commit.c

   1 #include "cache.h"
   2 #include "tag.h"
   3 #include "commit.h"
   4 #include "repository.h"
   5 #include "object-store.h"
   6 #include "pkt-line.h"
   7 #include "utf8.h"
   8 #include "diff.h"
   9 #include "revision.h"
  10 #include "notes.h"
  11 #include "alloc.h"
  12 #include "gpg-interface.h"
  13 #include "mergesort.h"
  14 #include "commit-slab.h"
  15 #include "prio-queue.h"
  16 #include "sha1-lookup.h"
  17 #include "wt-status.h"
  18
  19 static struct commit_extra_header *read_commit_extra_header_lines(const char *buf, size_t len, const char **);
  20
  21 int save_commit_buffer = 1;
  22
  23 const char *commit_type = "commit";
  24
  25 struct commit *lookup_commit_reference_gently(const struct object_id *oid,
  26                                               int quiet)
  27 {
  28         struct object *obj = deref_tag(parse_object(oid), NULL, 0);
  29
  30         if (!obj)
  31                 return NULL;
  32         return object_as_type(obj, OBJ_COMMIT, quiet);
  33 }
  34
  35 struct commit *lookup_commit_reference(const struct object_id *oid)
  36 {
  37         return lookup_commit_reference_gently(oid, 0);
  38 }
  39
  40 struct commit *lookup_commit_or_die(const struct object_id *oid, const char *ref_name)
  41 {
  42         struct commit *c = lookup_commit_reference(oid);
  43         if (!c)
  44                 die(_("could not parse %s"), ref_name);
  45         if (oidcmp(oid, &c->object.oid)) {
  46                 warning(_("%s %s is not a commit!"),
  47                         ref_name, oid_to_hex(oid));
  48         }
  49         return c;
  50 }
  51
  52 struct commit *lookup_commit(const struct object_id *oid)
  53 {
  54         struct object *obj = lookup_object(oid->hash);
  55         if (!obj)
  56                 return create_object(the_repository, oid->hash,
  57                                      alloc_commit_node(the_repository));
  58         return object_as_type(obj, OBJ_COMMIT, 0);
  59 }
  60
  61 struct commit *lookup_commit_reference_by_name(const char *name)
  62 {
  63         struct object_id oid;
  64         struct commit *commit;
  65
  66         if (get_oid_committish(name, &oid))
  67                 return NULL;
  68         commit = lookup_commit_reference(&oid);
  69         if (parse_commit(commit))
  70                 return NULL;
  71         return commit;
  72 }
  73
  74 static timestamp_t parse_commit_date(const char *buf, const char *tail)
  75 {
  76         const char *dateptr;
  77
  78         if (buf + 6 >= tail)
  79                 return 0;
  80         if (memcmp(buf, "author", 6))
  81                 return 0;
  82         while (buf < tail && *buf++ != '\n')
  83                 /* nada */;
  84         if (buf + 9 >= tail)
  85                 return 0;
  86         if (memcmp(buf, "committer", 9))
  87                 return 0;
  88         while (buf < tail && *buf++ != '>')
  89                 /* nada */;
  90         if (buf >= tail)
  91                 return 0;
  92         dateptr = buf;
  93         while (buf < tail && *buf++ != '\n')
  94                 /* nada */;
  95         if (buf >= tail)
  96                 return 0;
  97         /* dateptr < buf && buf[-1] == '\n', so parsing will stop at buf-1 */
  98         return parse_timestamp(dateptr, NULL, 10);
  99 }
 100
 101 static const unsigned char *commit_graft_sha1_access(size_t index, void *table)
 102 {
 103         struct commit_graft **commit_graft_table = table;
 104         return commit_graft_table[index]->oid.hash;
 105 }
 106
 107 #define commit_graft_pos(r, s) commit_graft_pos_##r(s)
 108 static int commit_graft_pos_the_repository(const unsigned char *sha1)
 109 {
 110         return sha1_pos(sha1, the_repository->parsed_objects->grafts,
 111                         the_repository->parsed_objects->grafts_nr,
 112                         commit_graft_sha1_access);
 113 }
 114
 115 int register_commit_graft_the_repository(struct commit_graft *graft, int ignore_dups)
 116 {
 117         int pos = commit_graft_pos(the_repository, graft->oid.hash);
 118
 119         if (0 <= pos) {
 120                 if (ignore_dups)
 121                         free(graft);
 122                 else {
 123                         free(the_repository->parsed_objects->grafts[pos]);
 124                         the_repository->parsed_objects->grafts[pos] = graft;
 125                 }
 126                 return 1;
 127         }
 128         pos = -pos - 1;
 129         ALLOC_GROW(the_repository->parsed_objects->grafts,
 130                    the_repository->parsed_objects->grafts_nr + 1,
 131                    the_repository->parsed_objects->grafts_alloc);
 132         the_repository->parsed_objects->grafts_nr++;
 133         if (pos < the_repository->parsed_objects->grafts_nr)
 134                 memmove(the_repository->parsed_objects->grafts + pos + 1,
 135                         the_repository->parsed_objects->grafts + pos,
 136                         (the_repository->parsed_objects->grafts_nr - pos - 1) *
 137                         sizeof(*the_repository->parsed_objects->grafts));
 138         the_repository->parsed_objects->grafts[pos] = graft;
 139         return 0;
 140 }
 141
 142 struct commit_graft *read_graft_line(struct strbuf *line)
 143 {
 144         /* The format is just "Commit Parent1 Parent2 ...\n" */
 145         int i, phase;
 146         const char *tail = NULL;
 147         struct commit_graft *graft = NULL;
 148         struct object_id dummy_oid, *oid;
 149
 150         strbuf_rtrim(line);
 151         if (!line->len || line->buf[0] == '#')
 152                 return NULL;
 153         /*
 154          * phase 0 verifies line, counts hashes in line and allocates graft
 155          * phase 1 fills graft
 156          */
 157         for (phase = 0; phase < 2; phase++) {
 158                 oid = graft ? &graft->oid : &dummy_oid;
 159                 if (parse_oid_hex(line->buf, oid, &tail))
 160                         goto bad_graft_data;
 161                 for (i = 0; *tail != '\0'; i++) {
 162                         oid = graft ? &graft->parent[i] : &dummy_oid;
 163                         if (!isspace(*tail++) || parse_oid_hex(tail, oid, &tail))
 164                                 goto bad_graft_data;
 165                 }
 166                 if (!graft) {
 167                         graft = xmalloc(st_add(sizeof(*graft),
 168                                                st_mult(sizeof(struct object_id), i)));
 169                         graft->nr_parent = i;
 170                 }
 171         }
 172         return graft;
 173
 174 bad_graft_data:
 175         error("bad graft data: %s", line->buf);
 176         assert(!graft);
 177         return NULL;
 178 }
 179
 180 #define read_graft_file(r, f) read_graft_file_##r(f)
 181 static int read_graft_file_the_repository(const char *graft_file)
 182 {
 183         FILE *fp = fopen_or_warn(graft_file, "r");
 184         struct strbuf buf = STRBUF_INIT;
 185         if (!fp)
 186                 return -1;
 187         while (!strbuf_getwholeline(&buf, fp, '\n')) {
 188                 /* The format is just "Commit Parent1 Parent2 ...\n" */
