git.oblomov.eu Git - git/blob - commit.c

   1 #include "cache.h"
   2 #include "tag.h"
   3 #include "commit.h"
   4 #include "commit-graph.h"
   5 #include "repository.h"
   6 #include "object-store.h"
   7 #include "pkt-line.h"
   8 #include "utf8.h"
   9 #include "diff.h"
  10 #include "revision.h"
  11 #include "notes.h"
  12 #include "alloc.h"
  13 #include "gpg-interface.h"
  14 #include "mergesort.h"
  15 #include "commit-slab.h"
  16 #include "prio-queue.h"
  17 #include "sha1-lookup.h"
  18 #include "wt-status.h"
  19 #include "advice.h"
  20
  21 static struct commit_extra_header *read_commit_extra_header_lines(const char *buf, size_t len, const char **);
  22
  23 int save_commit_buffer = 1;
  24
  25 const char *commit_type = "commit";
  26
  27 struct commit *lookup_commit_reference_gently_the_repository(
  28                 const struct object_id *oid, int quiet)
  29 {
  30         struct object *obj = deref_tag(the_repository,
  31                                        parse_object(the_repository, oid),
  32                                        NULL, 0);
  33
  34         if (!obj)
  35                 return NULL;
  36         return object_as_type(the_repository, obj, OBJ_COMMIT, quiet);
  37 }
  38
  39 struct commit *lookup_commit_reference_the_repository(const struct object_id *oid)
  40 {
  41         return lookup_commit_reference_gently(the_repository, oid, 0);
  42 }
  43
  44 struct commit *lookup_commit_or_die(const struct object_id *oid, const char *ref_name)
  45 {
  46         struct commit *c = lookup_commit_reference(the_repository, oid);
  47         if (!c)
  48                 die(_("could not parse %s"), ref_name);
  49         if (oidcmp(oid, &c->object.oid)) {
  50                 warning(_("%s %s is not a commit!"),
  51                         ref_name, oid_to_hex(oid));
  52         }
  53         return c;
  54 }
  55
  56 struct commit *lookup_commit(struct repository *r, const struct object_id *oid)
  57 {
  58         struct object *obj = lookup_object(r, oid->hash);
  59         if (!obj)
  60                 return create_object(r, oid->hash,
  61                                      alloc_commit_node(r));
  62         return object_as_type(r, obj, OBJ_COMMIT, 0);
  63 }
  64
  65 struct commit *lookup_commit_reference_by_name(const char *name)
  66 {
  67         struct object_id oid;
  68         struct commit *commit;
  69
  70         if (get_oid_committish(name, &oid))
  71                 return NULL;
  72         commit = lookup_commit_reference(the_repository, &oid);
  73         if (parse_commit(commit))
  74                 return NULL;
  75         return commit;
  76 }
  77
  78 static timestamp_t parse_commit_date(const char *buf, const char *tail)
  79 {
  80         const char *dateptr;
  81
  82         if (buf + 6 >= tail)
  83                 return 0;
  84         if (memcmp(buf, "author", 6))
  85                 return 0;
  86         while (buf < tail && *buf++ != '\n')
  87                 /* nada */;
  88         if (buf + 9 >= tail)
  89                 return 0;
  90         if (memcmp(buf, "committer", 9))
  91                 return 0;
  92         while (buf < tail && *buf++ != '>')
  93                 /* nada */;
  94         if (buf >= tail)
  95                 return 0;
  96         dateptr = buf;
  97         while (buf < tail && *buf++ != '\n')
  98                 /* nada */;
  99         if (buf >= tail)
 100                 return 0;
 101         /* dateptr < buf && buf[-1] == '\n', so parsing will stop at buf-1 */
 102         return parse_timestamp(dateptr, NULL, 10);
 103 }
 104
 105 static const unsigned char *commit_graft_sha1_access(size_t index, void *table)
 106 {
 107         struct commit_graft **commit_graft_table = table;
 108         return commit_graft_table[index]->oid.hash;
 109 }
 110
 111 static int commit_graft_pos(struct repository *r, const unsigned char *sha1)
 112 {
 113         return sha1_pos(sha1, r->parsed_objects->grafts,
 114                         r->parsed_objects->grafts_nr,
 115                         commit_graft_sha1_access);
 116 }
 117
 118 int register_commit_graft(struct repository *r, struct commit_graft *graft,
 119                           int ignore_dups)
 120 {
 121         int pos = commit_graft_pos(r, graft->oid.hash);
 122
 123         if (0 <= pos) {
 124                 if (ignore_dups)
 125                         free(graft);
 126                 else {
 127                         free(r->parsed_objects->grafts[pos]);
 128                         r->parsed_objects->grafts[pos] = graft;
 129                 }
 130                 return 1;
 131         }
 132         pos = -pos - 1;
 133         ALLOC_GROW(r->parsed_objects->grafts,
 134                    r->parsed_objects->grafts_nr + 1,
 135                    r->parsed_objects->grafts_alloc);
 136         r->parsed_objects->grafts_nr++;
 137         if (pos < r->parsed_objects->grafts_nr)
 138                 memmove(r->parsed_objects->grafts + pos + 1,
 139                         r->parsed_objects->grafts + pos,
 140                         (r->parsed_objects->grafts_nr - pos - 1) *
 141                         sizeof(*r->parsed_objects->grafts));
 142         r->parsed_objects->grafts[pos] = graft;
 143         return 0;
 144 }
 145
 146 struct commit_graft *read_graft_line(struct strbuf *line)
 147 {
 148         /* The format is just "Commit Parent1 Parent2 ...\n" */
 149         int i, phase;
 150         const char *tail = NULL;
 151         struct commit_graft *graft = NULL;
 152         struct object_id dummy_oid, *oid;
 153
 154         strbuf_rtrim(line);
 155         if (!line->len || line->buf[0] == '#')
 156                 return NULL;
 157         /*
 158          * phase 0 verifies line, counts hashes in line and allocates graft
 159          * phase 1 fills graft
 160          */
 161         for (phase = 0; phase < 2; phase++) {
 162                 oid = graft ? &graft->oid : &dummy_oid;
 163                 if (parse_oid_hex(line->buf, oid, &tail))
 164                         goto bad_graft_data;
 165                 for (i = 0; *tail != '\0'; i++) {
 166                         oid = graft ? &graft->parent[i] : &dummy_oid;
 167                         if (!isspace(*tail++) || parse_oid_hex(tail, oid, &tail))
 168                                 goto bad_graft_data;
 169                 }
 170                 if (!graft) {
 171                         graft = xmalloc(st_add(sizeof(*graft),
 172                                                st_mult(sizeof(struct object_id), i)));
 173                         graft->nr_parent = i;
 174                 }
 175         }
 176         return graft;
 177
 178 bad_graft_data:
 179         error("bad graft data: %s", line->buf);
 180         assert(!graft);
 181         return NULL;
 182 }
 183
 184 static int read_graft_file(struct repository *r, const char *graft_file)
 185 {
 186         FILE *fp = fopen_or_warn(graft_file, "r");
 187         struct strbuf buf = STRBUF_INIT;
 188         if (!fp)
 189                 return -1;
 190         if (advice_graft_file_deprecated)
 191                 advise(_("Support for <GIT_DIR>/info/grafts is deprecated\n"
 192                          "and will be removed in a future Git version.\n"
 193                          "\n"
 194                          "Please use \"git replace --convert-graft-file\"\n"
 195                          "to convert the grafts into replace refs.\n"
 196                          "\n"
 197                          "Turn this message off by running\n"
 198                          "\"git config advice.graftFileDeprecated false\""));
 199         while (!strbuf_getwholeline(&buf, fp, '\n')) {
 200                 /* The format is just "Commit Parent1 Parent2 ...\n" */
