commit.c

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