fast-import.c

   1 /*
   2 Format of STDIN stream:
   3
   4   stream ::= cmd*;
   5
   6   cmd ::= new_blob
   7         | new_commit
   8         | new_tag
   9         | reset_branch
  10         | checkpoint
  11         | progress
  12         ;
  13
  14   new_blob ::= 'blob' lf
  15     mark?
  16     file_content;
  17   file_content ::= data;
  18
  19   new_commit ::= 'commit' sp ref_str lf
  20     mark?
  21     ('author' sp name '<' email '>' when lf)?
  22     'committer' sp name '<' email '>' when lf
  23     commit_msg
  24     ('from' sp (ref_str | hexsha1 | sha1exp_str | idnum) lf)?
  25     ('merge' sp (ref_str | hexsha1 | sha1exp_str | idnum) lf)*
  26     file_change*
  27     lf?;
  28   commit_msg ::= data;
  29
  30   file_change ::= file_clr
  31     | file_del
  32     | file_rnm
  33     | file_cpy
  34     | file_obm
  35     | file_inm;
  36   file_clr ::= 'deleteall' lf;
  37   file_del ::= 'D' sp path_str lf;
  38   file_rnm ::= 'R' sp path_str sp path_str lf;
  39   file_cpy ::= 'C' sp path_str sp path_str lf;
  40   file_obm ::= 'M' sp mode sp (hexsha1 | idnum) sp path_str lf;
  41   file_inm ::= 'M' sp mode sp 'inline' sp path_str lf
  42     data;
  43
  44   new_tag ::= 'tag' sp tag_str lf
  45     'from' sp (ref_str | hexsha1 | sha1exp_str | idnum) lf
  46     'tagger' sp name '<' email '>' when lf
  47     tag_msg;
  48   tag_msg ::= data;
  49
  50   reset_branch ::= 'reset' sp ref_str lf
  51     ('from' sp (ref_str | hexsha1 | sha1exp_str | idnum) lf)?
  52     lf?;
  53
  54   checkpoint ::= 'checkpoint' lf
  55     lf?;
  56
  57   progress ::= 'progress' sp not_lf* lf
  58     lf?;
  59
  60      # note: the first idnum in a stream should be 1 and subsequent
  61      # idnums should not have gaps between values as this will cause
  62      # the stream parser to reserve space for the gapped values.  An
  63      # idnum can be updated in the future to a new object by issuing
  64      # a new mark directive with the old idnum.
  65      #
  66   mark ::= 'mark' sp idnum lf;
  67   data ::= (delimited_data | exact_data)
  68     lf?;
  69
  70     # note: delim may be any string but must not contain lf.
  71     # data_line may contain any data but must not be exactly
  72     # delim.
  73   delimited_data ::= 'data' sp '<<' delim lf
  74     (data_line lf)*
  75     delim lf;
  76
  77      # note: declen indicates the length of binary_data in bytes.
  78      # declen does not include the lf preceeding the binary data.
  79      #
  80   exact_data ::= 'data' sp declen lf
  81     binary_data;
  82
  83      # note: quoted strings are C-style quoting supporting \c for
  84      # common escapes of 'c' (e..g \n, \t, \\, \") or \nnn where nnn
  85      # is the signed byte value in octal.  Note that the only
  86      # characters which must actually be escaped to protect the
  87      # stream formatting is: \, " and LF.  Otherwise these values
  88      # are UTF8.
  89      #
  90   ref_str     ::= ref;
  91   sha1exp_str ::= sha1exp;
  92   tag_str     ::= tag;
  93   path_str    ::= path    | '"' quoted(path)    '"' ;
  94   mode        ::= '100644' | '644'
  95                 | '100755' | '755'
  96                 | '120000'
  97                 ;
  98
  99   declen ::= # unsigned 32 bit value, ascii base10 notation;
 100   bigint ::= # unsigned integer value, ascii base10 notation;
 101   binary_data ::= # file content, not interpreted;
 102
 103   when         ::= raw_when | rfc2822_when;
 104   raw_when     ::= ts sp tz;
 105   rfc2822_when ::= # Valid RFC 2822 date and time;
 106
 107   sp ::= # ASCII space character;
 108   lf ::= # ASCII newline (LF) character;
 109
 110      # note: a colon (':') must precede the numerical value assigned to
 111      # an idnum.  This is to distinguish it from a ref or tag name as
 112      # GIT does not permit ':' in ref or tag strings.
 113      #
 114   idnum   ::= ':' bigint;
 115   path    ::= # GIT style file path, e.g. "a/b/c";
 116   ref     ::= # GIT ref name, e.g. "refs/heads/MOZ_GECKO_EXPERIMENT";
 117   tag     ::= # GIT tag name, e.g. "FIREFOX_1_5";
 118   sha1exp ::= # Any valid GIT SHA1 expression;
 119   hexsha1 ::= # SHA1 in hexadecimal format;
 120
 121      # note: name and email are UTF8 strings, however name must not
 122      # contain '<' or lf and email must not contain any of the
 123      # following: '<', '>', lf.
 124      #
 125   name  ::= # valid GIT author/committer name;
 126   email ::= # valid GIT author/committer email;
 127   ts    ::= # time since the epoch in seconds, ascii base10 notation;
 128   tz    ::= # GIT style timezone;
 129
 130      # note: comments may appear anywhere in the input, except
 131      # within a data command.  Any form of the data command
 132      # always escapes the related input from comment processing.
 133      #
 134      # In case it is not clear, the '#' that starts the comment
 135      # must be the first character on that the line (an lf have
 136      # preceeded it).
 137      #
 138   comment ::= '#' not_lf* lf;
 139   not_lf  ::= # Any byte that is not ASCII newline (LF);
 140 */
 141
 142 #include "builtin.h"
 143 #include "cache.h"
 144 #include "object.h"
 145 #include "blob.h"
 146 #include "tree.h"
 147 #include "commit.h"
 148 #include "delta.h"
 149 #include "pack.h"
 150 #include "refs.h"
 151 #include "csum-file.h"
 152 #include "strbuf.h"
 153 #include "quote.h"
 154
 155 #define PACK_ID_BITS 16
 156 #define MAX_PACK_ID ((1<<PACK_ID_BITS)-1)
 157
 158 struct object_entry
 159 {
 160         struct object_entry *next;
 161         uint32_t offset;
 162         unsigned type : TYPE_BITS;
 163         unsigned pack_id : PACK_ID_BITS;
 164         unsigned char sha1[20];
 165 };
 166
 167 struct object_entry_pool
 168 {
 169         struct object_entry_pool *next_pool;
 170         struct object_entry *next_free;
 171         struct object_entry *end;
 172         struct object_entry entries[FLEX_ARRAY]; /* more */
 173 };
 174
 175 struct mark_set
 176 {
 177         union {
 178                 struct object_entry *marked[1024];
 179                 struct mark_set *sets[1024];
 180         } data;
 181         unsigned int shift;
 182 };
 183
 184 struct last_object
 185 {
 186         void *data;
 187         unsigned long len;
 188         uint32_t offset;
 189         unsigned int depth;
 190         unsigned no_free:1;
 191 };
 192
 193 struct mem_pool
 194 {
 195         struct mem_pool *next_pool;
 196         char *next_free;
 197         char *end;
 198         char space[FLEX_ARRAY]; /* more */
 199 };
 200
 201 struct atom_str
 202 {
 203         struct atom_str *next_atom;
 204         unsigned short str_len;
 205         char str_dat[FLEX_ARRAY]; /* more */
 206 };
 207
 208 struct tree_content;
 209 struct tree_entry
 210 {
 211         struct tree_content *tree;
 212         struct atom_str* name;
 213         struct tree_entry_ms
 214         {
 215                 uint16_t mode;
 216                 unsigned char sha1[20];
 217         } versions[2];
 218 };
 219
 220 struct tree_content
 221 {
 222         unsigned int entry_capacity; /* must match avail_tree_content */
 223         unsigned int entry_count;
 224         unsigned int delta_depth;
 225         struct tree_entry *entries[FLEX_ARRAY]; /* more */
 226 };
 227
 228 struct avail_tree_content
 229 {
 230         unsigned int entry_capacity; /* must match tree_content */
 231         struct avail_tree_content *next_avail;
 232 };
 233
 234 struct branch
 235 {
 236         struct branch *table_next_branch;
 237         struct branch *active_next_branch;
 238         const char *name;
 239         struct tree_entry branch_tree;
 240         uintmax_t last_commit;
 241         unsigned active : 1;
 242         unsigned pack_id : PACK_ID_BITS;
 243         unsigned char sha1[20];
 244 };
 245
 246 struct tag
 247 {
 248         struct tag *next_tag;
 249         const char *name;
 250         unsigned int pack_id;
 251         unsigned char sha1[20];
 252 };
 253
 254 struct dbuf
 255 {
 256         void *buffer;
 257         size_t capacity;
 258 };
 259
 260 struct hash_list
 261 {
 262         struct hash_list *next;
 263         unsigned char sha1[20];
 264 };
 265
 266 typedef enum {
 267         WHENSPEC_RAW = 1,
 268         WHENSPEC_RFC2822,
 269         WHENSPEC_NOW,
 270 } whenspec_type;
 271
 272 /* Configured limits on output */
 273 static unsigned long max_depth = 10;
 274 static off_t max_packsize = (1LL << 32) - 1;
 275 static int force_update;
 276
 277 /* Stats and misc. counters */
 278 static uintmax_t alloc_count;
 279 static uintmax_t marks_set_count;
 280 static uintmax_t object_count_by_type[1 << TYPE_BITS];
 281 static uintmax_t duplicate_count_by_type[1 << TYPE_BITS];
 282 static uintmax_t delta_count_by_type[1 << TYPE_BITS];
 283 static unsigned long object_count;
 284 static unsigned long branch_count;
 285 static unsigned long branch_load_count;
 286 static int failure;
 287 static FILE *pack_edges;
 288
 289 /* Memory pools */
 290 static size_t mem_pool_alloc = 2*1024*1024 - sizeof(struct mem_pool);
 291 static size_t total_allocd;
 292 static struct mem_pool *mem_pool;
 293
 294 /* Atom management */
 295 static unsigned int atom_table_sz = 4451;
 296 static unsigned int atom_cnt;
 297 static struct atom_str **atom_table;
 298
 299 /* The .pack file being generated */
 300 static unsigned int pack_id;
 301 static struct packed_git *pack_data;
 302 static struct packed_git **all_packs;
 303 static unsigned long pack_size;
 304
 305 /* Table of objects we've written. */
 306 static unsigned int object_entry_alloc = 5000;
 307 static struct object_entry_pool *blocks;
 308 static struct object_entry *object_table[1 << 16];
 309 static struct mark_set *marks;
 310 static const char* mark_file;
 311
 312 /* Our last blob */
 313 static struct last_object last_blob;
 314
 315 /* Tree management */
 316 static unsigned int tree_entry_alloc = 1000;
 317 static void *avail_tree_entry;
 318 static unsigned int avail_tree_table_sz = 100;
 319 static struct avail_tree_content **avail_tree_table;
 320 static struct dbuf old_tree;
 321 static struct dbuf new_tree;
 322
 323 /* Branch data */
 324 static unsigned long max_active_branches = 5;
 325 static unsigned long cur_active_branches;
 326 static unsigned long branch_table_sz = 1039;
 327 static struct branch **branch_table;
 328 static struct branch *active_branches;
 329
 330 /* Tag data */
 331 static struct tag *first_tag;
 332 static struct tag *last_tag;
 333
 334 /* Input stream parsing */
 335 static whenspec_type whenspec = WHENSPEC_RAW;
 336 static struct strbuf command_buf;
 337 static int unread_command_buf;
 338 static uintmax_t next_mark;
 339 static struct dbuf new_data;
 340
 341
