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         b->branch_tree.versions[0].mode = S_IFDIR;
 482         b->branch_tree.versions[1].mode = S_IFDIR;
 483         branch_table[hc] = b;
 484         branch_count++;
 485         return b;
 486 }
 487
 488 static unsigned int hc_entries(unsigned int cnt)
 489 {
 490         cnt = cnt & 7 ? (cnt / 8) + 1 : cnt / 8;
 491         return cnt < avail_tree_table_sz ? cnt : avail_tree_table_sz - 1;
 492 }
 493
 494 static struct tree_content* new_tree_content(unsigned int cnt)
 495 {
 496         struct avail_tree_content *f, *l = NULL;
 497         struct tree_content *t;
 498         unsigned int hc = hc_entries(cnt);
 499
 500         for (f = avail_tree_table[hc]; f; l = f, f = f->next_avail)
 501                 if (f->entry_capacity >= cnt)
 502                         break;
 503
 504         if (f) {
 505                 if (l)
 506                         l->next_avail = f->next_avail;
 507                 else
 508                         avail_tree_table[hc] = f->next_avail;
 509         } else {
 510                 cnt = cnt & 7 ? ((cnt / 8) + 1) * 8 : cnt;
 511                 f = pool_alloc(sizeof(*t) + sizeof(t->entries[0]) * cnt);
 512                 f->entry_capacity = cnt;
 513         }
 514
 515         t = (struct tree_content*)f;
 516         t->entry_count = 0;
 517         t->delta_depth = 0;
 518         return t;
 519 }
 520
 521 static void release_tree_entry(struct tree_entry *e);
 522 static void release_tree_content(struct tree_content *t)
 523 {
 524         struct avail_tree_content *f = (struct avail_tree_content*)t;
 525         unsigned int hc = hc_entries(f->entry_capacity);
 526         f->next_avail = avail_tree_table[hc];
 527         avail_tree_table[hc] = f;
 528 }
 529
 530 static void release_tree_content_recursive(struct tree_content *t)
 531 {
 532         unsigned int i;
 533         for (i = 0; i < t->entry_count; i++)
 534                 release_tree_entry(t->entries[i]);
 535         release_tree_content(t);
 536 }
 537
 538 static struct tree_content* grow_tree_content(
 539         struct tree_content *t,
 540         int amt)
 541 {
 542         struct tree_content *r = new_tree_content(t->entry_count + amt);
 543         r->entry_count = t->entry_count;
 544         r->delta_depth = t->delta_depth;
 545         memcpy(r->entries,t->entries,t->entry_count*sizeof(t->entries[0]));
 546         release_tree_content(t);
 547         return r;
 548 }
 549
 550 static struct tree_entry* new_tree_entry()
 551 {
 552         struct tree_entry *e;
 553
 554         if (!avail_tree_entry) {
 555                 unsigned int n = tree_entry_alloc;
 556                 total_allocd += n * sizeof(struct tree_entry);
 557                 avail_tree_entry = e = xmalloc(n * sizeof(struct tree_entry));
 558                 while (n-- > 1) {
 559                         *((void**)e) = e + 1;
 560                         e++;
 561                 }
 562                 *((void**)e) = NULL;
 563         }
 564
 565         e = avail_tree_entry;
 566         avail_tree_entry = *((void**)e);
 567         return e;
 568 }
 569
 570 static void release_tree_entry(struct tree_entry *e)
 571 {
 572         if (e->tree)
 573                 release_tree_content_recursive(e->tree);
 574         *((void**)e) = avail_tree_entry;
 575         avail_tree_entry = e;
 576 }
 577
 578 static void yread(int fd, void *buffer, size_t length)
 579 {
 580         ssize_t ret = 0;
 581         while (ret < length) {
 582                 ssize_t size = xread(fd, (char *) buffer + ret, length - ret);
 583                 if (!size)
 584                         die("Read from descriptor %i: end of stream", fd);
 585                 if (size < 0)
 586                         die("Read from descriptor %i: %s", fd, strerror(errno));
 587                 ret += size;
 588         }
 589 }
 590
 591 static size_t encode_header(
 592         enum object_type type,
 593         size_t size,
 594         unsigned char *hdr)
 595 {
 596         int n = 1;
 597         unsigned char c;
 598
 599         if (type < OBJ_COMMIT || type > OBJ_DELTA)
 600                 die("bad type %d", type);
