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
  15 static struct commit_extra_header *read_commit_extra_header_lines(const char *buf, size_t len, const char **);
  16
  17 int save_commit_buffer = 1;
  18
  19 const char *commit_type = "commit";
  20
  21 struct commit *lookup_commit_reference_gently(const unsigned char *sha1,
  22                                               int quiet)
  23 {
  24         struct object *obj = deref_tag(parse_object(sha1), NULL, 0);
  25
  26         if (!obj)
  27                 return NULL;
  28         return object_as_type(obj, OBJ_COMMIT, quiet);
  29 }
  30
  31 struct commit *lookup_commit_reference(const unsigned char *sha1)
  32 {
  33         return lookup_commit_reference_gently(sha1, 0);
  34 }
  35
  36 struct commit *lookup_commit_or_die(const unsigned char *sha1, const char *ref_name)
  37 {
  38         struct commit *c = lookup_commit_reference(sha1);
  39         if (!c)
  40                 die(_("could not parse %s"), ref_name);
  41         if (hashcmp(sha1, c->object.sha1)) {
  42                 warning(_("%s %s is not a commit!"),
  43                         ref_name, sha1_to_hex(sha1));
  44         }
  45         return c;
  46 }
  47
  48 struct commit *lookup_commit(const unsigned char *sha1)
  49 {
  50         struct object *obj = lookup_object(sha1);
  51         if (!obj)
  52                 return create_object(sha1, alloc_commit_node());
  53         return object_as_type(obj, OBJ_COMMIT, 0);
  54 }
  55
  56 struct commit *lookup_commit_reference_by_name(const char *name)
  57 {
  58         struct object_id oid;
  59         struct commit *commit;
  60
  61         if (get_sha1_committish(name, oid.hash))
  62                 return NULL;
  63         commit = lookup_commit_reference(oid.hash);
  64         if (parse_commit(commit))
  65                 return NULL;
  66         return commit;
  67 }
  68
  69 static unsigned long parse_commit_date(const char *buf, const char *tail)
  70 {
  71         const char *dateptr;
  72
  73         if (buf + 6 >= tail)
  74                 return 0;
  75         if (memcmp(buf, "author", 6))
  76                 return 0;
  77         while (buf < tail && *buf++ != '\n')
  78                 /* nada */;
  79         if (buf + 9 >= tail)
  80                 return 0;
  81         if (memcmp(buf, "committer", 9))
  82                 return 0;
  83         while (buf < tail && *buf++ != '>')
  84                 /* nada */;
  85         if (buf >= tail)
  86                 return 0;
  87         dateptr = buf;
  88         while (buf < tail && *buf++ != '\n')
  89                 /* nada */;
  90         if (buf >= tail)
  91                 return 0;
  92         /* dateptr < buf && buf[-1] == '\n', so strtoul will stop at buf-1 */
  93         return strtoul(dateptr, NULL, 10);
  94 }
  95
  96 static struct commit_graft **commit_graft;
  97 static int commit_graft_alloc, commit_graft_nr;
  98
  99 static const unsigned char *commit_graft_sha1_access(size_t index, void *table)
 100 {
 101         struct commit_graft **commit_graft_table = table;
 102         return commit_graft_table[index]->oid.hash;
 103 }
 104
 105 static int commit_graft_pos(const unsigned char *sha1)
 106 {
 107         return sha1_pos(sha1, commit_graft, commit_graft_nr,
 108                         commit_graft_sha1_access);
 109 }
 110
 111 int register_commit_graft(struct commit_graft *graft, int ignore_dups)
 112 {
 113         int pos = commit_graft_pos(graft->oid.hash);
 114
 115         if (0 <= pos) {
 116                 if (ignore_dups)
 117                         free(graft);
 118                 else {
 119                         free(commit_graft[pos]);
 120                         commit_graft[pos] = graft;
 121                 }
 122                 return 1;
 123         }
 124         pos = -pos - 1;
 125         ALLOC_GROW(commit_graft, commit_graft_nr + 1, commit_graft_alloc);
 126         commit_graft_nr++;
 127         if (pos < commit_graft_nr)
 128                 memmove(commit_graft + pos + 1,
 129                         commit_graft + pos,
 130                         (commit_graft_nr - pos - 1) *
 131                         sizeof(*commit_graft));
 132         commit_graft[pos] = graft;
 133         return 0;
 134 }
 135
 136 struct commit_graft *read_graft_line(char *buf, int len)
 137 {
 138         /* The format is just "Commit Parent1 Parent2 ...\n" */
 139         int i;
 140         struct commit_graft *graft = NULL;
 141         const int entry_size = GIT_SHA1_HEXSZ + 1;
 142
 143         while (len && isspace(buf[len-1]))
 144                 buf[--len] = '\0';
 145         if (buf[0] == '#' || buf[0] == '\0')
 146                 return NULL;
 147         if ((len + 1) % entry_size)
 148                 goto bad_graft_data;
 149         i = (len + 1) / entry_size - 1;
 150         graft = xmalloc(sizeof(*graft) + GIT_SHA1_RAWSZ * i);
 151         graft->nr_parent = i;
 152         if (get_oid_hex(buf, &graft->oid))
 153                 goto bad_graft_data;
 154         for (i = GIT_SHA1_HEXSZ; i < len; i += entry_size) {
 155                 if (buf[i] != ' ')
 156                         goto bad_graft_data;
 157                 if (get_sha1_hex(buf + i + 1, graft->parent[i/entry_size].hash))
 158                         goto bad_graft_data;
 159         }
 160         return graft;
 161
 162 bad_graft_data:
 163         error("bad graft data: %s", buf);
 164         free(graft);
 165         return NULL;
 166 }
 167
 168 static int read_graft_file(const char *graft_file)
 169 {
 170         FILE *fp = fopen(graft_file, "r");
 171         struct strbuf buf = STRBUF_INIT;
 172         if (!fp)
 173                 return -1;
 174         while (!strbuf_getwholeline(&buf, fp, '\n')) {
 175                 /* The format is just "Commit Parent1 Parent2 ...\n" */
 176                 struct commit_graft *graft = read_graft_line(buf.buf, buf.len);
 177                 if (!graft)
 178                         continue;
 179                 if (register_commit_graft(graft, 1))
 180                         error("duplicate graft data: %s", buf.buf);
 181         }
 182         fclose(fp);
 183         strbuf_release(&buf);
 184         return 0;
 185 }
 186
 187 static void prepare_commit_graft(void)
 188 {
 189         static int commit_graft_prepared;
 190         char *graft_file;
 191
 192         if (commit_graft_prepared)
 193                 return;
 194         graft_file = get_graft_file();
 195         read_graft_file(graft_file);
 196         /* make sure shallows are read */
 197         is_repository_shallow();
 198         commit_graft_prepared = 1;
 199 }
 200
 201 struct commit_graft *lookup_commit_graft(const unsigned char *sha1)
 202 {
 203         int pos;
 204         prepare_commit_graft();
 205         pos = commit_graft_pos(sha1);
 206         if (pos < 0)
 207                 return NULL;
 208         return commit_graft[pos];
 209 }
 210
 211 int for_each_commit_graft(each_commit_graft_fn fn, void *cb_data)
 212 {
 213         int i, ret;
 214         for (i = ret = 0; i < commit_graft_nr && !ret; i++)
 215                 ret = fn(commit_graft[i], cb_data);
 216         return ret;
 217 }
 218
 219 int unregister_shallow(const unsigned char *sha1)
 220 {
 221         int pos = commit_graft_pos(sha1);
 222         if (pos < 0)
 223                 return -1;
 224         if (pos + 1 < commit_graft_nr)
 225                 memmove(commit_graft + pos, commit_graft + pos + 1,