 201                 struct commit_graft *graft = read_graft_line(&buf);
 202                 if (!graft)
 203                         continue;
 204                 if (register_commit_graft(r, graft, 1))
 205                         error("duplicate graft data: %s", buf.buf);
 206         }
 207         fclose(fp);
 208         strbuf_release(&buf);
 209         return 0;
 210 }
 211
 212 static void prepare_commit_graft(struct repository *r)
 213 {
 214         char *graft_file;
 215
 216         if (r->parsed_objects->commit_graft_prepared)
 217                 return;
 218         if (!startup_info->have_repository)
 219                 return;
 220
 221         graft_file = get_graft_file(r);
 222         read_graft_file(r, graft_file);
 223         /* make sure shallows are read */
 224         is_repository_shallow(r);
 225         r->parsed_objects->commit_graft_prepared = 1;
 226 }
 227
 228 struct commit_graft *lookup_commit_graft(struct repository *r, const struct object_id *oid)
 229 {
 230         int pos;
 231         prepare_commit_graft(r);
 232         pos = commit_graft_pos(r, oid->hash);
 233         if (pos < 0)
 234                 return NULL;
 235         return r->parsed_objects->grafts[pos];
 236 }
 237
 238 int for_each_commit_graft(each_commit_graft_fn fn, void *cb_data)
 239 {
 240         int i, ret;
 241         for (i = ret = 0; i < the_repository->parsed_objects->grafts_nr && !ret; i++)
 242                 ret = fn(the_repository->parsed_objects->grafts[i], cb_data);
 243         return ret;
 244 }
 245
 246 int unregister_shallow(const struct object_id *oid)
 247 {
 248         int pos = commit_graft_pos(the_repository, oid->hash);
 249         if (pos < 0)
 250                 return -1;
 251         if (pos + 1 < the_repository->parsed_objects->grafts_nr)
 252                 MOVE_ARRAY(the_repository->parsed_objects->grafts + pos,
 253                            the_repository->parsed_objects->grafts + pos + 1,
 254                            the_repository->parsed_objects->grafts_nr - pos - 1);
 255         the_repository->parsed_objects->grafts_nr--;
 256         return 0;
 257 }
 258
 259 struct commit_buffer {
 260         void *buffer;
 261         unsigned long size;
 262 };
 263 define_commit_slab(buffer_slab, struct commit_buffer);
 264 static struct buffer_slab buffer_slab = COMMIT_SLAB_INIT(1, buffer_slab);
 265
 266 void set_commit_buffer_the_repository(struct commit *commit, void *buffer, unsigned long size)
 267 {
 268         struct commit_buffer *v = buffer_slab_at(&buffer_slab, commit);
 269         v->buffer = buffer;
 270         v->size = size;
 271 }
 272
 273 const void *get_cached_commit_buffer_the_repository(const struct commit *commit, unsigned long *sizep)
 274 {
 275         struct commit_buffer *v = buffer_slab_peek(&buffer_slab, commit);
 276         if (!v) {
 277                 if (sizep)
 278                         *sizep = 0;
 279                 return NULL;
 280         }
 281         if (sizep)
 282                 *sizep = v->size;
 283         return v->buffer;
 284 }
 285
 286 const void *get_commit_buffer(const struct commit *commit, unsigned long *sizep)
 287 {
 288         const void *ret = get_cached_commit_buffer(the_repository, commit, sizep);
 289         if (!ret) {
 290                 enum object_type type;
 291                 unsigned long size;
 292                 ret = read_object_file(&commit->object.oid, &type, &size);
 293                 if (!ret)
 294                         die("cannot read commit object %s",
 295                             oid_to_hex(&commit->object.oid));
 296                 if (type != OBJ_COMMIT)
 297                         die("expected commit for %s, got %s",
 298                             oid_to_hex(&commit->object.oid), type_name(type));
 299                 if (sizep)
 300                         *sizep = size;
 301         }
 302         return ret;
 303 }
 304
 305 void unuse_commit_buffer(const struct commit *commit, const void *buffer)
 306 {
 307         struct commit_buffer *v = buffer_slab_peek(&buffer_slab, commit);
 308         if (!(v && v->buffer == buffer))
 309                 free((void *)buffer);
 310 }
 311
 312 void free_commit_buffer(struct commit *commit)
 313 {
 314         struct commit_buffer *v = buffer_slab_peek(&buffer_slab, commit);
 315         if (v) {
 316                 FREE_AND_NULL(v->buffer);
 317                 v->size = 0;
 318         }
 319 }
 320
 321 struct tree *get_commit_tree(const struct commit *commit)
 322 {
 323         if (commit->maybe_tree || !commit->object.parsed)
 324                 return commit->maybe_tree;
 325
 326         if (commit->graph_pos == COMMIT_NOT_FROM_GRAPH)
 327                 BUG("commit has NULL tree, but was not loaded from commit-graph");
 328
 329         return get_commit_tree_in_graph(commit);
 330 }
 331
 332 struct object_id *get_commit_tree_oid(const struct commit *commit)
 333 {
 334         return &get_commit_tree(commit)->object.oid;
 335 }
 336
 337 void release_commit_memory(struct commit *c)
 338 {
 339         c->maybe_tree = NULL;
 340         c->index = 0;
 341         free_commit_buffer(c);
 342         free_commit_list(c->parents);
 343         /* TODO: what about commit->util? */
 344
 345         c->object.parsed = 0;
 346 }
 347
 348 const void *detach_commit_buffer(struct commit *commit, unsigned long *sizep)
 349 {
 350         struct commit_buffer *v = buffer_slab_peek(&buffer_slab, commit);
 351         void *ret;
 352
 353         if (!v) {
 354                 if (sizep)
 355                         *sizep = 0;
 356                 return NULL;
 357         }
 358         ret = v->buffer;
 359         if (sizep)
 360                 *sizep = v->size;
 361
 362         v->buffer = NULL;
 363         v->size = 0;
 364         return ret;
 365 }
 366
 367 int parse_commit_buffer(struct repository *r, struct commit *item, const void *buffer, unsigned long size, int check_graph)
 368 {
 369         const char *tail = buffer;
 370         const char *bufptr = buffer;
 371         struct object_id parent;
 372         struct commit_list **pptr;
 373         struct commit_graft *graft;
 374         const int tree_entry_len = GIT_SHA1_HEXSZ + 5;
 375         const int parent_entry_len = GIT_SHA1_HEXSZ + 7;
 376
 377         if (item->object.parsed)
 378                 return 0;
 379         item->object.parsed = 1;
 380         tail += size;
 381         if (tail <= bufptr + tree_entry_len + 1 || memcmp(bufptr, "tree ", 5) ||
 382                         bufptr[tree_entry_len] != '\n')
 383                 return error("bogus commit object %s", oid_to_hex(&item->object.oid));
 384         if (get_oid_hex(bufptr + 5, &parent) < 0)
 385                 return error("bad tree pointer in commit %s",
 386                              oid_to_hex(&item->object.oid));
 387         item->maybe_tree = lookup_tree(r, &parent);
 388         bufptr += tree_entry_len + 1; /* "tree " + "hex sha1" + "\n" */
 389         pptr = &item->parents;
 390
 391         graft = lookup_commit_graft(r, &item->object.oid);
 392         while (bufptr + parent_entry_len < tail && !memcmp(bufptr, "parent ", 7)) {
 393                 struct commit *new_parent;
 394
 395                 if (tail <= bufptr + parent_entry_len + 1 ||
 396                     get_oid_hex(bufptr + 7, &parent) ||
 397                     bufptr[parent_entry_len] != '\n')
 398                         return error("bad parents in commit %s", oid_to_hex(&item->object.oid));
 399                 bufptr += parent_entry_len + 1;
 400                 /*
 401                  * The clone is shallow if nr_parent < 0, and we must
 402                  * not traverse its real parents even when we unhide them.