 342 static void alloc_objects(unsigned int cnt)
 343 {
 344         struct object_entry_pool *b;
 345
 346         b = xmalloc(sizeof(struct object_entry_pool)
 347                 + cnt * sizeof(struct object_entry));
 348         b->next_pool = blocks;
 349         b->next_free = b->entries;
 350         b->end = b->entries + cnt;
 351         blocks = b;
 352         alloc_count += cnt;
 353 }
 354
 355 static struct object_entry *new_object(unsigned char *sha1)
 356 {
 357         struct object_entry *e;
 358
 359         if (blocks->next_free == blocks->end)
 360                 alloc_objects(object_entry_alloc);
 361
 362         e = blocks->next_free++;
 363         hashcpy(e->sha1, sha1);
 364         return e;
 365 }
 366
 367 static struct object_entry *find_object(unsigned char *sha1)
 368 {
 369         unsigned int h = sha1[0] << 8 | sha1[1];
 370         struct object_entry *e;
 371         for (e = object_table[h]; e; e = e->next)
 372                 if (!hashcmp(sha1, e->sha1))
 373                         return e;
 374         return NULL;
 375 }
 376
 377 static struct object_entry *insert_object(unsigned char *sha1)
 378 {
 379         unsigned int h = sha1[0] << 8 | sha1[1];
 380         struct object_entry *e = object_table[h];
 381         struct object_entry *p = NULL;
 382
 383         while (e) {
 384                 if (!hashcmp(sha1, e->sha1))
 385                         return e;
 386                 p = e;
 387                 e = e->next;
 388         }
 389
 390         e = new_object(sha1);
 391         e->next = NULL;
 392         e->offset = 0;
 393         if (p)
 394                 p->next = e;
 395         else
 396                 object_table[h] = e;
 397         return e;
 398 }
 399
 400 static unsigned int hc_str(const char *s, size_t len)
 401 {
 402         unsigned int r = 0;
 403         while (len-- > 0)
 404                 r = r * 31 + *s++;
 405         return r;
 406 }
 407
 408 static void *pool_alloc(size_t len)
 409 {
 410         struct mem_pool *p;
 411         void *r;
 412
 413         for (p = mem_pool; p; p = p->next_pool)
 414                 if ((p->end - p->next_free >= len))
 415                         break;
 416
 417         if (!p) {
 418                 if (len >= (mem_pool_alloc/2)) {
 419                         total_allocd += len;
 420                         return xmalloc(len);
 421                 }
 422                 total_allocd += sizeof(struct mem_pool) + mem_pool_alloc;
 423                 p = xmalloc(sizeof(struct mem_pool) + mem_pool_alloc);
 424                 p->next_pool = mem_pool;
 425                 p->next_free = p->space;
 426                 p->end = p->next_free + mem_pool_alloc;
 427                 mem_pool = p;
 428         }
 429
 430         r = p->next_free;
 431         /* round out to a pointer alignment */
 432         if (len & (sizeof(void*) - 1))
 433                 len += sizeof(void*) - (len & (sizeof(void*) - 1));
 434         p->next_free += len;
 435         return r;
 436 }
 437
 438 static void *pool_calloc(size_t count, size_t size)
 439 {
 440         size_t len = count * size;
 441         void *r = pool_alloc(len);
 442         memset(r, 0, len);
 443         return r;
 444 }
 445
 446 static char *pool_strdup(const char *s)
 447 {
 448         char *r = pool_alloc(strlen(s) + 1);
 449         strcpy(r, s);
 450         return r;
 451 }
 452
 453 static void size_dbuf(struct dbuf *b, size_t maxlen)
 454 {
 455         if (b->buffer) {
 456                 if (b->capacity >= maxlen)
 457                         return;
 458                 free(b->buffer);
 459         }
 460         b->capacity = ((maxlen / 1024) + 1) * 1024;
 461         b->buffer = xmalloc(b->capacity);
 462 }
 463
 464 static void insert_mark(uintmax_t idnum, struct object_entry *oe)
 465 {
 466         struct mark_set *s = marks;
 467         while ((idnum >> s->shift) >= 1024) {
 468                 s = pool_calloc(1, sizeof(struct mark_set));
 469                 s->shift = marks->shift + 10;
 470                 s->data.sets[0] = marks;
 471                 marks = s;
 472         }
 473         while (s->shift) {
 474                 uintmax_t i = idnum >> s->shift;
 475                 idnum -= i << s->shift;
 476                 if (!s->data.sets[i]) {
 477                         s->data.sets[i] = pool_calloc(1, sizeof(struct mark_set));
 478                         s->data.sets[i]->shift = s->shift - 10;
 479                 }
 480                 s = s->data.sets[i];
 481         }
 482         if (!s->data.marked[idnum])
 483                 marks_set_count++;
 484         s->data.marked[idnum] = oe;
 485 }
 486
 487 static struct object_entry *find_mark(uintmax_t idnum)
 488 {
 489         uintmax_t orig_idnum = idnum;
 490         struct mark_set *s = marks;
 491         struct object_entry *oe = NULL;
 492         if ((idnum >> s->shift) < 1024) {
 493                 while (s && s->shift) {
 494                         uintmax_t i = idnum >> s->shift;
 495                         idnum -= i << s->shift;
 496                         s = s->data.sets[i];
 497                 }
 498                 if (s)
 499                         oe = s->data.marked[idnum];
 500         }
 501         if (!oe)
 502                 die("mark :%" PRIuMAX " not declared", orig_idnum);
 503         return oe;
 504 }
 505
 506 static struct atom_str *to_atom(const char *s, unsigned short len)
 507 {
 508         unsigned int hc = hc_str(s, len) % atom_table_sz;
 509         struct atom_str *c;
 510
 511         for (c = atom_table[hc]; c; c = c->next_atom)
 512                 if (c->str_len == len && !strncmp(s, c->str_dat, len))
 513                         return c;
 514
 515         c = pool_alloc(sizeof(struct atom_str) + len + 1);
 516         c->str_len = len;
 517         strncpy(c->str_dat, s, len);
 518         c->str_dat[len] = 0;
 519         c->next_atom = atom_table[hc];
 520         atom_table[hc] = c;
 521         atom_cnt++;
 522         return c;
 523 }
 524
 525 static struct branch *lookup_branch(const char *name)
 526 {
 527         unsigned int hc = hc_str(name, strlen(name)) % branch_table_sz;
 528         struct branch *b;
 529
 530         for (b = branch_table[hc]; b; b = b->table_next_branch)
 531                 if (!strcmp(name, b->name))
 532                         return b;
 533         return NULL;
 534 }
 535
 536 static struct branch *new_branch(const char *name)
 537 {
 538         unsigned int hc = hc_str(name, strlen(name)) % branch_table_sz;
 539         struct branch* b = lookup_branch(name);
 540
 541         if (b)
 542                 die("Invalid attempt to create duplicate branch: %s", name);
 543         switch (check_ref_format(name)) {
 544         case  0: break; /* its valid */
 545         case -2: break; /* valid, but too few '/', allow anyway */
 546         default:
 547                 die("Branch name doesn't conform to GIT standards: %s", name);
 548         }
 549
 550         b = pool_calloc(1, sizeof(struct branch));
 551         b->name = pool_strdup(name);
 552         b->table_next_branch = branch_table[hc];
 553         b->branch_tree.versions[0].mode = S_IFDIR;
 554         b->branch_tree.versions[1].mode = S_IFDIR;
 555         b->active = 0;
 556         b->pack_id = MAX_PACK_ID;
 557         branch_table[hc] = b;
 558         branch_count++;
 559         return b;
 560 }
 561
 562 static unsigned int hc_entries(unsigned int cnt)
 563 {
 564         cnt = cnt & 7 ? (cnt / 8) + 1 : cnt / 8;
 565         return cnt < avail_tree_table_sz ? cnt : avail_tree_table_sz - 1;
 566 }
 567
 568 static struct tree_content *new_tree_content(unsigned int cnt)
 569 {
 570         struct avail_tree_content *f, *l = NULL;
 571         struct tree_content *t;
 572         unsigned int hc = hc_entries(cnt);
 573
 574         for (f = avail_tree_table[hc]; f; l = f, f = f->next_avail)
 575                 if (f->entry_capacity >= cnt)
 576                         break;
 577
 578         if (f) {
 579                 if (l)
 580                         l->next_avail = f->next_avail;
 581                 else
 582                         avail_tree_table[hc] = f->next_avail;
 583         } else {
 584                 cnt = cnt & 7 ? ((cnt / 8) + 1) * 8 : cnt;
 585                 f = pool_alloc(sizeof(*t) + sizeof(t->entries[0]) * cnt);
 586                 f->entry_capacity = cnt;
 587         }
 588
 589         t = (struct tree_content*)f;
 590         t->entry_count = 0;
 591         t->delta_depth = 0;
 592         return t;
 593 }
 594
 595 static void release_tree_entry(struct tree_entry *e);
 596 static void release_tree_content(struct tree_content *t)
 597 {
 598         struct avail_tree_content *f = (struct avail_tree_content*)t;
 599         unsigned int hc = hc_entries(f->entry_capacity);
 600         f->next_avail = avail_tree_table[hc];
 601         avail_tree_table[hc] = f;
 602 }
 603
 604 static void release_tree_content_recursive(struct tree_content *t)
 605 {
 606         unsigned int i;
 607         for (i = 0; i < t->entry_count; i++)
 608                 release_tree_entry(t->entries[i]);
 609         release_tree_content(t);
 610 }
 611
 612 static struct tree_content *grow_tree_content(
 613         struct tree_content *t,
 614         int amt)
 615 {
 616         struct tree_content *r = new_tree_content(t->entry_count + amt);
 617         r->entry_count = t->entry_count;
 618         r->delta_depth = t->delta_depth;
 619         memcpy(r->entries,t->entries,t->entry_count*sizeof(t->entries[0]));
 620         release_tree_content(t);
 621         return r;
 622 }
 623
 624 static struct tree_entry *new_tree_entry(void)
 625 {
 626         struct tree_entry *e;
 627
 628         if (!avail_tree_entry) {
 629                 unsigned int n = tree_entry_alloc;
 630                 total_allocd += n * sizeof(struct tree_entry);
 631                 avail_tree_entry = e = xmalloc(n * sizeof(struct tree_entry));
 632                 while (n-- > 1) {
 633                         *((void**)e) = e + 1;
 634                         e++;
 635                 }
 636                 *((void**)e) = NULL;
 637         }
 638
 639         e = avail_tree_entry;
 640         avail_tree_entry = *((void**)e);
 641         return e;
 642 }
 643
 644 static void release_tree_entry(struct tree_entry *e)
 645 {
 646         if (e->tree)
 647                 release_tree_content_recursive(e->tree);
 648         *((void**)e) = avail_tree_entry;
 649         avail_tree_entry = e;
 650 }
 651
 652 static struct tree_content *dup_tree_content(struct tree_content *s)
 653 {
 654         struct tree_content *d;
 655         struct tree_entry *a, *b;
 656         unsigned int i;
 657
 658         if (!s)
 659                 return NULL;