 226                                 sizeof(struct commit_graft *)
 227                                 * (commit_graft_nr - pos - 1));
 228         commit_graft_nr--;
 229         return 0;
 230 }
 231
 232 struct commit_buffer {
 233         void *buffer;
 234         unsigned long size;
 235 };
 236 define_commit_slab(buffer_slab, struct commit_buffer);
 237 static struct buffer_slab buffer_slab = COMMIT_SLAB_INIT(1, buffer_slab);
 238
 239 void set_commit_buffer(struct commit *commit, void *buffer, unsigned long size)
 240 {
 241         struct commit_buffer *v = buffer_slab_at(&buffer_slab, commit);
 242         v->buffer = buffer;
 243         v->size = size;
 244 }
 245
 246 const void *get_cached_commit_buffer(const struct commit *commit, unsigned long *sizep)
 247 {
 248         struct commit_buffer *v = buffer_slab_at(&buffer_slab, commit);
 249         if (sizep)
 250                 *sizep = v->size;
 251         return v->buffer;
 252 }
 253
 254 const void *get_commit_buffer(const struct commit *commit, unsigned long *sizep)
 255 {
 256         const void *ret = get_cached_commit_buffer(commit, sizep);
 257         if (!ret) {
 258                 enum object_type type;
 259                 unsigned long size;
 260                 ret = read_sha1_file(commit->object.sha1, &type, &size);
 261                 if (!ret)
 262                         die("cannot read commit object %s",
 263                             sha1_to_hex(commit->object.sha1));
 264                 if (type != OBJ_COMMIT)
 265                         die("expected commit for %s, got %s",
 266                             sha1_to_hex(commit->object.sha1), typename(type));
 267                 if (sizep)
 268                         *sizep = size;
 269         }
 270         return ret;
 271 }
 272
 273 void unuse_commit_buffer(const struct commit *commit, const void *buffer)
 274 {
 275         struct commit_buffer *v = buffer_slab_at(&buffer_slab, commit);
 276         if (v->buffer != buffer)
 277                 free((void *)buffer);
 278 }
 279
 280 void free_commit_buffer(struct commit *commit)
 281 {
 282         struct commit_buffer *v = buffer_slab_at(&buffer_slab, commit);
 283         free(v->buffer);
 284         v->buffer = NULL;
 285         v->size = 0;
 286 }
 287
 288 const void *detach_commit_buffer(struct commit *commit, unsigned long *sizep)
 289 {
 290         struct commit_buffer *v = buffer_slab_at(&buffer_slab, commit);
 291         void *ret;
 292
 293         ret = v->buffer;
 294         if (sizep)
 295                 *sizep = v->size;
 296
 297         v->buffer = NULL;
 298         v->size = 0;
 299         return ret;
 300 }
 301
 302 int parse_commit_buffer(struct commit *item, const void *buffer, unsigned long size)
 303 {
 304         const char *tail = buffer;
 305         const char *bufptr = buffer;
 306         struct object_id parent;
 307         struct commit_list **pptr;
 308         struct commit_graft *graft;
 309         const int tree_entry_len = GIT_SHA1_HEXSZ + 5;
 310         const int parent_entry_len = GIT_SHA1_HEXSZ + 7;
 311
 312         if (item->object.parsed)
 313                 return 0;
 314         item->object.parsed = 1;
 315         tail += size;
 316         if (tail <= bufptr + tree_entry_len + 1 || memcmp(bufptr, "tree ", 5) ||
 317                         bufptr[tree_entry_len] != '\n')
 318                 return error("bogus commit object %s", sha1_to_hex(item->object.sha1));
 319         if (get_sha1_hex(bufptr + 5, parent.hash) < 0)
 320                 return error("bad tree pointer in commit %s",
 321                              sha1_to_hex(item->object.sha1));
 322         item->tree = lookup_tree(parent.hash);
 323         bufptr += tree_entry_len + 1; /* "tree " + "hex sha1" + "\n" */
 324         pptr = &item->parents;
 325
 326         graft = lookup_commit_graft(item->object.sha1);
 327         while (bufptr + parent_entry_len < tail && !memcmp(bufptr, "parent ", 7)) {
 328                 struct commit *new_parent;
 329
 330                 if (tail <= bufptr + parent_entry_len + 1 ||
 331                     get_sha1_hex(bufptr + 7, parent.hash) ||
 332                     bufptr[parent_entry_len] != '\n')
 333                         return error("bad parents in commit %s", sha1_to_hex(item->object.sha1));
 334                 bufptr += parent_entry_len + 1;
 335                 /*
 336                  * The clone is shallow if nr_parent < 0, and we must
 337                  * not traverse its real parents even when we unhide them.
 338                  */
 339                 if (graft && (graft->nr_parent < 0 || grafts_replace_parents))
 340                         continue;
 341                 new_parent = lookup_commit(parent.hash);
 342                 if (new_parent)
 343                         pptr = &commit_list_insert(new_parent, pptr)->next;
 344         }
 345         if (graft) {
 346                 int i;
 347                 struct commit *new_parent;
 348                 for (i = 0; i < graft->nr_parent; i++) {
 349                         new_parent = lookup_commit(graft->parent[i].hash);
 350                         if (!new_parent)
 351                                 continue;
 352                         pptr = &commit_list_insert(new_parent, pptr)->next;
 353                 }
 354         }
 355         item->date = parse_commit_date(bufptr, tail);
 356
 357         return 0;
 358 }
 359
 360 int parse_commit(struct commit *item)
 361 {
 362         enum object_type type;
 363         void *buffer;
 364         unsigned long size;
 365         int ret;
 366
 367         if (!item)
 368                 return -1;
 369         if (item->object.parsed)
 370                 return 0;
 371         buffer = read_sha1_file(item->object.sha1, &type, &size);
 372         if (!buffer)
 373                 return error("Could not read %s",
 374                              sha1_to_hex(item->object.sha1));
 375         if (type != OBJ_COMMIT) {
 376                 free(buffer);
 377                 return error("Object %s not a commit",
 378                              sha1_to_hex(item->object.sha1));
 379         }
 380         ret = parse_commit_buffer(item, buffer, size);
 381         if (save_commit_buffer && !ret) {
 382                 set_commit_buffer(item, buffer, size);
 383                 return 0;
 384         }
 385         free(buffer);
 386         return ret;
 387 }
 388
 389 void parse_commit_or_die(struct commit *item)
 390 {
 391         if (parse_commit(item))
 392                 die("unable to parse commit %s",
 393                     item ? sha1_to_hex(item->object.sha1) : "(null)");
 394 }
 395
 396 int find_commit_subject(const char *commit_buffer, const char **subject)
 397 {
 398         const char *eol;
 399         const char *p = commit_buffer;
 400
 401         while (*p && (*p != '\n' || p[1] != '\n'))
 402                 p++;
 403         if (*p) {
 404                 p += 2;
 405                 for (eol = p; *eol && *eol != '\n'; eol++)
 406                         ; /* do nothing */
 407         } else
 408                 eol = p;
 409
 410         *subject = p;
 411
 412         return eol - p;
 413 }
 414
 415 struct commit_list *commit_list_insert(struct commit *item, struct commit_list **list_p)
 416 {
 417         struct commit_list *new_list = xmalloc(sizeof(struct commit_list));
 418         new_list->item = item;
 419         new_list->next = *list_p;