 403                  */
 404                 if (graft && (graft->nr_parent < 0 || grafts_replace_parents))
 405                         continue;
 406                 new_parent = lookup_commit(r, &parent);
 407                 if (new_parent)
 408                         pptr = &commit_list_insert(new_parent, pptr)->next;
 409         }
 410         if (graft) {
 411                 int i;
 412                 struct commit *new_parent;
 413                 for (i = 0; i < graft->nr_parent; i++) {
 414                         new_parent = lookup_commit(r,
 415                                                    &graft->parent[i]);
 416                         if (!new_parent)
 417                                 continue;
 418                         pptr = &commit_list_insert(new_parent, pptr)->next;
 419                 }
 420         }
 421         item->date = parse_commit_date(bufptr, tail);
 422
 423         if (check_graph)
 424                 load_commit_graph_info(item);
 425
 426         return 0;
 427 }
 428
 429 int parse_commit_gently(struct commit *item, int quiet_on_missing)
 430 {
 431         enum object_type type;
 432         void *buffer;
 433         unsigned long size;
 434         int ret;
 435
 436         if (!item)
 437                 return -1;
 438         if (item->object.parsed)
 439                 return 0;
 440         if (parse_commit_in_graph(item))
 441                 return 0;
 442         buffer = read_object_file(&item->object.oid, &type, &size);
 443         if (!buffer)
 444                 return quiet_on_missing ? -1 :
 445                         error("Could not read %s",
 446                              oid_to_hex(&item->object.oid));
 447         if (type != OBJ_COMMIT) {
 448                 free(buffer);
 449                 return error("Object %s not a commit",
 450                              oid_to_hex(&item->object.oid));
 451         }
 452         ret = parse_commit_buffer(the_repository, item, buffer, size, 0);
 453         if (save_commit_buffer && !ret) {
 454                 set_commit_buffer(the_repository, item, buffer, size);
 455                 return 0;
 456         }
 457         free(buffer);
 458         return ret;
 459 }
 460
 461 void parse_commit_or_die(struct commit *item)
 462 {
 463         if (parse_commit(item))
 464                 die("unable to parse commit %s",
 465                     item ? oid_to_hex(&item->object.oid) : "(null)");
 466 }
 467
 468 int find_commit_subject(const char *commit_buffer, const char **subject)
 469 {
 470         const char *eol;
 471         const char *p = commit_buffer;
 472
 473         while (*p && (*p != '\n' || p[1] != '\n'))
 474                 p++;
 475         if (*p) {
 476                 p = skip_blank_lines(p + 2);
 477                 eol = strchrnul(p, '\n');
 478         } else
 479                 eol = p;
 480
 481         *subject = p;
 482
 483         return eol - p;
 484 }
 485
 486 struct commit_list *commit_list_insert(struct commit *item, struct commit_list **list_p)
 487 {
 488         struct commit_list *new_list = xmalloc(sizeof(struct commit_list));
 489         new_list->item = item;
 490         new_list->next = *list_p;
 491         *list_p = new_list;
 492         return new_list;
 493 }
 494
 495 unsigned commit_list_count(const struct commit_list *l)
 496 {
 497         unsigned c = 0;
 498         for (; l; l = l->next )
 499                 c++;
 500         return c;
 501 }
 502
 503 struct commit_list *copy_commit_list(struct commit_list *list)
 504 {
 505         struct commit_list *head = NULL;
 506         struct commit_list **pp = &head;
 507         while (list) {
 508                 pp = commit_list_append(list->item, pp);
 509                 list = list->next;
 510         }
 511         return head;
 512 }
 513
 514 void free_commit_list(struct commit_list *list)
 515 {
 516         while (list)
 517                 pop_commit(&list);
 518 }
 519
 520 struct commit_list * commit_list_insert_by_date(struct commit *item, struct commit_list **list)
 521 {
 522         struct commit_list **pp = list;
 523         struct commit_list *p;
 524         while ((p = *pp) != NULL) {
 525                 if (p->item->date < item->date) {
 526                         break;
 527                 }
 528                 pp = &p->next;
 529         }
 530         return commit_list_insert(item, pp);
 531 }
 532
 533 static int commit_list_compare_by_date(const void *a, const void *b)
 534 {
 535         timestamp_t a_date = ((const struct commit_list *)a)->item->date;
 536         timestamp_t b_date = ((const struct commit_list *)b)->item->date;
 537         if (a_date < b_date)
 538                 return 1;
 539         if (a_date > b_date)
 540                 return -1;
 541         return 0;
 542 }
 543
 544 static void *commit_list_get_next(const void *a)
 545 {
 546         return ((const struct commit_list *)a)->next;
 547 }
 548
 549 static void commit_list_set_next(void *a, void *next)
 550 {
 551         ((struct commit_list *)a)->next = next;
 552 }
 553
 554 void commit_list_sort_by_date(struct commit_list **list)
 555 {
 556         *list = llist_mergesort(*list, commit_list_get_next, commit_list_set_next,
 557                                 commit_list_compare_by_date);
 558 }
 559
 560 struct commit *pop_most_recent_commit(struct commit_list **list,
 561                                       unsigned int mark)
 562 {
 563         struct commit *ret = pop_commit(list);
 564         struct commit_list *parents = ret->parents;
 565
 566         while (parents) {
 567                 struct commit *commit = parents->item;
 568                 if (!parse_commit(commit) && !(commit->object.flags & mark)) {
 569                         commit->object.flags |= mark;
 570                         commit_list_insert_by_date(commit, list);
 571                 }
 572                 parents = parents->next;
 573         }
 574         return ret;
 575 }
 576
 577 static void clear_commit_marks_1(struct commit_list **plist,
 578                                  struct commit *commit, unsigned int mark)
 579 {
 580         while (commit) {
 581                 struct commit_list *parents;
 582
 583                 if (!(mark & commit->object.flags))
 584                         return;
 585
 586                 commit->object.flags &= ~mark;
 587
 588                 parents = commit->parents;
 589                 if (!parents)
 590                         return;
 591
 592                 while ((parents = parents->next))
 593                         commit_list_insert(parents->item, plist);
 594
 595                 commit = commit->parents->item;
 596         }
 597 }
 598
 599 void clear_commit_marks_many(int nr, struct commit **commit, unsigned int mark)
 600 {
 601         struct commit_list *list = NULL;
 602
 603         while (nr--) {
 604                 clear_commit_marks_1(&list, *commit, mark);
 605                 commit++;
 606         }
 607         while (list)
 608                 clear_commit_marks_1(&list, pop_commit(&list), mark);
 609 }
 610
 611 void clear_commit_marks(struct commit *commit, unsigned int mark)
 612 {
 613         clear_commit_marks_many(1, &commit, mark);
 614 }
 615
 616 struct commit *pop_commit(struct commit_list **stack)
 617 {
 618         struct commit_list *top = *stack;
 619         struct commit *item = top ? top->item : NULL;
 620
 621         if (top) {
 622                 *stack = top->next;
 623                 free(top);
 624         }
 625         return item;
 626 }
 627
 628 /*
 629  * Topological sort support
 630  */
 631
 632 /* count number of children that have not been emitted */
 633 define_commit_slab(indegree_slab, int);
 634
 635 /* record author-date for each commit object */
 636 define_commit_slab(author_date_slab, unsigned long);
 637
 638 static void record_author_date(struct author_date_slab *author_date,
 639                                struct commit *commit)
 640 {
 641         const char *buffer = get_commit_buffer(commit, NULL);
 642         struct ident_split ident;
 643         const char *ident_line;
 644         size_t ident_len;
 645         char *date_end;
 646         timestamp_t date;
 647
 648         ident_line = find_commit_header(buffer, "author", &ident_len);
 649         if (!ident_line)
 650                 goto fail_exit; /* no author line */
 651         if (split_ident_line(&ident, ident_line, ident_len) ||
 652             !ident.date_begin || !ident.date_end)
 653                 goto fail_exit; /* malformed "author" line */
 654
 655         date = parse_timestamp(ident.date_begin, &date_end, 10);
 656         if (date_end != ident.date_end)
 657                 goto fail_exit; /* malformed date */
 658         *(author_date_slab_at(author_date, commit)) = date;
 659
 660 fail_exit:
 661         unuse_commit_buffer(commit, buffer);
 662 }
 663
 664 static int compare_commits_by_author_date(const void *a_, const void *b_,
 665                                           void *cb_data)
 666 {
 667         const struct commit *a = a_, *b = b_;
 668         struct author_date_slab *author_date = cb_data;
 669         timestamp_t a_date = *(author_date_slab_at(author_date, a));
 670         timestamp_t b_date = *(author_date_slab_at(author_date, b));
 671
 672         /* newer commits with larger date first */
 673         if (a_date < b_date)
 674                 return 1;
 675         else if (a_date > b_date)
 676                 return -1;
 677         return 0;
 678 }
 679
 680 int compare_commits_by_gen_then_commit_date(const void *a_, const void *b_, void *unused)
 681 {
 682         const struct commit *a = a_, *b = b_;
 683
 684         /* newer commits first */
 685         if (a->generation < b->generation)
 686                 return 1;
 687         else if (a->generation > b->generation)
 688                 return -1;
 689
 690         /* use date as a heuristic when generations are equal */
 691         if (a->date < b->date)
 692                 return 1;
 693         else if (a->date > b->date)
 694                 return -1;
 695         return 0;
 696 }
 697
 698 int compare_commits_by_commit_date(const void *a_, const void *b_, void *unused)
 699 {
 700         const struct commit *a = a_, *b = b_;
 701         /* newer commits with larger date first */
 702         if (a->date < b->date)
 703                 return 1;
 704         else if (a->date > b->date)
 705                 return -1;
 706         return 0;
 707 }
 708
 709 /*
 710  * Performs an in-place topological sort on the list supplied.