 189                 struct commit_graft *graft = read_graft_line(&buf);
 190                 if (!graft)
 191                         continue;
 192                 if (register_commit_graft(the_repository, graft, 1))
 193                         error("duplicate graft data: %s", buf.buf);
 194         }
 195         fclose(fp);
 196         strbuf_release(&buf);
 197         return 0;
 198 }
 199
 200 #define prepare_commit_graft(r) prepare_commit_graft_##r()
 201 static void prepare_commit_graft_the_repository(void)
 202 {
 203         static int commit_graft_prepared;
 204         char *graft_file;
 205
 206         if (commit_graft_prepared)
 207                 return;
 208         graft_file = get_graft_file();
 209         read_graft_file(the_repository, graft_file);
 210         /* make sure shallows are read */
 211         is_repository_shallow();
 212         commit_graft_prepared = 1;
 213 }
 214
 215 struct commit_graft *lookup_commit_graft_the_repository(const struct object_id *oid)
 216 {
 217         int pos;
 218         prepare_commit_graft(the_repository);
 219         pos = commit_graft_pos(the_repository, oid->hash);
 220         if (pos < 0)
 221                 return NULL;
 222         return the_repository->parsed_objects->grafts[pos];
 223 }
 224
 225 int for_each_commit_graft(each_commit_graft_fn fn, void *cb_data)
 226 {
 227         int i, ret;
 228         for (i = ret = 0; i < the_repository->parsed_objects->grafts_nr && !ret; i++)
 229                 ret = fn(the_repository->parsed_objects->grafts[i], cb_data);
 230         return ret;
 231 }
 232
 233 int unregister_shallow(const struct object_id *oid)
 234 {
 235         int pos = commit_graft_pos(the_repository, oid->hash);
 236         if (pos < 0)
 237                 return -1;
 238         if (pos + 1 < the_repository->parsed_objects->grafts_nr)
 239                 MOVE_ARRAY(the_repository->parsed_objects->grafts + pos,
 240                            the_repository->parsed_objects->grafts + pos + 1,
 241                            the_repository->parsed_objects->grafts_nr - pos - 1);
 242         the_repository->parsed_objects->grafts_nr--;
 243         return 0;
 244 }
 245
 246 struct commit_buffer {
 247         void *buffer;
 248         unsigned long size;
 249 };
 250 define_commit_slab(buffer_slab, struct commit_buffer);
 251 static struct buffer_slab buffer_slab = COMMIT_SLAB_INIT(1, buffer_slab);
 252
 253 void set_commit_buffer(struct commit *commit, void *buffer, unsigned long size)
 254 {
 255         struct commit_buffer *v = buffer_slab_at(&buffer_slab, commit);
 256         v->buffer = buffer;
 257         v->size = size;
 258 }
 259
 260 const void *get_cached_commit_buffer(const struct commit *commit, unsigned long *sizep)
 261 {
 262         struct commit_buffer *v = buffer_slab_peek(&buffer_slab, commit);
 263         if (!v) {
 264                 if (sizep)
 265                         *sizep = 0;
 266                 return NULL;
 267         }
 268         if (sizep)
 269                 *sizep = v->size;
 270         return v->buffer;
 271 }
 272
 273 const void *get_commit_buffer(const struct commit *commit, unsigned long *sizep)
 274 {
 275         const void *ret = get_cached_commit_buffer(commit, sizep);
 276         if (!ret) {
 277                 enum object_type type;
 278                 unsigned long size;
 279                 ret = read_object_file(&commit->object.oid, &type, &size);
 280                 if (!ret)
 281                         die("cannot read commit object %s",
 282                             oid_to_hex(&commit->object.oid));
 283                 if (type != OBJ_COMMIT)
 284                         die("expected commit for %s, got %s",
 285                             oid_to_hex(&commit->object.oid), type_name(type));
 286                 if (sizep)
 287                         *sizep = size;
 288         }
 289         return ret;
 290 }
 291
 292 void unuse_commit_buffer(const struct commit *commit, const void *buffer)
 293 {
 294         struct commit_buffer *v = buffer_slab_peek(&buffer_slab, commit);
 295         if (!(v && v->buffer == buffer))
 296                 free((void *)buffer);
 297 }
 298
 299 void free_commit_buffer(struct commit *commit)
 300 {
 301         struct commit_buffer *v = buffer_slab_peek(&buffer_slab, commit);
 302         if (v) {
 303                 FREE_AND_NULL(v->buffer);
 304                 v->size = 0;
 305         }
 306 }
 307
 308 void release_commit_memory(struct commit *c)
 309 {
 310         c->tree = NULL;
 311         c->index = 0;
 312         free_commit_buffer(c);
 313         free_commit_list(c->parents);
 314         /* TODO: what about commit->util? */
 315
 316         c->object.parsed = 0;
 317 }
 318
 319 const void *detach_commit_buffer(struct commit *commit, unsigned long *sizep)
 320 {
 321         struct commit_buffer *v = buffer_slab_peek(&buffer_slab, commit);
 322         void *ret;
 323
 324         if (!v) {
 325                 if (sizep)
 326                         *sizep = 0;
 327                 return NULL;
 328         }
 329         ret = v->buffer;
 330         if (sizep)
 331                 *sizep = v->size;
 332
 333         v->buffer = NULL;
 334         v->size = 0;
 335         return ret;
 336 }
 337
 338 int parse_commit_buffer(struct commit *item, const void *buffer, unsigned long size)
 339 {
 340         const char *tail = buffer;
 341         const char *bufptr = buffer;
 342         struct object_id parent;
 343         struct commit_list **pptr;
 344         struct commit_graft *graft;
 345         const int tree_entry_len = GIT_SHA1_HEXSZ + 5;
 346         const int parent_entry_len = GIT_SHA1_HEXSZ + 7;
 347
 348         if (item->object.parsed)
 349                 return 0;
 350         item->object.parsed = 1;
 351         tail += size;
 352         if (tail <= bufptr + tree_entry_len + 1 || memcmp(bufptr, "tree ", 5) ||
 353                         bufptr[tree_entry_len] != '\n')
 354                 return error("bogus commit object %s", oid_to_hex(&item->object.oid));
 355         if (get_sha1_hex(bufptr + 5, parent.hash) < 0)
 356                 return error("bad tree pointer in commit %s",
 357                              oid_to_hex(&item->object.oid));
 358         item->tree = lookup_tree(&parent);
 359         bufptr += tree_entry_len + 1; /* "tree " + "hex sha1" + "\n" */
 360         pptr = &item->parents;
 361
 362         graft = lookup_commit_graft(the_repository, &item->object.oid);
 363         while (bufptr + parent_entry_len < tail && !memcmp(bufptr, "parent ", 7)) {
 364                 struct commit *new_parent;
 365
 366                 if (tail <= bufptr + parent_entry_len + 1 ||
 367                     get_sha1_hex(bufptr + 7, parent.hash) ||
 368                     bufptr[parent_entry_len] != '\n')
 369                         return error("bad parents in commit %s", oid_to_hex(&item->object.oid));
 370                 bufptr += parent_entry_len + 1;
 371                 /*
 372                  * The clone is shallow if nr_parent < 0, and we must
 373                  * not traverse its real parents even when we unhide them.