 601
 602         c = (type << 4) | (size & 15);
 603         size >>= 4;
 604         while (size) {
 605                 *hdr++ = c | 0x80;
 606                 c = size & 0x7f;
 607                 size >>= 7;
 608                 n++;
 609         }
 610         *hdr = c;
 611         return n;
 612 }
 613
 614 static int store_object(
 615         enum object_type type,
 616         void *dat,
 617         size_t datlen,
 618         struct last_object *last,
 619         unsigned char *sha1out,
 620         unsigned long mark)
 621 {
 622         void *out, *delta;
 623         struct object_entry *e;
 624         unsigned char hdr[96];
 625         unsigned char sha1[20];
 626         unsigned long hdrlen, deltalen;
 627         SHA_CTX c;
 628         z_stream s;
 629
 630         hdrlen = sprintf((char*)hdr,"%s %lu",type_names[type],datlen) + 1;
 631         SHA1_Init(&c);
 632         SHA1_Update(&c, hdr, hdrlen);
 633         SHA1_Update(&c, dat, datlen);
 634         SHA1_Final(sha1, &c);
 635         if (sha1out)
 636                 hashcpy(sha1out, sha1);
 637
 638         e = insert_object(sha1);
 639         if (mark)
 640                 insert_mark(mark, e);
 641         if (e->offset) {
 642                 duplicate_count++;
 643                 duplicate_count_by_type[type]++;
 644                 return 1;
 645         }
 646         e->type = type;
 647         e->offset = pack_size;
 648         object_count++;
 649         object_count_by_type[type]++;
 650
 651         if (last && last->data && last->depth < max_depth)
 652                 delta = diff_delta(last->data, last->len,
 653                         dat, datlen,
 654                         &deltalen, 0);
 655         else
 656                 delta = 0;
 657
 658         memset(&s, 0, sizeof(s));
 659         deflateInit(&s, zlib_compression_level);
 660
 661         if (delta) {
 662                 delta_count_by_type[type]++;
 663                 last->depth++;
 664                 s.next_in = delta;
 665                 s.avail_in = deltalen;
 666                 hdrlen = encode_header(OBJ_DELTA, deltalen, hdr);
 667                 write_or_die(pack_fd, hdr, hdrlen);
 668                 write_or_die(pack_fd, last->sha1, sizeof(sha1));
 669                 pack_size += hdrlen + sizeof(sha1);
 670         } else {
 671                 if (last)
 672                         last->depth = 0;
 673                 s.next_in = dat;
 674                 s.avail_in = datlen;
 675                 hdrlen = encode_header(type, datlen, hdr);
 676                 write_or_die(pack_fd, hdr, hdrlen);
 677                 pack_size += hdrlen;
 678         }
 679
 680         s.avail_out = deflateBound(&s, s.avail_in);
 681         s.next_out = out = xmalloc(s.avail_out);
 682         while (deflate(&s, Z_FINISH) == Z_OK)
 683                 /* nothing */;
 684         deflateEnd(&s);
 685
 686         write_or_die(pack_fd, out, s.total_out);
 687         pack_size += s.total_out;
 688
 689         free(out);
 690         if (delta)
 691                 free(delta);
 692         if (last) {
 693                 if (last->data && !last->no_free)
 694                         free(last->data);
 695                 last->data = dat;
 696                 last->len = datlen;
 697                 hashcpy(last->sha1, sha1);
 698         }
 699         return 0;
 700 }
 701
 702 static unsigned char* map_pack(unsigned long offset, unsigned int *left)
 703 {
 704         if (offset >= pack_size)
 705                 die("object offset outside of pack file");
 706         if (!pack_base
 707                         || offset < pack_moff
 708                         || (offset + 20) >= (pack_moff + pack_mlen)) {
 709                 if (pack_base)
 710                         munmap(pack_base, pack_mlen);
 711                 pack_moff = (offset / page_size) * page_size;
 712                 pack_base = mmap(NULL,pack_mlen,PROT_READ,MAP_SHARED,
 713                         pack_fd,pack_moff);
 714                 if (pack_base == MAP_FAILED)
 715                         die("Failed to map generated pack: %s", strerror(errno));
 716                 remap_count++;
 717         }
