git.oblomov.eu Git - git/blob - 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         ;
  11
  12   new_blob ::= 'blob' lf
  13         mark?
  14     file_content;
  15   file_content ::= data;
  16
  17   new_commit ::= 'commit' sp ref_str lf
  18     mark?
  19     ('author' sp name '<' email '>' ts tz lf)?
  20     'committer' sp name '<' email '>' ts tz lf
  21     commit_msg
  22     ('from' sp (ref_str | hexsha1 | sha1exp_str | idnum) lf)?
  23     file_change*
  24     lf;
  25   commit_msg ::= data;
  26
  27   file_change ::= 'M' sp mode sp (hexsha1 | idnum) sp path_str lf
  28                 | 'D' sp path_str lf
  29                 ;
  30   mode ::= '644' | '755';
  31
  32   new_tag ::= 'tag' sp tag_str lf
  33     'from' sp (ref_str | hexsha1 | sha1exp_str | idnum) lf
  34         'tagger' sp name '<' email '>' ts tz lf
  35     tag_msg;
  36   tag_msg ::= data;
  37
  38   reset_branch ::= 'reset' sp ref_str lf;
  39
  40      # note: the first idnum in a stream should be 1 and subsequent
  41      # idnums should not have gaps between values as this will cause
  42      # the stream parser to reserve space for the gapped values.  An
  43          # idnum can be updated in the future to a new object by issuing
  44      # a new mark directive with the old idnum.
  45          #
  46   mark ::= 'mark' sp idnum lf;
  47
  48      # note: declen indicates the length of binary_data in bytes.
  49      # declen does not include the lf preceeding or trailing the
  50      # binary data.
  51      #
  52   data ::= 'data' sp declen lf
  53     binary_data
  54         lf;
  55
  56      # note: quoted strings are C-style quoting supporting \c for
  57      # common escapes of 'c' (e..g \n, \t, \\, \") or \nnn where nnn
  58          # is the signed byte value in octal.  Note that the only
  59      # characters which must actually be escaped to protect the
  60      # stream formatting is: \, " and LF.  Otherwise these values
  61          # are UTF8.
  62      #
  63   ref_str     ::= ref     | '"' quoted(ref)     '"' ;
  64   sha1exp_str ::= sha1exp | '"' quoted(sha1exp) '"' ;
  65   tag_str     ::= tag     | '"' quoted(tag)     '"' ;
  66   path_str    ::= path    | '"' quoted(path)    '"' ;
  67
  68   declen ::= # unsigned 32 bit value, ascii base10 notation;
  69   binary_data ::= # file content, not interpreted;
  70
  71   sp ::= # ASCII space character;
  72   lf ::= # ASCII newline (LF) character;
  73
  74      # note: a colon (':') must precede the numerical value assigned to
  75          # an idnum.  This is to distinguish it from a ref or tag name as
  76      # GIT does not permit ':' in ref or tag strings.
  77          #
  78   idnum   ::= ':' declen;
  79   path    ::= # GIT style file path, e.g. "a/b/c";
  80   ref     ::= # GIT ref name, e.g. "refs/heads/MOZ_GECKO_EXPERIMENT";
  81   tag     ::= # GIT tag name, e.g. "FIREFOX_1_5";
  82   sha1exp ::= # Any valid GIT SHA1 expression;
  83   hexsha1 ::= # SHA1 in hexadecimal format;
  84
  85      # note: name and email are UTF8 strings, however name must not
  86          # contain '<' or lf and email must not contain any of the
  87      # following: '<', '>', lf.
  88          #
  89   name  ::= # valid GIT author/committer name;
  90   email ::= # valid GIT author/committer email;
  91   ts    ::= # time since the epoch in seconds, ascii base10 notation;
  92   tz    ::= # GIT style timezone;
  93 */
  94
  95 #include "builtin.h"
  96 #include "cache.h"
  97 #include "object.h"
  98 #include "blob.h"
  99 #include "tree.h"
 100 #include "delta.h"
 101 #include "pack.h"
 102 #include "refs.h"
 103 #include "csum-file.h"
 104 #include "strbuf.h"
 105 #include "quote.h"
 106
 107 struct object_entry
 108 {
 109         struct object_entry *next;
 110         enum object_type type;
 111         unsigned long offset;
 112         unsigned char sha1[20];
 113 };
 114
 115 struct object_entry_pool
 116 {
 117         struct object_entry_pool *next_pool;
 118         struct object_entry *next_free;
 119         struct object_entry *end;
 120         struct object_entry entries[FLEX_ARRAY]; /* more */
 121 };
 122
 123 struct mark_set
 124 {
 125         int shift;
 126         union {
 127                 struct object_entry *marked[1024];
 128                 struct mark_set *sets[1024];
 129         } data;
 130 };
 131
 132 struct last_object
 133 {
 134         void *data;
 135         unsigned long len;
 136         unsigned int depth;
 137         int no_free;
 138         unsigned char sha1[20];
 139 };
 140
 141 struct mem_pool
 142 {
 143         struct mem_pool *next_pool;
 144         char *next_free;
 145         char *end;
 146         char space[FLEX_ARRAY]; /* more */
 147 };
 148
 149 struct atom_str
 150 {
 151         struct atom_str *next_atom;
 152         int str_len;
 153         char str_dat[FLEX_ARRAY]; /* more */
 154 };
 155
 156 struct tree_content;
 157 struct tree_entry
 158 {
 159         struct tree_content *tree;
 160         struct atom_str* name;
 161         struct tree_entry_ms
 162         {
 163                 unsigned int mode;
 164                 unsigned char sha1[20];
 165         } versions[2];
 166 };
 167
 168 struct tree_content
 169 {
 170         unsigned int entry_capacity; /* must match avail_tree_content */
 171         unsigned int entry_count;
 172         unsigned int delta_depth;
 173         struct tree_entry *entries[FLEX_ARRAY]; /* more */
 174 };
 175
 176 struct avail_tree_content
 177 {
 178         unsigned int entry_capacity; /* must match tree_content */
 179         struct avail_tree_content *next_avail;
 180 };
 181
 182 struct branch
 183 {
 184         struct branch *table_next_branch;
 185         struct branch *active_next_branch;
 186         const char *name;
 187         unsigned long last_commit;
 188         struct tree_entry branch_tree;
 189         unsigned char sha1[20];
 190 };
 191
 192 struct tag
 193 {
 194         struct tag *next_tag;
 195         const char *name;
 196         unsigned char sha1[20];
 197 };
 198
 199 struct dbuf
 200 {
 201         void *buffer;
 202         size_t capacity;
 203 };
 204
 205
 206 /* Stats and misc. counters */
 207 static unsigned long max_depth = 10;
 208 static unsigned long alloc_count;
 209 static unsigned long branch_count;
 210 static unsigned long branch_load_count;
 211 static unsigned long remap_count;
 212 static unsigned long object_count;
 213 static unsigned long duplicate_count;
 214 static unsigned long marks_set_count;
 215 static unsigned long object_count_by_type[9];
 216 static unsigned long duplicate_count_by_type[9];
 217 static unsigned long delta_count_by_type[9];
 218
 219 /* Memory pools */
 220 static size_t mem_pool_alloc = 2*1024*1024 - sizeof(struct mem_pool);
 221 static size_t total_allocd;
 222 static struct mem_pool *mem_pool;
 223
 224 /* Atom management */
 225 static unsigned int atom_table_sz = 4451;
 226 static unsigned int atom_cnt;
 227 static struct atom_str **atom_table;
 228
 229 /* The .pack file being generated */
 230 static int pack_fd;
 231 static unsigned long pack_size;
 232 static unsigned char pack_sha1[20];
 233 static unsigned char* pack_base;
 234 static unsigned long pack_moff;
 235 static unsigned long pack_mlen = 128*1024*1024;
 236 static unsigned long page_size;
 237
 238 /* Table of objects we've written. */
 239 static unsigned int object_entry_alloc = 5000;
 240 static struct object_entry_pool *blocks;
 241 static struct object_entry *object_table[1 << 16];
 242 static struct mark_set *marks;
 243 static const char* mark_file;
 244
 245 /* Our last blob */
 246 static struct last_object last_blob;
 247
 248 /* Tree management */
 249 static unsigned int tree_entry_alloc = 1000;
 250 static void *avail_tree_entry;
 251 static unsigned int avail_tree_table_sz = 100;
 252 static struct avail_tree_content **avail_tree_table;
 253 static struct dbuf old_tree;
 254 static struct dbuf new_tree;
 255
 256 /* Branch data */
 257 static unsigned long max_active_branches = 5;
 258 static unsigned long cur_active_branches;
 259 static unsigned long branch_table_sz = 1039;
 260 static struct branch **branch_table;
 261 static struct branch *active_branches;
 262
 263 /* Tag data */
 264 static struct tag *first_tag;
 265 static struct tag *last_tag;
 266
 267 /* Input stream parsing */
 268 static struct strbuf command_buf;
 269 static unsigned long next_mark;
 270 static struct dbuf new_data;
 271 static FILE* branch_log;