 711  */
 712 void sort_in_topological_order(struct commit_list **list, enum rev_sort_order sort_order)
 713 {
 714         struct commit_list *next, *orig = *list;
 715         struct commit_list **pptr;
 716         struct indegree_slab indegree;
 717         struct prio_queue queue;
 718         struct commit *commit;
 719         struct author_date_slab author_date;
 720
 721         if (!orig)
 722                 return;
 723         *list = NULL;
 724
 725         init_indegree_slab(&indegree);
 726         memset(&queue, '\0', sizeof(queue));
 727
 728         switch (sort_order) {
 729         default: /* REV_SORT_IN_GRAPH_ORDER */
 730                 queue.compare = NULL;
 731                 break;
 732         case REV_SORT_BY_COMMIT_DATE:
 733                 queue.compare = compare_commits_by_commit_date;
 734                 break;
 735         case REV_SORT_BY_AUTHOR_DATE:
 736                 init_author_date_slab(&author_date);
 737                 queue.compare = compare_commits_by_author_date;
 738                 queue.cb_data = &author_date;
 739                 break;
 740         }
 741
 742         /* Mark them and clear the indegree */
 743         for (next = orig; next; next = next->next) {
 744                 struct commit *commit = next->item;
 745                 *(indegree_slab_at(&indegree, commit)) = 1;
 746                 /* also record the author dates, if needed */
 747                 if (sort_order == REV_SORT_BY_AUTHOR_DATE)
 748                         record_author_date(&author_date, commit);
 749         }
 750
 751         /* update the indegree */
 752         for (next = orig; next; next = next->next) {
 753                 struct commit_list *parents = next->item->parents;
 754                 while (parents) {
 755                         struct commit *parent = parents->item;
 756                         int *pi = indegree_slab_at(&indegree, parent);
 757
 758                         if (*pi)
 759                                 (*pi)++;
 760                         parents = parents->next;
 761                 }
 762         }
 763
 764         /*
 765          * find the tips
 766          *
 767          * tips are nodes not reachable from any other node in the list
 768          *
 769          * the tips serve as a starting set for the work queue.
 770          */
 771         for (next = orig; next; next = next->next) {
 772                 struct commit *commit = next->item;
 773
 774                 if (*(indegree_slab_at(&indegree, commit)) == 1)
 775                         prio_queue_put(&queue, commit);
 776         }
 777
 778         /*
 779          * This is unfortunate; the initial tips need to be shown
 780          * in the order given from the revision traversal machinery.
 781          */
 782         if (sort_order == REV_SORT_IN_GRAPH_ORDER)
 783                 prio_queue_reverse(&queue);
 784
 785         /* We no longer need the commit list */
 786         free_commit_list(orig);
 787
 788         pptr = list;
 789         *list = NULL;
 790         while ((commit = prio_queue_get(&queue)) != NULL) {
 791                 struct commit_list *parents;
 792
 793                 for (parents = commit->parents; parents ; parents = parents->next) {
 794                         struct commit *parent = parents->item;
 795                         int *pi = indegree_slab_at(&indegree, parent);
 796
 797                         if (!*pi)
 798                                 continue;
 799
 800                         /*
 801                          * parents are only enqueued for emission
 802                          * when all their children have been emitted thereby
 803                          * guaranteeing topological order.
 804                          */
 805                         if (--(*pi) == 1)
 806                                 prio_queue_put(&queue, parent);
 807                 }
 808                 /*
 809                  * all children of commit have already been
 810                  * emitted. we can emit it now.
 811                  */
 812                 *(indegree_slab_at(&indegree, commit)) = 0;
 813
 814                 pptr = &commit_list_insert(commit, pptr)->next;
 815         }
 816
 817         clear_indegree_slab(&indegree);
 818         clear_prio_queue(&queue);
 819         if (sort_order == REV_SORT_BY_AUTHOR_DATE)
 820                 clear_author_date_slab(&author_date);
 821 }
 822
 823 /* merge-base stuff */
 824
 825 /* Remember to update object flag allocation in object.h */
 826 #define PARENT1         (1u<<16)
 827 #define PARENT2         (1u<<17)
 828 #define STALE           (1u<<18)
 829 #define RESULT          (1u<<19)
 830
 831 static const unsigned all_flags = (PARENT1 | PARENT2 | STALE | RESULT);
 832
 833 static int queue_has_nonstale(struct prio_queue *queue)
 834 {
 835         int i;
 836         for (i = 0; i < queue->nr; i++) {
 837                 struct commit *commit = queue->array[i].data;
 838                 if (!(commit->object.flags & STALE))
 839                         return 1;
 840         }
 841         return 0;
 842 }
 843
 844 /* all input commits in one and twos[] must have been parsed! */
 845 static struct commit_list *paint_down_to_common(struct commit *one, int n,
 846                                                 struct commit **twos,
 847                                                 int min_generation)
 848 {
 849         struct prio_queue queue = { compare_commits_by_gen_then_commit_date };
 850         struct commit_list *result = NULL;
 851         int i;
 852         uint32_t last_gen = GENERATION_NUMBER_INFINITY;
 853
 854         one->object.flags |= PARENT1;
 855         if (!n) {
 856                 commit_list_append(one, &result);
 857                 return result;
 858         }
 859         prio_queue_put(&queue, one);
 860
 861         for (i = 0; i < n; i++) {
 862                 twos[i]->object.flags |= PARENT2;
 863                 prio_queue_put(&queue, twos[i]);
 864         }
 865
 866         while (queue_has_nonstale(&queue)) {
 867                 struct commit *commit = prio_queue_get(&queue);
 868                 struct commit_list *parents;
 869                 int flags;
 870
 871                 if (commit->generation > last_gen)
 872                         BUG("bad generation skip %8x > %8x at %s",
 873                             commit->generation, last_gen,
 874                             oid_to_hex(&commit->object.oid));
 875                 last_gen = commit->generation;
 876
 877                 if (commit->generation < min_generation)
 878                         break;
 879
 880                 flags = commit->object.flags & (PARENT1 | PARENT2 | STALE);
 881                 if (flags == (PARENT1 | PARENT2)) {
 882                         if (!(commit->object.flags & RESULT)) {
 883                                 commit->object.flags |= RESULT;
 884                                 commit_list_insert_by_date(commit, &result);
 885                         }
 886                         /* Mark parents of a found merge stale */
 887                         flags |= STALE;
 888                 }
 889                 parents = commit->parents;
 890                 while (parents) {
 891                         struct commit *p = parents->item;
 892                         parents = parents->next;
 893                         if ((p->object.flags & flags) == flags)
 894                                 continue;
 895                         if (parse_commit(p))
 896                                 return NULL;
 897                         p->object.flags |= flags;
 898                         prio_queue_put(&queue, p);
 899                 }
 900         }
 901
 902         clear_prio_queue(&queue);
 903         return result;
 904 }
 905
 906 static struct commit_list *merge_bases_many(struct commit *one, int n, struct commit **twos)
 907 {
 908         struct commit_list *list = NULL;
 909         struct commit_list *result = NULL;
 910         int i;
 911
 912         for (i = 0; i < n; i++) {
 913                 if (one == twos[i])
 914                         /*
 915                          * We do not mark this even with RESULT so we do not
 916                          * have to clean it up.