 420         *list_p = new_list;
 421         return new_list;
 422 }
 423
 424 unsigned commit_list_count(const struct commit_list *l)
 425 {
 426         unsigned c = 0;
 427         for (; l; l = l->next )
 428                 c++;
 429         return c;
 430 }
 431
 432 struct commit_list *copy_commit_list(struct commit_list *list)
 433 {
 434         struct commit_list *head = NULL;
 435         struct commit_list **pp = &head;
 436         while (list) {
 437                 pp = commit_list_append(list->item, pp);
 438                 list = list->next;
 439         }
 440         return head;
 441 }
 442
 443 void free_commit_list(struct commit_list *list)
 444 {
 445         while (list) {
 446                 struct commit_list *temp = list;
 447                 list = temp->next;
 448                 free(temp);
 449         }
 450 }
 451
 452 struct commit_list * commit_list_insert_by_date(struct commit *item, struct commit_list **list)
 453 {
 454         struct commit_list **pp = list;
 455         struct commit_list *p;
 456         while ((p = *pp) != NULL) {
 457                 if (p->item->date < item->date) {
 458                         break;
 459                 }
 460                 pp = &p->next;
 461         }
 462         return commit_list_insert(item, pp);
 463 }
 464
 465 static int commit_list_compare_by_date(const void *a, const void *b)
 466 {
 467         unsigned long a_date = ((const struct commit_list *)a)->item->date;
 468         unsigned long b_date = ((const struct commit_list *)b)->item->date;
 469         if (a_date < b_date)
 470                 return 1;
 471         if (a_date > b_date)
 472                 return -1;
 473         return 0;
 474 }
 475
 476 static void *commit_list_get_next(const void *a)
 477 {
 478         return ((const struct commit_list *)a)->next;
 479 }
 480
 481 static void commit_list_set_next(void *a, void *next)
 482 {
 483         ((struct commit_list *)a)->next = next;
 484 }
 485
 486 void commit_list_sort_by_date(struct commit_list **list)
 487 {
 488         *list = llist_mergesort(*list, commit_list_get_next, commit_list_set_next,
 489                                 commit_list_compare_by_date);
 490 }
 491
 492 struct commit *pop_most_recent_commit(struct commit_list **list,
 493                                       unsigned int mark)
 494 {
 495         struct commit *ret = (*list)->item;
 496         struct commit_list *parents = ret->parents;
 497         struct commit_list *old = *list;
 498
 499         *list = (*list)->next;
 500         free(old);
 501
 502         while (parents) {
 503                 struct commit *commit = parents->item;
 504                 if (!parse_commit(commit) && !(commit->object.flags & mark)) {
 505                         commit->object.flags |= mark;
 506                         commit_list_insert_by_date(commit, list);
 507                 }
 508                 parents = parents->next;
 509         }
 510         return ret;
 511 }
 512
 513 static void clear_commit_marks_1(struct commit_list **plist,
 514                                  struct commit *commit, unsigned int mark)
 515 {
 516         while (commit) {
 517                 struct commit_list *parents;
 518
 519                 if (!(mark & commit->object.flags))
 520                         return;
 521
 522                 commit->object.flags &= ~mark;
 523
 524                 parents = commit->parents;
 525                 if (!parents)
 526                         return;
 527
 528                 while ((parents = parents->next))
 529                         commit_list_insert(parents->item, plist);
 530
 531                 commit = commit->parents->item;
 532         }
 533 }
 534
 535 void clear_commit_marks_many(int nr, struct commit **commit, unsigned int mark)
 536 {
 537         struct commit_list *list = NULL;
 538
 539         while (nr--) {
 540                 commit_list_insert(*commit, &list);
 541                 commit++;
 542         }
 543         while (list)
 544                 clear_commit_marks_1(&list, pop_commit(&list), mark);
 545 }
 546
 547 void clear_commit_marks(struct commit *commit, unsigned int mark)
 548 {
 549         clear_commit_marks_many(1, &commit, mark);
 550 }
 551
 552 void clear_commit_marks_for_object_array(struct object_array *a, unsigned mark)
 553 {
 554         struct object *object;
 555         struct commit *commit;
 556         unsigned int i;
 557
 558         for (i = 0; i < a->nr; i++) {
 559                 object = a->objects[i].item;
 560                 commit = lookup_commit_reference_gently(object->sha1, 1);
 561                 if (commit)
 562                         clear_commit_marks(commit, mark);
 563         }
 564 }
 565
 566 struct commit *pop_commit(struct commit_list **stack)
 567 {
 568         struct commit_list *top = *stack;
 569         struct commit *item = top ? top->item : NULL;
 570
 571         if (top) {
 572                 *stack = top->next;
 573                 free(top);
 574         }
 575         return item;
 576 }
 577
 578 /*
 579  * Topological sort support
 580  */
 581
 582 /* count number of children that have not been emitted */
 583 define_commit_slab(indegree_slab, int);
 584
 585 /* record author-date for each commit object */
 586 define_commit_slab(author_date_slab, unsigned long);
 587
 588 static void record_author_date(struct author_date_slab *author_date,
 589                                struct commit *commit)
 590 {
 591         const char *buffer = get_commit_buffer(commit, NULL);
 592         struct ident_split ident;
 593         const char *ident_line;
 594         size_t ident_len;
 595         char *date_end;
 596         unsigned long date;
 597
 598         ident_line = find_commit_header(buffer, "author", &ident_len);
 599         if (!ident_line)
 600                 goto fail_exit; /* no author line */
 601         if (split_ident_line(&ident, ident_line, ident_len) ||
 602             !ident.date_begin || !ident.date_end)
 603                 goto fail_exit; /* malformed "author" line */
 604
 605         date = strtoul(ident.date_begin, &date_end, 10);
 606         if (date_end != ident.date_end)
 607                 goto fail_exit; /* malformed date */
 608         *(author_date_slab_at(author_date, commit)) = date;
 609
 610 fail_exit:
 611         unuse_commit_buffer(commit, buffer);
 612 }
 613
 614 static int compare_commits_by_author_date(const void *a_, const void *b_,
 615                                           void *cb_data)
 616 {
 617         const struct commit *a = a_, *b = b_;
 618         struct author_date_slab *author_date = cb_data;
 619         unsigned long a_date = *(author_date_slab_at(author_date, a));
 620         unsigned long b_date = *(author_date_slab_at(author_date, b));
 621
 622         /* newer commits with larger date first */
 623         if (a_date < b_date)
 624                 return 1;
 625         else if (a_date > b_date)
 626                 return -1;
 627         return 0;
 628 }
 629
 630 int compare_commits_by_commit_date(const void *a_, const void *b_, void *unused)
 631 {
 632         const struct commit *a = a_, *b = b_;
 633         /* newer commits with larger date first */
 634         if (a->date < b->date)
 635                 return 1;
 636         else if (a->date > b->date)
 637                 return -1;
 638         return 0;
 639 }
 640
 641 /*
 642  * Performs an in-place topological sort on the list supplied.
