bytes);
1459
1460                 /* Reject codepoints that are out of range for the sequence length. */
1461                 if (codepoint < min_val || codepoint > max_val)
1462                         return bad_offset;
1463                 /* Surrogates are only for UTF-16 and cannot be encoded in UTF-8. */
1464                 if ((codepoint & 0x1ff800) == 0xd800)
1465                         return bad_offset;
1466                 /* U+xxFFFE and U+xxFFFF are guaranteed non-characters. */
1467                 if ((codepoint & 0xfffe) == 0xfffe)
1468                         return bad_offset;
1469                 /* So are anything in the range U+FDD0..U+FDEF. */
1470                 if (codepoint >= 0xfdd0 && codepoint <= 0xfdef)
1471                         return bad_offset;
1472         }
1473         return -1;
1474 }
1475
1476 /*
1477  * This verifies that the buffer is in proper utf8 format.
1478  *
1479  * If it isn't, it assumes any non-utf8 characters are Latin1,
1480  * and does the conversion.
1481  */
1482 static int verify_utf8(struct strbuf *buf)
1483 {
1484         int ok = 1;
1485         long pos = 0;
1486
1487         for (;;) {
1488                 int bad;
1489                 unsigned char c;
1490                 unsigned char replace[2];
1491
1492                 bad = find_invalid_utf8(buf->buf + pos, buf->len - pos);
1493                 if (bad < 0)
1494                         return ok;
1495                 pos += bad;
1496                 ok = 0;
1497                 c = buf->buf[pos];
1498                 strbuf_remove(buf, pos, 1);
1499
1500                 /* We know 'c' must be in the range 128-255 */
1501                 replace[0] = 0xc0 + (c >> 6);
1502                 replace[1] = 0x80 + (c & 0x3f);
1503                 strbuf_insert(buf, pos, replace, 2);
1504                 pos += 2;
1505         }
1506 }
1507
1508 static const char commit_utf8_warn[] =
1509 N_("Warning: commit message did not conform to UTF-8.\n"
1510    "You may want to amend it after fixing the message, or set the config\n"
1511    "variable i18n.commitencoding to the encoding your project uses.\n");
1512
1513 int commit_tree_extended(const char *msg, size_t msg_len,
1514                          const unsigned char *tree,
1515                          struct commit_list *parents, unsigned char *ret,
1516                          const char *author, const char *sign_commit,
1517                          struct commit_extra_header *extra)
1518 {
1519         int result;
1520         int encoding_is_utf8;
1521         struct strbuf buffer;
1522
1523         assert_sha1_type(tree, OBJ_TREE);
1524
1525         if (memchr(msg, '\0', msg_len))
1526                 return error("a NUL byte in commit log message not allowed.");
1527
1528         /* Not having i18n.commitencoding is the same as having utf-8 */
1529         encoding_is_utf8 = is_encoding_utf8(git_commit_encoding);
1530
1531         strbuf_init(&buffer, 8192); /* should avoid reallocs for the headers */
1532         strbuf_addf(&buffer, "tree %s\n", sha1_to_hex(tree));
1533
1534         /*
1535          * NOTE! This ordering means that the same exact tree merged with a
1536          * different order of parents will be a _different_ changeset even
1537          * if everything else stays the same.
1538          */
1539         while (parents) {
1540                 struct commit *parent = pop_commit(&parents);
1541                 strbuf_addf(&buffer, "parent %s\n",
1542                             oid_to_hex(&parent->object.oid));
1543         }
1544
1545         /* Person/date information */
1546         if (!author)
1547                 author = git_author_info(IDENT_STRICT);
1548         strbuf_addf(&buffer, "author %s\n", author);
1549         strbuf_addf(&buffer, "committer %s\n", git_committer_info(IDENT_STRICT));
1550         if (!encoding_is_utf8)
1551                 strbuf_addf(&buffer, "encoding %s\n", git_commit_encoding);
1552
1553         while (extra) {
1554                 add_extra_header(&buffer, extra);
1555                 extra = extra->next;
1556         }
1557         strbuf_addch(&buffer, '\n');
1558
1559         /* And add the comment */
1560         strbuf_add(&buffer, msg, msg_len);
1561
1562         /* And check the encoding */
1563         if (encoding_is_utf8 && !verify_utf8(&buffer))
1564                 fprintf(stderr, _(commit_utf8_warn));
1565
1566         if (sign_commit && do_sign_commit(&buffer, sign_commit)) {
1567                 result = -1;
1568                 goto out;
1569         }
1570
1571         result = write_sha1_file(buffer.buf, buffer.len, commit_type, ret);
1572 out:
1573         strbuf_release(&buffer);
1574         return result;
1575 }
1576
1577 void set_merge_remote_desc(struct commit *commit,
1578                            const char *name, struct object *obj)
1579 {
1580         struct merge_remote_desc *desc;
1581         FLEX_ALLOC_STR(desc, name, name);
1582         desc->obj = obj;
1583         commit->util = desc;
1584 }
1585
1586 struct commit *get_merge_parent(const char *name)
1587 {
1588         struct object *obj;
1589         struct commit *commit;
1590         struct object_id oid;
1591         if (get_oid(name, &oid))
1592                 return NULL;
1593         obj = parse_object(&oid);
1594         commit = (struct commit *)peel_to_type(name, 0, obj, OBJ_COMMIT);
1595         if (commit && !commit->util)
1596                 set_merge_remote_desc(commit, name, obj);
1597         return commit;
1598 }
1599
1600 /*
1601  * Append a commit to the end of the commit_list.