 660         d = new_tree_content(s->entry_count);
 661         for (i = 0; i < s->entry_count; i++) {
 662                 a = s->entries[i];
 663                 b = new_tree_entry();
 664                 memcpy(b, a, sizeof(*a));
 665                 if (a->tree && is_null_sha1(b->versions[1].sha1))
 666                         b->tree = dup_tree_content(a->tree);
 667                 else
 668                         b->tree = NULL;
 669                 d->entries[i] = b;
 670         }
 671         d->entry_count = s->entry_count;
 672         d->delta_depth = s->delta_depth;
 673
 674         return d;
 675 }
 676
 677 static void start_packfile(void)
 678 {
 679         static char tmpfile[PATH_MAX];
 680         struct packed_git *p;
 681         struct pack_header hdr;
 682         int pack_fd;
 683
 684         snprintf(tmpfile, sizeof(tmpfile),
 685                 "%s/tmp_pack_XXXXXX", get_object_directory());
 686         pack_fd = xmkstemp(tmpfile);
 687         p = xcalloc(1, sizeof(*p) + strlen(tmpfile) + 2);
 688         strcpy(p->pack_name, tmpfile);
 689         p->pack_fd = pack_fd;
 690
 691         hdr.hdr_signature = htonl(PACK_SIGNATURE);
 692         hdr.hdr_version = htonl(2);
 693         hdr.hdr_entries = 0;
 694         write_or_die(p->pack_fd, &hdr, sizeof(hdr));
 695
 696         pack_data = p;
 697         pack_size = sizeof(hdr);
 698         object_count = 0;
 699
 700         all_packs = xrealloc(all_packs, sizeof(*all_packs) * (pack_id + 1));
 701         all_packs[pack_id] = p;
 702 }
 703
 704 static int oecmp (const void *a_, const void *b_)
 705 {
 706         struct object_entry *a = *((struct object_entry**)a_);
 707         struct object_entry *b = *((struct object_entry**)b_);
 708         return hashcmp(a->sha1, b->sha1);
 709 }
 710
 711 static char *create_index(void)
 712 {
 713         static char tmpfile[PATH_MAX];
 714         SHA_CTX ctx;
 715         struct sha1file *f;
 716         struct object_entry **idx, **c, **last, *e;
 717         struct object_entry_pool *o;
 718         uint32_t array[256];
 719         int i, idx_fd;
 720
 721         /* Build the sorted table of object IDs. */
 722         idx = xmalloc(object_count * sizeof(struct object_entry*));
 723         c = idx;
 724         for (o = blocks; o; o = o->next_pool)
 725                 for (e = o->next_free; e-- != o->entries;)
 726                         if (pack_id == e->pack_id)
 727                                 *c++ = e;
 728         last = idx + object_count;
 729         if (c != last)
 730                 die("internal consistency error creating the index");
 731         qsort(idx, object_count, sizeof(struct object_entry*), oecmp);
 732
 733         /* Generate the fan-out array. */
 734         c = idx;
 735         for (i = 0; i < 256; i++) {
 736                 struct object_entry **next = c;;
 737                 while (next < last) {
 738                         if ((*next)->sha1[0] != i)
 739                                 break;
 740                         next++;
 741                 }
 742                 array[i] = htonl(next - idx);
 743                 c = next;
 744         }
 745
 746         snprintf(tmpfile, sizeof(tmpfile),
 747                 "%s/tmp_idx_XXXXXX", get_object_directory());
 748         idx_fd = xmkstemp(tmpfile);
 749         f = sha1fd(idx_fd, tmpfile);
 750         sha1write(f, array, 256 * sizeof(int));
 751         SHA1_Init(&ctx);
 752         for (c = idx; c != last; c++) {
 753                 uint32_t offset = htonl((*c)->offset);
 754                 sha1write(f, &offset, 4);
 755                 sha1write(f, (*c)->sha1, sizeof((*c)->sha1));
 756                 SHA1_Update(&ctx, (*c)->sha1, 20);
 757         }
 758         sha1write(f, pack_data->sha1, sizeof(pack_data->sha1));
 759         sha1close(f, NULL, 1);
 760         free(idx);
 761         SHA1_Final(pack_data->sha1, &ctx);
 762         return tmpfile;
 763 }
 764
 765 static char *keep_pack(char *curr_index_name)
 766 {
 767         static char name[PATH_MAX];
 768         static const char *keep_msg = "fast-import";
 769         int keep_fd;
 770
 771         chmod(pack_data->pack_name, 0444);
 772         chmod(curr_index_name, 0444);
 773
 774         snprintf(name, sizeof(name), "%s/pack/pack-%s.keep",
 775                  get_object_directory(), sha1_to_hex(pack_data->sha1));
 776         keep_fd = open(name, O_RDWR|O_CREAT|O_EXCL, 0600);
 777         if (keep_fd < 0)
 778                 die("cannot create keep file");
 779         write(keep_fd, keep_msg, strlen(keep_msg));
 780         close(keep_fd);
 781
 782         snprintf(name, sizeof(name), "%s/pack/pack-%s.pack",
 783                  get_object_directory(), sha1_to_hex(pack_data->sha1));
 784         if (move_temp_to_file(pack_data->pack_name, name))
 785                 die("cannot store pack file");
 786
 787         snprintf(name, sizeof(name), "%s/pack/pack-%s.idx",
 788                  get_object_directory(), sha1_to_hex(pack_data->sha1));
 789         if (move_temp_to_file(curr_index_name, name))
 790                 die("cannot store index file");
 791         return name;
 792 }
 793
 794 static void unkeep_all_packs(void)
 795 {
 796         static char name[PATH_MAX];
 797         int k;
 798
 799         for (k = 0; k < pack_id; k++) {
 800                 struct packed_git *p = all_packs[k];
 801                 snprintf(name, sizeof(name), "%s/pack/pack-%s.keep",
 802                          get_object_directory(), sha1_to_hex(p->sha1));
 803                 unlink(name);
 804         }
 805 }
 806
 807 static void end_packfile(void)
 808 {
 809         struct packed_git *old_p = pack_data, *new_p;
 810
 811         if (object_count) {
 812                 char *idx_name;
 813                 int i;
 814                 struct branch *b;
 815                 struct tag *t;
 816
 817                 fixup_pack_header_footer(pack_data->pack_fd, pack_data->sha1,
 818                                     pack_data->pack_name, object_count);
 819                 close(pack_data->pack_fd);
 820                 idx_name = keep_pack(create_index());
 821
 822                 /* Register the packfile with core git's machinary. */
 823                 new_p = add_packed_git(idx_name, strlen(idx_name), 1);
 824                 if (!new_p)
 825                         die("core git rejected index %s", idx_name);
 826                 new_p->windows = old_p->windows;
 827                 all_packs[pack_id] = new_p;
 828                 install_packed_git(new_p);
 829
 830                 /* Print the boundary */
 831                 if (pack_edges) {
 832                         fprintf(pack_edges, "%s:", new_p->pack_name);
 833                         for (i = 0; i < branch_table_sz; i++) {
 834                                 for (b = branch_table[i]; b; b = b->table_next_branch) {
 835                                         if (b->pack_id == pack_id)
 836                                                 fprintf(pack_edges, " %s", sha1_to_hex(b->sha1));
 837                                 }
 838                         }
 839                         for (t = first_tag; t; t = t->next_tag) {
 840                                 if (t->pack_id == pack_id)
 841                                         fprintf(pack_edges, " %s", sha1_to_hex(t->sha1));
 842                         }
 843                         fputc('\n', pack_edges);
 844                         fflush(pack_edges);
 845                 }
 846
 847                 pack_id++;
 848         }
 849         else
 850                 unlink(old_p->pack_name);
 851         free(old_p);
 852
 853         /* We can't carry a delta across packfiles. */
 854         free(last_blob.data);
 855         last_blob.data = NULL;
 856         last_blob.len = 0;
 857         last_blob.offset = 0;
 858         last_blob.depth = 0;
 859 }
 860
 861 static void cycle_packfile(void)
 862 {
 863         end_packfile();
 864         start_packfile();
 865 }
 866
 867 static size_t encode_header(
 868         enum object_type type,
 869         size_t size,
 870         unsigned char *hdr)
 871 {
 872         int n = 1;
 873         unsigned char c;
 874
 875         if (type < OBJ_COMMIT || type > OBJ_REF_DELTA)
 876                 die("bad type %d", type);
 877
 878         c = (type << 4) | (size & 15);
 879         size >>= 4;
 880         while (size) {
 881                 *hdr++ = c | 0x80;
 882                 c = size & 0x7f;
 883                 size >>= 7;
 884                 n++;
 885         }
 886         *hdr = c;
 887         return n;
 888 }
 889
 890 static int store_object(
 891         enum object_type type,
 892         void *dat,
 893         size_t datlen,
 894         struct last_object *last,
 895         unsigned char *sha1out,
 896         uintmax_t mark)
 897 {
 898         void *out, *delta;
 899         struct object_entry *e;
 900         unsigned char hdr[96];
 901         unsigned char sha1[20];
 902         unsigned long hdrlen, deltalen;
 903         SHA_CTX c;
 904         z_stream s;
 905
 906         hdrlen = sprintf((char*)hdr,"%s %lu", typename(type),
 907                 (unsigned long)datlen) + 1;
 908         SHA1_Init(&c);
 909         SHA1_Update(&c, hdr, hdrlen);
 910         SHA1_Update(&c, dat, datlen);
 911         SHA1_Final(sha1, &c);
 912         if (sha1out)
 913                 hashcpy(sha1out, sha1);
 914
 915         e = insert_object(sha1);
 916         if (mark)
 917                 insert_mark(mark, e);
 918         if (e->offset) {
 919                 duplicate_count_by_type[type]++;
 920                 return 1;
 921         } else if (find_sha1_pack(sha1, packed_git)) {
 922                 e->type = type;
 923                 e->pack_id = MAX_PACK_ID;
 924                 e->offset = 1; /* just not zero! */
 925                 duplicate_count_by_type[type]++;
 926                 return 1;
 927         }
 928
 929         if (last && last->data && last->depth < max_depth) {
 930                 delta = diff_delta(last->data, last->len,
 931                         dat, datlen,
 932                         &deltalen, 0);
 933                 if (delta && deltalen >= datlen) {
 934                         free(delta);
 935                         delta = NULL;
 936                 }
 937         } else
 938                 delta = NULL;
 939
 940         memset(&s, 0, sizeof(s));
 941         deflateInit(&s, zlib_compression_level);
 942         if (delta) {
 943                 s.next_in = delta;
 944                 s.avail_in = deltalen;
 945         } else {
 946                 s.next_in = dat;
 947                 s.avail_in = datlen;
 948         }
 949         s.avail_out = deflateBound(&s, s.avail_in);