 718         offset -= pack_moff;
 719         if (left)
 720                 *left = pack_mlen - offset;
 721         return pack_base + offset;
 722 }
 723
 724 static unsigned long unpack_object_header(unsigned long offset,
 725         enum object_type *type,
 726         unsigned long *sizep)
 727 {
 728         unsigned shift;
 729         unsigned char c;
 730         unsigned long size;
 731
 732         c = *map_pack(offset++, NULL);
 733         *type = (c >> 4) & 7;
 734         size = c & 15;
 735         shift = 4;
 736         while (c & 0x80) {
 737                 c = *map_pack(offset++, NULL);
 738                 size += (c & 0x7f) << shift;
 739                 shift += 7;
 740         }
 741         *sizep = size;
 742         return offset;
 743 }
 744
 745 static void *unpack_non_delta_entry(unsigned long o, unsigned long sz)
 746 {
 747         z_stream stream;
 748         unsigned char *result;
 749
 750         result = xmalloc(sz + 1);
 751         result[sz] = 0;
 752
 753         memset(&stream, 0, sizeof(stream));
 754         stream.next_in = map_pack(o, &stream.avail_in);
 755         stream.next_out = result;
 756         stream.avail_out = sz;
 757
 758         inflateInit(&stream);
 759         for (;;) {
 760                 int st = inflate(&stream, Z_FINISH);
 761                 if (st == Z_STREAM_END)
 762                         break;
 763                 if (st == Z_OK || st == Z_BUF_ERROR) {
 764                         o = stream.next_in - pack_base + pack_moff;
 765                         stream.next_in = map_pack(o, &stream.avail_in);
 766                         continue;
 767                 }
 768                 die("Error %i from zlib during inflate.", st);
 769         }
 770         inflateEnd(&stream);
 771         if (stream.total_out != sz)
 772                 die("Error after inflate: sizes mismatch");
 773         return result;
 774 }
 775
 776 static void *unpack_entry(unsigned long offset,
 777         unsigned long *sizep,
 778         unsigned int *delta_depth);
 779
 780 static void *unpack_delta_entry(unsigned long offset,
 781         unsigned long delta_size,
 782         unsigned long *sizep,
 783         unsigned int *delta_depth)
 784 {
 785         struct object_entry *base_oe;
 786         unsigned char *base_sha1;
 787         void *delta_data, *base, *result;
 788         unsigned long base_size, result_size;
 789
 790         base_sha1 = map_pack(offset, NULL);
 791         base_oe = find_object(base_sha1);
 792         if (!base_oe)
 793                 die("I'm broken; I can't find a base I know must be here.");
 794         base = unpack_entry(base_oe->offset, &base_size, delta_depth);
 795         delta_data = unpack_non_delta_entry(offset + 20, delta_size);
 796         result = patch_delta(base, base_size,
 797                              delta_data, delta_size,
 798                              &result_size);
 799         if (!result)
 800                 die("failed to apply delta");
 801         free(delta_data);
 802         free(base);
 803         *sizep = result_size;
 804         (*delta_depth)++;
 805         return result;
 806 }
 807
 808 static void *unpack_entry(unsigned long offset,
 809         unsigned long *sizep,
 810         unsigned int *delta_depth)
 811 {
 812         unsigned long size;
 813         enum object_type kind;
 814
 815         offset = unpack_object_header(offset, &kind, &size);
 816         switch (kind) {
 817         case OBJ_DELTA:
 818                 return unpack_delta_entry(offset, size, sizep, delta_depth);
 819         case OBJ_COMMIT:
 820         case OBJ_TREE:
 821         case OBJ_BLOB:
 822         case OBJ_TAG:
 823                 *sizep = size;
 824                 *delta_depth = 0;
 825                 return unpack_non_delta_entry(offset, size);
 826         default:
 827                 die("I created an object I can't read!");
 828         }
 829 }
 830
 831 static const char *get_mode(const char *str, unsigned int *modep)
 832 {
 833         unsigned char c;
 834         unsigned int mode = 0;
 835
 836         while ((c = *str++) != ' ') {
 837                 if (c < '0' || c > '7')
 838                         return NULL;