 272
 273
 274 static void alloc_objects(int cnt)
 275 {
 276         struct object_entry_pool *b;
 277
 278         b = xmalloc(sizeof(struct object_entry_pool)
 279                 + cnt * sizeof(struct object_entry));
 280         b->next_pool = blocks;
 281         b->next_free = b->entries;
 282         b->end = b->entries + cnt;
 283         blocks = b;
 284         alloc_count += cnt;
 285 }
 286
 287 static struct object_entry* new_object(unsigned char *sha1)
 288 {
 289         struct object_entry *e;
 290
 291         if (blocks->next_free == blocks->end)
 292                 alloc_objects(object_entry_alloc);
 293
 294         e = blocks->next_free++;
 295         hashcpy(e->sha1, sha1);
 296         return e;
 297 }
 298
 299 static struct object_entry* find_object(unsigned char *sha1)
 300 {
 301         unsigned int h = sha1[0] << 8 | sha1[1];
 302         struct object_entry *e;
 303         for (e = object_table[h]; e; e = e->next)
 304                 if (!hashcmp(sha1, e->sha1))
 305                         return e;
 306         return NULL;
 307 }
 308
 309 static struct object_entry* insert_object(unsigned char *sha1)
 310 {
 311         unsigned int h = sha1[0] << 8 | sha1[1];
 312         struct object_entry *e = object_table[h];
 313         struct object_entry *p = NULL;
 314
 315         while (e) {
 316                 if (!hashcmp(sha1, e->sha1))
 317                         return e;
 318                 p = e;
 319                 e = e->next;
 320         }
 321
 322         e = new_object(sha1);
 323         e->next = NULL;
 324         e->offset = 0;
 325         if (p)
 326                 p->next = e;
 327         else
 328                 object_table[h] = e;
 329         return e;
 330 }
 331
 332 static unsigned int hc_str(const char *s, size_t len)
 333 {
 334         unsigned int r = 0;
 335         while (len-- > 0)
 336                 r = r * 31 + *s++;
 337         return r;
 338 }
 339
 340 static void* pool_alloc(size_t len)
 341 {
 342         struct mem_pool *p;
 343         void *r;
 344
 345         for (p = mem_pool; p; p = p->next_pool)
 346                 if ((p->end - p->next_free >= len))
 347                         break;
 348
 349         if (!p) {
 350                 if (len >= (mem_pool_alloc/2)) {
 351                         total_allocd += len;
 352                         return xmalloc(len);
 353                 }
 354                 total_allocd += sizeof(struct mem_pool) + mem_pool_alloc;
 355                 p = xmalloc(sizeof(struct mem_pool) + mem_pool_alloc);
 356                 p->next_pool = mem_pool;
 357                 p->next_free = p->space;
 358                 p->end = p->next_free + mem_pool_alloc;
 359                 mem_pool = p;
 360         }
 361
 362         r = p->next_free;
 363         /* round out to a pointer alignment */
 364         if (len & (sizeof(void*) - 1))
 365                 len += sizeof(void*) - (len & (sizeof(void*) - 1));
 366         p->next_free += len;
 367         return r;
 368 }
 369
 370 static void* pool_calloc(size_t count, size_t size)
 371 {
 372         size_t len = count * size;
 373         void *r = pool_alloc(len);
 374         memset(r, 0, len);
 375         return r;
 376 }
 377
 378 static char* pool_strdup(const char *s)
 379 {
 380         char *r = pool_alloc(strlen(s) + 1);
 381         strcpy(r, s);
 382         return r;
 383 }
 384
 385 static void size_dbuf(struct dbuf *b, size_t maxlen)
 386 {
 387         if (b->buffer) {
 388                 if (b->capacity >= maxlen)
 389                         return;
 390                 free(b->buffer);
 391         }
 392         b->capacity = ((maxlen / 1024) + 1) * 1024;
 393         b->buffer = xmalloc(b->capacity);
 394 }
 395
 396 static void insert_mark(unsigned long idnum, struct object_entry *oe)
 397 {
 398         struct mark_set *s = marks;
 399         while ((idnum >> s->shift) >= 1024) {
 400                 s = pool_calloc(1, sizeof(struct mark_set));
 401                 s->shift = marks->shift + 10;
 402                 s->data.sets[0] = marks;
 403                 marks = s;
 404         }
 405         while (s->shift) {
 406                 unsigned long i = idnum >> s->shift;
 407                 idnum -= i << s->shift;
 408                 if (!s->data.sets[i]) {
 409                         s->data.sets[i] = pool_calloc(1, sizeof(struct mark_set));
 410                         s->data.sets[i]->shift = s->shift - 10;
 411                 }
 412                 s = s->data.sets[i];
 413         }
 414         if (!s->data.marked[idnum])
 415                 marks_set_count++;
 416         s->data.marked[idnum] = oe;
 417 }
 418
 419 static struct object_entry* find_mark(unsigned long idnum)
 420 {
 421         unsigned long orig_idnum = idnum;
 422         struct mark_set *s = marks;
 423         struct object_entry *oe = NULL;
 424         if ((idnum >> s->shift) < 1024) {
 425                 while (s && s->shift) {
 426                         unsigned long i = idnum >> s->shift;
 427                         idnum -= i << s->shift;
 428                         s = s->data.sets[i];
 429                 }
 430                 if (s)
 431                         oe = s->data.marked[idnum];
 432         }
 433         if (!oe)
 434                 die("mark :%lu not declared", orig_idnum);
 435         return oe;
 436 }
 437
 438 static struct atom_str* to_atom(const char *s, size_t len)
 439 {
 440         unsigned int hc = hc_str(s, len) % atom_table_sz;
 441         struct atom_str *c;
 442
 443         for (c = atom_table[hc]; c; c = c->next_atom)
 444                 if (c->str_len == len && !strncmp(s, c->str_dat, len))
 445                         return c;
 446
 447         c = pool_alloc(sizeof(struct atom_str) + len + 1);
 448         c->str_len = len;
 449         strncpy(c->str_dat, s, len);
 450         c->str_dat[len] = 0;
 451         c->next_atom = atom_table[hc];
 452         atom_table[hc] = c;
 453         atom_cnt++;
 454         return c;
 455 }
 456
 457 static struct branch* lookup_branch(const char *name)
 458 {
 459         unsigned int hc = hc_str(name, strlen(name)) % branch_table_sz;
 460         struct branch *b;
 461
 462         for (b = branch_table[hc]; b; b = b->table_next_branch)
 463                 if (!strcmp(name, b->name))
 464                         return b;
 465         return NULL;
 466 }
 467
 468 static struct branch* new_branch(const char *name)
 469 {
 470         unsigned int hc = hc_str(name, strlen(name)) % branch_table_sz;
 471         struct branch* b = lookup_branch(name);
 472
 473         if (b)
 474                 die("Invalid attempt to create duplicate branch: %s", name);
 475         if (check_ref_format(name))
 476                 die("Branch name doesn't conform to GIT standards: %s", name);
 477
 478         b = pool_calloc(1, sizeof(struct branch));
 479         b->name = pool_strdup(name);
 480         b->table_next_branch = branch_table[hc];
 481         branch_table[hc] = b;
 482         branch_count++;
 483         return b;
 484 }
 485
 486 static unsigned int hc_entries(unsigned int cnt)
 487 {
 488         cnt = cnt & 7 ? (cnt / 8) + 1 : cnt / 8;
 489         return cnt < avail_tree_table_sz ? cnt : avail_tree_table_sz - 1;
 490 }
 491
 492 static struct tree_content* new_tree_content(unsigned int cnt)
 493 {
 494         struct avail_tree_content *f, *l = NULL;
 495         struct tree_content *t;
 496         unsigned int hc = hc_entries(cnt);
 497
 498         for (f = avail_tree_table[hc]; f; l = f, f = f->next_avail)
 499                 if (f->entry_capacity >= cnt)
 500                         break;
 501
 502         if (f) {
 503                 if (l)
 504                         l->next_avail = f->next_avail;
 505                 else
 506                         avail_tree_table[hc] = f->next_avail;
 507         } else {
 508                 cnt = cnt & 7 ? ((cnt / 8) + 1) * 8 : cnt;
 509                 f = pool_alloc(sizeof(*t) + sizeof(t->entries[0]) * cnt);
 510                 f->entry_capacity = cnt;
 511         }
 512
 513         t = (struct tree_content*)f;
 514         t->entry_count = 0;
 515         t->delta_depth = 0;
 516         return t;
 517 }
 518
 519 static void release_tree_entry(struct tree_entry *e);
 520 static void release_tree_content(struct tree_content *t)
 521 {
 522         struct avail_tree_content *f = (struct avail_tree_content*)t;