1602  *
1603  * next starts by pointing to the variable that holds the head of an
1604  * empty commit_list, and is updated to point to the "next" field of
1605  * the last item on the list as new commits are appended.
1606  *
1607  * Usage example:
1608  *
1609  *     struct commit_list *list;
1610  *     struct commit_list **next = &list;
1611  *
1612  *     next = commit_list_append(c1, next);
1613  *     next = commit_list_append(c2, next);
1614  *     assert(commit_list_count(list) == 2);
1615  *     return list;
1616  */
1617 struct commit_list **commit_list_append(struct commit *commit,
1618                                         struct commit_list **next)
1619 {
1620         struct commit_list *new = xmalloc(sizeof(struct commit_list));
1621         new->item = commit;
1622         *next = new;
1623         new->next = NULL;
1624         return &new->next;
1625 }
1626
1627 const char *find_commit_header(const char *msg, const char *key, size_t *out_len)
1628 {
1629         int key_len = strlen(key);
1630         const char *line = msg;
1631
1632         while (line) {
1633                 const char *eol = strchrnul(line, '\n');
1634
1635                 if (line == eol)
1636                         return NULL;
1637
1638                 if (eol - line > key_len &&
1639                     !strncmp(line, key, key_len) &&
1640                     line[key_len] == ' ') {
1641                         *out_len = eol - line - key_len - 1;
1642                         return line + key_len + 1;
1643                 }
1644                 line = *eol ? eol + 1 : NULL;
1645         }
1646         return NULL;
1647 }
1648
1649 /*
1650  * Inspect the given string and determine the true "end" of the log message, in
1651  * order to find where to put a new Signed-off-by: line.  Ignored are
1652  * trailing comment lines and blank lines.  To support "git commit -s
1653  * --amend" on an existing commit, we also ignore "Conflicts:".  To
1654  * support "git commit -v", we truncate at cut lines.
1655  *
1656  * Returns the number of bytes from the tail to ignore, to be fed as
1657  * the second parameter to append_signoff().
1658  */
1659 int ignore_non_trailer(const char *buf, size_t len)
1660 {
1661         int boc = 0;
1662         int bol = 0;
1663         int in_old_conflicts_block = 0;
1664         size_t cutoff = wt_status_locate_end(buf, len);
1665
1666         while (bol < cutoff) {
1667                 const char *next_line = memchr(buf + bol, '\n', len - bol);
1668
1669                 if (!next_line)
1670                         next_line = buf + len;
1671                 else
1672                         next_line++;
1673
1674                 if (buf[bol] == comment_line_char || buf[bol] == '\n') {
1675                         /* is this the first of the run of comments? */
1676                         if (!boc)
1677                                 boc = bol;
1678                         /* otherwise, it is just continuing */
1679                 } else if (starts_with(buf + bol, "Conflicts:\n")) {
1680                         in_old_conflicts_block = 1;
1681                         if (!boc)
1682                                 boc = bol;
1683                 } else if (in_old_conflicts_block && buf[bol] == '\t') {
1684                         ; /* a pathname in the conflicts block */
1685                 } else if (boc) {
1686                         /* the previous was not trailing comment */
1687                         boc = 0;
1688                         in_old_conflicts_block = 0;
1689                 }
1690                 bol = next_line - buf;
1691         }
1692         return boc ? len - boc : len - cutoff;
1693 }