 839                 mode = (mode << 3) + (c - '0');
 840         }
 841         *modep = mode;
 842         return str;
 843 }
 844
 845 static void load_tree(struct tree_entry *root)
 846 {
 847         unsigned char* sha1 = root->versions[1].sha1;
 848         struct object_entry *myoe;
 849         struct tree_content *t;
 850         unsigned long size;
 851         char *buf;
 852         const char *c;
 853
 854         root->tree = t = new_tree_content(8);
 855         if (is_null_sha1(sha1))
 856                 return;
 857
 858         myoe = find_object(sha1);
 859         if (myoe) {
 860                 if (myoe->type != OBJ_TREE)
 861                         die("Not a tree: %s", sha1_to_hex(sha1));
 862                 buf = unpack_entry(myoe->offset, &size, &t->delta_depth);
 863         } else {
 864                 char type[20];
 865                 buf = read_sha1_file(sha1, type, &size);
 866                 if (!buf || strcmp(type, tree_type))
 867                         die("Can't load tree %s", sha1_to_hex(sha1));
 868         }
 869
 870         c = buf;
 871         while (c != (buf + size)) {
 872                 struct tree_entry *e = new_tree_entry();
 873
 874                 if (t->entry_count == t->entry_capacity)
 875                         root->tree = t = grow_tree_content(t, 8);
 876                 t->entries[t->entry_count++] = e;
 877
 878                 e->tree = NULL;
 879                 c = get_mode(c, &e->versions[1].mode);
 880                 if (!c)
 881                         die("Corrupt mode in %s", sha1_to_hex(sha1));
 882                 e->versions[0].mode = e->versions[1].mode;
 883                 e->name = to_atom(c, strlen(c));
 884                 c += e->name->str_len + 1;
 885                 hashcpy(e->versions[0].sha1, (unsigned char*)c);
 886                 hashcpy(e->versions[1].sha1, (unsigned char*)c);
 887                 c += 20;
 888         }
 889         free(buf);
 890 }
 891
 892 static int tecmp0 (const void *_a, const void *_b)
 893 {
 894         struct tree_entry *a = *((struct tree_entry**)_a);
 895         struct tree_entry *b = *((struct tree_entry**)_b);
 896         return base_name_compare(
 897                 a->name->str_dat, a->name->str_len, a->versions[0].mode,
 898                 b->name->str_dat, b->name->str_len, b->versions[0].mode);
 899 }
 900
 901 static int tecmp1 (const void *_a, const void *_b)
 902 {
 903         struct tree_entry *a = *((struct tree_entry**)_a);
 904         struct tree_entry *b = *((struct tree_entry**)_b);
 905         return base_name_compare(
 906                 a->name->str_dat, a->name->str_len, a->versions[1].mode,
 907                 b->name->str_dat, b->name->str_len, b->versions[1].mode);
 908 }
 909
 910 static void mktree(struct tree_content *t,
 911         int v,
 912         unsigned long *szp,
 913         struct dbuf *b)
 914 {
 915         size_t maxlen = 0;
 916         unsigned int i;
 917         char *c;
 918
 919         if (!v)
 920                 qsort(t->entries,t->entry_count,sizeof(t->entries[0]),tecmp0);
 921         else
 922                 qsort(t->entries,t->entry_count,sizeof(t->entries[0]),tecmp1);
 923
 924         for (i = 0; i < t->entry_count; i++) {
 925                 if (t->entries[i]->versions[v].mode)
 926                         maxlen += t->entries[i]->name->str_len + 34;
 927         }
 928
 929         size_dbuf(b, maxlen);
 930         c = b->buffer;
 931         for (i = 0; i < t->entry_count; i++) {
 932                 struct tree_entry *e = t->entries[i];
 933                 if (!e->versions[v].mode)
 934                         continue;
 935                 c += sprintf(c, "%o", e->versions[v].mode);
 936                 *c++ = ' ';
 937                 strcpy(c, e->name->str_dat);
 938                 c += e->name->str_len + 1;
 939                 hashcpy((unsigned char*)c, e->versions[v].sha1);
 940                 c += 20;
 941         }
 942         *szp = c - (char*)b->buffer;
 943 }
 944
 945 static void store_tree(struct tree_entry *root)
 946 {
 947         struct tree_content *t = root->tree;
 948         unsigned int i, j, del;