 523         unsigned int hc = hc_entries(f->entry_capacity);
 524         f->next_avail = avail_tree_table[hc];
 525         avail_tree_table[hc] = f;
 526 }
 527
 528 static void release_tree_content_recursive(struct tree_content *t)
 529 {
 530         unsigned int i;
 531         for (i = 0; i < t->entry_count; i++)
 532                 release_tree_entry(t->entries[i]);
 533         release_tree_content(t);
 534 }
 535
 536 static struct tree_content* grow_tree_content(
 537         struct tree_content *t,
 538         int amt)
 539 {
 540         struct tree_content *r = new_tree_content(t->entry_count + amt);
 541         r->entry_count = t->entry_count;
 542         r->delta_depth = t->delta_depth;
 543         memcpy(r->entries,t->entries,t->entry_count*sizeof(t->entries[0]));
 544         release_tree_content(t);
 545         return r;
 546 }
 547
 548 static struct tree_entry* new_tree_entry()
 549 {
 550         struct tree_entry *e;
 551
 552         if (!avail_tree_entry) {
 553                 unsigned int n = tree_entry_alloc;
 554                 total_allocd += n * sizeof(struct tree_entry);
 555                 avail_tree_entry = e = xmalloc(n * sizeof(struct tree_entry));
 556                 while (n-- > 1) {
 557                         *((void**)e) = e + 1;
 558                         e++;
 559                 }
 560                 *((void**)e) = NULL;
 561         }
 562
 563         e = avail_tree_entry;
 564         avail_tree_entry = *((void**)e);
 565         return e;
 566 }
 567
 568 static void release_tree_entry(struct tree_entry *e)
 569 {
 570         if (e->tree)
 571                 release_tree_content_recursive(e->tree);
 572         *((void**)e) = avail_tree_entry;
 573         avail_tree_entry = e;
 574 }
 575
 576 static void yread(int fd, void *buffer, size_t length)
 577 {
 578         ssize_t ret = 0;
 579         while (ret < length) {
 580                 ssize_t size = xread(fd, (char *) buffer + ret, length - ret);
 581                 if (!size)
 582                         die("Read from descriptor %i: end of stream", fd);
 583                 if (size < 0)
 584                         die("Read from descriptor %i: %s", fd, strerror(errno));
 585                 ret += size;
 586         }
 587 }
 588
 589 static void ywrite(int fd, void *buffer, size_t length)
 590 {
 591         ssize_t ret = 0;
 592         while (ret < length) {
 593                 ssize_t size = xwrite(fd, (char *) buffer + ret, length - ret);
 594                 if (!size)
 595                         die("Write to descriptor %i: end of file", fd);
 596                 if (size < 0)
 597                         die("Write to descriptor %i: %s", fd, strerror(errno));
 598                 ret += size;
 599         }
 600 }
 601
 602 static size_t encode_header(
 603         enum object_type type,
 604         size_t size,
 605         unsigned char *hdr)
 606 {
 607         int n = 1;
 608         unsigned char c;
 609
 610         if (type < OBJ_COMMIT || type > OBJ_DELTA)
 611                 die("bad type %d", type);
 612
 613         c = (type << 4) | (size & 15);
 614         size >>= 4;
 615         while (size) {
 616                 *hdr++ = c | 0x80;
 617                 c = size & 0x7f;
 618                 size >>= 7;
 619                 n++;
 620         }
 621         *hdr = c;
 622         return n;
 623 }
 624
 625 static int store_object(
 626         enum object_type type,
 627         void *dat,
 628         size_t datlen,
 629         struct last_object *last,
 630         unsigned char *sha1out,
 631         unsigned long mark)
 632 {
 633         void *out, *delta;
 634         struct object_entry *e;
 635         unsigned char hdr[96];
 636         unsigned char sha1[20];
 637         unsigned long hdrlen, deltalen;
 638         SHA_CTX c;
 639         z_stream s;
 640
 641         hdrlen = sprintf((char*)hdr,"%s %lu",type_names[type],datlen) + 1;
 642         SHA1_Init(&c);
 643         SHA1_Update(&c, hdr, hdrlen);
 644         SHA1_Update(&c, dat, datlen);
 645         SHA1_Final(sha1, &c);
 646         if (sha1out)
 647                 hashcpy(sha1out, sha1);
 648
 649         e = insert_object(sha1);
 650         if (mark)
 651                 insert_mark(mark, e);
 652         if (e->offset) {
 653                 duplicate_count++;
 654                 duplicate_count_by_type[type]++;
 655                 return 1;
 656         }
 657         e->type = type;
 658         e->offset = pack_size;
 659         object_count++;
 660         object_count_by_type[type]++;
 661
 662         if (last && last->data && last->depth < max_depth)
 663                 delta = diff_delta(last->data, last->len,
 664                         dat, datlen,
 665                         &deltalen, 0);
 666         else
 667                 delta = 0;
 668
 669         memset(&s, 0, sizeof(s));
 670         deflateInit(&s, zlib_compression_level);
 671
 672         if (delta) {
 673                 delta_count_by_type[type]++;
 674                 last->depth++;
 675                 s.next_in = delta;
 676                 s.avail_in = deltalen;
 677                 hdrlen = encode_header(OBJ_DELTA, deltalen, hdr);
 678                 ywrite(pack_fd, hdr, hdrlen);
 679                 ywrite(pack_fd, last->sha1, sizeof(sha1));
 680                 pack_size += hdrlen + sizeof(sha1);
 681         } else {
 682                 if (last)
 683                         last->depth = 0;
 684                 s.next_in = dat;
 685                 s.avail_in = datlen;
 686                 hdrlen = encode_header(type, datlen, hdr);
 687                 ywrite(pack_fd, hdr, hdrlen);
 688                 pack_size += hdrlen;
 689         }
 690
 691         s.avail_out = deflateBound(&s, s.avail_in);
 692         s.next_out = out = xmalloc(s.avail_out);
 693         while (deflate(&s, Z_FINISH) == Z_OK)
 694                 /* nothing */;
 695         deflateEnd(&s);
 696
 697         ywrite(pack_fd, out, s.total_out);
 698         pack_size += s.total_out;
 699
 700         free(out);
 701         if (delta)
 702                 free(delta);
 703         if (last) {
 704                 if (last->data && !last->no_free)
 705                         free(last->data);
 706                 last->data = dat;
 707                 last->len = datlen;
 708                 hashcpy(last->sha1, sha1);
 709         }
 710         return 0;
 711 }
 712
 713 static unsigned char* map_pack(unsigned long offset, unsigned int *left)
 714 {
 715         if (offset >= pack_size)
 716                 die("object offset outside of pack file");
 717         if (!pack_base
 718                         || offset < pack_moff
 719                         || (offset + 20) >= (pack_moff + pack_mlen)) {
 720                 if (pack_base)
 721                         munmap(pack_base, pack_mlen);
 722                 pack_moff = (offset / page_size) * page_size;
 723                 pack_base = mmap(NULL,pack_mlen,PROT_READ,MAP_SHARED,
 724                         pack_fd,pack_moff);
 725                 if (pack_base == MAP_FAILED)
 726                         die("Failed to map generated pack: %s", strerror(errno));
 727                 remap_count++;
 728         }
 729         offset -= pack_moff;
 730         if (left)
 731                 *left = pack_mlen - offset;
 732         return pack_base + offset;
 733 }
 734
 735 static unsigned long unpack_object_header(unsigned long offset,
 736         enum object_type *type,
 737         unsigned long *sizep)
 738 {
 739         unsigned shift;
 740         unsigned char c;
 741         unsigned long size;
 742
 743         c = *map_pack(offset++, NULL);
 744         *type = (c >> 4) & 7;
 745         size = c & 15;
 746         shift = 4;
 747         while (c & 0x80) {
 748                 c = *map_pack(offset++, NULL);
 749                 size += (c & 0x7f) << shift;
 750                 shift += 7;
 751         }
 752         *sizep = size;
 753         return offset;
 754 }
 755
 756 static void *unpack_non_delta_entry(unsigned long o, unsigned long sz)
 757 {
 758         z_stream stream;
 759         unsigned char *result;
 760
 761         result = xmalloc(sz + 1);
 762         result[sz] = 0;
 763
 764         memset(&stream, 0, sizeof(stream));
 765         stream.next_in = map_pack(o, &stream.avail_in);
 766         stream.next_out = result;
 767         stream.avail_out = sz;
 768
 769         inflateInit(&stream);