 917                          */
 918                         return commit_list_insert(one, &result);
 919         }
 920
 921         if (parse_commit(one))
 922                 return NULL;
 923         for (i = 0; i < n; i++) {
 924                 if (parse_commit(twos[i]))
 925                         return NULL;
 926         }
 927
 928         list = paint_down_to_common(one, n, twos, 0);
 929
 930         while (list) {
 931                 struct commit *commit = pop_commit(&list);
 932                 if (!(commit->object.flags & STALE))
 933                         commit_list_insert_by_date(commit, &result);
 934         }
 935         return result;
 936 }
 937
 938 struct commit_list *get_octopus_merge_bases(struct commit_list *in)
 939 {
 940         struct commit_list *i, *j, *k, *ret = NULL;
 941
 942         if (!in)
 943                 return ret;
 944
 945         commit_list_insert(in->item, &ret);
 946
 947         for (i = in->next; i; i = i->next) {
 948                 struct commit_list *new_commits = NULL, *end = NULL;
 949
 950                 for (j = ret; j; j = j->next) {
 951                         struct commit_list *bases;
 952                         bases = get_merge_bases(i->item, j->item);
 953                         if (!new_commits)
 954                                 new_commits = bases;
 955                         else
 956                                 end->next = bases;
 957                         for (k = bases; k; k = k->next)
 958                                 end = k;
 959                 }
 960                 ret = new_commits;
 961         }
 962         return ret;
 963 }
 964
 965 static int remove_redundant(struct commit **array, int cnt)
 966 {
 967         /*
 968          * Some commit in the array may be an ancestor of
 969          * another commit.  Move such commit to the end of
 970          * the array, and return the number of commits that
 971          * are independent from each other.
 972          */
 973         struct commit **work;
 974         unsigned char *redundant;
 975         int *filled_index;
 976         int i, j, filled;
 977
 978         work = xcalloc(cnt, sizeof(*work));
 979         redundant = xcalloc(cnt, 1);
 980         ALLOC_ARRAY(filled_index, cnt - 1);
 981
 982         for (i = 0; i < cnt; i++)
 983                 parse_commit(array[i]);
 984         for (i = 0; i < cnt; i++) {
 985                 struct commit_list *common;
 986                 uint32_t min_generation = array[i]->generation;
 987
 988                 if (redundant[i])
 989                         continue;
 990                 for (j = filled = 0; j < cnt; j++) {
 991                         if (i == j || redundant[j])
 992                                 continue;
 993                         filled_index[filled] = j;
 994                         work[filled++] = array[j];
 995
 996                         if (array[j]->generation < min_generation)
 997                                 min_generation = array[j]->generation;
 998                 }
 999                 common = paint_down_to_common(array[i], filled, work,
1000                                               min_generation);
1001                 if (array[i]->object.flags & PARENT2)
1002                         redundant[i] = 1;
1003                 for (j = 0; j < filled; j++)
1004                         if (work[j]->object.flags & PARENT1)
1005                                 redundant[filled_index[j]] = 1;
1006                 clear_commit_marks(array[i], all_flags);
1007                 clear_commit_marks_many(filled, work, all_flags);
1008                 free_commit_list(common);
1009         }
1010
1011         /* Now collect the result */
1012         COPY_ARRAY(work, array, cnt);
1013         for (i = filled = 0; i < cnt; i++)
1014                 if (!redundant[i])
1015                         array[filled++] = work[i];
1016         for (j = filled, i = 0; i < cnt; i++)
1017                 if (redundant[i])
1018                         array[j++] = work[i];
1019         free(work);
1020         free(redundant);
1021         free(filled_index);
1022         return filled;
1023 }
1024
1025 static struct commit_list *get_merge_bases_many_0(struct commit *one,
1026                                                   int n,
1027                                                   struct commit **twos,
1028                                                   int cleanup)
1029 {
1030         struct commit_list *list;
1031         struct commit **rslt;
1032         struct commit_list *result;
1033         int cnt, i;
1034
1035         result = merge_bases_many(one, n, twos);
1036         for (i = 0; i < n; i++) {
1037                 if (one == twos[i])
1038                         return result;
1039         }
1040         if (!result || !result->next) {
1041                 if (cleanup) {
1042                         clear_commit_marks(one, all_flags);
1043                         clear_commit_marks_many(n, twos, all_flags);
1044                 }
1045                 return result;
1046         }
1047
1048         /* There are more than one */
1049         cnt = commit_list_count(result);
1050         rslt = xcalloc(cnt, sizeof(*rslt));
1051         for (list = result, i = 0; list; list = list->next)
1052                 rslt[i++] = list->item;
1053         free_commit_list(result);
1054
1055         clear_commit_marks(one, all_flags);
1056         clear_commit_marks_many(n, twos, all_flags);
1057
1058         cnt = remove_redundant(rslt, cnt);
1059         result = NULL;
1060         for (i = 0; i < cnt; i++)
1061                 commit_list_insert_by_date(rslt[i], &result);
1062         free(rslt);
1063         return result;
1064 }
1065
1066 struct commit_list *get_merge_bases_many(struct commit *one,
1067                                          int n,
1068                                          struct commit **twos)
1069 {
1070         return get_merge_bases_many_0(one, n, twos, 1);
1071 }
1072
1073 struct commit_list *get_merge_bases_many_dirty(struct commit *one,
1074                                                int n,
1075                                                struct commit **twos)
1076 {
1077         return get_merge_bases_many_0(one, n, twos, 0);
1078 }
1079
1080 struct commit_list *get_merge_bases(struct commit *one, struct commit *two)
1081 {
1082         return get_merge_bases_many_0(one, 1, &two, 1);
1083 }
1084
1085 /*
1086  * Is "commit" a descendant of one of the elements on the "with_commit" list?
1087  */
1088 int is_descendant_of(struct commit *commit, struct commit_list *with_commit)
1089 {
1090         if (!with_commit)
1091                 return 1;
1092         while (with_commit) {
1093                 struct commit *other;
1094
1095                 other = with_commit->item;
1096                 with_commit = with_commit->next;
1097                 if (in_merge_bases(other, commit))
1098                         return 1;
1099         }
1100         return 0;
1101 }
1102
1103 /*
1104  * Is "commit" an ancestor of one of the "references"?
1105  */
1106 int in_merge_bases_many(struct commit *commit, int nr_reference, struct commit **reference)
1107 {
1108         struct commit_list *bases;
1109         int ret = 0, i;
1110         uint32_t min_generation = GENERATION_NUMBER_INFINITY;
1111
1112         if (parse_commit(commit))
1113                 return ret;
1114         for (i = 0; i < nr_reference; i++) {
1115                 if (parse_commit(reference[i]))
1116                         return ret;
1117                 if (reference[i]->generation < min_generation)
1118                         min_generation = reference[i]->generation;
1119         }
1120
1121         if (commit->generation > min_generation)
1122                 return ret;
1123
1124         bases = paint_down_to_common(commit, nr_reference, reference, commit->generation);
1125         if (commit->object.flags & PARENT2)
1126                 ret = 1;
1127         clear_commit_marks(commit, all_flags);
1128         clear_commit_marks_many(nr_reference, reference, all_flags);
1129         free_commit_list(bases);
1130         return ret;
1131 }
1132
1133 /*
1134  * Is "commit" an ancestor of (i.e. reachable from) the "reference"?