 643  */
 644 void sort_in_topological_order(struct commit_list **list, enum rev_sort_order sort_order)
 645 {
 646         struct commit_list *next, *orig = *list;
 647         struct commit_list **pptr;
 648         struct indegree_slab indegree;
 649         struct prio_queue queue;
 650         struct commit *commit;
 651         struct author_date_slab author_date;
 652
 653         if (!orig)
 654                 return;
 655         *list = NULL;
 656
 657         init_indegree_slab(&indegree);
 658         memset(&queue, '\0', sizeof(queue));
 659
 660         switch (sort_order) {
 661         default: /* REV_SORT_IN_GRAPH_ORDER */
 662                 queue.compare = NULL;
 663                 break;
 664         case REV_SORT_BY_COMMIT_DATE:
 665                 queue.compare = compare_commits_by_commit_date;
 666                 break;
 667         case REV_SORT_BY_AUTHOR_DATE:
 668                 init_author_date_slab(&author_date);
 669                 queue.compare = compare_commits_by_author_date;
 670                 queue.cb_data = &author_date;
 671                 break;
 672         }
 673
 674         /* Mark them and clear the indegree */
 675         for (next = orig; next; next = next->next) {
 676                 struct commit *commit = next->item;
 677                 *(indegree_slab_at(&indegree, commit)) = 1;
 678                 /* also record the author dates, if needed */
 679                 if (sort_order == REV_SORT_BY_AUTHOR_DATE)
 680                         record_author_date(&author_date, commit);
 681         }
 682
 683         /* update the indegree */
 684         for (next = orig; next; next = next->next) {
 685                 struct commit_list *parents = next->item->parents;
 686                 while (parents) {
 687                         struct commit *parent = parents->item;
 688                         int *pi = indegree_slab_at(&indegree, parent);
 689
 690                         if (*pi)
 691                                 (*pi)++;
 692                         parents = parents->next;
 693                 }
 694         }
 695
 696         /*
 697          * find the tips
 698          *
 699          * tips are nodes not reachable from any other node in the list
 700          *
 701          * the tips serve as a starting set for the work queue.
 702          */
 703         for (next = orig; next; next = next->next) {
 704                 struct commit *commit = next->item;
 705
 706                 if (*(indegree_slab_at(&indegree, commit)) == 1)
 707                         prio_queue_put(&queue, commit);
 708         }
 709
 710         /*
 711          * This is unfortunate; the initial tips need to be shown
 712          * in the order given from the revision traversal machinery.
 713          */
 714         if (sort_order == REV_SORT_IN_GRAPH_ORDER)
 715                 prio_queue_reverse(&queue);
 716
 717         /* We no longer need the commit list */
 718         free_commit_list(orig);
 719
 720         pptr = list;
 721         *list = NULL;
 722         while ((commit = prio_queue_get(&queue)) != NULL) {
 723                 struct commit_list *parents;
 724
 725                 for (parents = commit->parents; parents ; parents = parents->next) {
 726                         struct commit *parent = parents->item;
 727                         int *pi = indegree_slab_at(&indegree, parent);
 728
 729                         if (!*pi)
 730                                 continue;
 731
 732                         /*
 733                          * parents are only enqueued for emission
 734                          * when all their children have been emitted thereby
 735                          * guaranteeing topological order.
 736                          */
 737                         if (--(*pi) == 1)
 738                                 prio_queue_put(&queue, parent);
 739                 }
 740                 /*
 741                  * all children of commit have already been
 742                  * emitted. we can emit it now.
 743                  */
 744                 *(indegree_slab_at(&indegree, commit)) = 0;
 745
 746                 pptr = &commit_list_insert(commit, pptr)->next;
 747         }
 748
 749         clear_indegree_slab(&indegree);
 750         clear_prio_queue(&queue);
 751         if (sort_order == REV_SORT_BY_AUTHOR_DATE)
 752                 clear_author_date_slab(&author_date);
 753 }
 754
 755 /* merge-base stuff */
 756
 757 /* Remember to update object flag allocation in object.h */
 758 #define PARENT1         (1u<<16)
 759 #define PARENT2         (1u<<17)
 760 #define STALE           (1u<<18)
 761 #define RESULT          (1u<<19)
 762
 763 static const unsigned all_flags = (PARENT1 | PARENT2 | STALE | RESULT);
 764
 765 static int queue_has_nonstale(struct prio_queue *queue)
 766 {
 767         int i;
 768         for (i = 0; i < queue->nr; i++) {
 769                 struct commit *commit = queue->array[i].data;
 770                 if (!(commit->object.flags & STALE))
 771                         return 1;
 772         }
 773         return 0;
 774 }
 775
 776 /* all input commits in one and twos[] must have been parsed! */
 777 static struct commit_list *paint_down_to_common(struct commit *one, int n, struct commit **twos)
 778 {
 779         struct prio_queue queue = { compare_commits_by_commit_date };
 780         struct commit_list *result = NULL;
 781         int i;
 782
 783         one->object.flags |= PARENT1;
 784         if (!n) {
 785                 commit_list_append(one, &result);
 786                 return result;
 787         }
 788         prio_queue_put(&queue, one);
 789
 790         for (i = 0; i < n; i++) {
 791                 twos[i]->object.flags |= PARENT2;
 792                 prio_queue_put(&queue, twos[i]);
 793         }
 794
 795         while (queue_has_nonstale(&queue)) {
 796                 struct commit *commit = prio_queue_get(&queue);
 797                 struct commit_list *parents;
 798                 int flags;
 799
 800                 flags = commit->object.flags & (PARENT1 | PARENT2 | STALE);
 801                 if (flags == (PARENT1 | PARENT2)) {
 802                         if (!(commit->object.flags & RESULT)) {
 803                                 commit->object.flags |= RESULT;
 804                                 commit_list_insert_by_date(commit, &result);
 805                         }
 806                         /* Mark parents of a found merge stale */
 807                         flags |= STALE;
 808                 }
 809                 parents = commit->parents;
 810                 while (parents) {
 811                         struct commit *p = parents->item;
 812                         parents = parents->next;
 813                         if ((p->object.flags & flags) == flags)
 814                                 continue;
 815                         if (parse_commit(p))
 816                                 return NULL;
 817                         p->object.flags |= flags;
 818                         prio_queue_put(&queue, p);
 819                 }
 820         }
 821
 822         clear_prio_queue(&queue);
 823         return result;
 824 }
 825
 826 static struct commit_list *merge_bases_many(struct commit *one, int n, struct commit **twos)
 827 {
 828         struct commit_list *list = NULL;
 829         struct commit_list *result = NULL;
 830         int i;
 831
 832         for (i = 0; i < n; i++) {
 833                 if (one == twos[i])
 834                         /*
 835                          * We do not mark this even with RESULT so we do not
 836                          * have to clean it up.
 837                          */
 838                         return commit_list_insert(one, &result);
 839         }
 840
 841         if (parse_commit(one))
 842                 return NULL;
 843         for (i = 0; i < n; i++) {
 844                 if (parse_commit(twos[i]))
 845                         return NULL;
 846         }
 847
 848         list = paint_down_to_common(one, n, twos);
 849
 850         while (list) {
 851                 struct commit_list *next = list->next;
 852                 if (!(list->item->object.flags & STALE))
 853                         commit_list_insert_by_date(list->item, &result);
 854                 free(list);
 855                 list = next;
 856         }
 857         return result;
 858 }
 859
 860 struct commit_list *get_octopus_merge_bases(struct commit_list *in)
 861 {
 862         struct commit_list *i, *j, *k, *ret = NULL;
 863
 864         if (!in)
 865                 return ret;
 866
 867         commit_list_insert(in->item, &ret);
 868
 869         for (i = in->next; i; i = i->next) {
 870                 struct commit_list *new = NULL, *end = NULL;
 871
 872                 for (j = ret; j; j = j->next) {
 873                         struct commit_list *bases;
 874                         bases = get_merge_bases(i->item, j->item);
 875                         if (!new)
 876                                 new = bases;
 877                         else
 878                                 end->next = bases;
 879                         for (k = bases; k; k = k->next)
 880                                 end = k;
 881                 }
 882                 ret = new;
 883         }
 884         return ret;
 885 }
 886
 887 static int remove_redundant(struct commit **array, int cnt)
 888 {
 889         /*
 890          * Some commit in the array may be an ancestor of
 891          * another commit.  Move such commit to the end of
 892          * the array, and return the number of commits that
 893          * are independent from each other.