 949         unsigned long new_len;
 950         struct last_object lo;
 951
 952         if (!is_null_sha1(root->versions[1].sha1))
 953                 return;
 954
 955         for (i = 0; i < t->entry_count; i++) {
 956                 if (t->entries[i]->tree)
 957                         store_tree(t->entries[i]);
 958         }
 959
 960         if (!S_ISDIR(root->versions[0].mode)
 961                         || is_null_sha1(root->versions[0].sha1)
 962                         || !find_object(root->versions[0].sha1)) {
 963                 lo.data = NULL;
 964                 lo.depth = 0;
 965         } else {
 966                 mktree(t, 0, &lo.len, &old_tree);
 967                 lo.data = old_tree.buffer;
 968                 lo.depth = t->delta_depth;
 969                 lo.no_free = 1;
 970                 hashcpy(lo.sha1, root->versions[0].sha1);
 971         }
 972
 973         mktree(t, 1, &new_len, &new_tree);
 974         store_object(OBJ_TREE, new_tree.buffer, new_len,
 975                 &lo, root->versions[1].sha1, 0);
 976
 977         t->delta_depth = lo.depth;
 978         for (i = 0, j = 0, del = 0; i < t->entry_count; i++) {
 979                 struct tree_entry *e = t->entries[i];
 980                 if (e->versions[1].mode) {
 981                         e->versions[0].mode = e->versions[1].mode;
 982                         hashcpy(e->versions[0].sha1, e->versions[1].sha1);
 983                         t->entries[j++] = e;
 984                 } else {
 985                         release_tree_entry(e);
 986                         del++;
 987                 }
 988         }
 989         t->entry_count -= del;
 990 }
 991
 992 static int tree_content_set(
 993         struct tree_entry *root,
 994         const char *p,
 995         const unsigned char *sha1,
 996         const unsigned int mode)
 997 {
 998         struct tree_content *t = root->tree;
 999         const char *slash1;
1000         unsigned int i, n;
1001         struct tree_entry *e;
1002
1003         slash1 = strchr(p, '/');
1004         if (slash1)
1005                 n = slash1 - p;
1006         else
1007                 n = strlen(p);
1008
1009         for (i = 0; i < t->entry_count; i++) {
1010                 e = t->entries[i];
1011                 if (e->name->str_len == n && !strncmp(p, e->name->str_dat, n)) {
1012                         if (!slash1) {
1013                                 if (e->versions[1].mode == mode
1014                                                 && !hashcmp(e->versions[1].sha1, sha1))
1015                                         return 0;
1016                                 e->versions[1].mode = mode;
1017                                 hashcpy(e->versions[1].sha1, sha1);
1018                                 if (e->tree) {
1019                                         release_tree_content_recursive(e->tree);
1020                                         e->tree = NULL;
1021                                 }
1022                                 hashclr(root->versions[1].sha1);
1023                                 return 1;
1024                         }
1025                         if (!S_ISDIR(e->versions[1].mode)) {
1026                                 e->tree = new_tree_content(8);
1027                                 e->versions[1].mode = S_IFDIR;
1028                         }
1029                         if (!e->tree)
1030                                 load_tree(e);
1031                         if (tree_content_set(e, slash1 + 1, sha1, mode)) {
1032                                 hashclr(root->versions[1].sha1);
1033                                 return 1;
1034                         }
1035                         return 0;
1036                 }
1037         }
1038
1039         if (t->entry_count == t->entry_capacity)
1040                 root->tree = t = grow_tree_content(t, 8);
1041         e = new_tree_entry();
1042         e->name = to_atom(p, n);
1043         e->versions[0].mode = 0;
1044         hashclr(e->versions[0].sha1);
1045         t->entries[t->entry_count++] = e;
1046         if (slash1) {
1047                 e->tree = new_tree_content(8);
1048                 e->versions[1].mode = S_IFDIR;