 950         s.next_out = out = xmalloc(s.avail_out);
 951         while (deflate(&s, Z_FINISH) == Z_OK)
 952                 /* nothing */;
 953         deflateEnd(&s);
 954
 955         /* Determine if we should auto-checkpoint. */
 956         if ((pack_size + 60 + s.total_out) > max_packsize
 957                 || (pack_size + 60 + s.total_out) < pack_size) {
 958
 959                 /* This new object needs to *not* have the current pack_id. */
 960                 e->pack_id = pack_id + 1;
 961                 cycle_packfile();
 962
 963                 /* We cannot carry a delta into the new pack. */
 964                 if (delta) {
 965                         free(delta);
 966                         delta = NULL;
 967
 968                         memset(&s, 0, sizeof(s));
 969                         deflateInit(&s, zlib_compression_level);
 970                         s.next_in = dat;
 971                         s.avail_in = datlen;
 972                         s.avail_out = deflateBound(&s, s.avail_in);
 973                         s.next_out = out = xrealloc(out, s.avail_out);
 974                         while (deflate(&s, Z_FINISH) == Z_OK)
 975                                 /* nothing */;
 976                         deflateEnd(&s);
 977                 }
 978         }
 979
 980         e->type = type;
 981         e->pack_id = pack_id;
 982         e->offset = pack_size;
 983         object_count++;
 984         object_count_by_type[type]++;
 985
 986         if (delta) {
 987                 unsigned long ofs = e->offset - last->offset;
 988                 unsigned pos = sizeof(hdr) - 1;
 989
 990                 delta_count_by_type[type]++;
 991                 last->depth++;
 992
 993                 hdrlen = encode_header(OBJ_OFS_DELTA, deltalen, hdr);
 994                 write_or_die(pack_data->pack_fd, hdr, hdrlen);
 995                 pack_size += hdrlen;
 996
 997                 hdr[pos] = ofs & 127;
 998                 while (ofs >>= 7)
 999                         hdr[--pos] = 128 | (--ofs & 127);
1000                 write_or_die(pack_data->pack_fd, hdr + pos, sizeof(hdr) - pos);
1001                 pack_size += sizeof(hdr) - pos;
1002         } else {
1003                 if (last)
1004                         last->depth = 0;
1005                 hdrlen = encode_header(type, datlen, hdr);
1006                 write_or_die(pack_data->pack_fd, hdr, hdrlen);
1007                 pack_size += hdrlen;
1008         }
1009
1010         write_or_die(pack_data->pack_fd, out, s.total_out);
1011         pack_size += s.total_out;
1012
1013         free(out);
1014         free(delta);
1015         if (last) {
1016                 if (!last->no_free)
1017                         free(last->data);
1018                 last->data = dat;
1019                 last->offset = e->offset;
1020                 last->len = datlen;
1021         }
1022         return 0;
1023 }
1024
1025 static void *gfi_unpack_entry(
1026         struct object_entry *oe,
1027         unsigned long *sizep)
1028 {
1029         enum object_type type;
1030         struct packed_git *p = all_packs[oe->pack_id];
1031         if (p == pack_data)
1032                 p->pack_size = pack_size + 20;
1033         return unpack_entry(p, oe->offset, &type, sizep);
1034 }
1035
1036 static const char *get_mode(const char *str, uint16_t *modep)
1037 {
1038         unsigned char c;
1039         uint16_t mode = 0;
1040
1041         while ((c = *str++) != ' ') {
1042                 if (c < '0' || c > '7')
1043                         return NULL;
1044                 mode = (mode << 3) + (c - '0');
1045         }
1046         *modep = mode;
1047         return str;
1048 }
1049
1050 static void load_tree(struct tree_entry *root)
1051 {
1052         unsigned char* sha1 = root->versions[1].sha1;
1053         struct object_entry *myoe;
1054         struct tree_content *t;
1055         unsigned long size;
1056         char *buf;
1057         const char *c;
1058
1059         root->tree = t = new_tree_content(8);
1060         if (is_null_sha1(sha1))
1061                 return;
1062
1063         myoe = find_object(sha1);
1064         if (myoe && myoe->pack_id != MAX_PACK_ID) {
1065                 if (myoe->type != OBJ_TREE)
1066                         die("Not a tree: %s", sha1_to_hex(sha1));
1067                 t->delta_depth = 0;
1068                 buf = gfi_unpack_entry(myoe, &size);
1069         } else {
1070                 enum object_type type;
1071                 buf = read_sha1_file(sha1, &type, &size);
1072                 if (!buf || type != OBJ_TREE)
1073                         die("Can't load tree %s", sha1_to_hex(sha1));
1074         }
1075
1076         c = buf;
1077         while (c != (buf + size)) {
1078                 struct tree_entry *e = new_tree_entry();
1079
1080                 if (t->entry_count == t->entry_capacity)
1081                         root->tree = t = grow_tree_content(t, t->entry_count);
1082                 t->entries[t->entry_count++] = e;
1083
1084                 e->tree = NULL;
1085                 c = get_mode(c, &e->versions[1].mode);
1086                 if (!c)
1087                         die("Corrupt mode in %s", sha1_to_hex(sha1));
1088                 e->versions[0].mode = e->versions[1].mode;
1089                 e->name = to_atom(c, strlen(c));
1090                 c += e->name->str_len + 1;
1091                 hashcpy(e->versions[0].sha1, (unsigned char*)c);
1092                 hashcpy(e->versions[1].sha1, (unsigned char*)c);
1093                 c += 20;
1094         }
1095         free(buf);
1096 }
1097
1098 static int tecmp0 (const void *_a, const void *_b)
1099 {
1100         struct tree_entry *a = *((struct tree_entry**)_a);
1101         struct tree_entry *b = *((struct tree_entry**)_b);
1102         return base_name_compare(
1103                 a->name->str_dat, a->name->str_len, a->versions[0].mode,
1104                 b->name->str_dat, b->name->str_len, b->versions[0].mode);
1105 }
1106
1107 static int tecmp1 (const void *_a, const void *_b)
1108 {
1109         struct tree_entry *a = *((struct tree_entry**)_a);
1110         struct tree_entry *b = *((struct tree_entry**)_b);
1111         return base_name_compare(
1112                 a->name->str_dat, a->name->str_len, a->versions[1].mode,
1113                 b->name->str_dat, b->name->str_len, b->versions[1].mode);
1114 }
1115
1116 static void mktree(struct tree_content *t,
1117         int v,
1118         unsigned long *szp,
1119         struct dbuf *b)
1120 {
1121         size_t maxlen = 0;
1122         unsigned int i;
1123         char *c;
1124
1125         if (!v)
1126                 qsort(t->entries,t->entry_count,sizeof(t->entries[0]),tecmp0);
1127         else
1128                 qsort(t->entries,t->entry_count,sizeof(t->entries[0]),tecmp1);
1129
1130         for (i = 0; i < t->entry_count; i++) {
1131                 if (t->entries[i]->versions[v].mode)
1132                         maxlen += t->entries[i]->name->str_len + 34;
1133         }
1134
1135         size_dbuf(b, maxlen);
1136         c = b->buffer;
1137         for (i = 0; i < t->entry_count; i++) {
1138                 struct tree_entry *e = t->entries[i];
1139                 if (!e->versions[v].mode)
1140                         continue;
1141                 c += sprintf(c, "%o", (unsigned int)e->versions[v].mode);
1142                 *c++ = ' ';
1143                 strcpy(c, e->name->str_dat);
1144                 c += e->name->str_len + 1;
1145                 hashcpy((unsigned char*)c, e->versions[v].sha1);
1146                 c += 20;
1147         }
1148         *szp = c - (char*)b->buffer;
1149 }
1150
1151 static void store_tree(struct tree_entry *root)
1152 {
1153         struct tree_content *t = root->tree;
1154         unsigned int i, j, del;
1155         unsigned long new_len;
1156         struct last_object lo;
1157         struct object_entry *le;
1158
1159         if (!is_null_sha1(root->versions[1].sha1))
1160                 return;
1161
1162         for (i = 0; i < t->entry_count; i++) {
1163                 if (t->entries[i]->tree)
1164                         store_tree(t->entries[i]);
1165         }
1166
1167         le = find_object(root->versions[0].sha1);
1168         if (!S_ISDIR(root->versions[0].mode)
1169                 || !le
1170                 || le->pack_id != pack_id) {
1171                 lo.data = NULL;
1172                 lo.depth = 0;
1173                 lo.no_free = 0;
1174         } else {
1175                 mktree(t, 0, &lo.len, &old_tree);
1176                 lo.data = old_tree.buffer;
1177                 lo.offset = le->offset;
1178                 lo.depth = t->delta_depth;
1179                 lo.no_free = 1;
1180         }
1181
1182         mktree(t, 1, &new_len, &new_tree);
1183         store_object(OBJ_TREE, new_tree.buffer, new_len,
1184                 &lo, root->versions[1].sha1, 0);
1185
1186         t->delta_depth = lo.depth;
1187         for (i = 0, j = 0, del = 0; i < t->entry_count; i++) {
1188                 struct tree_entry *e = t->entries[i];
1189                 if (e->versions[1].mode) {
1190                         e->versions[0].mode = e->versions[1].mode;
1191                         hashcpy(e->versions[0].sha1, e->versions[1].sha1);
1192                         t->entries[j++] = e;
1193                 } else {
1194                         release_tree_entry(e);
1195                         del++;
1196                 }
1197         }
1198         t->entry_count -= del;
1199 }
1200
1201 static int tree_content_set(
1202         struct tree_entry *root,
1203         const char *p,
1204         const unsigned char *sha1,
1205         const uint16_t mode,
1206         struct tree_content *subtree)
1207 {
1208         struct tree_content *t = root->tree;
1209         const char *slash1;
1210         unsigned int i, n;
1211         struct tree_entry *e;
1212
1213         slash1 = strchr(p, '/');
1214         if (slash1)
1215                 n = slash1 - p;
1216         else
1217                 n = strlen(p);
1218         if (!n)
1219                 die("Empty path component found in input");
1220         if (!slash1 && !S_ISDIR(mode) && subtree)
1221                 die("Non-directories cannot have subtrees");
1222
1223         for (i = 0; i < t->entry_count; i++) {
1224                 e = t->entries[i];
1225                 if (e->name->str_len == n && !strncmp(p, e->name->str_dat, n)) {
1226                         if (!slash1) {
1227                                 if (!S_ISDIR(mode)
1228                                                 && e->versions[1].mode == mode
1229                                                 && !hashcmp(e->versions[1].sha1, sha1))
1230                                         return 0;