 770         for (;;) {
 771                 int st = inflate(&stream, Z_FINISH);
 772                 if (st == Z_STREAM_END)
 773                         break;
 774                 if (st == Z_OK || st == Z_BUF_ERROR) {
 775                         o = stream.next_in - pack_base + pack_moff;
 776                         stream.next_in = map_pack(o, &stream.avail_in);
 777                         continue;
 778                 }
 779                 die("Error %i from zlib during inflate.", st);
 780         }
 781         inflateEnd(&stream);
 782         if (stream.total_out != sz)
 783                 die("Error after inflate: sizes mismatch");
 784         return result;
 785 }
 786
 787 static void *unpack_entry(unsigned long offset,
 788         unsigned long *sizep,
 789         unsigned int *delta_depth);
 790
 791 static void *unpack_delta_entry(unsigned long offset,
 792         unsigned long delta_size,
 793         unsigned long *sizep,
 794         unsigned int *delta_depth)
 795 {
 796         struct object_entry *base_oe;
 797         unsigned char *base_sha1;
 798         void *delta_data, *base, *result;
 799         unsigned long base_size, result_size;
 800
 801         base_sha1 = map_pack(offset, NULL);
 802         base_oe = find_object(base_sha1);
 803         if (!base_oe)
 804                 die("I'm broken; I can't find a base I know must be here.");
 805         base = unpack_entry(base_oe->offset, &base_size, delta_depth);
 806         delta_data = unpack_non_delta_entry(offset + 20, delta_size);
 807         result = patch_delta(base, base_size,
 808                              delta_data, delta_size,
 809                              &result_size);
 810         if (!result)
 811                 die("failed to apply delta");
 812         free(delta_data);
 813         free(base);
 814         *sizep = result_size;
 815         (*delta_depth)++;
 816         return result;
 817 }
 818
 819 static void *unpack_entry(unsigned long offset,
 820         unsigned long *sizep,
 821         unsigned int *delta_depth)
 822 {
 823         unsigned long size;
 824         enum object_type kind;
 825
 826         offset = unpack_object_header(offset, &kind, &size);
 827         switch (kind) {
 828         case OBJ_DELTA:
 829                 return unpack_delta_entry(offset, size, sizep, delta_depth);
 830         case OBJ_COMMIT:
 831         case OBJ_TREE:
 832         case OBJ_BLOB:
 833         case OBJ_TAG:
 834                 *sizep = size;
 835                 *delta_depth = 0;
 836                 return unpack_non_delta_entry(offset, size);
 837         default:
 838                 die("I created an object I can't read!");
 839         }
 840 }
 841
 842 static const char *get_mode(const char *str, unsigned int *modep)
 843 {
 844         unsigned char c;
 845         unsigned int mode = 0;
 846
 847         while ((c = *str++) != ' ') {
 848                 if (c < '0' || c > '7')
 849                         return NULL;
 850                 mode = (mode << 3) + (c - '0');
 851         }
 852         *modep = mode;
 853         return str;
 854 }
 855
 856 static void load_tree(struct tree_entry *root)
 857 {
 858         unsigned char* sha1 = root->versions[1].sha1;
 859         struct object_entry *myoe;
 860         struct tree_content *t;
 861         unsigned long size;
 862         char *buf;
 863         const char *c;
 864
 865         root->tree = t = new_tree_content(8);
 866         if (is_null_sha1(sha1))
 867                 return;
 868
 869         myoe = find_object(sha1);
 870         if (myoe) {
 871                 if (myoe->type != OBJ_TREE)
 872                         die("Not a tree: %s", sha1_to_hex(sha1));
 873                 buf = unpack_entry(myoe->offset, &size, &t->delta_depth);
 874         } else {
 875                 char type[20];
 876                 buf = read_sha1_file(sha1, type, &size);
 877                 if (!buf || strcmp(type, tree_type))
 878                         die("Can't load tree %s", sha1_to_hex(sha1));
 879         }
 880
 881         c = buf;
 882         while (c != (buf + size)) {
 883                 struct tree_entry *e = new_tree_entry();
 884
 885                 if (t->entry_count == t->entry_capacity)
 886                         root->tree = t = grow_tree_content(t, 8);
 887                 t->entries[t->entry_count++] = e;
 888
 889                 e->tree = NULL;
 890                 c = get_mode(c, &e->versions[1].mode);
 891                 if (!c)
 892                         die("Corrupt mode in %s", sha1_to_hex(sha1));
 893                 e->versions[0].mode = e->versions[1].mode;
 894                 e->name = to_atom(c, strlen(c));
 895                 c += e->name->str_len + 1;
 896                 hashcpy(e->versions[0].sha1, (unsigned char*)c);
 897                 hashcpy(e->versions[1].sha1, (unsigned char*)c);
 898                 c += 20;
 899         }
 900         free(buf);
 901 }
 902
 903 static int tecmp0 (const void *_a, const void *_b)
 904 {
 905         struct tree_entry *a = *((struct tree_entry**)_a);
 906         struct tree_entry *b = *((struct tree_entry**)_b);
 907         return base_name_compare(
 908                 a->name->str_dat, a->name->str_len, a->versions[0].mode,
 909                 b->name->str_dat, b->name->str_len, b->versions[0].mode);
 910 }
 911
 912 static int tecmp1 (const void *_a, const void *_b)
 913 {
 914         struct tree_entry *a = *((struct tree_entry**)_a);
 915         struct tree_entry *b = *((struct tree_entry**)_b);
 916         return base_name_compare(
 917                 a->name->str_dat, a->name->str_len, a->versions[1].mode,
 918                 b->name->str_dat, b->name->str_len, b->versions[1].mode);
 919 }
 920
 921 static void mktree(struct tree_content *t,
 922         int v,
 923         unsigned long *szp,
 924         struct dbuf *b)
 925 {
 926         size_t maxlen = 0;
 927         unsigned int i;
 928         char *c;
 929
 930         if (!v)
 931                 qsort(t->entries,t->entry_count,sizeof(t->entries[0]),tecmp0);
 932         else
 933                 qsort(t->entries,t->entry_count,sizeof(t->entries[0]),tecmp1);
 934
 935         for (i = 0; i < t->entry_count; i++) {
 936                 if (t->entries[i]->versions[v].mode)
 937                         maxlen += t->entries[i]->name->str_len + 34;
 938         }
 939
 940         size_dbuf(b, maxlen);
 941         c = b->buffer;
 942         for (i = 0; i < t->entry_count; i++) {
 943                 struct tree_entry *e = t->entries[i];
 944                 if (!e->versions[v].mode)
 945                         continue;
 946                 c += sprintf(c, "%o", e->versions[v].mode);
 947                 *c++ = ' ';
 948                 strcpy(c, e->name->str_dat);
 949                 c += e->name->str_len + 1;
 950                 hashcpy((unsigned char*)c, e->versions[v].sha1);
 951                 c += 20;
 952         }
 953         *szp = c - (char*)b->buffer;
 954 }
 955
 956 static void store_tree(struct tree_entry *root)
 957 {
 958         struct tree_content *t = root->tree;
 959         unsigned int i, j, del;
 960         unsigned long new_len;
 961         struct last_object lo;
 962
 963         if (!is_null_sha1(root->versions[1].sha1))
 964                 return;
 965
 966         for (i = 0; i < t->entry_count; i++) {
 967                 if (t->entries[i]->tree)
 968                         store_tree(t->entries[i]);
 969         }
 970
 971         if (is_null_sha1(root->versions[0].sha1)
 972                         || !find_object(root->versions[0].sha1)) {
 973                 lo.data = NULL;
 974                 lo.depth = 0;
 975         } else {
 976                 mktree(t, 0, &lo.len, &old_tree);
 977                 lo.data = old_tree.buffer;
 978                 lo.depth = t->delta_depth;
 979                 lo.no_free = 1;
 980                 hashcpy(lo.sha1, root->versions[0].sha1);
 981         }
 982         mktree(t, 1, &new_len, &new_tree);
 983
 984         store_object(OBJ_TREE, new_tree.buffer, new_len,
 985                 &lo, root->versions[1].sha1, 0);
 986
 987         t->delta_depth = lo.depth;
 988         hashcpy(root->versions[0].sha1, root->versions[1].sha1);
 989         for (i = 0, j = 0, del = 0; i < t->entry_count; i++) {
 990                 struct tree_entry *e = t->entries[i];
 991                 if (e->versions[1].mode) {
 992                         e->versions[0].mode = e->versions[1].mode;