 894          */
 895         struct commit **work;
 896         unsigned char *redundant;
 897         int *filled_index;
 898         int i, j, filled;
 899
 900         work = xcalloc(cnt, sizeof(*work));
 901         redundant = xcalloc(cnt, 1);
 902         filled_index = xmalloc(sizeof(*filled_index) * (cnt - 1));
 903
 904         for (i = 0; i < cnt; i++)
 905                 parse_commit(array[i]);
 906         for (i = 0; i < cnt; i++) {
 907                 struct commit_list *common;
 908
 909                 if (redundant[i])
 910                         continue;
 911                 for (j = filled = 0; j < cnt; j++) {
 912                         if (i == j || redundant[j])
 913                                 continue;
 914                         filled_index[filled] = j;
 915                         work[filled++] = array[j];
 916                 }
 917                 common = paint_down_to_common(array[i], filled, work);
 918                 if (array[i]->object.flags & PARENT2)
 919                         redundant[i] = 1;
 920                 for (j = 0; j < filled; j++)
 921                         if (work[j]->object.flags & PARENT1)
 922                                 redundant[filled_index[j]] = 1;
 923                 clear_commit_marks(array[i], all_flags);
 924                 for (j = 0; j < filled; j++)
 925                         clear_commit_marks(work[j], all_flags);
 926                 free_commit_list(common);
 927         }
 928
 929         /* Now collect the result */
 930         memcpy(work, array, sizeof(*array) * cnt);
 931         for (i = filled = 0; i < cnt; i++)
 932                 if (!redundant[i])
 933                         array[filled++] = work[i];
 934         for (j = filled, i = 0; i < cnt; i++)
 935                 if (redundant[i])
 936                         array[j++] = work[i];
 937         free(work);
 938         free(redundant);
 939         free(filled_index);
 940         return filled;
 941 }
 942
 943 static struct commit_list *get_merge_bases_many_0(struct commit *one,
 944                                                   int n,
 945                                                   struct commit **twos,
 946                                                   int cleanup)
 947 {
 948         struct commit_list *list;
 949         struct commit **rslt;
 950         struct commit_list *result;
 951         int cnt, i;
 952
 953         result = merge_bases_many(one, n, twos);
 954         for (i = 0; i < n; i++) {
 955                 if (one == twos[i])
 956                         return result;
 957         }
 958         if (!result || !result->next) {
 959                 if (cleanup) {
 960                         clear_commit_marks(one, all_flags);
 961                         clear_commit_marks_many(n, twos, all_flags);
 962                 }
 963                 return result;
 964         }
 965
 966         /* There are more than one */
 967         cnt = commit_list_count(result);
 968         rslt = xcalloc(cnt, sizeof(*rslt));
 969         for (list = result, i = 0; list; list = list->next)
 970                 rslt[i++] = list->item;
 971         free_commit_list(result);
 972
 973         clear_commit_marks(one, all_flags);
 974         clear_commit_marks_many(n, twos, all_flags);
 975
 976         cnt = remove_redundant(rslt, cnt);
 977         result = NULL;
 978         for (i = 0; i < cnt; i++)
 979                 commit_list_insert_by_date(rslt[i], &result);
 980         free(rslt);
 981         return result;
 982 }
 983
 984 struct commit_list *get_merge_bases_many(struct commit *one,
 985                                          int n,
 986                                          struct commit **twos)
 987 {
 988         return get_merge_bases_many_0(one, n, twos, 1);
 989 }
 990
 991 struct commit_list *get_merge_bases_many_dirty(struct commit *one,
 992                                                int n,
 993                                                struct commit **twos)
 994 {
 995         return get_merge_bases_many_0(one, n, twos, 0);
 996 }
 997
 998 struct commit_list *get_merge_bases(struct commit *one, struct commit *two)
 999 {
1000         return get_merge_bases_many_0(one, 1, &two, 1);
1001 }
1002
1003 /*
1004  * Is "commit" a descendant of one of the elements on the "with_commit" list?
1005  */
1006 int is_descendant_of(struct commit *commit, struct commit_list *with_commit)
1007 {
1008         if (!with_commit)
1009                 return 1;
1010         while (with_commit) {
1011                 struct commit *other;
1012
1013                 other = with_commit->item;
1014                 with_commit = with_commit->next;
1015                 if (in_merge_bases(other, commit))
1016                         return 1;
1017         }
1018         return 0;
1019 }
1020
1021 /*
1022  * Is "commit" an ancestor of one of the "references"?
1023  */
1024 int in_merge_bases_many(struct commit *commit, int nr_reference, struct commit **reference)
1025 {
1026         struct commit_list *bases;
1027         int ret = 0, i;
1028
1029         if (parse_commit(commit))
1030                 return ret;
1031         for (i = 0; i < nr_reference; i++)
1032                 if (parse_commit(reference[i]))
1033                         return ret;
1034
1035         bases = paint_down_to_common(commit, nr_reference, reference);
1036         if (commit->object.flags & PARENT2)
1037                 ret = 1;
1038         clear_commit_marks(commit, all_flags);
1039         clear_commit_marks_many(nr_reference, reference, all_flags);
1040         free_commit_list(bases);
1041         return ret;
1042 }
1043
1044 /*
1045  * Is "commit" an ancestor of (i.e. reachable from) the "reference"?