1049                 tree_content_set(e, slash1 + 1, sha1, mode);
1050         } else {
1051                 e->tree = NULL;
1052                 e->versions[1].mode = mode;
1053                 hashcpy(e->versions[1].sha1, sha1);
1054         }
1055         hashclr(root->versions[1].sha1);
1056         return 1;
1057 }
1058
1059 static int tree_content_remove(struct tree_entry *root, const char *p)
1060 {
1061         struct tree_content *t = root->tree;
1062         const char *slash1;
1063         unsigned int i, n;
1064         struct tree_entry *e;
1065
1066         slash1 = strchr(p, '/');
1067         if (slash1)
1068                 n = slash1 - p;
1069         else
1070                 n = strlen(p);
1071
1072         for (i = 0; i < t->entry_count; i++) {
1073                 e = t->entries[i];
1074                 if (e->name->str_len == n && !strncmp(p, e->name->str_dat, n)) {
1075                         if (!slash1 || !S_ISDIR(e->versions[1].mode))
1076                                 goto del_entry;
1077                         if (!e->tree)
1078                                 load_tree(e);
1079                         if (tree_content_remove(e, slash1 + 1)) {
1080                                 for (n = 0; n < e->tree->entry_count; n++) {
1081                                         if (e->tree->entries[n]->versions[1].mode) {
1082                                                 hashclr(root->versions[1].sha1);
1083                                                 return 1;
1084                                         }
1085                                 }
1086                                 goto del_entry;
1087                         }
1088                         return 0;
1089                 }
1090         }
1091         return 0;
1092
1093 del_entry:
1094         if (e->tree) {
1095                 release_tree_content_recursive(e->tree);
1096                 e->tree = NULL;
1097         }
1098         e->versions[1].mode = 0;
1099         hashclr(e->versions[1].sha1);
1100         hashclr(root->versions[1].sha1);
1101         return 1;
1102 }
1103
1104 static void init_pack_header()
1105 {
1106         struct pack_header hdr;
1107
1108         hdr.hdr_signature = htonl(PACK_SIGNATURE);
1109         hdr.hdr_version = htonl(2);
1110         hdr.hdr_entries = 0;
1111
1112         write_or_die(pack_fd, &hdr, sizeof(hdr));
1113         pack_size = sizeof(hdr);
1114 }
1115
1116 static void fixup_header_footer()
1117 {
1118         SHA_CTX c;
1119         char hdr[8];
1120         unsigned long cnt;
1121         char *buf;
1122         size_t n;
1123
1124         if (lseek(pack_fd, 0, SEEK_SET) != 0)
1125                 die("Failed seeking to start: %s", strerror(errno));
1126
1127         SHA1_Init(&c);
1128         yread(pack_fd, hdr, 8);
1129         SHA1_Update(&c, hdr, 8);
1130
1131         cnt = htonl(object_count);
1132         SHA1_Update(&c, &cnt, 4);
1133         write_or_die(pack_fd, &cnt, 4);
1134
1135         buf = xmalloc(128 * 1024);
1136         for (;;) {
1137                 n = xread(pack_fd, buf, 128 * 1024);
1138                 if (n <= 0)
1139                         break;
1140                 SHA1_Update(&c, buf, n);
1141         }
1142         free(buf);
1143
1144         SHA1_Final(pack_sha1, &c);
1145         write_or_die(pack_fd, pack_sha1, sizeof(pack_sha1));
1146 }
1147
1148 static int oecmp (const void *_a, const void *_b)
1149 {
1150         struct object_entry *a = *((struct object_entry**)_a);
1151         struct object_entry *b = *((struct object_entry**)_b);
1152         return hashcmp(a->sha1, b->sha1);
1153 }
1154
1155 static void write_index(const char *idx_name)
1156 {
1157         struct sha1file *f;
1158         struct object_entry **idx, **c, **last;
1159         struct object_entry *e;
1160         struct object_entry_pool *o;
1161         unsigned int array[256];
1162         int i;
1163
1164         /* Build the sorted table of object IDs. */
1165         idx = xmalloc(object_count * sizeof(struct object_entry*));
1166         c = idx;
1167         for (o = blocks; o; o = o->next_pool)
1168                 for (e = o->entries; e != o->next_free; e++)
1169                         *c++ = e;