 374                  */
 375                 if (graft && (graft->nr_parent < 0 || grafts_replace_parents))
 376                         continue;
 377                 new_parent = lookup_commit(&parent);
 378                 if (new_parent)
 379                         pptr = &commit_list_insert(new_parent, pptr)->next;
 380         }
 381         if (graft) {
 382                 int i;
 383                 struct commit *new_parent;
 384                 for (i = 0; i < graft->nr_parent; i++) {
 385                         new_parent = lookup_commit(&graft->parent[i]);
 386                         if (!new_parent)
 387                                 continue;
 388                         pptr = &commit_list_insert(new_parent, pptr)->next;
 389                 }
 390         }
 391         item->date = parse_commit_date(bufptr, tail);
 392
 393         return 0;
 394 }
 395
 396 int parse_commit_gently(struct commit *item, int quiet_on_missing)
 397 {
 398         enum object_type type;
 399         void *buffer;
 400         unsigned long size;
 401         int ret;
 402
 403         if (!item)
 404                 return -1;
 405         if (item->object.parsed)
 406                 return 0;
 407         buffer = read_object_file(&item->object.oid, &type, &size);
 408         if (!buffer)
 409                 return quiet_on_missing ? -1 :
 410                         error("Could not read %s",
 411                              oid_to_hex(&item->object.oid));
 412         if (type != OBJ_COMMIT) {
 413                 free(buffer);
 414                 return error("Object %s not a commit",
 415                              oid_to_hex(&item->object.oid));
 416         }
 417         ret = parse_commit_buffer(item, buffer, size);
 418         if (save_commit_buffer && !ret) {
 419                 set_commit_buffer(item, buffer, size);
 420                 return 0;
 421         }
 422         free(buffer);
 423         return ret;
 424 }
 425
 426 void parse_commit_or_die(struct commit *item)
 427 {
 428         if (parse_commit(item))
 429                 die("unable to parse commit %s",
 430                     item ? oid_to_hex(&item->object.oid) : "(null)");
 431 }
 432
 433 int find_commit_subject(const char *commit_buffer, const char **subject)
 434 {
 435         const char *eol;
 436         const char *p = commit_buffer;
 437
 438         while (*p && (*p != '\n' || p[1] != '\n'))
 439                 p++;
 440         if (*p) {
 441                 p = skip_blank_lines(p + 2);
 442                 eol = strchrnul(p, '\n');
 443         } else
 444                 eol = p;
 445
 446         *subject = p;
 447
 448         return eol - p;
 449 }
 450
 451 struct commit_list *commit_list_insert(struct commit *item, struct commit_list **list_p)
 452 {
 453         struct commit_list *new_list = xmalloc(sizeof(struct commit_list));
 454         new_list->item = item;
 455         new_list->next = *list_p;
 456         *list_p = new_list;
 457         return new_list;
 458 }
 459
 460 unsigned commit_list_count(const struct commit_list *l)
 461 {
 462         unsigned c = 0;
 463         for (; l; l = l->next )
 464                 c++;
 465         return c;
 466 }
 467
 468 struct commit_list *copy_commit_list(struct commit_list *list)
 469 {
 470         struct commit_list *head = NULL;
 471         struct commit_list **pp = &head;
 472         while (list) {
 473                 pp = commit_list_append(list->item, pp);
 474                 list = list->next;
 475         }
 476         return head;
 477 }
 478
 479 void free_commit_list(struct commit_list *list)
 480 {
 481         while (list)
 482                 pop_commit(&list);
 483 }
 484
 485 struct commit_list * commit_list_insert_by_date(struct commit *item, struct commit_list **list)
 486 {
 487         struct commit_list **pp = list;
 488         struct commit_list *p;
 489         while ((p = *pp) != NULL) {
 490                 if (p->item->date < item->date) {
 491                         break;
 492                 }
 493                 pp = &p->next;
 494         }
 495         return commit_list_insert(item, pp);
 496 }
 497
 498 static int commit_list_compare_by_date(const void *a, const void *b)
 499 {
 500         timestamp_t a_date = ((const struct commit_list *)a)->item->date;
 501         timestamp_t b_date = ((const struct commit_list *)b)->item->date;
 502         if (a_date < b_date)
 503                 return 1;
 504         if (a_date > b_date)
 505                 return -1;
 506         return 0;
 507 }
 508
 509 static void *commit_list_get_next(const void *a)
 510 {
 511         return ((const struct commit_list *)a)->next;
 512 }
 513
 514 static void commit_list_set_next(void *a, void *next)
 515 {
 516         ((struct commit_list *)a)->next = next;
 517 }
 518
 519 void commit_list_sort_by_date(struct commit_list **list)
 520 {
 521         *list = llist_mergesort(*list, commit_list_get_next, commit_list_set_next,
 522                                 commit_list_compare_by_date);
 523 }
 524
 525 struct commit *pop_most_recent_commit(struct commit_list **list,
 526                                       unsigned int mark)
 527 {
 528         struct commit *ret = pop_commit(list);
 529         struct commit_list *parents = ret->parents;
 530
 531         while (parents) {
 532                 struct commit *commit = parents->item;
 533                 if (!parse_commit(commit) && !(commit->object.flags & mark)) {
 534                         commit->object.flags |= mark;
 535                         commit_list_insert_by_date(commit, list);
 536                 }
 537                 parents = parents->next;
 538         }
 539         return ret;
 540 }
 541
 542 static void clear_commit_marks_1(struct commit_list **plist,
 543                                  struct commit *commit, unsigned int mark)
 544 {
 545         while (commit) {
 546                 struct commit_list *parents;
 547
 548                 if (!(mark & commit->object.flags))
 549                         return;
 550
 551                 commit->object.flags &= ~mark;
 552
 553                 parents = commit->parents;
 554                 if (!parents)
 555                         return;
 556
 557                 while ((parents = parents->next))
 558                         commit_list_insert(parents->item, plist);
 559
 560                 commit = commit->parents->item;
 561         }
 562 }
 563
 564 void clear_commit_marks_many(int nr, struct commit **commit, unsigned int mark)
 565 {
 566         struct commit_list *list = NULL;
 567
 568         while (nr--) {
 569                 clear_commit_marks_1(&list, *commit, mark);
 570                 commit++;
 571         }
 572         while (list)
 573                 clear_commit_marks_1(&list, pop_commit(&list), mark);
 574 }
 575
 576 void clear_commit_marks(struct commit *commit, unsigned int mark)
 577 {
 578         clear_commit_marks_many(1, &commit, mark);
 579 }
 580
 581 struct commit *pop_commit(struct commit_list **stack)
 582 {
 583         struct commit_list *top = *stack;
 584         struct commit *item = top ? top->item : NULL;
 585
 586         if (top) {
 587                 *stack = top->next;
 588                 free(top);
 589         }
 590         return item;
 591 }
 592
 593 /*
 594  * Topological sort support
 595  */
 596
 597 /* count number of children that have not been emitted */
 598 define_commit_slab(indegree_slab, int);
 599
 600 /* record author-date for each commit object */
 601 define_commit_slab(author_date_slab, unsigned long);
 602
 603 static void record_author_date(struct author_date_slab *author_date,
 604                                struct commit *commit)
 605 {
 606         const char *buffer = get_commit_buffer(commit, NULL);
 607         struct ident_split ident;
 608         const char *ident_line;
 609         size_t ident_len;
 610         char *date_end;
 611         timestamp_t date;
 612
 613         ident_line = find_commit_header(buffer, "author", &ident_len);
 614         if (!ident_line)
 615                 goto fail_exit; /* no author line */
 616         if (split_ident_line(&ident, ident_line, ident_len) ||
 617             !ident.date_begin || !ident.date_end)
 618                 goto fail_exit; /* malformed "author" line */
 619
 620         date = parse_timestamp(ident.date_begin, &date_end, 10);
 621         if (date_end != ident.date_end)
 622                 goto fail_exit; /* malformed date */
 623         *(author_date_slab_at(author_date, commit)) = date;
 624
 625 fail_exit:
 626         unuse_commit_buffer(commit, buffer);
 627 }
 628
 629 static int compare_commits_by_author_date(const void *a_, const void *b_,
 630                                           void *cb_data)
 631 {
 632         const struct commit *a = a_, *b = b_;
 633         struct author_date_slab *author_date = cb_data;
 634         timestamp_t a_date = *(author_date_slab_at(author_date, a));
 635         timestamp_t b_date = *(author_date_slab_at(author_date, b));
 636
 637         /* newer commits with larger date first */
 638         if (a_date < b_date)
 639                 return 1;
 640         else if (a_date > b_date)
 641                 return -1;
 642         return 0;
 643 }
 644
 645 int compare_commits_by_commit_date(const void *a_, const void *b_, void *unused)
 646 {
 647         const struct commit *a = a_, *b = b_;
 648         /* newer commits with larger date first */
 649         if (a->date < b->date)
 650                 return 1;
 651         else if (a->date > b->date)
 652                 return -1;
 653         return 0;
 654 }
 655
 656 /*
 657  * Performs an in-place topological sort on the list supplied.