1046  */
1047 int in_merge_bases(struct commit *commit, struct commit *reference)
1048 {
1049         return in_merge_bases_many(commit, 1, &reference);
1050 }
1051
1052 struct commit_list *reduce_heads(struct commit_list *heads)
1053 {
1054         struct commit_list *p;
1055         struct commit_list *result = NULL, **tail = &result;
1056         struct commit **array;
1057         int num_head, i;
1058
1059         if (!heads)
1060                 return NULL;
1061
1062         /* Uniquify */
1063         for (p = heads; p; p = p->next)
1064                 p->item->object.flags &= ~STALE;
1065         for (p = heads, num_head = 0; p; p = p->next) {
1066                 if (p->item->object.flags & STALE)
1067                         continue;
1068                 p->item->object.flags |= STALE;
1069                 num_head++;
1070         }
1071         array = xcalloc(num_head, sizeof(*array));
1072         for (p = heads, i = 0; p; p = p->next) {
1073                 if (p->item->object.flags & STALE) {
1074                         array[i++] = p->item;
1075                         p->item->object.flags &= ~STALE;
1076                 }
1077         }
1078         num_head = remove_redundant(array, num_head);
1079         for (i = 0; i < num_head; i++)
1080                 tail = &commit_list_insert(array[i], tail)->next;
1081         return 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
1092         /* find the end of the header */
1093         inspos = strstr(buf->buf, "\n\n") - buf->buf + 1;
1094
1095         if (!keyid || !*keyid)
1096                 keyid = get_signing_key();
1097         if (sign_buffer(buf, &sig, keyid)) {
1098                 strbuf_release(&sig);
1099                 return -1;
1100         }
1101
1102         for (copypos = 0; sig.buf[copypos]; ) {
1103                 const char *bol = sig.buf + copypos;
1104                 const char *eol = strchrnul(bol, '\n');
1105                 int len = (eol - bol) + !!*eol;
1106
1107                 if (!copypos) {
1108                         strbuf_insert(buf, inspos, gpg_sig_header, gpg_sig_header_len);
1109                         inspos += gpg_sig_header_len;
1110                 }
1111                 strbuf_insert(buf, inspos++, " ", 1);
1112                 strbuf_insert(buf, inspos, bol, len);
1113                 inspos += len;
1114                 copypos += len;
1115         }
1116         strbuf_release(&sig);
1117         return 0;
1118 }
1119
1120 int parse_signed_commit(const struct commit *commit,
1121                         struct strbuf *payload, struct strbuf *signature)
1122 {
1123
1124         unsigned long size;
1125         const char *buffer = get_commit_buffer(commit, &size);
1126         int in_signature, saw_signature = -1;
1127         const char *line, *tail;
1128
1129         line = buffer;
1130         tail = buffer + size;
1131         in_signature = 0;
1132         saw_signature = 0;
1133         while (line < tail) {
1134                 const char *sig = NULL;
1135                 const char *next = memchr(line, '\n', tail - line);
1136
1137                 next = next ? next + 1 : tail;
1138                 if (in_signature && line[0] == ' ')
1139                         sig = line + 1;
1140                 else if (starts_with(line, gpg_sig_header) &&
1141                          line[gpg_sig_header_len] == ' ')
1142                         sig = line + gpg_sig_header_len + 1;
1143                 if (sig) {
1144                         strbuf_add(signature, sig, next - sig);
1145                         saw_signature = 1;
1146                         in_signature = 1;
1147                 } else {
1148                         if (*line == '\n')
1149                                 /* dump the whole remainder of the buffer */
1150                                 next = tail;
1151                         strbuf_add(payload, line, next - line);
1152                         in_signature = 0;
1153                 }
1154                 line = next;
1155         }
1156         unuse_commit_buffer(commit, buffer);
1157         return saw_signature;
1158 }
1159
1160 int remove_signature(struct strbuf *buf)
1161 {
1162         const char *line = buf->buf;
1163         const char *tail = buf->buf + buf->len;
1164         int in_signature = 0;
1165         const char *sig_start = NULL;
1166         const char *sig_end = NULL;
1167
1168         while (line < tail) {
1169                 const char *next = memchr(line, '\n', tail - line);
1170                 next = next ? next + 1 : tail;
1171
1172                 if (in_signature && line[0] == ' ')
1173                         sig_end = next;
1174                 else if (starts_with(line, gpg_sig_header) &&
1175                          line[gpg_sig_header_len] == ' ') {
1176                         sig_start = line;
1177                         sig_end = next;
1178                         in_signature = 1;
1179                 } else {
1180                         if (*line == '\n')
1181                                 /* dump the whole remainder of the buffer */
1182                                 next = tail;
1183                         in_signature = 0;
1184                 }
1185                 line = next;
1186         }
1187
1188         if (sig_start)
1189                 strbuf_remove(buf, sig_start - buf->buf, sig_end - sig_start);
1190
1191         return sig_start != NULL;
1192 }
1193
1194 static void handle_signed_tag(struct commit *parent, struct commit_extra_header ***tail)
1195 {
1196         struct merge_remote_desc *desc;
1197         struct commit_extra_header *mergetag;
1198         char *buf;
1199         unsigned long size, len;
1200         enum object_type type;
1201
1202         desc = merge_remote_util(parent);
1203         if (!desc || !desc->obj)
1204                 return;
1205         buf = read_sha1_file(desc->obj->sha1, &type, &size);
1206         if (!buf || type != OBJ_TAG)
1207                 goto free_return;
1208         len = parse_signature(buf, size);
1209         if (size == len)
1210                 goto free_return;
1211         /*
1212          * We could verify this signature and either omit the tag when
1213          * it does not validate, but the integrator may not have the
1214          * public key of the signer of the tag he is merging, while a
1215          * later auditor may have it while auditing, so let's not run
1216          * verify-signed-buffer here for now...
1217          *
1218          * if (verify_signed_buffer(buf, len, buf + len, size - len, ...))
1219          *      warn("warning: signed tag unverified.");
1220          */
1221         mergetag = xcalloc(1, sizeof(*mergetag));
1222         mergetag->key = xstrdup("mergetag");
1223         mergetag->value = buf;
1224         mergetag->len = size;
1225
1226         **tail = mergetag;
1227         *tail = &mergetag->next;
1228         return;
1229
1230 free_return:
1231         free(buf);
1232 }
1233
1234 void check_commit_signature(const struct commit *commit, struct signature_check *sigc)
1235 {
1236         struct strbuf payload = STRBUF_INIT;
1237         struct strbuf signature = STRBUF_INIT;
1238         struct strbuf gpg_output = STRBUF_INIT;
1239         struct strbuf gpg_status = STRBUF_INIT;
1240         int status;
1241
1242         sigc->result = 'N';
1243
1244         if (parse_signed_commit(commit, &payload, &signature) <= 0)
1245                 goto out;
1246         status = verify_signed_buffer(payload.buf, payload.len,
1247                                       signature.buf, signature.len,
1248                                       &gpg_output, &gpg_status);
1249         if (status && !gpg_output.len)
1250                 goto out;
1251         sigc->payload = strbuf_detach(&payload, NULL);
1252         sigc->gpg_output = strbuf_detach(&gpg_output, NULL);
1253         sigc->gpg_status = strbuf_detach(&gpg_status, NULL);
1254         parse_gpg_output(sigc);
1255
1256  out:
1257         strbuf_release(&gpg_status);
1258         strbuf_release(&gpg_output);
1259         strbuf_release(&payload);
1260         strbuf_release(&signature);
1261 }
1262
1263
1264
1265 void append_merge_tag_headers(struct commit_list *parents,
1266                               struct commit_extra_header ***tail)
1267 {
1268         while (parents) {
1269                 struct commit *parent = parents->item;
1270                 handle_signed_tag(parent, tail);