1135  */
1136 int in_merge_bases(struct commit *commit, struct commit *reference)
1137 {
1138         return in_merge_bases_many(commit, 1, &reference);
1139 }
1140
1141 struct commit_list *reduce_heads(struct commit_list *heads)
1142 {
1143         struct commit_list *p;
1144         struct commit_list *result = NULL, **tail = &result;
1145         struct commit **array;
1146         int num_head, i;
1147
1148         if (!heads)
1149                 return NULL;
1150
1151         /* Uniquify */
1152         for (p = heads; p; p = p->next)
1153                 p->item->object.flags &= ~STALE;
1154         for (p = heads, num_head = 0; p; p = p->next) {
1155                 if (p->item->object.flags & STALE)
1156                         continue;
1157                 p->item->object.flags |= STALE;
1158                 num_head++;
1159         }
1160         array = xcalloc(num_head, sizeof(*array));
1161         for (p = heads, i = 0; p; p = p->next) {
1162                 if (p->item->object.flags & STALE) {
1163                         array[i++] = p->item;
1164                         p->item->object.flags &= ~STALE;
1165                 }
1166         }
1167         num_head = remove_redundant(array, num_head);
1168         for (i = 0; i < num_head; i++)
1169                 tail = &commit_list_insert(array[i], tail)->next;
1170         free(array);
1171         return result;
1172 }
1173
1174 void reduce_heads_replace(struct commit_list **heads)
1175 {
1176         struct commit_list *result = reduce_heads(*heads);
1177         free_commit_list(*heads);
1178         *heads = result;
1179 }
1180
1181 static const char gpg_sig_header[] = "gpgsig";
1182 static const int gpg_sig_header_len = sizeof(gpg_sig_header) - 1;
1183
1184 static int do_sign_commit(struct strbuf *buf, const char *keyid)
1185 {
1186         struct strbuf sig = STRBUF_INIT;
1187         int inspos, copypos;
1188         const char *eoh;
1189
1190         /* find the end of the header */
1191         eoh = strstr(buf->buf, "\n\n");
1192         if (!eoh)
1193                 inspos = buf->len;
1194         else
1195                 inspos = eoh - buf->buf + 1;
1196
1197         if (!keyid || !*keyid)
1198                 keyid = get_signing_key();
1199         if (sign_buffer(buf, &sig, keyid)) {
1200                 strbuf_release(&sig);
1201                 return -1;
1202         }
1203
1204         for (copypos = 0; sig.buf[copypos]; ) {
1205                 const char *bol = sig.buf + copypos;
1206                 const char *eol = strchrnul(bol, '\n');
1207                 int len = (eol - bol) + !!*eol;
1208
1209                 if (!copypos) {
1210                         strbuf_insert(buf, inspos, gpg_sig_header, gpg_sig_header_len);
1211                         inspos += gpg_sig_header_len;
1212                 }
1213                 strbuf_insert(buf, inspos++, " ", 1);
1214                 strbuf_insert(buf, inspos, bol, len);
1215                 inspos += len;
1216                 copypos += len;
1217         }
1218         strbuf_release(&sig);
1219         return 0;
1220 }
1221
1222 int parse_signed_commit(const struct commit *commit,
1223                         struct strbuf *payload, struct strbuf *signature)
1224 {
1225
1226         unsigned long size;
1227         const char *buffer = get_commit_buffer(commit, &size);
1228         int in_signature, saw_signature = -1;
1229         const char *line, *tail;
1230
1231         line = buffer;
1232         tail = buffer + size;
1233         in_signature = 0;
1234         saw_signature = 0;
1235         while (line < tail) {
1236                 const char *sig = NULL;
1237                 const char *next = memchr(line, '\n', tail - line);
1238
1239                 next = next ? next + 1 : tail;
1240                 if (in_signature && line[0] == ' ')
1241                         sig = line + 1;
1242                 else if (starts_with(line, gpg_sig_header) &&
1243                          line[gpg_sig_header_len] == ' ')
1244                         sig = line + gpg_sig_header_len + 1;
1245                 if (sig) {
1246                         strbuf_add(signature, sig, next - sig);
1247                         saw_signature = 1;
1248                         in_signature = 1;
1249                 } else {
1250                         if (*line == '\n')
1251                                 /* dump the whole remainder of the buffer */
1252                                 next = tail;
1253                         strbuf_add(payload, line, next - line);
1254                         in_signature = 0;
1255                 }
1256                 line = next;
1257         }
1258         unuse_commit_buffer(commit, buffer);
1259         return saw_signature;
1260 }
1261
1262 int remove_signature(struct strbuf *buf)
1263 {
1264         const char *line = buf->buf;
1265         const char *tail = buf->buf + buf->len;
1266         int in_signature = 0;
1267         const char *sig_start = NULL;
1268         const char *sig_end = NULL;
1269
1270         while (line < tail) {
1271                 const char *next = memchr(line, '\n', tail - line);
1272                 next = next ? next + 1 : tail;
1273
1274                 if (in_signature && line[0] == ' ')
1275                         sig_end = next;
1276                 else if (starts_with(line, gpg_sig_header) &&
1277                          line[gpg_sig_header_len] == ' ') {
1278                         sig_start = line;
1279                         sig_end = next;
1280                         in_signature = 1;
1281                 } else {
1282                         if (*line == '\n')
1283                                 /* dump the whole remainder of the buffer */
1284                                 next = tail;
1285                         in_signature = 0;
1286                 }
1287                 line = next;
1288         }
1289
1290         if (sig_start)
1291                 strbuf_remove(buf, sig_start - buf->buf, sig_end - sig_start);
1292
1293         return sig_start != NULL;
1294 }
1295
1296 static void handle_signed_tag(struct commit *parent, struct commit_extra_header ***tail)
1297 {
1298         struct merge_remote_desc *desc;
1299         struct commit_extra_header *mergetag;
1300         char *buf;
1301         unsigned long size, len;
1302         enum object_type type;
1303
1304         desc = merge_remote_util(parent);
1305         if (!desc || !desc->obj)
1306                 return;
1307         buf = read_object_file(&desc->obj->oid, &type, &size);
1308         if (!buf || type != OBJ_TAG)
1309                 goto free_return;
1310         len = parse_signature(buf, size);
1311         if (size == len)
1312                 goto free_return;
1313         /*
1314          * We could verify this signature and either omit the tag when
1315          * it does not validate, but the integrator may not have the
1316          * public key of the signer of the tag he is merging, while a
1317          * later auditor may have it while auditing, so let's not run
1318          * verify-signed-buffer here for now...
1319          *
1320          * if (verify_signed_buffer(buf, len, buf + len, size - len, ...))