 658  */
 659 void sort_in_topological_order(struct commit_list **list, enum rev_sort_order sort_order)
 660 {
 661         struct commit_list *next, *orig = *list;
 662         struct commit_list **pptr;
 663         struct indegree_slab indegree;
 664         struct prio_queue queue;
 665         struct commit *commit;
 666         struct author_date_slab author_date;
 667
 668         if (!orig)
 669                 return;
 670         *list = NULL;
 671
 672         init_indegree_slab(&indegree);
 673         memset(&queue, '\0', sizeof(queue));
 674
 675         switch (sort_order) {
 676         default: /* REV_SORT_IN_GRAPH_ORDER */
 677                 queue.compare = NULL;
 678                 break;
 679         case REV_SORT_BY_COMMIT_DATE:
 680                 queue.compare = compare_commits_by_commit_date;
 681                 break;
 682         case REV_SORT_BY_AUTHOR_DATE:
 683                 init_author_date_slab(&author_date);
 684                 queue.compare = compare_commits_by_author_date;
 685                 queue.cb_data = &author_date;
 686                 break;
 687         }
 688
 689         /* Mark them and clear the indegree */
 690         for (next = orig; next; next = next->next) {
 691                 struct commit *commit = next->item;
 692                 *(indegree_slab_at(&indegree, commit)) = 1;
 693                 /* also record the author dates, if needed */
 694                 if (sort_order == REV_SORT_BY_AUTHOR_DATE)
 695                         record_author_date(&author_date, commit);
 696         }
 697
 698         /* update the indegree */
 699         for (next = orig; next; next = next->next) {
 700                 struct commit_list *parents = next->item->parents;
 701                 while (parents) {
 702                         struct commit *parent = parents->item;
 703                         int *pi = indegree_slab_at(&indegree, parent);
 704
 705                         if (*pi)
 706                                 (*pi)++;
 707                         parents = parents->next;
 708                 }
 709         }
 710
 711         /*
 712          * find the tips
 713          *
 714          * tips are nodes not reachable from any other node in the list
 715          *
 716          * the tips serve as a starting set for the work queue.
 717          */
 718         for (next = orig; next; next = next->next) {
 719                 struct commit *commit = next->item;
 720
 721                 if (*(indegree_slab_at(&indegree, commit)) == 1)
 722                         prio_queue_put(&queue, commit);
 723         }
 724
 725         /*
 726          * This is unfortunate; the initial tips need to be shown
 727          * in the order given from the revision traversal machinery.
 728          */
 729         if (sort_order == REV_SORT_IN_GRAPH_ORDER)
 730                 prio_queue_reverse(&queue);
 731
 732         /* We no longer need the commit list */
 733         free_commit_list(orig);
 734
 735         pptr = list;
 736         *list = NULL;
 737         while ((commit = prio_queue_get(&queue)) != NULL) {
 738                 struct commit_list *parents;
 739
 740                 for (parents = commit->parents; parents ; parents = parents->next) {
 741                         struct commit *parent = parents->item;
 742                         int *pi = indegree_slab_at(&indegree, parent);
 743
 744                         if (!*pi)
 745                                 continue;
 746
 747                         /*
 748                          * parents are only enqueued for emission
 749                          * when all their children have been emitted thereby
 750                          * guaranteeing topological order.
 751                          */
 752                         if (--(*pi) == 1)
 753                                 prio_queue_put(&queue, parent);
 754                 }
 755                 /*
 756                  * all children of commit have already been
 757                  * emitted. we can emit it now.
 758                  */
 759                 *(indegree_slab_at(&indegree, commit)) = 0;
 760
 761                 pptr = &commit_list_insert(commit, pptr)->next;
 762         }
 763
 764         clear_indegree_slab(&indegree);
 765         clear_prio_queue(&queue);
 766         if (sort_order == REV_SORT_BY_AUTHOR_DATE)
 767                 clear_author_date_slab(&author_date);
 768 }
 769
 770 /* merge-base stuff */
 771
 772 /* Remember to update object flag allocation in object.h */
 773 #define PARENT1         (1u<<16)
 774 #define PARENT2         (1u<<17)
 775 #define STALE           (1u<<18)
 776 #define RESULT          (1u<<19)
 777
 778 static const unsigned all_flags = (PARENT1 | PARENT2 | STALE | RESULT);
 779
 780 static int queue_has_nonstale(struct prio_queue *queue)
 781 {
 782         int i;
 783         for (i = 0; i < queue->nr; i++) {
 784                 struct commit *commit = queue->array[i].data;
 785                 if (!(commit->object.flags & STALE))
 786                         return 1;
 787         }
 788         return 0;
 789 }
 790
 791 /* all input commits in one and twos[] must have been parsed! */
 792 static struct commit_list *paint_down_to_common(struct commit *one, int n, struct commit **twos)
 793 {
 794         struct prio_queue queue = { compare_commits_by_commit_date };
 795         struct commit_list *result = NULL;
 796         int i;
 797
 798         one->object.flags |= PARENT1;
 799         if (!n) {
 800                 commit_list_append(one, &result);
 801                 return result;
 802         }
 803         prio_queue_put(&queue, one);
 804
 805         for (i = 0; i < n; i++) {
 806                 twos[i]->object.flags |= PARENT2;
 807                 prio_queue_put(&queue, twos[i]);
 808         }
 809
 810         while (queue_has_nonstale(&queue)) {
 811                 struct commit *commit = prio_queue_get(&queue);
 812                 struct commit_list *parents;
 813                 int flags;
 814
 815                 flags = commit->object.flags & (PARENT1 | PARENT2 | STALE);
 816                 if (flags == (PARENT1 | PARENT2)) {
 817                         if (!(commit->object.flags & RESULT)) {
 818                                 commit->object.flags |= RESULT;
 819                                 commit_list_insert_by_date(commit, &result);
 820                         }
 821                         /* Mark parents of a found merge stale */
 822                         flags |= STALE;
 823                 }
 824                 parents = commit->parents;
 825                 while (parents) {
 826                         struct commit *p = parents->item;
 827                         parents = parents->next;
 828                         if ((p->object.flags & flags) == flags)
 829                                 continue;
 830                         if (parse_commit(p))
 831                                 return NULL;
 832                         p->object.flags |= flags;
 833                         prio_queue_put(&queue, p);
 834                 }
 835         }
 836
 837         clear_prio_queue(&queue);
 838         return result;
 839 }
 840
 841 static struct commit_list *merge_bases_many(struct commit *one, int n, struct commit **twos)
 842 {
 843         struct commit_list *list = NULL;
 844         struct commit_list *result = NULL;
 845         int i;
 846
 847         for (i = 0; i < n; i++) {
 848                 if (one == twos[i])
 849                         /*
 850                          * We do not mark this even with RESULT so we do not
 851                          * have to clean it up.