1231                                 e->versions[1].mode = mode;
1232                                 hashcpy(e->versions[1].sha1, sha1);
1233                                 if (e->tree)
1234                                         release_tree_content_recursive(e->tree);
1235                                 e->tree = subtree;
1236                                 hashclr(root->versions[1].sha1);
1237                                 return 1;
1238                         }
1239                         if (!S_ISDIR(e->versions[1].mode)) {
1240                                 e->tree = new_tree_content(8);
1241                                 e->versions[1].mode = S_IFDIR;
1242                         }
1243                         if (!e->tree)
1244                                 load_tree(e);
1245                         if (tree_content_set(e, slash1 + 1, sha1, mode, subtree)) {
1246                                 hashclr(root->versions[1].sha1);
1247                                 return 1;
1248                         }
1249                         return 0;
1250                 }
1251         }
1252
1253         if (t->entry_count == t->entry_capacity)
1254                 root->tree = t = grow_tree_content(t, t->entry_count);
1255         e = new_tree_entry();
1256         e->name = to_atom(p, n);
1257         e->versions[0].mode = 0;
1258         hashclr(e->versions[0].sha1);
1259         t->entries[t->entry_count++] = e;
1260         if (slash1) {
1261                 e->tree = new_tree_content(8);
1262                 e->versions[1].mode = S_IFDIR;
1263                 tree_content_set(e, slash1 + 1, sha1, mode, subtree);
1264         } else {
1265                 e->tree = subtree;
1266                 e->versions[1].mode = mode;
1267                 hashcpy(e->versions[1].sha1, sha1);
1268         }
1269         hashclr(root->versions[1].sha1);
1270         return 1;
1271 }
1272
1273 static int tree_content_remove(
1274         struct tree_entry *root,
1275         const char *p,
1276         struct tree_entry *backup_leaf)
1277 {
1278         struct tree_content *t = root->tree;
1279         const char *slash1;
1280         unsigned int i, n;
1281         struct tree_entry *e;
1282
1283         slash1 = strchr(p, '/');
1284         if (slash1)
1285                 n = slash1 - p;
1286         else
1287                 n = strlen(p);
1288
1289         for (i = 0; i < t->entry_count; i++) {
1290                 e = t->entries[i];
1291                 if (e->name->str_len == n && !strncmp(p, e->name->str_dat, n)) {
1292                         if (!slash1 || !S_ISDIR(e->versions[1].mode))
1293                                 goto del_entry;
1294                         if (!e->tree)
1295                                 load_tree(e);
1296                         if (tree_content_remove(e, slash1 + 1, backup_leaf)) {
1297                                 for (n = 0; n < e->tree->entry_count; n++) {
1298                                         if (e->tree->entries[n]->versions[1].mode) {
1299                                                 hashclr(root->versions[1].sha1);
1300                                                 return 1;
1301                                         }
1302                                 }
1303                                 backup_leaf = NULL;
1304                                 goto del_entry;
1305                         }
1306                         return 0;
1307                 }
1308         }
1309         return 0;
1310
1311 del_entry:
1312         if (backup_leaf)
1313                 memcpy(backup_leaf, e, sizeof(*backup_leaf));
1314         else if (e->tree)
1315                 release_tree_content_recursive(e->tree);
1316         e->tree = NULL;
1317         e->versions[1].mode = 0;
1318         hashclr(e->versions[1].sha1);
1319         hashclr(root->versions[1].sha1);
1320         return 1;
1321 }
1322
1323 static int tree_content_get(
1324         struct tree_entry *root,
1325         const char *p,
1326         struct tree_entry *leaf)
1327 {
1328         struct tree_content *t = root->tree;
1329         const char *slash1;
1330         unsigned int i, n;
1331         struct tree_entry *e;
1332
1333         slash1 = strchr(p, '/');
1334         if (slash1)
1335                 n = slash1 - p;
1336         else
1337                 n = strlen(p);
1338
1339         for (i = 0; i < t->entry_count; i++) {
1340                 e = t->entries[i];
1341                 if (e->name->str_len == n && !strncmp(p, e->name->str_dat, n)) {
1342                         if (!slash1) {
1343                                 memcpy(leaf, e, sizeof(*leaf));
1344                                 if (e->tree && is_null_sha1(e->versions[1].sha1))
1345                                         leaf->tree = dup_tree_content(e->tree);
1346                                 else
1347                                         leaf->tree = NULL;
1348                                 return 1;
1349                         }
1350                         if (!S_ISDIR(e->versions[1].mode))
1351                                 return 0;
1352                         if (!e->tree)
1353                                 load_tree(e);
1354                         return tree_content_get(e, slash1 + 1, leaf);
1355                 }
1356         }
1357         return 0;
1358 }
1359
1360 static int update_branch(struct branch *b)
1361 {
1362         static const char *msg = "fast-import";
1363         struct ref_lock *lock;
1364         unsigned char old_sha1[20];
1365
1366         if (read_ref(b->name, old_sha1))
1367                 hashclr(old_sha1);
1368         lock = lock_any_ref_for_update(b->name, old_sha1, 0);
1369         if (!lock)
1370                 return error("Unable to lock %s", b->name);
1371         if (!force_update && !is_null_sha1(old_sha1)) {
1372                 struct commit *old_cmit, *new_cmit;
1373
1374                 old_cmit = lookup_commit_reference_gently(old_sha1, 0);
1375                 new_cmit = lookup_commit_reference_gently(b->sha1, 0);
1376                 if (!old_cmit || !new_cmit) {
1377                         unlock_ref(lock);
1378                         return error("Branch %s is missing commits.", b->name);
1379                 }
1380
1381                 if (!in_merge_bases(old_cmit, &new_cmit, 1)) {
1382                         unlock_ref(lock);
1383                         warning("Not updating %s"
1384                                 " (new tip %s does not contain %s)",
1385                                 b->name, sha1_to_hex(b->sha1), sha1_to_hex(old_sha1));
1386                         return -1;
1387                 }
1388         }
1389         if (write_ref_sha1(lock, b->sha1, msg) < 0)
1390                 return error("Unable to update %s", b->name);
1391         return 0;
1392 }
1393
1394 static void dump_branches(void)
1395 {
1396         unsigned int i;
1397         struct branch *b;
1398
1399         for (i = 0; i < branch_table_sz; i++) {
1400                 for (b = branch_table[i]; b; b = b->table_next_branch)
1401                         failure |= update_branch(b);
1402         }
1403 }
1404
1405 static void dump_tags(void)
1406 {
1407         static const char *msg = "fast-import";
1408         struct tag *t;
1409         struct ref_lock *lock;
1410         char ref_name[PATH_MAX];
1411
1412         for (t = first_tag; t; t = t->next_tag) {
1413                 sprintf(ref_name, "tags/%s", t->name);
1414                 lock = lock_ref_sha1(ref_name, NULL);
1415                 if (!lock || write_ref_sha1(lock, t->sha1, msg) < 0)
1416                         failure |= error("Unable to update %s", ref_name);
1417         }
1418 }
1419
1420 static void dump_marks_helper(FILE *f,
1421         uintmax_t base,
1422         struct mark_set *m)
1423 {
1424         uintmax_t k;
1425         if (m->shift) {
1426                 for (k = 0; k < 1024; k++) {
1427                         if (m->data.sets[k])
1428                                 dump_marks_helper(f, (base + k) << m->shift,
1429                                         m->data.sets[k]);
1430                 }
1431         } else {
1432                 for (k = 0; k < 1024; k++) {
1433                         if (m->data.marked[k])
1434                                 fprintf(f, ":%" PRIuMAX " %s\n", base + k,
1435                                         sha1_to_hex(m->data.marked[k]->sha1));
1436                 }
1437         }
1438 }
1439
1440 static void dump_marks(void)
1441 {
1442         static struct lock_file mark_lock;
1443         int mark_fd;
1444         FILE *f;
1445
1446         if (!mark_file)
1447                 return;
1448
1449         mark_fd = hold_lock_file_for_update(&mark_lock, mark_file, 0);
1450         if (mark_fd < 0) {
1451                 failure |= error("Unable to write marks file %s: %s",
1452                         mark_file, strerror(errno));
1453                 return;
1454         }
1455
1456         f = fdopen(mark_fd, "w");
1457         if (!f) {
1458                 rollback_lock_file(&mark_lock);
1459                 failure |= error("Unable to write marks file %s: %s",
1460                         mark_file, strerror(errno));
1461                 return;
1462         }
1463
1464         dump_marks_helper(f, 0, marks);
1465         fclose(f);
1466         if (commit_lock_file(&mark_lock))
1467                 failure |= error("Unable to write marks file %s: %s",
1468                         mark_file, strerror(errno));
1469 }
1470
1471 static void read_next_command(void)
1472 {
1473         do {
1474                 if (unread_command_buf)
1475                         unread_command_buf = 0;
1476                 else
1477                         read_line(&command_buf, stdin, '\n');
1478         } while (!command_buf.eof && command_buf.buf[0] == '#');
1479 }
1480
1481 static void skip_optional_lf()
1482 {
1483         int term_char = fgetc(stdin);
1484         if (term_char != '\n' && term_char != EOF)
1485                 ungetc(term_char, stdin);
1486 }
1487
1488 static void cmd_mark(void)
1489 {
1490         if (!prefixcmp(command_buf.buf, "mark :")) {
1491                 next_mark = strtoumax(command_buf.buf + 6, NULL, 10);
1492                 read_next_command();
1493         }
1494         else
1495                 next_mark = 0;
1496 }
1497
1498 static void *cmd_data (size_t *size)
1499 {
1500         size_t length;
1501         char *buffer;
1502
1503         if (prefixcmp(command_buf.buf, "data "))
1504                 die("Expected 'data n' command, found: %s", command_buf.buf);
1505
1506         if (!prefixcmp(command_buf.buf + 5, "<<")) {
1507                 char *term = xstrdup(command_buf.buf + 5 + 2);