1170         last = idx + object_count;
1171         qsort(idx, object_count, sizeof(struct object_entry*), oecmp);
1172
1173         /* Generate the fan-out array. */
1174         c = idx;
1175         for (i = 0; i < 256; i++) {
1176                 struct object_entry **next = c;;
1177                 while (next < last) {
1178                         if ((*next)->sha1[0] != i)
1179                                 break;
1180                         next++;
1181                 }
1182                 array[i] = htonl(next - idx);
1183                 c = next;
1184         }
1185
1186         f = sha1create("%s", idx_name);
1187         sha1write(f, array, 256 * sizeof(int));
1188         for (c = idx; c != last; c++) {
1189                 unsigned int offset = htonl((*c)->offset);
1190                 sha1write(f, &offset, 4);
1191                 sha1write(f, (*c)->sha1, sizeof((*c)->sha1));
1192         }
1193         sha1write(f, pack_sha1, sizeof(pack_sha1));
1194         sha1close(f, NULL, 1);
1195         free(idx);
1196 }
1197
1198 static void dump_branches()
1199 {
1200         static const char *msg = "fast-import";
1201         unsigned int i;
1202         struct branch *b;
1203         struct ref_lock *lock;
1204
1205         for (i = 0; i < branch_table_sz; i++) {
1206                 for (b = branch_table[i]; b; b = b->table_next_branch) {
1207                         lock = lock_any_ref_for_update(b->name, NULL, 0);
1208                         if (!lock || write_ref_sha1(lock, b->sha1, msg) < 0)
1209                                 die("Can't write %s", b->name);
1210                 }
1211         }
1212 }
1213
1214 static void dump_tags()
1215 {
1216         static const char *msg = "fast-import";
1217         struct tag *t;
1218         struct ref_lock *lock;
1219         char path[PATH_MAX];
1220
1221         for (t = first_tag; t; t = t->next_tag) {
1222                 sprintf(path, "refs/tags/%s", t->name);
1223                 lock = lock_any_ref_for_update(path, NULL, 0);
1224                 if (!lock || write_ref_sha1(lock, t->sha1, msg) < 0)
1225                         die("Can't write %s", path);
1226         }
1227 }
1228
1229 static void dump_marks_helper(FILE *f,
1230         unsigned long base,
1231         struct mark_set *m)
1232 {
1233         int k;
1234         if (m->shift) {
1235                 for (k = 0; k < 1024; k++) {
1236                         if (m->data.sets[k])
1237                                 dump_marks_helper(f, (base + k) << m->shift,
1238                                         m->data.sets[k]);
1239                 }
1240         } else {
1241                 for (k = 0; k < 1024; k++) {
1242                         if (m->data.marked[k])
1243                                 fprintf(f, ":%lu %s\n", base + k,
1244                                         sha1_to_hex(m->data.marked[k]->sha1));
1245                 }
1246         }
1247 }
1248
1249 static void dump_marks()
1250 {
1251         if (mark_file)
1252         {
1253                 FILE *f = fopen(mark_file, "w");
1254                 dump_marks_helper(f, 0, marks);
1255                 fclose(f);
1256         }
1257 }
1258
1259 static void read_next_command()
1260 {
1261         read_line(&command_buf, stdin, '\n');
1262 }
1263
1264 static void cmd_mark()
1265 {
1266         if (!strncmp("mark :", command_buf.buf, 6)) {
1267                 next_mark = strtoul(command_buf.buf + 6, NULL, 10);
1268                 read_next_command();
1269         }
1270         else
1271                 next_mark = 0;
1272 }
1273
1274 static void* cmd_data (size_t *size)
1275 {
1276         size_t n = 0;
1277         void *buffer;
1278         size_t length;
1279
1280         if (strncmp("data ", command_buf.buf, 5))
1281                 die("Expected 'data n' command, found: %s", command_buf.buf);
1282
1283         length = strtoul(command_buf.buf + 5, NULL, 10);
1284         buffer = xmalloc(length);
1285
1286         while (n < length) {
1287                 size_t s = fread((char*)buffer + n, 1, length - n, stdin);
1288                 if (!s && feof(stdin))
1289                         die("EOF in data (%lu bytes remaining)", length - n);