 852                          */
 853                         return commit_list_insert(one, &result);
 854         }
 855
 856         if (parse_commit(one))
 857                 return NULL;
 858         for (i = 0; i < n; i++) {
 859                 if (parse_commit(twos[i]))
 860                         return NULL;
 861         }
 862
 863         list = paint_down_to_common(one, n, twos);
 864
 865         while (list) {
 866                 struct commit *commit = pop_commit(&list);
 867                 if (!(commit->object.flags & STALE))
 868                         commit_list_insert_by_date(commit, &result);
 869         }
 870         return result;
 871 }
 872
 873 struct commit_list *get_octopus_merge_bases(struct commit_list *in)
 874 {
 875         struct commit_list *i, *j, *k, *ret = NULL;
 876
 877         if (!in)
 878                 return ret;
 879
 880         commit_list_insert(in->item, &ret);
 881
 882         for (i = in->next; i; i = i->next) {
 883                 struct commit_list *new_commits = NULL, *end = NULL;
 884
 885                 for (j = ret; j; j = j->next) {
 886                         struct commit_list *bases;
 887                         bases = get_merge_bases(i->item, j->item);
 888                         if (!new_commits)
 889                                 new_commits = bases;
 890                         else
 891                                 end->next = bases;
 892                         for (k = bases; k; k = k->next)
 893                                 end = k;
 894                 }
 895                 ret = new_commits;
 896         }
 897         return ret;
 898 }
 899
 900 static int remove_redundant(struct commit **array, int cnt)
 901 {
 902         /*
 903          * Some commit in the array may be an ancestor of
 904          * another commit.  Move such commit to the end of
 905          * the array, and return the number of commits that
 906          * are independent from each other.
 907          */
 908         struct commit **work;
 909         unsigned char *redundant;
 910         int *filled_index;
 911         int i, j, filled;
 912
 913         work = xcalloc(cnt, sizeof(*work));
 914         redundant = xcalloc(cnt, 1);
 915         ALLOC_ARRAY(filled_index, cnt - 1);
 916
 917         for (i = 0; i < cnt; i++)
 918                 parse_commit(array[i]);
 919         for (i = 0; i < cnt; i++) {
 920                 struct commit_list *common;
 921
 922                 if (redundant[i])
 923                         continue;
 924                 for (j = filled = 0; j < cnt; j++) {
 925                         if (i == j || redundant[j])
 926                                 continue;
 927                         filled_index[filled] = j;
 928                         work[filled++] = array[j];
 929                 }
 930                 common = paint_down_to_common(array[i], filled, work);
 931                 if (array[i]->object.flags & PARENT2)
 932                         redundant[i] = 1;
 933                 for (j = 0; j < filled; j++)
 934                         if (work[j]->object.flags & PARENT1)
 935                                 redundant[filled_index[j]] = 1;
 936                 clear_commit_marks(array[i], all_flags);
 937                 clear_commit_marks_many(filled, work, all_flags);
 938                 free_commit_list(common);
 939         }
 940
 941         /* Now collect the result */
 942         COPY_ARRAY(work, array, cnt);
 943         for (i = filled = 0; i < cnt; i++)
 944                 if (!redundant[i])
 945                         array[filled++] = work[i];
 946         for (j = filled, i = 0; i < cnt; i++)
 947                 if (redundant[i])
 948                         array[j++] = work[i];
 949         free(work);
 950         free(redundant);
 951         free(filled_index);
 952         return filled;
 953 }
 954
 955 static struct commit_list *get_merge_bases_many_0(struct commit *one,
 956                                                   int n,
 957                                                   struct commit **twos,
 958                                                   int cleanup)
 959 {
 960         struct commit_list *list;
 961         struct commit **rslt;
 962         struct commit_list *result;
 963         int cnt, i;
 964
 965         result = merge_bases_many(one, n, twos);
 966         for (i = 0; i < n; i++) {
 967                 if (one == twos[i])
 968                         return result;
 969         }
 970         if (!result || !result->next) {
 971                 if (cleanup) {
 972                         clear_commit_marks(one, all_flags);
 973                         clear_commit_marks_many(n, twos, all_flags);
 974                 }
 975                 return result;
 976         }
 977
 978         /* There are more than one */
 979         cnt = commit_list_count(result);
 980         rslt = xcalloc(cnt, sizeof(*rslt));
 981         for (list = result, i = 0; list; list = list->next)
 982                 rslt[i++] = list->item;
 983         free_commit_list(result);
 984
 985         clear_commit_marks(one, all_flags);
 986         clear_commit_marks_many(n, twos, all_flags);
 987
 988         cnt = remove_redundant(rslt, cnt);
 989         result = NULL;
 990         for (i = 0; i < cnt; i++)
 991                 commit_list_insert_by_date(rslt[i], &result);
 992         free(rslt);
 993         return result;
 994 }
 995
 996 struct commit_list *get_merge_bases_many(struct commit *one,
 997                                          int n,
 998                                          struct commit **twos)
 999 {
1000         return get_merge_bases_many_0(one, n, twos, 1);
1001 }
1002
1003 struct commit_list *get_merge_bases_many_dirty(struct commit *one,
1004                                                int n,
1005                                                struct commit **twos)
1006 {
1007         return get_merge_bases_many_0(one, n, twos, 0);
1008 }
1009
1010 struct commit_list *get_merge_bases(struct commit *one, struct commit *two)
1011 {
1012         return get_merge_bases_many_0(one, 1, &two, 1);
1013 }
1014
1015 /*
1016  * Is "commit" a descendant of one of the elements on the "with_commit" list?
1017  */
1018 int is_descendant_of(struct commit *commit, struct commit_list *with_commit)
1019 {
1020         if (!with_commit)
1021                 return 1;
1022         while (with_commit) {
1023                 struct commit *other;
1024
1025                 other = with_commit->item;
1026                 with_commit = with_commit->next;
1027                 if (in_merge_bases(other, commit))
1028                         return 1;
1029         }
1030         return 0;
1031 }
1032
1033 /*
1034  * Is "commit" an ancestor of one of the "references"?
1035  */
1036 int in_merge_bases_many(struct commit *commit, int nr_reference, struct commit **reference)
1037 {
1038         struct commit_list *bases;
1039         int ret = 0, i;
1040
1041         if (parse_commit(commit))
1042                 return ret;
1043         for (i = 0; i < nr_reference; i++)
1044                 if (parse_commit(reference[i]))
1045                         return ret;
1046
1047         bases = paint_down_to_common(commit, nr_reference, reference);
1048         if (commit->object.flags & PARENT2)
1049                 ret = 1;
1050         clear_commit_marks(commit, all_flags);
1051         clear_commit_marks_many(nr_reference, reference, all_flags);
1052         free_commit_list(bases);
1053         return ret;
1054 }
1055
1056 /*
1057  * Is "commit" an ancestor of (i.e. reachable from) the "reference"?