1508                 size_t sz = 8192, term_len = command_buf.len - 5 - 2;
1509                 length = 0;
1510                 buffer = xmalloc(sz);
1511                 for (;;) {
1512                         read_line(&command_buf, stdin, '\n');
1513                         if (command_buf.eof)
1514                                 die("EOF in data (terminator '%s' not found)", term);
1515                         if (term_len == command_buf.len
1516                                 && !strcmp(term, command_buf.buf))
1517                                 break;
1518                         ALLOC_GROW(buffer, length + command_buf.len, sz);
1519                         memcpy(buffer + length,
1520                                 command_buf.buf,
1521                                 command_buf.len - 1);
1522                         length += command_buf.len - 1;
1523                         buffer[length++] = '\n';
1524                 }
1525                 free(term);
1526         }
1527         else {
1528                 size_t n = 0;
1529                 length = strtoul(command_buf.buf + 5, NULL, 10);
1530                 buffer = xmalloc(length);
1531                 while (n < length) {
1532                         size_t s = fread(buffer + n, 1, length - n, stdin);
1533                         if (!s && feof(stdin))
1534                                 die("EOF in data (%lu bytes remaining)",
1535                                         (unsigned long)(length - n));
1536                         n += s;
1537                 }
1538         }
1539
1540         skip_optional_lf();
1541         *size = length;
1542         return buffer;
1543 }
1544
1545 static int validate_raw_date(const char *src, char *result, int maxlen)
1546 {
1547         const char *orig_src = src;
1548         char *endp, sign;
1549
1550         strtoul(src, &endp, 10);
1551         if (endp == src || *endp != ' ')
1552                 return -1;
1553
1554         src = endp + 1;
1555         if (*src != '-' && *src != '+')
1556                 return -1;
1557         sign = *src;
1558
1559         strtoul(src + 1, &endp, 10);
1560         if (endp == src || *endp || (endp - orig_src) >= maxlen)
1561                 return -1;
1562
1563         strcpy(result, orig_src);
1564         return 0;
1565 }
1566
1567 static char *parse_ident(const char *buf)
1568 {
1569         const char *gt;
1570         size_t name_len;
1571         char *ident;
1572
1573         gt = strrchr(buf, '>');
1574         if (!gt)
1575                 die("Missing > in ident string: %s", buf);
1576         gt++;
1577         if (*gt != ' ')
1578                 die("Missing space after > in ident string: %s", buf);
1579         gt++;
1580         name_len = gt - buf;
1581         ident = xmalloc(name_len + 24);
1582         strncpy(ident, buf, name_len);
1583
1584         switch (whenspec) {
1585         case WHENSPEC_RAW:
1586                 if (validate_raw_date(gt, ident + name_len, 24) < 0)
1587                         die("Invalid raw date \"%s\" in ident: %s", gt, buf);
1588                 break;
1589         case WHENSPEC_RFC2822:
1590                 if (parse_date(gt, ident + name_len, 24) < 0)
1591                         die("Invalid rfc2822 date \"%s\" in ident: %s", gt, buf);
1592                 break;
1593         case WHENSPEC_NOW:
1594                 if (strcmp("now", gt))
1595                         die("Date in ident must be 'now': %s", buf);
1596                 datestamp(ident + name_len, 24);
1597                 break;
1598         }
1599
1600         return ident;
1601 }
1602
1603 static void cmd_new_blob(void)
1604 {
1605         size_t l;
1606         void *d;
1607
1608         read_next_command();
1609         cmd_mark();
1610         d = cmd_data(&l);
1611
1612         if (store_object(OBJ_BLOB, d, l, &last_blob, NULL, next_mark))
1613                 free(d);
1614 }
1615
1616 static void unload_one_branch(void)
1617 {
1618         while (cur_active_branches
1619                 && cur_active_branches >= max_active_branches) {
1620                 uintmax_t min_commit = ULONG_MAX;
1621                 struct branch *e, *l = NULL, *p = NULL;
1622
1623                 for (e = active_branches; e; e = e->active_next_branch) {
1624                         if (e->last_commit < min_commit) {
1625                                 p = l;
1626                                 min_commit = e->last_commit;
1627                         }
1628                         l = e;
1629                 }
1630
1631                 if (p) {
1632                         e = p->active_next_branch;
1633                         p->active_next_branch = e->active_next_branch;
1634                 } else {
1635                         e = active_branches;
1636                         active_branches = e->active_next_branch;
1637                 }
1638                 e->active = 0;
1639                 e->active_next_branch = NULL;
1640                 if (e->branch_tree.tree) {
1641                         release_tree_content_recursive(e->branch_tree.tree);
1642                         e->branch_tree.tree = NULL;
1643                 }
1644                 cur_active_branches--;
1645         }
1646 }
1647
1648 static void load_branch(struct branch *b)
1649 {
1650         load_tree(&b->branch_tree);
1651         if (!b->active) {
1652                 b->active = 1;
1653                 b->active_next_branch = active_branches;
1654                 active_branches = b;
1655                 cur_active_branches++;
1656                 branch_load_count++;
1657         }
1658 }
1659
1660 static void file_change_m(struct branch *b)
1661 {
1662         const char *p = command_buf.buf + 2;
1663         char *p_uq;
1664         const char *endp;
1665         struct object_entry *oe = oe;
1666         unsigned char sha1[20];
1667         uint16_t mode, inline_data = 0;
1668
1669         p = get_mode(p, &mode);
1670         if (!p)
1671                 die("Corrupt mode: %s", command_buf.buf);
1672         switch (mode) {
1673         case S_IFREG | 0644:
1674         case S_IFREG | 0755:
1675         case S_IFLNK:
1676         case 0644:
1677         case 0755:
1678                 /* ok */
1679                 break;
1680         default:
1681                 die("Corrupt mode: %s", command_buf.buf);
1682         }
1683
1684         if (*p == ':') {
1685                 char *x;
1686                 oe = find_mark(strtoumax(p + 1, &x, 10));
1687                 hashcpy(sha1, oe->sha1);
1688                 p = x;
1689         } else if (!prefixcmp(p, "inline")) {
1690                 inline_data = 1;
1691                 p += 6;
1692         } else {
1693                 if (get_sha1_hex(p, sha1))
1694                         die("Invalid SHA1: %s", command_buf.buf);
1695                 oe = find_object(sha1);
1696                 p += 40;
1697         }
1698         if (*p++ != ' ')
1699                 die("Missing space after SHA1: %s", command_buf.buf);
1700
1701         p_uq = unquote_c_style(p, &endp);
1702         if (p_uq) {
1703                 if (*endp)
1704                         die("Garbage after path in: %s", command_buf.buf);
1705                 p = p_uq;
1706         }
1707
1708         if (inline_data) {
1709                 size_t l;
1710                 void *d;
1711                 if (!p_uq)
1712                         p = p_uq = xstrdup(p);
1713                 read_next_command();
1714                 d = cmd_data(&l);
1715                 if (store_object(OBJ_BLOB, d, l, &last_blob, sha1, 0))
1716                         free(d);
1717         } else if (oe) {
1718                 if (oe->type != OBJ_BLOB)
1719                         die("Not a blob (actually a %s): %s",
1720                                 command_buf.buf, typename(oe->type));
1721         } else {
1722                 enum object_type type = sha1_object_info(sha1, NULL);
1723                 if (type < 0)
1724                         die("Blob not found: %s", command_buf.buf);
1725                 if (type != OBJ_BLOB)
1726                         die("Not a blob (actually a %s): %s",
1727                             typename(type), command_buf.buf);
1728         }
1729
1730         tree_content_set(&b->branch_tree, p, sha1, S_IFREG | mode, NULL);
1731         free(p_uq);
1732 }
1733
1734 static void file_change_d(struct branch *b)
1735 {
1736         const char *p = command_buf.buf + 2;
1737         char *p_uq;
1738         const char *endp;
1739
1740         p_uq = unquote_c_style(p, &endp);
1741         if (p_uq) {
1742                 if (*endp)
1743                         die("Garbage after path in: %s", command_buf.buf);
1744                 p = p_uq;
1745         }
1746         tree_content_remove(&b->branch_tree, p, NULL);
1747         free(p_uq);
1748 }
1749
1750 static void file_change_cr(struct branch *b, int rename)
1751 {
1752         const char *s, *d;
1753         char *s_uq, *d_uq;
1754         const char *endp;
1755         struct tree_entry leaf;
1756
1757         s = command_buf.buf + 2;
1758         s_uq = unquote_c_style(s, &endp);
1759         if (s_uq) {
1760                 if (*endp != ' ')
1761                         die("Missing space after source: %s", command_buf.buf);
1762         }
1763         else {
1764                 endp = strchr(s, ' ');
1765                 if (!endp)
1766                         die("Missing space after source: %s", command_buf.buf);
1767                 s_uq = xmalloc(endp - s + 1);
1768                 memcpy(s_uq, s, endp - s);
1769                 s_uq[endp - s] = 0;
1770         }
1771         s = s_uq;
1772
1773         endp++;
1774         if (!*endp)
1775                 die("Missing dest: %s", command_buf.buf);
1776
1777         d = endp;
1778         d_uq = unquote_c_style(d, &endp);
1779         if (d_uq) {
1780                 if (*endp)
1781                         die("Garbage after dest in: %s", command_buf.buf);
1782                 d = d_uq;
1783         }
1784
1785         memset(&leaf, 0, sizeof(leaf));
1786         if (rename)
1787                 tree_content_remove(&b->branch_tree, s, &leaf);
1788         else
1789                 tree_content_get(&b->branch_tree, s, &leaf);
1790         if (!leaf.versions[1].mode)
1791                 die("Path %s not in branch", s);
1792         tree_content_set(&b->branch_tree, d,
1793                 leaf.versions[1].sha1,
1794                 leaf.versions[1].mode,
1795                 leaf.tree);
1796
1797         free(s_uq);
1798         free(d_uq);
1799 }
1800
1801 static void file_change_deleteall(struct branch *b)
1802 {
1803         release_tree_content_recursive(b->branch_tree.tree);
1804         hashclr(b->branch_tree.versions[0].sha1);
1805         hashclr(b->branch_tree.versions[1].sha1);