1321          *      warn("warning: signed tag unverified.");
1322          */
1323         mergetag = xcalloc(1, sizeof(*mergetag));
1324         mergetag->key = xstrdup("mergetag");
1325         mergetag->value = buf;
1326         mergetag->len = size;
1327
1328         **tail = mergetag;
1329         *tail = &mergetag->next;
1330         return;
1331
1332 free_return:
1333         free(buf);
1334 }
1335
1336 int check_commit_signature(const struct commit *commit, struct signature_check *sigc)
1337 {
1338         struct strbuf payload = STRBUF_INIT;
1339         struct strbuf signature = STRBUF_INIT;
1340         int ret = 1;
1341
1342         sigc->result = 'N';
1343
1344         if (parse_signed_commit(commit, &payload, &signature) <= 0)
1345                 goto out;
1346         ret = check_signature(payload.buf, payload.len, signature.buf,
1347                 signature.len, sigc);
1348
1349  out:
1350         strbuf_release(&payload);
1351         strbuf_release(&signature);
1352
1353         return ret;
1354 }
1355
1356
1357
1358 void append_merge_tag_headers(struct commit_list *parents,
1359                               struct commit_extra_header ***tail)
1360 {
1361         while (parents) {
1362                 struct commit *parent = parents->item;
1363                 handle_signed_tag(parent, tail);
1364                 parents = parents->next;
1365         }
1366 }
1367
1368 static void add_extra_header(struct strbuf *buffer,
1369                              struct commit_extra_header *extra)
1370 {
1371         strbuf_addstr(buffer, extra->key);
1372         if (extra->len)
1373                 strbuf_add_lines(buffer, " ", extra->value, extra->len);
1374         else
1375                 strbuf_addch(buffer, '\n');
1376 }
1377
1378 struct commit_extra_header *read_commit_extra_headers(struct commit *commit,
1379                                                       const char **exclude)
1380 {
1381         struct commit_extra_header *extra = NULL;
1382         unsigned long size;
1383         const char *buffer = get_commit_buffer(commit, &size);
1384         extra = read_commit_extra_header_lines(buffer, size, exclude);
1385         unuse_commit_buffer(commit, buffer);
1386         return extra;
1387 }
1388
1389 int for_each_mergetag(each_mergetag_fn fn, struct commit *commit, void *data)
1390 {
1391         struct commit_extra_header *extra, *to_free;
1392         int res = 0;
1393
1394         to_free = read_commit_extra_headers(commit, NULL);
1395         for (extra = to_free; !res && extra; extra = extra->next) {
1396                 if (strcmp(extra->key, "mergetag"))
1397                         continue; /* not a merge tag */
1398                 res = fn(commit, extra, data);
1399         }
1400         free_commit_extra_headers(to_free);
1401         return res;
1402 }
1403
1404 static inline int standard_header_field(const char *field, size_t len)
1405 {
1406         return ((len == 4 && !memcmp(field, "tree", 4)) ||
1407                 (len == 6 && !memcmp(field, "parent", 6)) ||
1408                 (len == 6 && !memcmp(field, "author", 6)) ||
1409                 (len == 9 && !memcmp(field, "committer", 9)) ||
1410                 (len == 8 && !memcmp(field, "encoding", 8)));
1411 }
1412
1413 static int excluded_header_field(const char *field, size_t len, const char **exclude)
1414 {
1415         if (!exclude)
1416                 return 0;
1417
1418         while (*exclude) {
1419                 size_t xlen = strlen(*exclude);
1420                 if (len == xlen && !memcmp(field, *exclude, xlen))
1421                         return 1;
1422                 exclude++;
1423         }
1424         return 0;
1425 }
1426
1427 static struct commit_extra_header *read_commit_extra_header_lines(
1428         const char *buffer, size_t size,
1429         const char **exclude)
1430 {
1431         struct commit_extra_header *extra = NULL, **tail = &extra, *it = NULL;
1432         const char *line, *next, *eof, *eob;
1433         struct strbuf buf = STRBUF_INIT;
1434
1435         for (line = buffer, eob = line + size;
1436              line < eob && *line != '\n';
1437              line = next) {
1438                 next = memchr(line, '\n', eob - line);
1439                 next = next ? next + 1 : eob;
1440                 if (*line == ' ') {
1441                         /* continuation */
1442                         if (it)
1443                                 strbuf_add(&buf, line + 1, next - (line + 1));
1444                         continue;
1445                 }
1446                 if (it)
1447                         it->value = strbuf_detach(&buf, &it->len);
1448                 strbuf_reset(&buf);
1449                 it = NULL;
1450
1451                 eof = memchr(line, ' ', next - line);
1452                 if (!eof)
1453                         eof = next;
1454                 else if (standard_header_field(line, eof - line) ||
1455                          excluded_header_field(line, eof - line, exclude))
1456                         continue;
1457
1458                 it = xcalloc(1, sizeof(*it));
1459                 it->key = xmemdupz(line, eof-line);
1460                 *tail = it;
1461                 tail = &it->next;
1462                 if (eof + 1 < next)
1463                         strbuf_add(&buf, eof + 1, next - (eof + 1));
1464         }
1465         if (it)
1466                 it->value = strbuf_detach(&buf, &it->len);
1467         return extra;
1468 }
1469
1470 void free_commit_extra_headers(struct commit_extra_header *extra)
1471 {
1472         while (extra) {
1473                 struct commit_extra_header *next = extra->next;
1474                 free(extra->key);
1475                 free(extra->value);
1476                 free(extra);
1477                 extra = next;
1478         }
1479 }
1480
1481 int commit_tree(const char *msg, size_t msg_len, const struct object_id *tree,
1482                 struct commit_list *parents, struct object_id *ret,
1483                 const char *author, const char *sign_commit)
1484 {
1485         struct commit_extra_header *extra = NULL, **tail = &extra;
1486         int result;
1487
1488         append_merge_tag_headers(parents, &tail);
1489         result = commit_tree_extended(msg, msg_len, tree, parents, ret,
1490                                       author, sign_commit, extra);
1491         free_commit_extra_headers(extra);
1492         return result;
1493 }
1494
1495 static int find_invalid_utf8(const char *buf, int len)
1496 {
1497         int offset = 0;
1498         static const unsigned int max_codepoint[] = {
1499                 0x7f, 0x7ff, 0xffff, 0x10ffff
1500         };
1501
1502         while (len) {
1503                 unsigned char c = *buf++;
1504                 int bytes, bad_offset;
1505                 unsigned int codepoint;
1506                 unsigned int min_val, max_val;
1507
1508                 len--;
1509                 offset++;
1510
1511                 /* Simple US-ASCII? No worries. */
1512                 if (c < 0x80)
1513                         continue;
1514
1515                 bad_offset = offset-1;
1516
1517                 /*
1518                  * Count how many more high bits set: that's how
1519                  * many more bytes this sequence should have.
1520                  */
1521                 bytes = 0;
1522                 while (c & 0x40) {
1523                         c <<= 1;
1524                         bytes++;
1525                 }
1526
1527                 /*
1528                  * Must be between 1 and 3 more bytes.  Longer sequences result in
1529                  * codepoints beyond U+10FFFF, which are guaranteed never to exist.
1530                  */
1531                 if (bytes < 1 || 3 < bytes)
1532                         return bad_offset;
1533
1534                 /* Do we *have* that many bytes? */
1535                 if (len < bytes)
1536                         return bad_offset;
1537
1538                 /*
1539                  * Place the encoded bits at the bottom of the value and compute the
1540                  * valid range.
1541                  */
1542                 codepoint = (c & 0x7f) >> bytes;
1543                 min_val = max_codepoint[bytes-1] + 1;
1544                 max_val = max_codepoint[bytes];
1545
1546                 offset += bytes;
1547                 len -= bytes;
1548
1549                 /* And verify that they are good continuation bytes */
1550                 do {
1551                         codepoint <<= 6;
1552                         codepoint |= *buf & 0x3f;
1553                         if ((*buf++ & 0xc0) != 0x80)
1554                                 return bad_offset;
1555                 } while (--bytes);
1556
1557                 /* Reject codepoints that are out of range for the sequence length. */
1558                 if (codepoint < min_val || codepoint > max_val)
1559                         return bad_offset;
1560                 /* Surrogates are only for UTF-16 and cannot be encoded in UTF-8. */
1561                 if ((codepoint & 0x1ff800) == 0xd800)
1562                         return bad_offset;
1563                 /* U+xxFFFE and U+xxFFFF are guaranteed non-characters. */
1564                 if ((codepoint & 0xfffe) == 0xfffe)
1565                         return bad_offset;
1566                 /* So are anything in the range U+FDD0..U+FDEF. */
1567                 if (codepoint >= 0xfdd0 && codepoint <= 0xfdef)
1568                         return bad_offset;
1569         }
1570         return -1;
1571 }
1572
1573 /*
1574  * This verifies that the buffer is in proper utf8 format.
1575  *
1576  * If it isn't, it assumes any non-utf8 characters are Latin1,
1577  * and does the conversion.