1058  */
1059 int in_merge_bases(struct commit *commit, struct commit *reference)
1060 {
1061         return in_merge_bases_many(commit, 1, &reference);
1062 }
1063
1064 struct commit_list *reduce_heads(struct commit_list *heads)
1065 {
1066         struct commit_list *p;
1067         struct commit_list *result = NULL, **tail = &result;
1068         struct commit **array;
1069         int num_head, i;
1070
1071         if (!heads)
1072                 return NULL;
1073
1074         /* Uniquify */
1075         for (p = heads; p; p = p->next)
1076                 p->item->object.flags &= ~STALE;
1077         for (p = heads, num_head = 0; p; p = p->next) {
1078                 if (p->item->object.flags & STALE)
1079                         continue;
1080                 p->item->object.flags |= STALE;
1081                 num_head++;
1082         }
1083         array = xcalloc(num_head, sizeof(*array));
1084         for (p = heads, i = 0; p; p = p->next) {
1085                 if (p->item->object.flags & STALE) {
1086                         array[i++] = p->item;
1087                         p->item->object.flags &= ~STALE;
1088                 }
1089         }
1090         num_head = remove_redundant(array, num_head);
1091         for (i = 0; i < num_head; i++)
1092                 tail = &commit_list_insert(array[i], tail)->next;
1093         free(array);
1094         return result;
1095 }
1096
1097 void reduce_heads_replace(struct commit_list **heads)
1098 {
1099         struct commit_list *result = reduce_heads(*heads);
1100         free_commit_list(*heads);
1101         *heads = result;
1102 }
1103
1104 static const char gpg_sig_header[] = "gpgsig";
1105 static const int gpg_sig_header_len = sizeof(gpg_sig_header) - 1;
1106
1107 static int do_sign_commit(struct strbuf *buf, const char *keyid)
1108 {
1109         struct strbuf sig = STRBUF_INIT;
1110         int inspos, copypos;
1111         const char *eoh;
1112
1113         /* find the end of the header */
1114         eoh = strstr(buf->buf, "\n\n");
1115         if (!eoh)
1116                 inspos = buf->len;
1117         else
1118                 inspos = eoh - buf->buf + 1;
1119
1120         if (!keyid || !*keyid)
1121                 keyid = get_signing_key();
1122         if (sign_buffer(buf, &sig, keyid)) {
1123                 strbuf_release(&sig);
1124                 return -1;
1125         }
1126
1127         for (copypos = 0; sig.buf[copypos]; ) {
1128                 const char *bol = sig.buf + copypos;
1129                 const char *eol = strchrnul(bol, '\n');
1130                 int len = (eol - bol) + !!*eol;
1131
1132                 if (!copypos) {
1133                         strbuf_insert(buf, inspos, gpg_sig_header, gpg_sig_header_len);
1134                         inspos += gpg_sig_header_len;
1135                 }
1136                 strbuf_insert(buf, inspos++, " ", 1);
1137                 strbuf_insert(buf, inspos, bol, len);
1138                 inspos += len;
1139                 copypos += len;
1140         }
1141         strbuf_release(&sig);
1142         return 0;
1143 }
1144
1145 int parse_signed_commit(const struct commit *commit,
1146                         struct strbuf *payload, struct strbuf *signature)
1147 {
1148
1149         unsigned long size;
1150         const char *buffer = get_commit_buffer(commit, &size);
1151         int in_signature, saw_signature = -1;
1152         const char *line, *tail;
1153
1154         line = buffer;
1155         tail = buffer + size;
1156         in_signature = 0;
1157         saw_signature = 0;
1158         while (line < tail) {
1159                 const char *sig = NULL;
1160                 const char *next = memchr(line, '\n', tail - line);
1161
1162                 next = next ? next + 1 : tail;
1163                 if (in_signature && line[0] == ' ')
1164                         sig = line + 1;
1165                 else if (starts_with(line, gpg_sig_header) &&
1166                          line[gpg_sig_header_len] == ' ')
1167                         sig = line + gpg_sig_header_len + 1;
1168                 if (sig) {
1169                         strbuf_add(signature, sig, next - sig);
1170                         saw_signature = 1;
1171                         in_signature = 1;
1172                 } else {
1173                         if (*line == '\n')
1174                                 /* dump the whole remainder of the buffer */
1175                                 next = tail;
1176                         strbuf_add(payload, line, next - line);
1177                         in_signature = 0;
1178                 }
1179                 line = next;
1180         }
1181         unuse_commit_buffer(commit, buffer);
1182         return saw_signature;
1183 }
1184
1185 int remove_signature(struct strbuf *buf)
1186 {
1187         const char *line = buf->buf;
1188         const char *tail = buf->buf + buf->len;
1189         int in_signature = 0;
1190         const char *sig_start = NULL;
1191         const char *sig_end = NULL;
1192
1193         while (line < tail) {
1194                 const char *next = memchr(line, '\n', tail - line);
1195                 next = next ? next + 1 : tail;
1196
1197                 if (in_signature && line[0] == ' ')
1198                         sig_end = next;
1199                 else if (starts_with(line, gpg_sig_header) &&
1200                          line[gpg_sig_header_len] == ' ') {
1201                         sig_start = line;
1202                         sig_end = next;
1203                         in_signature = 1;
1204                 } else {
1205                         if (*line == '\n')
1206                                 /* dump the whole remainder of the buffer */
1207                                 next = tail;
1208                         in_signature = 0;
1209                 }
1210                 line = next;
1211         }
1212
1213         if (sig_start)
1214                 strbuf_remove(buf, sig_start - buf->buf, sig_end - sig_start);
1215
1216         return sig_start != NULL;
1217 }
1218
1219 static void handle_signed_tag(struct commit *parent, struct commit_extra_header ***tail)
1220 {
1221         struct merge_remote_desc *desc;
1222         struct commit_extra_header *mergetag;
1223         char *buf;
1224         unsigned long size, len;
1225         enum object_type type;
1226
1227         desc = merge_remote_util(parent);
1228         if (!desc || !desc->obj)
1229                 return;
1230         buf = read_object_file(&desc->obj->oid, &type, &size);
1231         if (!buf || type != OBJ_TAG)
1232                 goto free_return;
1233         len = parse_signature(buf, size);
1234         if (size == len)
1235                 goto free_return;
1236         /*
1237          * We could verify this signature and either omit the tag when
1238          * it does not validate, but the integrator may not have the
1239          * public key of the signer of the tag he is merging, while a
1240          * later auditor may have it while auditing, so let's not run
1241          * verify-signed-buffer here for now...
1242          *
1243          * if (verify_signed_buffer(buf, len, buf + len, size - len, ...))