 993                         hashcpy(e->versions[0].sha1, e->versions[1].sha1);
 994                         t->entries[j++] = e;
 995                 } else {
 996                         release_tree_entry(e);
 997                         del++;
 998                 }
 999         }
1000         t->entry_count -= del;
1001 }
1002
1003 static int tree_content_set(
1004         struct tree_entry *root,
1005         const char *p,
1006         const unsigned char *sha1,
1007         const unsigned int mode)
1008 {
1009         struct tree_content *t = root->tree;
1010         const char *slash1;
1011         unsigned int i, n;
1012         struct tree_entry *e;
1013
1014         slash1 = strchr(p, '/');
1015         if (slash1)
1016                 n = slash1 - p;
1017         else
1018                 n = strlen(p);
1019
1020         for (i = 0; i < t->entry_count; i++) {
1021                 e = t->entries[i];
1022                 if (e->name->str_len == n && !strncmp(p, e->name->str_dat, n)) {
1023                         if (!slash1) {
1024                                 if (e->versions[1].mode == mode
1025                                                 && !hashcmp(e->versions[1].sha1, sha1))
1026                                         return 0;
1027                                 e->versions[1].mode = mode;
1028                                 hashcpy(e->versions[1].sha1, sha1);
1029                                 if (e->tree) {
1030                                         release_tree_content_recursive(e->tree);
1031                                         e->tree = NULL;
1032                                 }
1033                                 hashclr(root->versions[1].sha1);
1034                                 return 1;
1035                         }
1036                         if (!S_ISDIR(e->versions[1].mode)) {
1037                                 e->tree = new_tree_content(8);
1038                                 e->versions[1].mode = S_IFDIR;
1039                         }
1040                         if (!e->tree)
1041                                 load_tree(e);
1042                         if (tree_content_set(e, slash1 + 1, sha1, mode)) {
1043                                 hashclr(root->versions[1].sha1);
1044                                 return 1;
1045                         }
1046                         return 0;
1047                 }
1048         }
1049
1050         if (t->entry_count == t->entry_capacity)
1051                 root->tree = t = grow_tree_content(t, 8);
1052         e = new_tree_entry();
1053         e->name = to_atom(p, n);
1054         e->versions[0].mode = 0;
1055         hashclr(e->versions[0].sha1);
1056         t->entries[t->entry_count++] = e;
1057         if (slash1) {
1058                 e->tree = new_tree_content(8);
1059                 e->versions[1].mode = S_IFDIR;
1060                 tree_content_set(e, slash1 + 1, sha1, mode);
1061         } else {
1062                 e->tree = NULL;
1063                 e->versions[1].mode = mode;
1064                 hashcpy(e->versions[1].sha1, sha1);
1065         }
1066         hashclr(root->versions[1].sha1);
1067         return 1;
1068 }
1069
1070 static int tree_content_remove(struct tree_entry *root, const char *p)
1071 {
1072         struct tree_content *t = root->tree;
1073         const char *slash1;
1074         unsigned int i, n;
1075         struct tree_entry *e;
1076
1077         slash1 = strchr(p, '/');
1078         if (slash1)
1079                 n = slash1 - p;
1080         else
1081                 n = strlen(p);
1082
1083         for (i = 0; i < t->entry_count; i++) {
1084                 e = t->entries[i];
1085                 if (e->name->str_len == n && !strncmp(p, e->name->str_dat, n)) {
1086                         if (!slash1 || !S_ISDIR(e->versions[1].mode))
1087                                 goto del_entry;
1088                         if (!e->tree)
1089                                 load_tree(e);
1090                         if (tree_content_remove(e, slash1 + 1)) {
1091                                 if (!e->tree->entry_count)
1092                                         goto del_entry;
1093                                 hashclr(root->versions[1].sha1);
1094                                 return 1;
1095                         }
1096                         return 0;
1097                 }
1098         }
1099         return 0;
1100
1101 del_entry:
1102         if (e->tree) {
1103                 release_tree_content_recursive(e->tree);
1104                 e->tree = NULL;
1105         }
1106         e->versions[1].mode = 0;
1107         hashclr(e->versions[1].sha1);
1108         hashclr(root->versions[1].sha1);
1109         return 1;
1110 }
1111
1112 static void init_pack_header()
1113 {
1114         struct pack_header hdr;
1115
1116         hdr.hdr_signature = htonl(PACK_SIGNATURE);
1117         hdr.hdr_version = htonl(2);
1118         hdr.hdr_entries = 0;
1119
1120         ywrite(pack_fd, &hdr, sizeof(hdr));
1121         pack_size = sizeof(hdr);
1122 }
1123
1124 static void fixup_header_footer()
1125 {
1126         SHA_CTX c;
1127         char hdr[8];
1128         unsigned long cnt;
1129         char *buf;
1130         size_t n;
1131
1132         if (lseek(pack_fd, 0, SEEK_SET) != 0)
1133                 die("Failed seeking to start: %s", strerror(errno));
1134
1135         SHA1_Init(&c);
1136         yread(pack_fd, hdr, 8);
1137         SHA1_Update(&c, hdr, 8);
1138
1139         cnt = htonl(object_count);
1140         SHA1_Update(&c, &cnt, 4);
1141         ywrite(pack_fd, &cnt, 4);
1142
1143         buf = xmalloc(128 * 1024);
1144         for (;;) {
1145                 n = xread(pack_fd, buf, 128 * 1024);
1146                 if (n <= 0)
1147                         break;
1148                 SHA1_Update(&c, buf, n);
1149         }
1150         free(buf);
1151
1152         SHA1_Final(pack_sha1, &c);
1153         ywrite(pack_fd, pack_sha1, sizeof(pack_sha1));
1154 }
1155
1156 static int oecmp (const void *_a, const void *_b)
1157 {
1158         struct object_entry *a = *((struct object_entry**)_a);
1159         struct object_entry *b = *((struct object_entry**)_b);
1160         return hashcmp(a->sha1, b->sha1);
1161 }
1162
1163 static void write_index(const char *idx_name)
1164 {
1165         struct sha1file *f;
1166         struct object_entry **idx, **c, **last;
1167         struct object_entry *e;
1168         struct object_entry_pool *o;
1169         unsigned int array[256];
1170         int i;
1171
1172         /* Build the sorted table of object IDs. */
1173         idx = xmalloc(object_count * sizeof(struct object_entry*));
1174         c = idx;
1175         for (o = blocks; o; o = o->next_pool)
1176                 for (e = o->entries; e != o->next_free; e++)
1177                         *c++ = e;
1178         last = idx + object_count;
1179         qsort(idx, object_count, sizeof(struct object_entry*), oecmp);
1180
1181         /* Generate the fan-out array. */
1182         c = idx;
1183         for (i = 0; i < 256; i++) {
1184                 struct object_entry **next = c;;
1185                 while (next < last) {
1186                         if ((*next)->sha1[0] != i)
1187                                 break;
1188                         next++;
1189                 }
1190                 array[i] = htonl(next - idx);
1191                 c = next;
1192         }
1193
1194         f = sha1create("%s", idx_name);
1195         sha1write(f, array, 256 * sizeof(int));
1196         for (c = idx; c != last; c++) {
1197                 unsigned int offset = htonl((*c)->offset);
1198                 sha1write(f, &offset, 4);
1199                 sha1write(f, (*c)->sha1, sizeof((*c)->sha1));
1200         }
1201         sha1write(f, pack_sha1, sizeof(pack_sha1));
1202         sha1close(f, NULL, 1);
1203         free(idx);
1204 }
1205
1206 static void dump_branches()
1207 {
1208         static const char *msg = "fast-import";
1209         unsigned int i;
1210         struct branch *b;
1211         struct ref_lock *lock;
1212
1213         for (i = 0; i < branch_table_sz; i++) {
1214                 for (b = branch_table[i]; b; b = b->table_next_branch) {
1215                         lock = lock_any_ref_for_update(b->name, NULL, 0);
1216                         if (!lock || write_ref_sha1(lock, b->sha1, msg) < 0)
1217                                 die("Can't write %s", b->name);