1806         load_tree(&b->branch_tree);
1807 }
1808
1809 static void cmd_from_commit(struct branch *b, char *buf, unsigned long size)
1810 {
1811         if (!buf || size < 46)
1812                 die("Not a valid commit: %s", sha1_to_hex(b->sha1));
1813         if (memcmp("tree ", buf, 5)
1814                 || get_sha1_hex(buf + 5, b->branch_tree.versions[1].sha1))
1815                 die("The commit %s is corrupt", sha1_to_hex(b->sha1));
1816         hashcpy(b->branch_tree.versions[0].sha1,
1817                 b->branch_tree.versions[1].sha1);
1818 }
1819
1820 static void cmd_from_existing(struct branch *b)
1821 {
1822         if (is_null_sha1(b->sha1)) {
1823                 hashclr(b->branch_tree.versions[0].sha1);
1824                 hashclr(b->branch_tree.versions[1].sha1);
1825         } else {
1826                 unsigned long size;
1827                 char *buf;
1828
1829                 buf = read_object_with_reference(b->sha1,
1830                         commit_type, &size, b->sha1);
1831                 cmd_from_commit(b, buf, size);
1832                 free(buf);
1833         }
1834 }
1835
1836 static int cmd_from(struct branch *b)
1837 {
1838         const char *from;
1839         struct branch *s;
1840
1841         if (prefixcmp(command_buf.buf, "from "))
1842                 return 0;
1843
1844         if (b->branch_tree.tree) {
1845                 release_tree_content_recursive(b->branch_tree.tree);
1846                 b->branch_tree.tree = NULL;
1847         }
1848
1849         from = strchr(command_buf.buf, ' ') + 1;
1850         s = lookup_branch(from);
1851         if (b == s)
1852                 die("Can't create a branch from itself: %s", b->name);
1853         else if (s) {
1854                 unsigned char *t = s->branch_tree.versions[1].sha1;
1855                 hashcpy(b->sha1, s->sha1);
1856                 hashcpy(b->branch_tree.versions[0].sha1, t);
1857                 hashcpy(b->branch_tree.versions[1].sha1, t);
1858         } else if (*from == ':') {
1859                 uintmax_t idnum = strtoumax(from + 1, NULL, 10);
1860                 struct object_entry *oe = find_mark(idnum);
1861                 if (oe->type != OBJ_COMMIT)
1862                         die("Mark :%" PRIuMAX " not a commit", idnum);
1863                 hashcpy(b->sha1, oe->sha1);
1864                 if (oe->pack_id != MAX_PACK_ID) {
1865                         unsigned long size;
1866                         char *buf = gfi_unpack_entry(oe, &size);
1867                         cmd_from_commit(b, buf, size);
1868                         free(buf);
1869                 } else
1870                         cmd_from_existing(b);
1871         } else if (!get_sha1(from, b->sha1))
1872                 cmd_from_existing(b);
1873         else
1874                 die("Invalid ref name or SHA1 expression: %s", from);
1875
1876         read_next_command();
1877         return 1;
1878 }
1879
1880 static struct hash_list *cmd_merge(unsigned int *count)
1881 {
1882         struct hash_list *list = NULL, *n, *e = e;
1883         const char *from;
1884         struct branch *s;
1885
1886         *count = 0;
1887         while (!prefixcmp(command_buf.buf, "merge ")) {
1888                 from = strchr(command_buf.buf, ' ') + 1;
1889                 n = xmalloc(sizeof(*n));
1890                 s = lookup_branch(from);
1891                 if (s)
1892                         hashcpy(n->sha1, s->sha1);
1893                 else if (*from == ':') {
1894                         uintmax_t idnum = strtoumax(from + 1, NULL, 10);
1895                         struct object_entry *oe = find_mark(idnum);
1896                         if (oe->type != OBJ_COMMIT)
1897                                 die("Mark :%" PRIuMAX " not a commit", idnum);
1898                         hashcpy(n->sha1, oe->sha1);
1899                 } else if (!get_sha1(from, n->sha1)) {
1900                         unsigned long size;
1901                         char *buf = read_object_with_reference(n->sha1,
1902                                 commit_type, &size, n->sha1);
1903                         if (!buf || size < 46)
1904                                 die("Not a valid commit: %s", from);
1905                         free(buf);
1906                 } else
1907                         die("Invalid ref name or SHA1 expression: %s", from);
1908
1909                 n->next = NULL;
1910                 if (list)
1911                         e->next = n;
1912                 else
1913                         list = n;
1914                 e = n;
1915                 (*count)++;
1916                 read_next_command();
1917         }
1918         return list;
1919 }
1920
1921 static void cmd_new_commit(void)
1922 {
1923         struct branch *b;
1924         void *msg;
1925         size_t msglen;
1926         char *sp;
1927         char *author = NULL;
1928         char *committer = NULL;
1929         struct hash_list *merge_list = NULL;
1930         unsigned int merge_count;
1931
1932         /* Obtain the branch name from the rest of our command */
1933         sp = strchr(command_buf.buf, ' ') + 1;
1934         b = lookup_branch(sp);
1935         if (!b)
1936                 b = new_branch(sp);
1937
1938         read_next_command();
1939         cmd_mark();
1940         if (!prefixcmp(command_buf.buf, "author ")) {
1941                 author = parse_ident(command_buf.buf + 7);
1942                 read_next_command();
1943         }
1944         if (!prefixcmp(command_buf.buf, "committer ")) {
1945                 committer = parse_ident(command_buf.buf + 10);
1946                 read_next_command();
1947         }
1948         if (!committer)
1949                 die("Expected committer but didn't get one");
1950         msg = cmd_data(&msglen);
1951         read_next_command();
1952         cmd_from(b);
1953         merge_list = cmd_merge(&merge_count);
1954
1955         /* ensure the branch is active/loaded */
1956         if (!b->branch_tree.tree || !max_active_branches) {
1957                 unload_one_branch();
1958                 load_branch(b);
1959         }
1960
1961         /* file_change* */
1962         while (!command_buf.eof && command_buf.len > 1) {
1963                 if (!prefixcmp(command_buf.buf, "M "))
1964                         file_change_m(b);
1965                 else if (!prefixcmp(command_buf.buf, "D "))
1966                         file_change_d(b);
1967                 else if (!prefixcmp(command_buf.buf, "R "))
1968                         file_change_cr(b, 1);
1969                 else if (!prefixcmp(command_buf.buf, "C "))
1970                         file_change_cr(b, 0);
1971                 else if (!strcmp("deleteall", command_buf.buf))
1972                         file_change_deleteall(b);
1973                 else {
1974                         unread_command_buf = 1;
1975                         break;
1976                 }
1977                 read_next_command();
1978         }
1979
1980         /* build the tree and the commit */
1981         store_tree(&b->branch_tree);
1982         hashcpy(b->branch_tree.versions[0].sha1,
1983                 b->branch_tree.versions[1].sha1);
1984         size_dbuf(&new_data, 114 + msglen
1985                 + merge_count * 49
1986                 + (author
1987                         ? strlen(author) + strlen(committer)
1988                         : 2 * strlen(committer)));
1989         sp = new_data.buffer;
1990         sp += sprintf(sp, "tree %s\n",
1991                 sha1_to_hex(b->branch_tree.versions[1].sha1));
1992         if (!is_null_sha1(b->sha1))
1993                 sp += sprintf(sp, "parent %s\n", sha1_to_hex(b->sha1));
1994         while (merge_list) {
1995                 struct hash_list *next = merge_list->next;
1996                 sp += sprintf(sp, "parent %s\n", sha1_to_hex(merge_list->sha1));
1997                 free(merge_list);
1998                 merge_list = next;
1999         }
2000         sp += sprintf(sp, "author %s\n", author ? author : committer);
2001         sp += sprintf(sp, "committer %s\n", committer);
2002         *sp++ = '\n';
2003         memcpy(sp, msg, msglen);
2004         sp += msglen;
2005         free(author);
2006         free(committer);
2007         free(msg);
2008
2009         if (!store_object(OBJ_COMMIT,
2010                 new_data.buffer, sp - (char*)new_data.buffer,
2011                 NULL, b->sha1, next_mark))
2012                 b->pack_id = pack_id;
2013         b->last_commit = object_count_by_type[OBJ_COMMIT];
2014 }
2015
2016 static void cmd_new_tag(void)
2017 {
2018         char *sp;
2019         const char *from;
2020         char *tagger;
2021         struct branch *s;
2022         void *msg;
2023         size_t msglen;
2024         struct tag *t;
2025         uintmax_t from_mark = 0;
2026         unsigned char sha1[20];
2027
2028         /* Obtain the new tag name from the rest of our command */
2029         sp = strchr(command_buf.buf, ' ') + 1;
2030         t = pool_alloc(sizeof(struct tag));
2031         t->next_tag = NULL;
2032         t->name = pool_strdup(sp);
2033         if (last_tag)
2034                 last_tag->next_tag = t;
2035         else
2036                 first_tag = t;
2037         last_tag = t;
2038         read_next_command();
2039
2040         /* from ... */
2041         if (prefixcmp(command_buf.buf, "from "))
2042                 die("Expected from command, got %s", command_buf.buf);
2043         from = strchr(command_buf.buf, ' ') + 1;
2044         s = lookup_branch(from);
2045         if (s) {
2046                 hashcpy(sha1, s->sha1);
2047         } else if (*from == ':') {
2048                 struct object_entry *oe;
2049                 from_mark = strtoumax(from + 1, NULL, 10);
2050                 oe = find_mark(from_mark);
2051                 if (oe->type != OBJ_COMMIT)
2052                         die("Mark :%" PRIuMAX " not a commit", from_mark);
2053                 hashcpy(sha1, oe->sha1);
2054         } else if (!get_sha1(from, sha1)) {
2055                 unsigned long size;
2056                 char *buf;
2057
2058                 buf = read_object_with_reference(sha1,
2059                         commit_type, &size, sha1);
2060                 if (!buf || size < 46)
2061                         die("Not a valid commit: %s", from);
2062                 free(buf);
2063         } else
2064                 die("Invalid ref name or SHA1 expression: %s", from);
2065         read_next_command();
2066
2067         /* tagger ... */
2068         if (prefixcmp(command_buf.buf, "tagger "))
2069                 die("Expected tagger command, got %s", command_buf.buf);
2070         tagger = parse_ident(command_buf.buf + 7);
2071
2072         /* tag payload/message */