1244          *      warn("warning: signed tag unverified.");
1245          */
1246         mergetag = xcalloc(1, sizeof(*mergetag));
1247         mergetag->key = xstrdup("mergetag");
1248         mergetag->value = buf;
1249         mergetag->len = size;
1250
1251         **tail = mergetag;
1252         *tail = &mergetag->next;
1253         return;
1254
1255 free_return:
1256         free(buf);
1257 }
1258
1259 int check_commit_signature(const struct commit *commit, struct signature_check *sigc)
1260 {
1261         struct strbuf payload = STRBUF_INIT;
1262         struct strbuf signature = STRBUF_INIT;
1263         int ret = 1;
1264
1265         sigc->result = 'N';
1266
1267         if (parse_signed_commit(commit, &payload, &signature) <= 0)
1268                 goto out;
1269         ret = check_signature(payload.buf, payload.len, signature.buf,
1270                 signature.len, sigc);
1271
1272  out:
1273         strbuf_release(&payload);
1274         strbuf_release(&signature);
1275
1276         return ret;
1277 }
1278
1279
1280
1281 void append_merge_tag_headers(struct commit_list *parents,
1282                               struct commit_extra_header ***tail)
1283 {
1284         while (parents) {
1285                 struct commit *parent = parents->item;
1286                 handle_signed_tag(parent, tail);
1287                 parents = parents->next;
1288         }
1289 }
1290
1291 static void add_extra_header(struct strbuf *buffer,
1292                              struct commit_extra_header *extra)
1293 {
1294         strbuf_addstr(buffer, extra->key);
1295         if (extra->len)
1296                 strbuf_add_lines(buffer, " ", extra->value, extra->len);
1297         else
1298                 strbuf_addch(buffer, '\n');
1299 }
1300
1301 struct commit_extra_header *read_commit_extra_headers(struct commit *commit,
1302                                                       const char **exclude)
1303 {
1304         struct commit_extra_header *extra = NULL;
1305         unsigned long size;
1306         const char *buffer = get_commit_buffer(commit, &size);
1307         extra = read_commit_extra_header_lines(buffer, size, exclude);
1308         unuse_commit_buffer(commit, buffer);
1309         return extra;
1310 }
1311
1312 void for_each_mergetag(each_mergetag_fn fn, struct commit *commit, void *data)
1313 {
1314         struct commit_extra_header *extra, *to_free;
1315
1316         to_free = read_commit_extra_headers(commit, NULL);
1317         for (extra = to_free; extra; extra = extra->next) {
1318                 if (strcmp(extra->key, "mergetag"))
1319                         continue; /* not a merge tag */
1320                 fn(commit, extra, data);
1321         }
1322         free_commit_extra_headers(to_free);
1323 }
1324
1325 static inline int standard_header_field(const char *field, size_t len)
1326 {
1327         return ((len == 4 && !memcmp(field, "tree", 4)) ||
1328                 (len == 6 && !memcmp(field, "parent", 6)) ||
1329                 (len == 6 && !memcmp(field, "author", 6)) ||
1330                 (len == 9 && !memcmp(field, "committer", 9)) ||
1331                 (len == 8 && !memcmp(field, "encoding", 8)));
1332 }
1333
1334 static int excluded_header_field(const char *field, size_t len, const char **exclude)
1335 {
1336         if (!exclude)
1337                 return 0;
1338
1339         while (*exclude) {
1340                 size_t xlen = strlen(*exclude);
1341                 if (len == xlen && !memcmp(field, *exclude, xlen))
1342                         return 1;
1343                 exclude++;
1344         }
1345         return 0;
1346 }
1347
1348 static struct commit_extra_header *read_commit_extra_header_lines(
1349         const char *buffer, size_t size,
1350         const char **exclude)
1351 {
1352         struct commit_extra_header *extra = NULL, **tail = &extra, *it = NULL;
1353         const char *line, *next, *eof, *eob;
1354         struct strbuf buf = STRBUF_INIT;
1355
1356         for (line = buffer, eob = line + size;
1357              line < eob && *line != '\n';
1358              line = next) {
1359                 next = memchr(line, '\n', eob - line);
1360                 next = next ? next + 1 : eob;
1361                 if (*line == ' ') {
1362                         /* continuation */
1363                         if (it)
1364                                 strbuf_add(&buf, line + 1, next - (line + 1));
1365                         continue;
1366                 }
1367                 if (it)
1368                         it->value = strbuf_detach(&buf, &it->len);
1369                 strbuf_reset(&buf);
1370                 it = NULL;
1371
1372                 eof = memchr(line, ' ', next - line);
1373                 if (!eof)
1374                         eof = next;
1375                 else if (standard_header_field(line, eof - line) ||
1376                          excluded_header_field(line, eof - line, exclude))
1377                         continue;
1378
1379                 it = xcalloc(1, sizeof(*it));
1380                 it->key = xmemdupz(line, eof-line);
1381                 *tail = it;
1382                 tail = &it->next;
1383                 if (eof + 1 < next)
1384                         strbuf_add(&buf, eof + 1, next - (eof + 1));
1385         }
1386         if (it)
1387                 it->value = strbuf_detach(&buf, &it->len);
1388         return extra;
1389 }
1390
1391 void free_commit_extra_headers(struct commit_extra_header *extra)
1392 {
1393         while (extra) {
1394                 struct commit_extra_header *next = extra->next;
1395                 free(extra->key);
1396                 free(extra->value);
1397                 free(extra);
1398                 extra = next;
1399         }
1400 }
1401
1402 int commit_tree(const char *msg, size_t msg_len, const struct object_id *tree,
1403                 struct commit_list *parents, struct object_id *ret,
1404                 const char *author, const char *sign_commit)
1405 {
1406         struct commit_extra_header *extra = NULL, **tail = &extra;
1407         int result;
1408
1409         append_merge_tag_headers(parents, &tail);
1410         result = commit_tree_extended(msg, msg_len, tree, parents, ret,
1411                                       author, sign_commit, extra);
1412         free_commit_extra_headers(extra);
1413         return result;
1414 }
1415
1416 static int find_invalid_utf8(const char *buf, int len)
1417 {
1418         int offset = 0;
1419         static const unsigned int max_codepoint[] = {
1420                 0x7f, 0x7ff, 0xffff, 0x10ffff
1421         };
1422
1423         while (len) {
1424                 unsigned char c = *buf++;
1425                 int bytes, bad_offset;
1426                 unsigned int codepoint;
1427                 unsigned int min_val, max_val;
1428
1429                 len--;
1430                 offset++;
1431
1432                 /* Simple US-ASCII? No worries. */
1433                 if (c < 0x80)
1434                         continue;
1435
1436                 bad_offset = offset-1;
1437
1438                 /*
1439                  * Count how many more high bits set: that's how
1440                  * many more bytes this sequence should have.
1441                  */
1442                 bytes = 0;
1443                 while (c & 0x40) {
1444                         c <<= 1;
1445                         bytes++;
1446                 }
1447
1448                 /*
1449                  * Must be between 1 and 3 more bytes.  Longer sequences result in
1450                  * codepoints beyond U+10FFFF, which are guaranteed never to exist.
1451                  */
1452                 if (bytes < 1 || 3 < bytes)
1453                         return bad_offset;
1454
1455                 /* Do we *have* that many bytes? */
1456                 if (len < bytes)
1457                         return bad_offset;
1458
1459                 /*
1460                  * Place the encoded bits at the bottom of the value and compute the
1461                  * valid range.
1462                  */
1463                 codepoint = (c & 0x7f) >> bytes;
1464                 min_val = max_codepoint[bytes-1] + 1;
1465                 max_val = max_codepoint[bytes];
1466
1467                 offset += bytes;
1468                 len -= bytes;
1469
1470                 /* And verify that they are good continuation bytes */
1471                 do {
1472                         codepoint <<= 6;
1473                         codepoint |= *buf & 0x3f;
1474                         if ((*buf++ & 0xc0) != 0x80)
1475                                 return bad_offset;
1476                 } while (--bytes);
1477
1478                 /* Reject codepoints that are out of range for the sequence length. */
1479                 if (codepoint < min_val || codepoint > max_val)
1480                         return bad_offset;
1481                 /* Surrogates are only for UTF-16 and cannot be encoded in UTF-8. */
1482                 if ((codepoint & 0x1ff800) == 0xd800)
1483                         return bad_offset;
1484                 /* U+xxFFFE and U+xxFFFF are guaranteed non-characters. */
1485                 if ((codepoint & 0xfffe) == 0xfffe)
1486                         return bad_offset;
1487                 /* So are anything in the range U+FDD0..U+FDEF. */
1488                 if (codepoint >= 0xfdd0 && codepoint <= 0xfdef)
1489                         return bad_offset;
1490         }
1491         return -1;
1492 }
1493
1494 /*
1495  * This verifies that the buffer is in proper utf8 format.
1496  *
1497  * If it isn't, it assumes any non-utf8 characters are Latin1,
1498  * and does the conversion.