1218                 }
1219         }
1220 }
1221
1222 static void dump_tags()
1223 {
1224         static const char *msg = "fast-import";
1225         struct tag *t;
1226         struct ref_lock *lock;
1227         char path[PATH_MAX];
1228
1229         for (t = first_tag; t; t = t->next_tag) {
1230                 sprintf(path, "refs/tags/%s", t->name);
1231                 lock = lock_any_ref_for_update(path, NULL, 0);
1232                 if (!lock || write_ref_sha1(lock, t->sha1, msg) < 0)
1233                         die("Can't write %s", path);
1234         }
1235 }
1236
1237 static void dump_marks_helper(FILE *f,
1238         unsigned long base,
1239         struct mark_set *m)
1240 {
1241         int k;
1242         if (m->shift) {
1243                 for (k = 0; k < 1024; k++) {
1244                         if (m->data.sets[k])
1245                                 dump_marks_helper(f, (base + k) << m->shift,
1246                                         m->data.sets[k]);
1247                 }
1248         } else {
1249                 for (k = 0; k < 1024; k++) {
1250                         if (m->data.marked[k])
1251                                 fprintf(f, ":%lu %s\n", base + k,
1252                                         sha1_to_hex(m->data.marked[k]->sha1));
1253                 }
1254         }
1255 }
1256
1257 static void dump_marks()
1258 {
1259         if (mark_file)
1260         {
1261                 FILE *f = fopen(mark_file, "w");
1262                 dump_marks_helper(f, 0, marks);
1263                 fclose(f);
1264         }
1265 }
1266
1267 static void read_next_command()
1268 {
1269         read_line(&command_buf, stdin, '\n');
1270 }
1271
1272 static void cmd_mark()
1273 {
1274         if (!strncmp("mark :", command_buf.buf, 6)) {
1275                 next_mark = strtoul(command_buf.buf + 6, NULL, 10);
1276                 read_next_command();
1277         }
1278         else
1279                 next_mark = 0;
1280 }
1281
1282 static void* cmd_data (size_t *size)
1283 {
1284         size_t n = 0;
1285         void *buffer;
1286         size_t length;
1287
1288         if (strncmp("data ", command_buf.buf, 5))
1289                 die("Expected 'data n' command, found: %s", command_buf.buf);
1290
1291         length = strtoul(command_buf.buf + 5, NULL, 10);
1292         buffer = xmalloc(length);
1293
1294         while (n < length) {
1295                 size_t s = fread((char*)buffer + n, 1, length - n, stdin);
1296                 if (!s && feof(stdin))
1297                         die("EOF in data (%lu bytes remaining)", length - n);
1298                 n += s;
1299         }
1300
1301         if (fgetc(stdin) != '\n')
1302                 die("An lf did not trail the binary data as expected.");
1303
1304         *size = length;
1305         return buffer;
1306 }
1307
1308 static void cmd_new_blob()
1309 {
1310         size_t l;
1311         void *d;
1312
1313         read_next_command();
1314         cmd_mark();
1315         d = cmd_data(&l);
1316
1317         if (store_object(OBJ_BLOB, d, l, &last_blob, NULL, next_mark))
1318                 free(d);
1319 }
1320
1321 static void unload_one_branch()
1322 {
1323         while (cur_active_branches
1324                 && cur_active_branches >= max_active_branches) {
1325                 unsigned long min_commit = ULONG_MAX;
1326                 struct branch *e, *l = NULL, *p = NULL;
1327
1328                 for (e = active_branches; e; e = e->active_next_branch) {
1329                         if (e->last_commit < min_commit) {
1330                                 p = l;
1331                                 min_commit = e->last_commit;
1332                         }
1333                         l = e;
1334                 }
1335
1336                 if (p) {
1337                         e = p->active_next_branch;
1338                         p->active_next_branch = e->active_next_branch;
1339                 } else {
1340                         e = active_branches;
1341                         active_branches = e->active_next_branch;
1342                 }
1343                 e->active_next_branch = NULL;
1344                 if (e->branch_tree.tree) {
1345                         release_tree_content_recursive(e->branch_tree.tree);
1346                         e->branch_tree.tree = NULL;
1347                 }
1348                 cur_active_branches--;
1349         }
1350 }
1351
1352 static void load_branch(struct branch *b)
1353 {
1354         load_tree(&b->branch_tree);
1355         b->active_next_branch = active_branches;
1356         active_branches = b;
1357         cur_active_branches++;
1358         branch_load_count++;
1359 }
1360
1361 static void file_change_m(struct branch *b)
1362 {
1363         const char *p = command_buf.buf + 2;
1364         char *p_uq;
1365         const char *endp;
1366         struct object_entry *oe;
1367         unsigned char sha1[20];
1368         unsigned int mode;
1369         char type[20];
1370
1371         p = get_mode(p, &mode);
1372         if (!p)
1373                 die("Corrupt mode: %s", command_buf.buf);
1374         switch (mode) {
1375         case S_IFREG | 0644:
1376         case S_IFREG | 0755:
1377         case S_IFLNK:
1378         case 0644:
1379         case 0755:
1380                 /* ok */
1381                 break;
1382         default:
1383                 die("Corrupt mode: %s", command_buf.buf);
1384         }
1385
1386         if (*p == ':') {
1387                 char *x;
1388                 oe = find_mark(strtoul(p + 1, &x, 10));
1389                 p = x;
1390         } else {
1391                 if (get_sha1_hex(p, sha1))
1392                         die("Invalid SHA1: %s", command_buf.buf);
1393                 oe = find_object(sha1);
1394                 p += 40;
1395         }
1396         if (*p++ != ' ')
1397                 die("Missing space after SHA1: %s", command_buf.buf);
1398
1399         p_uq = unquote_c_style(p, &endp);
1400         if (p_uq) {
1401                 if (*endp)
1402                         die("Garbage after path in: %s", command_buf.buf);
1403                 p = p_uq;
1404         }
1405
1406         if (oe) {
1407                 if (oe->type != OBJ_BLOB)
1408                         die("Not a blob (actually a %s): %s",
1409                                 command_buf.buf, type_names[oe->type]);
1410         } else {
1411                 if (sha1_object_info(sha1, type, NULL))
1412                         die("Blob not found: %s", command_buf.buf);
1413                 if (strcmp(blob_type, type))
1414                         die("Not a blob (actually a %s): %s",
1415                                 command_buf.buf, type);
1416         }
1417
1418         tree_content_set(&b->branch_tree, p, sha1, S_IFREG | mode);
1419
1420         if (p_uq)
1421                 free(p_uq);
1422 }
1423
1424 static void file_change_d(struct branch *b)
1425 {
1426         const char *p = command_buf.buf + 2;
1427         char *p_uq;
1428         const char *endp;
1429
1430         p_uq = unquote_c_style(p, &endp);
1431         if (p_uq) {
1432                 if (*endp)
1433                         die("Garbage after path in: %s", command_buf.buf);
1434                 p = p_uq;
1435         }
1436         tree_content_remove(&b->branch_tree, p);
1437         if (p_uq)
1438                 free(p_uq);
1439 }
1440
1441 static void cmd_from(struct branch *b)
1442 {
1443         const char *from, *endp;
1444         char *str_uq;
1445         struct branch *s;
1446
1447         if (strncmp("from ", command_buf.buf, 5))
1448                 return;
1449
1450         if (b->last_commit)
1451                 die("Can't reinitailize branch %s", b->name);
1452
1453         from = strchr(command_buf.buf, ' ') + 1;
1454         str_uq = unquote_c_style(from, &endp);
1455         if (str_uq) {
1456                 if (*endp)
1457                         die("Garbage after string in: %s", command_buf.buf);
1458                 from = str_uq;
1459         }
1460
1461         s = lookup_branch(from);
1462         if (b == s)
1463                 die("Can't create a branch from itself: %s", b->name);
1464         else if (s) {
1465                 unsigned char *t = s->branch_tree.versions[1].sha1;
1466                 hashcpy(b->sha1, s->sha1);
1467                 hashcpy(b->branch_tree.versions[0].sha1, t);
1468                 hashcpy(b->branch_tree.versions[1].sha1, t);
1469         } else if (*from == ':') {