1271                 parents = parents->next;
1272         }
1273 }
1274
1275 static void add_extra_header(struct strbuf *buffer,
1276                              struct commit_extra_header *extra)
1277 {
1278         strbuf_addstr(buffer, extra->key);
1279         if (extra->len)
1280                 strbuf_add_lines(buffer, " ", extra->value, extra->len);
1281         else
1282                 strbuf_addch(buffer, '\n');
1283 }
1284
1285 struct commit_extra_header *read_commit_extra_headers(struct commit *commit,
1286                                                       const char **exclude)
1287 {
1288         struct commit_extra_header *extra = NULL;
1289         unsigned long size;
1290         const char *buffer = get_commit_buffer(commit, &size);
1291         extra = read_commit_extra_header_lines(buffer, size, exclude);
1292         unuse_commit_buffer(commit, buffer);
1293         return extra;
1294 }
1295
1296 void for_each_mergetag(each_mergetag_fn fn, struct commit *commit, void *data)
1297 {
1298         struct commit_extra_header *extra, *to_free;
1299
1300         to_free = read_commit_extra_headers(commit, NULL);
1301         for (extra = to_free; extra; extra = extra->next) {
1302                 if (strcmp(extra->key, "mergetag"))
1303                         continue; /* not a merge tag */
1304                 fn(commit, extra, data);
1305         }
1306         free_commit_extra_headers(to_free);
1307 }
1308
1309 static inline int standard_header_field(const char *field, size_t len)
1310 {
1311         return ((len == 4 && !memcmp(field, "tree ", 5)) ||
1312                 (len == 6 && !memcmp(field, "parent ", 7)) ||
1313                 (len == 6 && !memcmp(field, "author ", 7)) ||
1314                 (len == 9 && !memcmp(field, "committer ", 10)) ||
1315                 (len == 8 && !memcmp(field, "encoding ", 9)));
1316 }
1317
1318 static int excluded_header_field(const char *field, size_t len, const char **exclude)
1319 {
1320         if (!exclude)
1321                 return 0;
1322
1323         while (*exclude) {
1324                 size_t xlen = strlen(*exclude);
1325                 if (len == xlen &&
1326                     !memcmp(field, *exclude, xlen) && field[xlen] == ' ')
1327                         return 1;
1328                 exclude++;
1329         }
1330         return 0;
1331 }
1332
1333 static struct commit_extra_header *read_commit_extra_header_lines(
1334         const char *buffer, size_t size,
1335         const char **exclude)
1336 {
1337         struct commit_extra_header *extra = NULL, **tail = &extra, *it = NULL;
1338         const char *line, *next, *eof, *eob;
1339         struct strbuf buf = STRBUF_INIT;
1340
1341         for (line = buffer, eob = line + size;
1342              line < eob && *line != '\n';
1343              line = next) {
1344                 next = memchr(line, '\n', eob - line);
1345                 next = next ? next + 1 : eob;
1346                 if (*line == ' ') {
1347                         /* continuation */
1348                         if (it)
1349                                 strbuf_add(&buf, line + 1, next - (line + 1));
1350                         continue;
1351                 }
1352                 if (it)
1353                         it->value = strbuf_detach(&buf, &it->len);
1354                 strbuf_reset(&buf);
1355                 it = NULL;
1356
1357                 eof = strchr(line, ' ');
1358                 if (next <= eof)
1359                         eof = next;
1360
1361                 if (standard_header_field(line, eof - line) ||
1362                     excluded_header_field(line, eof - line, exclude))
1363                         continue;
1364
1365                 it = xcalloc(1, sizeof(*it));
1366                 it->key = xmemdupz(line, eof-line);
1367                 *tail = it;
1368                 tail = &it->next;
1369                 if (eof + 1 < next)
1370                         strbuf_add(&buf, eof + 1, next - (eof + 1));
1371         }
1372         if (it)
1373                 it->value = strbuf_detach(&buf, &it->len);
1374         return extra;
1375 }
1376
1377 void free_commit_extra_headers(struct commit_extra_header *extra)
1378 {
1379         while (extra) {
1380                 struct commit_extra_header *next = extra->next;
1381                 free(extra->key);
1382                 free(extra->value);
1383                 free(extra);
1384                 extra = next;
1385         }
1386 }
1387
1388 int commit_tree(const char *msg, size_t msg_len,
1389                 const unsigned char *tree,
1390                 struct commit_list *parents, unsigned char *ret,
1391                 const char *author, const char *sign_commit)
1392 {
1393         struct commit_extra_header *extra = NULL, **tail = &extra;
1394         int result;
1395
1396         append_merge_tag_headers(parents, &tail);
1397         result = commit_tree_extended(msg, msg_len, tree, parents, ret,
1398                                       author, sign_commit, extra);
1399         free_commit_extra_headers(extra);
1400         return result;
1401 }
1402
1403 static int find_invalid_utf8(const char *buf, int len)
1404 {
1405         int offset = 0;
1406         static const unsigned int max_codepoint[] = {
1407                 0x7f, 0x7ff, 0xffff, 0x10ffff
1408         };
1409
1410         while (len) {
1411                 unsigned char c = *buf++;
1412                 int bytes, bad_offset;
1413                 unsigned int codepoint;
1414                 unsigned int min_val, max_val;
1415
1416                 len--;
1417                 offset++;
1418
1419                 /* Simple US-ASCII? No worries. */
1420                 if (c < 0x80)
1421                         continue;
1422
1423                 bad_offset = offset-1;
1424
1425                 /*
1426                  * Count how many more high bits set: that's how
1427                  * many more bytes this sequence should have.
1428                  */
1429                 bytes = 0;
1430                 while (c & 0x40) {
1431                         c <<= 1;
1432                         bytes++;
1433                 }
1434
1435                 /*
1436                  * Must be between 1 and 3 more bytes.  Longer sequences result in
1437                  * codepoints beyond U+10FFFF, which are guaranteed never to exist.
1438                  */
1439                 if (bytes < 1 || 3 < bytes)
1440                         return bad_offset;
1441
1442                 /* Do we *have* that many bytes? */
1443                 if (len < bytes)
1444                         return bad_offset;
1445
1446                 /*
1447                  * Place the encoded bits at the bottom of the value and compute the
1448                  * valid range.
1449                  */
1450                 codepoint = (c & 0x7f) >> bytes;
1451                 min_val = max_codepoint[bytes-1] + 1;
1452                 max_val = max_codepoint[bytes];
1453
1454                 offset += bytes;
1455                 len -= bytes;
1456
1457                 /* And verify that they are good continuation bytes */
1458                 do {
1459                         codepoint <<= 6;
1460                         codepoint |= *buf & 0x3f;
1461                         if ((*buf++ & 0xc0) != 0x80)
1462                                 return bad_offset;
1463                 } while (--bytes);
1464
1465                 /* Reject codepoints that are out of range for the sequence length. */
1466                 if (codepoint < min_val || codepoint > max_val)
1467                         return bad_offset;
1468                 /* Surrogates are only for UTF-16 and cannot be encoded in UTF-8. */
1469                 if ((codepoint & 0x1ff800) == 0xd800)
1470                         return bad_offset;
1471                 /* U+xxFFFE and U+xxFFFF are guaranteed non-characters. */
1472                 if ((codepoint & 0xfffe) == 0xfffe)
1473                         return bad_offset;
1474                 /* So are anything in the range U+FDD0..U+FDEF. */
1475                 if (codepoint >= 0xfdd0 && codepoint <= 0xfdef)
1476                         return bad_offset;
1477         }
1478         return -1;
1479 }
1480
1481 /*
1482  * This verifies that the buffer is in proper utf8 format.
1483  *
1484  * If it isn't, it assumes any non-utf8 characters are Latin1,
1485  * and does the conversion.