2073         read_next_command();
2074         msg = cmd_data(&msglen);
2075
2076         /* build the tag object */
2077         size_dbuf(&new_data, 67+strlen(t->name)+strlen(tagger)+msglen);
2078         sp = new_data.buffer;
2079         sp += sprintf(sp, "object %s\n", sha1_to_hex(sha1));
2080         sp += sprintf(sp, "type %s\n", commit_type);
2081         sp += sprintf(sp, "tag %s\n", t->name);
2082         sp += sprintf(sp, "tagger %s\n", tagger);
2083         *sp++ = '\n';
2084         memcpy(sp, msg, msglen);
2085         sp += msglen;
2086         free(tagger);
2087         free(msg);
2088
2089         if (store_object(OBJ_TAG, new_data.buffer,
2090                 sp - (char*)new_data.buffer,
2091                 NULL, t->sha1, 0))
2092                 t->pack_id = MAX_PACK_ID;
2093         else
2094                 t->pack_id = pack_id;
2095 }
2096
2097 static void cmd_reset_branch(void)
2098 {
2099         struct branch *b;
2100         char *sp;
2101
2102         /* Obtain the branch name from the rest of our command */
2103         sp = strchr(command_buf.buf, ' ') + 1;
2104         b = lookup_branch(sp);
2105         if (b) {
2106                 hashclr(b->sha1);
2107                 hashclr(b->branch_tree.versions[0].sha1);
2108                 hashclr(b->branch_tree.versions[1].sha1);
2109                 if (b->branch_tree.tree) {
2110                         release_tree_content_recursive(b->branch_tree.tree);
2111                         b->branch_tree.tree = NULL;
2112                 }
2113         }
2114         else
2115                 b = new_branch(sp);
2116         read_next_command();
2117         if (!cmd_from(b) && command_buf.len > 1)
2118                 unread_command_buf = 1;
2119 }
2120
2121 static void cmd_checkpoint(void)
2122 {
2123         if (object_count) {
2124                 cycle_packfile();
2125                 dump_branches();
2126                 dump_tags();
2127                 dump_marks();
2128         }
2129         skip_optional_lf();
2130 }
2131
2132 static void cmd_progress(void)
2133 {
2134         fwrite(command_buf.buf, 1, command_buf.len - 1, stdout);
2135         fputc('\n', stdout);
2136         fflush(stdout);
2137         skip_optional_lf();
2138 }
2139
2140 static void import_marks(const char *input_file)
2141 {
2142         char line[512];
2143         FILE *f = fopen(input_file, "r");
2144         if (!f)
2145                 die("cannot read %s: %s", input_file, strerror(errno));
2146         while (fgets(line, sizeof(line), f)) {
2147                 uintmax_t mark;
2148                 char *end;
2149                 unsigned char sha1[20];
2150                 struct object_entry *e;
2151
2152                 end = strchr(line, '\n');
2153                 if (line[0] != ':' || !end)
2154                         die("corrupt mark line: %s", line);
2155                 *end = 0;
2156                 mark = strtoumax(line + 1, &end, 10);
2157                 if (!mark || end == line + 1
2158                         || *end != ' ' || get_sha1(end + 1, sha1))
2159                         die("corrupt mark line: %s", line);
2160                 e = find_object(sha1);
2161                 if (!e) {
2162                         enum object_type type = sha1_object_info(sha1, NULL);
2163                         if (type < 0)
2164                                 die("object not found: %s", sha1_to_hex(sha1));
2165                         e = insert_object(sha1);
2166                         e->type = type;
2167                         e->pack_id = MAX_PACK_ID;
2168                         e->offset = 1; /* just not zero! */
2169                 }
2170                 insert_mark(mark, e);
2171         }
2172         fclose(f);
2173 }
2174
2175 static const char fast_import_usage[] =
2176 "git-fast-import [--date-format=f] [--max-pack-size=n] [--depth=n] [--active-branches=n] [--export-marks=marks.file]";
2177
2178 int main(int argc, const char **argv)
2179 {
2180         int i, show_stats = 1;
2181
2182         git_config(git_default_config);
2183         alloc_objects(object_entry_alloc);
2184         strbuf_init(&command_buf);
2185         atom_table = xcalloc(atom_table_sz, sizeof(struct atom_str*));
2186         branch_table = xcalloc(branch_table_sz, sizeof(struct branch*));
2187         avail_tree_table = xcalloc(avail_tree_table_sz, sizeof(struct avail_tree_content*));
2188         marks = pool_calloc(1, sizeof(struct mark_set));
2189
2190         for (i = 1; i < argc; i++) {
2191                 const char *a = argv[i];
2192
2193                 if (*a != '-' || !strcmp(a, "--"))
2194                         break;
2195                 else if (!prefixcmp(a, "--date-format=")) {
2196                         const char *fmt = a + 14;
2197                         if (!strcmp(fmt, "raw"))
2198                                 whenspec = WHENSPEC_RAW;
2199                         else if (!strcmp(fmt, "rfc2822"))
2200                                 whenspec = WHENSPEC_RFC2822;
2201                         else if (!strcmp(fmt, "now"))
2202                                 whenspec = WHENSPEC_NOW;
2203                         else
2204                                 die("unknown --date-format argument %s", fmt);
2205                 }
2206                 else if (!prefixcmp(a, "--max-pack-size="))
2207                         max_packsize = strtoumax(a + 16, NULL, 0) * 1024 * 1024;
2208                 else if (!prefixcmp(a, "--depth="))
2209                         max_depth = strtoul(a + 8, NULL, 0);
2210                 else if (!prefixcmp(a, "--active-branches="))
2211                         max_active_branches = strtoul(a + 18, NULL, 0);
2212                 else if (!prefixcmp(a, "--import-marks="))
2213                         import_marks(a + 15);
2214                 else if (!prefixcmp(a, "--export-marks="))
2215                         mark_file = a + 15;
2216                 else if (!prefixcmp(a, "--export-pack-edges=")) {
2217                         if (pack_edges)
2218                                 fclose(pack_edges);
2219                         pack_edges = fopen(a + 20, "a");
2220                         if (!pack_edges)
2221                                 die("Cannot open %s: %s", a + 20, strerror(errno));
2222                 } else if (!strcmp(a, "--force"))
2223                         force_update = 1;
2224                 else if (!strcmp(a, "--quiet"))
2225                         show_stats = 0;
2226                 else if (!strcmp(a, "--stats"))
2227                         show_stats = 1;
2228                 else
2229                         die("unknown option %s", a);
2230         }
2231         if (i != argc)
2232                 usage(fast_import_usage);
2233
2234         prepare_packed_git();
2235         start_packfile();
2236         for (;;) {
2237                 read_next_command();
2238                 if (command_buf.eof)
2239                         break;
2240                 else if (!strcmp("blob", command_buf.buf))
2241                         cmd_new_blob();
2242                 else if (!prefixcmp(command_buf.buf, "commit "))
2243                         cmd_new_commit();
2244                 else if (!prefixcmp(command_buf.buf, "tag "))
2245                         cmd_new_tag();
2246                 else if (!prefixcmp(command_buf.buf, "reset "))
2247                         cmd_reset_branch();
2248                 else if (!strcmp("checkpoint", command_buf.buf))
2249                         cmd_checkpoint();
2250                 else if (!prefixcmp(command_buf.buf, "progress "))
2251                         cmd_progress();
2252                 else
2253                         die("Unsupported command: %s", command_buf.buf);
2254         }
2255         end_packfile();
2256
2257         dump_branches();
2258         dump_tags();
2259         unkeep_all_packs();
2260         dump_marks();
2261
2262         if (pack_edges)
2263                 fclose(pack_edges);
2264
2265         if (show_stats) {
2266                 uintmax_t total_count = 0, duplicate_count = 0;
2267                 for (i = 0; i < ARRAY_SIZE(object_count_by_type); i++)
2268                         total_count += object_count_by_type[i];
2269                 for (i = 0; i < ARRAY_SIZE(duplicate_count_by_type); i++)
2270                         duplicate_count += duplicate_count_by_type[i];
2271
2272                 fprintf(stderr, "%s statistics:\n", argv[0]);
2273                 fprintf(stderr, "---------------------------------------------------------------------\n");
2274                 fprintf(stderr, "Alloc'd objects: %10" PRIuMAX "\n", alloc_count);
2275                 fprintf(stderr, "Total objects:   %10" PRIuMAX " (%10" PRIuMAX " duplicates                  )\n", total_count, duplicate_count);
2276                 fprintf(stderr, "      blobs  :   %10" PRIuMAX " (%10" PRIuMAX " duplicates %10" PRIuMAX " deltas)\n", object_count_by_type[OBJ_BLOB], duplicate_count_by_type[OBJ_BLOB], delta_count_by_type[OBJ_BLOB]);
2277                 fprintf(stderr, "      trees  :   %10" PRIuMAX " (%10" PRIuMAX " duplicates %10" PRIuMAX " deltas)\n", object_count_by_type[OBJ_TREE], duplicate_count_by_type[OBJ_TREE], delta_count_by_type[OBJ_TREE]);
2278                 fprintf(stderr, "      commits:   %10" PRIuMAX " (%10" PRIuMAX " duplicates %10" PRIuMAX " deltas)\n", object_count_by_type[OBJ_COMMIT], duplicate_count_by_type[OBJ_COMMIT], delta_count_by_type[OBJ_COMMIT]);
2279                 fprintf(stderr, "      tags   :   %10" PRIuMAX " (%10" PRIuMAX " duplicates %10" PRIuMAX " deltas)\n", object_count_by_type[OBJ_TAG], duplicate_count_by_type[OBJ_TAG], delta_count_by_type[OBJ_TAG]);
2280                 fprintf(stderr, "Total branches:  %10lu (%10lu loads     )\n", branch_count, branch_load_count);
2281                 fprintf(stderr, "      marks:     %10" PRIuMAX " (%10" PRIuMAX " unique    )\n", (((uintmax_t)1) << marks->shift) * 1024, marks_set_count);
2282                 fprintf(stderr, "      atoms:     %10u\n", atom_cnt);
2283                 fprintf(stderr, "Memory total:    %10" PRIuMAX " KiB\n", (total_allocd + alloc_count*sizeof(struct object_entry))/1024);
2284                 fprintf(stderr, "       pools:    %10lu KiB\n", (unsigned long)(total_allocd/1024));
2285                 fprintf(stderr, "     objects:    %10" PRIuMAX " KiB\n", (alloc_count*sizeof(struct object_entry))/1024);
2286                 fprintf(stderr, "---------------------------------------------------------------------\n");
2287                 pack_report();
2288                 fprintf(stderr, "---------------------------------------------------------------------\n");
2289                 fprintf(stderr, "\n");
2290         }
2291
2292         return failure ? 1 : 0;
2293 }