1470                 unsigned long idnum = strtoul(from + 1, NULL, 10);
1471                 struct object_entry *oe = find_mark(idnum);
1472                 unsigned long size;
1473                 unsigned int depth;
1474                 char *buf;
1475                 if (oe->type != OBJ_COMMIT)
1476                         die("Mark :%lu not a commit", idnum);
1477                 hashcpy(b->sha1, oe->sha1);
1478                 buf = unpack_entry(oe->offset, &size, &depth);
1479                 if (!buf || size < 46)
1480                         die("Not a valid commit: %s", from);
1481                 if (memcmp("tree ", buf, 5)
1482                         || get_sha1_hex(buf + 5, b->branch_tree.versions[1].sha1))
1483                         die("The commit %s is corrupt", sha1_to_hex(b->sha1));
1484                 free(buf);
1485                 hashcpy(b->branch_tree.versions[0].sha1,
1486                         b->branch_tree.versions[1].sha1);
1487         } else if (!get_sha1(from, b->sha1)) {
1488                 if (is_null_sha1(b->sha1)) {
1489                         hashclr(b->branch_tree.versions[0].sha1);
1490                         hashclr(b->branch_tree.versions[1].sha1);
1491                 } else {
1492                         unsigned long size;
1493                         char *buf;
1494
1495                         buf = read_object_with_reference(b->sha1,
1496                                 type_names[OBJ_COMMIT], &size, b->sha1);
1497                         if (!buf || size < 46)
1498                                 die("Not a valid commit: %s", from);
1499                         if (memcmp("tree ", buf, 5)
1500                                 || get_sha1_hex(buf + 5, b->branch_tree.versions[1].sha1))
1501                                 die("The commit %s is corrupt", sha1_to_hex(b->sha1));
1502                         free(buf);
1503                         hashcpy(b->branch_tree.versions[0].sha1,
1504                                 b->branch_tree.versions[1].sha1);
1505                 }
1506         } else
1507                 die("Invalid ref name or SHA1 expression: %s", from);
1508
1509         read_next_command();
1510 }
1511
1512 static void cmd_new_commit()
1513 {
1514         struct branch *b;
1515         void *msg;
1516         size_t msglen;
1517         char *str_uq;
1518         const char *endp;
1519         char *sp;
1520         char *author = NULL;
1521         char *committer = NULL;
1522
1523         /* Obtain the branch name from the rest of our command */
1524         sp = strchr(command_buf.buf, ' ') + 1;
1525         str_uq = unquote_c_style(sp, &endp);
1526         if (str_uq) {
1527                 if (*endp)
1528                         die("Garbage after ref in: %s", command_buf.buf);
1529                 sp = str_uq;
1530         }
1531         b = lookup_branch(sp);
1532         if (!b)
1533                 b = new_branch(sp);
1534         if (str_uq)
1535                 free(str_uq);
1536
1537         read_next_command();
1538         cmd_mark();
1539         if (!strncmp("author ", command_buf.buf, 7)) {
1540                 author = strdup(command_buf.buf);
1541                 read_next_command();
1542         }
1543         if (!strncmp("committer ", command_buf.buf, 10)) {
1544                 committer = strdup(command_buf.buf);
1545                 read_next_command();
1546         }
1547         if (!committer)
1548                 die("Expected committer but didn't get one");
1549         msg = cmd_data(&msglen);
1550         read_next_command();
1551         cmd_from(b);
1552
1553         /* ensure the branch is active/loaded */
1554         if (!b->branch_tree.tree || !max_active_branches) {
1555                 unload_one_branch();
1556                 load_branch(b);
1557         }
1558
1559         /* file_change* */
1560         for (;;) {
1561                 if (1 == command_buf.len)
1562                         break;
1563                 else if (!strncmp("M ", command_buf.buf, 2))
1564                         file_change_m(b);
1565                 else if (!strncmp("D ", command_buf.buf, 2))
1566                         file_change_d(b);
1567                 else
1568                         die("Unsupported file_change: %s", command_buf.buf);
1569                 read_next_command();
1570         }
1571
1572         /* build the tree and the commit */
1573         store_tree(&b->branch_tree);
1574         size_dbuf(&new_data, 97 + msglen
1575                 + (author
1576                         ? strlen(author) + strlen(committer)
1577                         : 2 * strlen(committer)));
1578         sp = new_data.buffer;
1579         sp += sprintf(sp, "tree %s\n",
1580                 sha1_to_hex(b->branch_tree.versions[1].sha1));
1581         if (!is_null_sha1(b->sha1))
1582                 sp += sprintf(sp, "parent %s\n", sha1_to_hex(b->sha1));
1583         if (author)
1584                 sp += sprintf(sp, "%s\n", author);
1585         else
1586                 sp += sprintf(sp, "author %s\n", committer + 10);
1587         sp += sprintf(sp, "%s\n\n", committer);
1588         memcpy(sp, msg, msglen);
1589         sp += msglen;
1590         if (author)
1591                 free(author);
1592         free(committer);
1593         free(msg);
1594
1595         store_object(OBJ_COMMIT,
1596                 new_data.buffer, sp - (char*)new_data.buffer,
1597                 NULL, b->sha1, next_mark);
1598         b->last_commit = object_count_by_type[OBJ_COMMIT];
1599
1600         if (branch_log) {
1601                 int need_dq = quote_c_style(b->name, NULL, NULL, 0);
1602                 fprintf(branch_log, "commit ");
1603                 if (need_dq) {
1604                         fputc('"', branch_log);
1605                         quote_c_style(b->name, NULL, branch_log, 0);
1606                         fputc('"', branch_log);
1607                 } else
1608                         fprintf(branch_log, "%s", b->name);
1609                 fprintf(branch_log," :%lu %s\n",next_mark,sha1_to_hex(b->sha1));
1610         }
1611 }
1612
1613 static void cmd_new_tag()
1614 {
1615         char *str_uq;
1616         const char *endp;
1617         char *sp;
1618         const char *from;
1619         char *tagger;
1620         struct branch *s;
1621         void *msg;
1622         size_t msglen;
1623         struct tag *t;
1624         unsigned long from_mark = 0;
1625         unsigned char sha1[20];
1626
1627         /* Obtain the new tag name from the rest of our command */
1628         sp = strchr(command_buf.buf, ' ') + 1;
1629         str_uq = unquote_c_style(sp, &endp);
1630         if (str_uq) {
1631                 if (*endp)
1632                         die("Garbage after tag name in: %s", command_buf.buf);
1633                 sp = str_uq;
1634         }
1635         t = pool_alloc(sizeof(struct tag));
1636         t->next_tag = NULL;
1637         t->name = pool_strdup(sp);
1638         if (last_tag)
1639                 last_tag->next_tag = t;
1640         else
1641                 first_tag = t;
1642         last_tag = t;
1643         if (str_uq)
1644                 free(str_uq);
1645         read_next_command();
1646
1647         /* from ... */
1648         if (strncmp("from ", command_buf.buf, 5))
1649                 die("Expected from command, got %s", command_buf.buf);
1650
1651         from = strchr(command_buf.buf, ' ') + 1;
1652         str_uq = unquote_c_style(from, &endp);
1653         if (str_uq) {
1654                 if (*endp)
1655                         die("Garbage after string in: %s", command_buf.buf);
1656                 from = str_uq;
1657         }
1658
1659         s = lookup_branch(from);
1660         if (s) {
1661                 hashcpy(sha1, s->sha1);
1662         } else if (*from == ':') {
1663                 from_mark = strtoul(from + 1, NULL, 10);
1664                 struct object_entry *oe = find_mark(from_mark);
1665                 if (oe->type != OBJ_COMMIT)
1666                         die("Mark :%lu not a commit", from_mark);
1667                 hashcpy(sha1, oe->sha1);
1668         } else if (!get_sha1(from, sha1)) {
1669                 unsigned long size;
1670                 char *buf;
1671
1672                 buf = read_object_with_reference(sha1,
1673                         type_names[OBJ_COMMIT], &size, sha1);
1674                 if (!buf || size < 46)
1675                         die("Not a valid commit: %s", from);