1486  */
1487 static int verify_utf8(struct strbuf *buf)
1488 {
1489         int ok = 1;
1490         long pos = 0;
1491
1492         for (;;) {
1493                 int bad;
1494                 unsigned char c;
1495                 unsigned char replace[2];
1496
1497                 bad = find_invalid_utf8(buf->buf + pos, buf->len - pos);
1498                 if (bad < 0)
1499                         return ok;
1500                 pos += bad;
1501                 ok = 0;
1502                 c = buf->buf[pos];
1503                 strbuf_remove(buf, pos, 1);
1504
1505                 /* We know 'c' must be in the range 128-255 */
1506                 replace[0] = 0xc0 + (c >> 6);
1507                 replace[1] = 0x80 + (c & 0x3f);
1508                 strbuf_insert(buf, pos, replace, 2);
1509                 pos += 2;
1510         }
1511 }
1512
1513 static const char commit_utf8_warn[] =
1514 "Warning: commit message did not conform to UTF-8.\n"
1515 "You may want to amend it after fixing the message, or set the config\n"
1516 "variable i18n.commitencoding to the encoding your project uses.\n";
1517
1518 int commit_tree_extended(const char *msg, size_t msg_len,
1519                          const unsigned char *tree,
1520                          struct commit_list *parents, unsigned char *ret,
1521                          const char *author, const char *sign_commit,
1522                          struct commit_extra_header *extra)
1523 {
1524         int result;
1525         int encoding_is_utf8;
1526         struct strbuf buffer;
1527
1528         assert_sha1_type(tree, OBJ_TREE);
1529
1530         if (memchr(msg, '\0', msg_len))
1531                 return error("a NUL byte in commit log message not allowed.");
1532
1533         /* Not having i18n.commitencoding is the same as having utf-8 */
1534         encoding_is_utf8 = is_encoding_utf8(git_commit_encoding);
1535
1536         strbuf_init(&buffer, 8192); /* should avoid reallocs for the headers */
1537         strbuf_addf(&buffer, "tree %s\n", sha1_to_hex(tree));
1538
1539         /*
1540          * NOTE! This ordering means that the same exact tree merged with a
1541          * different order of parents will be a _different_ changeset even
1542          * if everything else stays the same.
1543          */
1544         while (parents) {
1545                 struct commit_list *next = parents->next;
1546                 struct commit *parent = parents->item;
1547
1548                 strbuf_addf(&buffer, "parent %s\n",
1549                             sha1_to_hex(parent->object.sha1));
1550                 free(parents);
1551                 parents = next;
1552         }
1553
1554         /* Person/date information */
1555         if (!author)
1556                 author = git_author_info(IDENT_STRICT);
1557         strbuf_addf(&buffer, "author %s\n", author);
1558         strbuf_addf(&buffer, "committer %s\n", git_committer_info(IDENT_STRICT));
1559         if (!encoding_is_utf8)
1560                 strbuf_addf(&buffer, "encoding %s\n", git_commit_encoding);
1561
1562         while (extra) {
1563                 add_extra_header(&buffer, extra);
1564                 extra = extra->next;
1565         }
1566         strbuf_addch(&buffer, '\n');
1567
1568         /* And add the comment */
1569         strbuf_add(&buffer, msg, msg_len);
1570
1571         /* And check the encoding */
1572         if (encoding_is_utf8 && !verify_utf8(&buffer))
1573                 fprintf(stderr, commit_utf8_warn);
1574
1575         if (sign_commit && do_sign_commit(&buffer, sign_commit))
1576                 return -1;
1577
1578         result = write_sha1_file(buffer.buf, buffer.len, commit_type, ret);
1579         strbuf_release(&buffer);
1580         return result;
1581 }
1582
1583 struct commit *get_merge_parent(const char *name)
1584 {
1585         struct object *obj;
1586         struct commit *commit;
1587         struct object_id oid;
1588         if (get_sha1(name, oid.hash))
1589                 return NULL;
1590         obj = parse_object(oid.hash);
1591         commit = (struct commit *)peel_to_type(name, 0, obj, OBJ_COMMIT);
1592         if (commit && !commit->util) {
1593                 struct merge_remote_desc *desc;
1594                 desc = xmalloc(sizeof(*desc));
1595                 desc->obj = obj;
1596                 desc->name = strdup(name);
1597                 commit->util = desc;
1598         }
1599         return commit;
1600 }
1601
1602 /*
1603  * Append a commit to the end of the commit_list.
1604  *
1605  * next starts by pointing to the variable that holds the head of an
1606  * empty commit_list, and is updated to point to the "next" field of
1607  * the last item on the list as new commits are appended.
1608  *
1609  * Usage example:
1610  *
1611  *     struct commit_list *list;
1612  *     struct commit_list **next = &list;
1613  *
1614  *     next = commit_list_append(c1, next);
1615  *     next = commit_list_append(c2, next);
1616  *     assert(commit_list_count(list) == 2);
1617  *     return list;
1618  */
1619 struct commit_list **commit_list_append(struct commit *commit,
1620                                         struct commit_list **next)
1621 {
1622         struct commit_list *new = xmalloc(sizeof(struct commit_list));
1623         new->item = commit;
1624         *next = new;
1625         new->next = NULL;
1626         return &new->next;
1627 }
1628
1629 void print_commit_list(struct commit_list *list,
1630                        const char *format_cur,
1631                        const char *format_last)
1632 {
1633         for ( ; list; list = list->next) {
1634                 const char *format = list->next ? format_cur : format_last;
1635                 printf(format, sha1_to_hex(list->item->object.sha1));
1636         }
1637 }
1638
1639 const char *find_commit_header(const char *msg, const char *key, size_t *out_len)
1640 {
1641         int key_len = strlen(key);
1642         const char *line = msg;
1643
1644         while (line) {
1645                 const char *eol = strchrnul(line, '\n');
1646
1647                 if (line == eol)
1648                         return NULL;
1649
1650                 if (eol - line > key_len &&
1651                     !strncmp(line, key, key_len) &&
1652                     line[key_len] == ' ') {
1653                         *out_len = eol - line - key_len - 1;
1654                         return line + key_len + 1;
1655                 }
1656                 line = *eol ? eol + 1 : NULL;
1657         }
1658         return NULL;
1659 }
1660
1661 /*
1662  * Inspect sb and determine the true "end" of the log message, in
1663  * order to find where to put a new Signed-off-by: line.  Ignored are
1664  * trailing comment lines and blank lines, and also the traditional
1665  * "Conflicts:" block that is not commented out, so that we can use
1666  * "git commit -s --amend" on an existing commit that forgot to remove
1667  * it.
1668  *
1669  * Returns the number of bytes from the tail to ignore, to be fed as
1670  * the second parameter to append_signoff().
1671  */
1672 int ignore_non_trailer(struct strbuf *sb)
1673 {
1674         int boc = 0;
1675         int bol = 0;
1676         int in_old_conflicts_block = 0;
1677
1678         while (bol < sb->len) {
1679                 char *next_line;
1680
1681                 if (!(next_line = memchr(sb->buf + bol, '\n', sb->len - bol)))
1682                         next_line = sb->buf + sb->len;
1683                 else
1684                         next_line++;
1685
1686                 if (sb->buf[bol] == comment_line_char || sb->buf[bol] == '\n') {
1687                         /* is this the first of the run of comments? */
1688                         if (!boc)
1689                                 boc = bol;
1690                         /* otherwise, it is just continuing */
1691                 } else if (starts_with(sb->buf + bol, "Conflicts:\n")) {
1692                         in_old_conflicts_block = 1;
1693                         if (!boc)
1694                                 boc = bol;
1695                 } else if (in_old_conflicts_block && sb->buf[bol] == '\t') {
1696                         ; /* a pathname in the conflicts block */
1697                 } else if (boc) {
1698                         /* the previous was not trailing comment */
1699                         boc = 0;
1700                         in_old_conflicts_block = 0;
1701                 }
1702                 bol = next_line - sb->buf;
1703         }
1704         return boc ? sb->len - boc : 0;
1705 }