1499  */
1500 static int verify_utf8(struct strbuf *buf)
1501 {
1502         int ok = 1;
1503         long pos = 0;
1504
1505         for (;;) {
1506                 int bad;
1507                 unsigned char c;
1508                 unsigned char replace[2];
1509
1510                 bad = find_invalid_utf8(buf->buf + pos, buf->len - pos);
1511                 if (bad < 0)
1512                         return ok;
1513                 pos += bad;
1514                 ok = 0;
1515                 c = buf->buf[pos];
1516                 strbuf_remove(buf, pos, 1);
1517
1518                 /* We know 'c' must be in the range 128-255 */
1519                 replace[0] = 0xc0 + (c >> 6);
1520                 replace[1] = 0x80 + (c & 0x3f);
1521                 strbuf_insert(buf, pos, replace, 2);
1522                 pos += 2;
1523         }
1524 }
1525
1526 static const char commit_utf8_warn[] =
1527 N_("Warning: commit message did not conform to UTF-8.\n"
1528    "You may want to amend it after fixing the message, or set the config\n"
1529    "variable i18n.commitencoding to the encoding your project uses.\n");
1530
1531 int commit_tree_extended(const char *msg, size_t msg_len,
1532                          const struct object_id *tree,
1533                          struct commit_list *parents, struct object_id *ret,
1534                          const char *author, const char *sign_commit,
1535                          struct commit_extra_header *extra)
1536 {
1537         int result;
1538         int encoding_is_utf8;
1539         struct strbuf buffer;
1540
1541         assert_oid_type(tree, OBJ_TREE);
1542
1543         if (memchr(msg, '\0', msg_len))
1544                 return error("a NUL byte in commit log message not allowed.");
1545
1546         /* Not having i18n.commitencoding is the same as having utf-8 */
1547         encoding_is_utf8 = is_encoding_utf8(git_commit_encoding);
1548
1549         strbuf_init(&buffer, 8192); /* should avoid reallocs for the headers */
1550         strbuf_addf(&buffer, "tree %s\n", oid_to_hex(tree));
1551
1552         /*
1553          * NOTE! This ordering means that the same exact tree merged with a
1554          * different order of parents will be a _different_ changeset even
1555          * if everything else stays the same.
1556          */
1557         while (parents) {
1558                 struct commit *parent = pop_commit(&parents);
1559                 strbuf_addf(&buffer, "parent %s\n",
1560                             oid_to_hex(&parent->object.oid));
1561         }
1562
1563         /* Person/date information */
1564         if (!author)
1565                 author = git_author_info(IDENT_STRICT);
1566         strbuf_addf(&buffer, "author %s\n", author);
1567         strbuf_addf(&buffer, "committer %s\n", git_committer_info(IDENT_STRICT));
1568         if (!encoding_is_utf8)
1569                 strbuf_addf(&buffer, "encoding %s\n", git_commit_encoding);
1570
1571         while (extra) {
1572                 add_extra_header(&buffer, extra);
1573                 extra = extra->next;
1574         }
1575         strbuf_addch(&buffer, '\n');
1576
1577         /* And add the comment */
1578         strbuf_add(&buffer, msg, msg_len);
1579
1580         /* And check the encoding */
1581         if (encoding_is_utf8 && !verify_utf8(&buffer))
1582                 fprintf(stderr, _(commit_utf8_warn));
1583
1584         if (sign_commit && do_sign_commit(&buffer, sign_commit)) {
1585                 result = -1;
1586                 goto out;
1587         }
1588
1589         result = write_object_file(buffer.buf, buffer.len, commit_type, ret);
1590 out:
1591         strbuf_release(&buffer);
1592         return result;
1593 }
1594
1595 void set_merge_remote_desc(struct commit *commit,
1596                            const char *name, struct object *obj)
1597 {
1598         struct merge_remote_desc *desc;
1599         FLEX_ALLOC_STR(desc, name, name);
1600         desc->obj = obj;
1601         commit->util = desc;
1602 }
1603
1604 struct commit *get_merge_parent(const char *name)
1605 {
1606         struct object *obj;
1607         struct commit *commit;
1608         struct object_id oid;
1609         if (get_oid(name, &oid))
1610                 return NULL;
1611         obj = parse_object(&oid);
1612         commit = (struct commit *)peel_to_type(name, 0, obj, OBJ_COMMIT);
1613         if (commit && !commit->util)
1614                 set_merge_remote_desc(commit, name, obj);
1615         return commit;
1616 }
1617
1618 /*
1619  * Append a commit to the end of the commit_list.
1620  *
1621  * next starts by pointing to the variable that holds the head of an
1622  * empty commit_list, and is updated to point to the "next" field of
1623  * the last item on the list as new commits are appended.
1624  *
1625  * Usage example:
1626  *
1627  *     struct commit_list *list;
1628  *     struct commit_list **next = &list;
1629  *
1630  *     next = commit_list_append(c1, next);
1631  *     next = commit_list_append(c2, next);
1632  *     assert(commit_list_count(list) == 2);
1633  *     return list;
1634  */
1635 struct commit_list **commit_list_append(struct commit *commit,
1636                                         struct commit_list **next)
1637 {
1638         struct commit_list *new_commit = xmalloc(sizeof(struct commit_list));
1639         new_commit->item = commit;
1640         *next = new_commit;
1641         new_commit->next = NULL;
1642         return &new_commit->next;
1643 }
1644
1645 const char *find_commit_header(const char *msg, const char *key, size_t *out_len)
1646 {
1647         int key_len = strlen(key);
1648         const char *line = msg;
1649
1650         while (line) {
1651                 const char *eol = strchrnul(line, '\n');
1652
1653                 if (line == eol)
1654                         return NULL;
1655
1656                 if (eol - line > key_len &&
1657                     !strncmp(line, key, key_len) &&
1658                     line[key_len] == ' ') {
1659                         *out_len = eol - line - key_len - 1;
1660                         return line + key_len + 1;
1661                 }
1662                 line = *eol ? eol + 1 : NULL;
1663         }
1664         return NULL;
1665 }
1666
1667 /*
1668  * Inspect the given string and determine the true "end" of the log message, in
1669  * order to find where to put a new Signed-off-by: line.  Ignored are
1670  * trailing comment lines and blank lines.  To support "git commit -s
1671  * --amend" on an existing commit, we also ignore "Conflicts:".  To
1672  * support "git commit -v", we truncate at cut lines.
1673  *
1674  * Returns the number of bytes from the tail to ignore, to be fed as
1675  * the second parameter to append_signoff().
1676  */
1677 int ignore_non_trailer(const char *buf, size_t len)
1678 {
1679         int boc = 0;
1680         int bol = 0;
1681         int in_old_conflicts_block = 0;
1682         size_t cutoff = wt_status_locate_end(buf, len);
1683
1684         while (bol < cutoff) {
1685                 const char *next_line = memchr(buf + bol, '\n', len - bol);
1686
1687                 if (!next_line)
1688                         next_line = buf + len;
1689                 else
1690                         next_line++;
1691
1692                 if (buf[bol] == comment_line_char || buf[bol] == '\n') {
1693                         /* is this the first of the run of comments? */
1694                         if (!boc)
1695                                 boc = bol;
1696                         /* otherwise, it is just continuing */
1697                 } else if (starts_with(buf + bol, "Conflicts:\n")) {
1698                         in_old_conflicts_block = 1;
1699                         if (!boc)
1700                                 boc = bol;
1701                 } else if (in_old_conflicts_block && buf[bol] == '\t') {
1702                         ; /* a pathname in the conflicts block */
1703                 } else if (boc) {
1704                         /* the previous was not trailing comment */
1705                         boc = 0;
1706                         in_old_conflicts_block = 0;
1707                 }
1708                 bol = next_line - buf;
1709         }
1710         return boc ? len - boc : len - cutoff;
1711 }