1676                 free(buf);
1677         } else
1678                 die("Invalid ref name or SHA1 expression: %s", from);
1679
1680         if (str_uq)
1681                 free(str_uq);
1682         read_next_command();
1683
1684         /* tagger ... */
1685         if (strncmp("tagger ", command_buf.buf, 7))
1686                 die("Expected tagger command, got %s", command_buf.buf);
1687         tagger = strdup(command_buf.buf);
1688
1689         /* tag payload/message */
1690         read_next_command();
1691         msg = cmd_data(&msglen);
1692
1693         /* build the tag object */
1694         size_dbuf(&new_data, 67+strlen(t->name)+strlen(tagger)+msglen);
1695         sp = new_data.buffer;
1696         sp += sprintf(sp, "object %s\n", sha1_to_hex(sha1));
1697         sp += sprintf(sp, "type %s\n", type_names[OBJ_COMMIT]);
1698         sp += sprintf(sp, "tag %s\n", t->name);
1699         sp += sprintf(sp, "%s\n\n", tagger);
1700         memcpy(sp, msg, msglen);
1701         sp += msglen;
1702         free(tagger);
1703         free(msg);
1704
1705         store_object(OBJ_TAG, new_data.buffer, sp - (char*)new_data.buffer,
1706                 NULL, t->sha1, 0);
1707
1708         if (branch_log) {
1709                 int need_dq = quote_c_style(t->name, NULL, NULL, 0);
1710                 fprintf(branch_log, "tag ");
1711                 if (need_dq) {
1712                         fputc('"', branch_log);
1713                         quote_c_style(t->name, NULL, branch_log, 0);
1714                         fputc('"', branch_log);
1715                 } else
1716                         fprintf(branch_log, "%s", t->name);
1717                 fprintf(branch_log," :%lu %s\n",from_mark,sha1_to_hex(t->sha1));
1718         }
1719 }
1720
1721 static void cmd_reset_branch()
1722 {
1723         struct branch *b;
1724         char *str_uq;
1725         const char *endp;
1726         char *sp;
1727
1728         /* Obtain the branch name from the rest of our command */
1729         sp = strchr(command_buf.buf, ' ') + 1;
1730         str_uq = unquote_c_style(sp, &endp);
1731         if (str_uq) {
1732                 if (*endp)
1733                         die("Garbage after ref in: %s", command_buf.buf);
1734                 sp = str_uq;
1735         }
1736         b = lookup_branch(sp);
1737         if (b) {
1738                 b->last_commit = 0;
1739                 if (b->branch_tree.tree) {
1740                         release_tree_content_recursive(b->branch_tree.tree);
1741                         b->branch_tree.tree = NULL;
1742                 }
1743         }
1744         if (str_uq)
1745                 free(str_uq);
1746 }
1747
1748 static const char fast_import_usage[] =
1749 "git-fast-import [--objects=n] [--depth=n] [--active-branches=n] [--export-marks=marks.file] [--branch-log=log] temp.pack";
1750
1751 int main(int argc, const char **argv)
1752 {
1753         const char *base_name;
1754         int i;
1755         unsigned long est_obj_cnt = object_entry_alloc;
1756         char *pack_name;
1757         char *idx_name;
1758         struct stat sb;
1759
1760         setup_ident();
1761         git_config(git_default_config);
1762         page_size = getpagesize();
1763
1764         for (i = 1; i < argc; i++) {
1765                 const char *a = argv[i];
1766
1767                 if (*a != '-' || !strcmp(a, "--"))
1768                         break;
1769                 else if (!strncmp(a, "--objects=", 10))
1770                         est_obj_cnt = strtoul(a + 10, NULL, 0);
1771                 else if (!strncmp(a, "--depth=", 8))
1772                         max_depth = strtoul(a + 8, NULL, 0);
1773                 else if (!strncmp(a, "--active-branches=", 18))
1774                         max_active_branches = strtoul(a + 18, NULL, 0);
1775                 else if (!strncmp(a, "--export-marks=", 15))
1776                         mark_file = a + 15;
1777                 else if (!strncmp(a, "--branch-log=", 13)) {
1778                         branch_log = fopen(a + 13, "w");
1779                         if (!branch_log)
1780                                 die("Can't create %s: %s", a + 13, strerror(errno));
1781                 }
1782                 else
1783                         die("unknown option %s", a);
1784         }
1785         if ((i+1) != argc)
1786                 usage(fast_import_usage);
1787         base_name = argv[i];
1788
1789         pack_name = xmalloc(strlen(base_name) + 6);
1790         sprintf(pack_name, "%s.pack", base_name);
1791         idx_name = xmalloc(strlen(base_name) + 5);
1792         sprintf(idx_name, "%s.idx", base_name);
1793
1794         pack_fd = open(pack_name, O_RDWR|O_CREAT|O_EXCL, 0666);
1795         if (pack_fd < 0)
1796                 die("Can't create %s: %s", pack_name, strerror(errno));
1797
1798         init_pack_header();
1799         alloc_objects(est_obj_cnt);
1800         strbuf_init(&command_buf);
1801
1802         atom_table = xcalloc(atom_table_sz, sizeof(struct atom_str*));
1803         branch_table = xcalloc(branch_table_sz, sizeof(struct branch*));
1804         avail_tree_table = xcalloc(avail_tree_table_sz, sizeof(struct avail_tree_content*));
1805         marks = pool_calloc(1, sizeof(struct mark_set));
1806
1807         for (;;) {
1808                 read_next_command();
1809                 if (command_buf.eof)
1810                         break;
1811                 else if (!strcmp("blob", command_buf.buf))
1812                         cmd_new_blob();
1813                 else if (!strncmp("commit ", command_buf.buf, 7))
1814                         cmd_new_commit();
1815                 else if (!strncmp("tag ", command_buf.buf, 4))
1816                         cmd_new_tag();
1817                 else if (!strncmp("reset ", command_buf.buf, 6))
1818                         cmd_reset_branch();
1819                 else
1820                         die("Unsupported command: %s", command_buf.buf);
1821         }
1822
1823         fixup_header_footer();
1824         close(pack_fd);
1825         write_index(idx_name);
1826         dump_branches();
1827         dump_tags();
1828         dump_marks();
1829         if (branch_log)
1830                 fclose(branch_log);
1831
1832         fprintf(stderr, "%s statistics:\n", argv[0]);
1833         fprintf(stderr, "---------------------------------------------------\n");
1834         fprintf(stderr, "Alloc'd objects: %10lu (%10lu overflow  )\n", alloc_count, alloc_count - est_obj_cnt);
1835         fprintf(stderr, "Total objects:   %10lu (%10lu duplicates)\n", object_count, duplicate_count);
1836         fprintf(stderr, "      blobs  :   %10lu (%10lu duplicates %10lu deltas)\n", object_count_by_type[OBJ_BLOB], duplicate_count_by_type[OBJ_BLOB], delta_count_by_type[OBJ_BLOB]);
1837         fprintf(stderr, "      trees  :   %10lu (%10lu duplicates %10lu deltas)\n", object_count_by_type[OBJ_TREE], duplicate_count_by_type[OBJ_TREE], delta_count_by_type[OBJ_TREE]);
1838         fprintf(stderr, "      commits:   %10lu (%10lu duplicates %10lu deltas)\n", object_count_by_type[OBJ_COMMIT], duplicate_count_by_type[OBJ_COMMIT], delta_count_by_type[OBJ_COMMIT]);
1839         fprintf(stderr, "      tags   :   %10lu (%10lu duplicates %10lu deltas)\n", object_count_by_type[OBJ_TAG], duplicate_count_by_type[OBJ_TAG], delta_count_by_type[OBJ_TAG]);
1840         fprintf(stderr, "Total branches:  %10lu (%10lu loads     )\n", branch_count, branch_load_count);
1841         fprintf(stderr, "      marks:     %10u (%10lu unique    )\n", (1 << marks->shift) * 1024, marks_set_count);
1842         fprintf(stderr, "      atoms:     %10u\n", atom_cnt);
1843         fprintf(stderr, "Memory total:    %10lu KiB\n", (total_allocd + alloc_count*sizeof(struct object_entry))/1024);
1844         fprintf(stderr, "       pools:    %10lu KiB\n", total_allocd/1024);
1845         fprintf(stderr, "     objects:    %10lu KiB\n", (alloc_count*sizeof(struct object_entry))/1024);
1846         fprintf(stderr, "Pack remaps:     %10lu\n", remap_count);
1847         fprintf(stderr, "---------------------------------------------------\n");
1848
1849         stat(pack_name, &sb);
1850         fprintf(stderr, "Pack size:       %10lu KiB\n", (unsigned long)(sb.st_size/1024));
1851         stat(idx_name, &sb);
1852         fprintf(stderr, "Index size:      %10lu KiB\n", (unsigned long)(sb.st_size/1024));
1853
1854         fprintf(stderr, "\n");
1855
1856         return 0;
1857 }