1578  */
1579 static int verify_utf8(struct strbuf *buf)
1580 {
1581         int ok = 1;
1582         long pos = 0;
1583
1584         for (;;) {
1585                 int bad;
1586                 unsigned char c;
1587                 unsigned char replace[2];
1588
1589                 bad = find_invalid_utf8(buf->buf + pos, buf->len - pos);
1590                 if (bad < 0)
1591                         return ok;
1592                 pos += bad;
1593                 ok = 0;
1594                 c = buf->buf[pos];
1595                 strbuf_remove(buf, pos, 1);
1596
1597                 /* We know 'c' must be in the range 128-255 */
1598                 replace[0] = 0xc0 + (c >> 6);
1599                 replace[1] = 0x80 + (c & 0x3f);
1600                 strbuf_insert(buf, pos, replace, 2);
1601                 pos += 2;
1602         }
1603 }
1604
1605 static const char commit_utf8_warn[] =
1606 N_("Warning: commit message did not conform to UTF-8.\n"
1607    "You may want to amend it after fixing the message, or set the config\n"
1608    "variable i18n.commitencoding to the encoding your project uses.\n");
1609
1610 int commit_tree_extended(const char *msg, size_t msg_len,
1611                          const struct object_id *tree,
1612                          struct commit_list *parents, struct object_id *ret,
1613                          const char *author, const char *sign_commit,
1614                          struct commit_extra_header *extra)
1615 {
1616         int result;
1617         int encoding_is_utf8;
1618         struct strbuf buffer;
1619
1620         assert_oid_type(tree, OBJ_TREE);
1621
1622         if (memchr(msg, '\0', msg_len))
1623                 return error("a NUL byte in commit log message not allowed.");
1624
1625         /* Not having i18n.commitencoding is the same as having utf-8 */
1626         encoding_is_utf8 = is_encoding_utf8(git_commit_encoding);
1627
1628         strbuf_init(&buffer, 8192); /* should avoid reallocs for the headers */
1629         strbuf_addf(&buffer, "tree %s\n", oid_to_hex(tree));
1630
1631         /*
1632          * NOTE! This ordering means that the same exact tree merged with a
1633          * different order of parents will be a _different_ changeset even
1634          * if everything else stays the same.
1635          */
1636         while (parents) {
1637                 struct commit *parent = pop_commit(&parents);
1638                 strbuf_addf(&buffer, "parent %s\n",
1639                             oid_to_hex(&parent->object.oid));
1640         }
1641
1642         /* Person/date information */
1643         if (!author)
1644                 author = git_author_info(IDENT_STRICT);
1645         strbuf_addf(&buffer, "author %s\n", author);
1646         strbuf_addf(&buffer, "committer %s\n", git_committer_info(IDENT_STRICT));
1647         if (!encoding_is_utf8)
1648                 strbuf_addf(&buffer, "encoding %s\n", git_commit_encoding);
1649
1650         while (extra) {
1651                 add_extra_header(&buffer, extra);
1652                 extra = extra->next;
1653         }
1654         strbuf_addch(&buffer, '\n');
1655
1656         /* And add the comment */
1657         strbuf_add(&buffer, msg, msg_len);
1658
1659         /* And check the encoding */
1660         if (encoding_is_utf8 && !verify_utf8(&buffer))
1661                 fprintf(stderr, _(commit_utf8_warn));
1662
1663         if (sign_commit && do_sign_commit(&buffer, sign_commit)) {
1664                 result = -1;
1665                 goto out;
1666         }
1667
1668         result = write_object_file(buffer.buf, buffer.len, commit_type, ret);
1669 out:
1670         strbuf_release(&buffer);
1671         return result;
1672 }
1673
1674 define_commit_slab(merge_desc_slab, struct merge_remote_desc *);
1675 static struct merge_desc_slab merge_desc_slab = COMMIT_SLAB_INIT(1, merge_desc_slab);
1676
1677 struct merge_remote_desc *merge_remote_util(struct commit *commit)
1678 {
1679         return *merge_desc_slab_at(&merge_desc_slab, commit);
1680 }
1681
1682 void set_merge_remote_desc(struct commit *commit,
1683                            const char *name, struct object *obj)
1684 {
1685         struct merge_remote_desc *desc;
1686         FLEX_ALLOC_STR(desc, name, name);
1687         desc->obj = obj;
1688         *merge_desc_slab_at(&merge_desc_slab, commit) = desc;
1689 }
1690
1691 struct commit *get_merge_parent(const char *name)
1692 {
1693         struct object *obj;
1694         struct commit *commit;
1695         struct object_id oid;
1696         if (get_oid(name, &oid))
1697                 return NULL;
1698         obj = parse_object(the_repository, &oid);
1699         commit = (struct commit *)peel_to_type(name, 0, obj, OBJ_COMMIT);
1700         if (commit && !merge_remote_util(commit))
1701                 set_merge_remote_desc(commit, name, obj);
1702         return commit;
1703 }
1704
1705 /*
1706  * Append a commit to the end of the commit_list.
1707  *
1708  * next starts by pointing to the variable that holds the head of an
1709  * empty commit_list, and is updated to point to the "next" field of
1710  * the last item on the list as new commits are appended.
1711  *
1712  * Usage example:
1713  *
1714  *     struct commit_list *list;
1715  *     struct commit_list **next = &list;
1716  *
1717  *     next = commit_list_append(c1, next);
1718  *     next = commit_list_append(c2, next);
1719  *     assert(commit_list_count(list) == 2);
1720  *     return list;
1721  */
1722 struct commit_list **commit_list_append(struct commit *commit,
1723                                         struct commit_list **next)
1724 {
1725         struct commit_list *new_commit = xmalloc(sizeof(struct commit_list));
1726         new_commit->item = commit;
1727         *next = new_commit;
1728         new_commit->next = NULL;
1729         return &new_commit->next;
1730 }
1731
1732 const char *find_commit_header(const char *msg, const char *key, size_t *out_len)
1733 {
1734         int key_len = strlen(key);
1735         const char *line = msg;
1736
1737         while (line) {
1738                 const char *eol = strchrnul(line, '\n');
1739
1740                 if (line == eol)
1741                         return NULL;
1742
1743                 if (eol - line > key_len &&
1744                     !strncmp(line, key, key_len) &&
1745                     line[key_len] == ' ') {
1746                         *out_len = eol - line - key_len - 1;
1747                         return line + key_len + 1;
1748                 }
1749                 line = *eol ? eol + 1 : NULL;
1750         }
1751         return NULL;
1752 }
1753
1754 /*
1755  * Inspect the given string and determine the true "end" of the log message, in
1756  * order to find where to put a new Signed-off-by: line.  Ignored are
1757  * trailing comment lines and blank lines.  To support "git commit -s
1758  * --amend" on an existing commit, we also ignore "Conflicts:".  To
1759  * support "git commit -v", we truncate at cut lines.
1760  *
1761  * Returns the number of bytes from the tail to ignore, to be fed as
1762  * the second parameter to append_signoff().
1763  */
1764 int ignore_non_trailer(const char *buf, size_t len)
1765 {
1766         int boc = 0;
1767         int bol = 0;
1768         int in_old_conflicts_block = 0;
1769         size_t cutoff = wt_status_locate_end(buf, len);
1770
1771         while (bol < cutoff) {
1772                 const char *next_line = memchr(buf + bol, '\n', len - bol);
1773
1774                 if (!next_line)
1775                         next_line = buf + len;
1776                 else
1777                         next_line++;
1778
1779                 if (buf[bol] == comment_line_char || buf[bol] == '\n') {
1780                         /* is this the first of the run of comments? */
1781                         if (!boc)
1782                                 boc = bol;
1783                         /* otherwise, it is just continuing */
1784                 } else if (starts_with(buf + bol, "Conflicts:\n")) {
1785                         in_old_conflicts_block = 1;
1786                         if (!boc)
1787                                 boc = bol;
1788                 } else if (in_old_conflicts_block && buf[bol] == '\t') {
1789                         ; /* a pathname in the conflicts block */
1790                 } else if (boc) {
1791                         /* the previous was not trailing comment */
1792                         boc = 0;
1793                         in_old_conflicts_block = 0;
1794                 }
1795                 bol = next_line - buf;
1796         }
1797         return boc ? len - boc : len - cutoff;
1798 }