1290                 n += s;
1291         }
1292
1293         if (fgetc(stdin) != '\n')
1294                 die("An lf did not trail the binary data as expected.");
1295
1296         *size = length;
1297         return buffer;
1298 }
1299
1300 static void cmd_new_blob()
1301 {
1302         size_t l;
1303         void *d;
1304
1305         read_next_command();
1306         cmd_mark();
1307         d = cmd_data(&l);
1308
1309         if (store_object(OBJ_BLOB, d, l, &last_blob, NULL, next_mark))
1310                 free(d);
1311 }
1312
1313 static void unload_one_branch()
1314 {
1315         while (cur_active_branches
1316                 && cur_active_branches >= max_active_branches) {
1317                 unsigned long min_commit = ULONG_MAX;
1318                 struct branch *e, *l = NULL, *p = NULL;
1319
1320                 for (e = active_branches; e; e = e->active_next_branch) {
1321                         if (e->last_commit < min_commit) {
1322                                 p = l;
1323                                 min_commit = e->last_commit;
1324                         }
1325                         l = e;
1326                 }
1327
1328                 if (p) {
1329                         e = p->active_next_branch;
1330                         p->active_next_branch = e->active_next_branch;
1331                 } else {
1332                         e = active_branches;
1333                         active_branches = e->active_next_branch;
1334                 }
1335                 e->active_next_branch = NULL;
1336                 if (e->branch_tree.tree) {
1337                         release_tree_content_recursive(e->branch_tree.tree);
1338                         e->branch_tree.tree = NULL;
1339                 }
1340                 cur_active_branches--;
1341         }
1342 }
1343
1344 static void load_branch(struct branch *b)
1345 {
1346         load_tree(&b->branch_tree);
1347         b->active_next_branch = active_branches;
1348         active_branches = b;
1349         cur_active_branches++;
1350         branch_load_count++;
1351 }
1352
1353 static void file_change_m(struct branch *b)
1354 {
1355         const char *p = command_buf.buf + 2;
1356         char *p_uq;
1357         const char *endp;
1358         struct object_entry *oe;
1359         unsigned char sha1[20];
1360         unsigned int mode;
1361         char type[20];
1362
1363         p = get_mode(p, &mode);
1364         if (!p)
1365                 die("Corrupt mode: %s", command_buf.buf);
1366         switch (mode) {
1367         case S_IFREG | 0644:
1368         case S_IFREG | 0755:
1369         case S_IFLNK:
1370         case 0644:
1371         case 0755:
1372                 /* ok */
1373                 break;
1374         default:
1375                 die("Corrupt mode: %s", command_buf.buf);
1376         }
1377
1378         if (*p == ':') {
1379                 char *x;
1380                 oe = find_mark(strtoul(p + 1, &x, 10));
1381                 p = x;
1382         } else {
1383                 if (get_sha1_hex(p, sha1))
1384                         die("Invalid SHA1: %s", command_buf.buf);
1385                 oe = find_object(sha1);
1386                 p += 40;
1387         }
1388         if (*p++ != ' ')
1389                 die("Missing space after SHA1: %s", command_buf.buf);
1390
1391         p_uq = unquote_c_style(p, &endp);
1392         if (p_uq) {
1393                 if (*endp)
1394                         die("Garbage after path in: %s", command_buf.buf);
1395                 p = p_uq;
1396         }
1397
1398         if (oe) {
1399                 if (oe->type != OBJ_BLOB)
1400                         die("Not a blob (actually a %s): %s",
1401                                 command_buf.buf, type_names[oe->type]);
1402         } else {
1403                 if (sha1_object_info(sha1, type, NULL))
1404                         die("Blob not found: %s", command_buf.buf);
1405                 if (strcmp(blob_type, type))
1406                         die("Not a blob (actually a %s): %s",
1407                                 command_buf.buf, type);
1408         }
1409
1410         tree_content_set(&b->branch_tree, p, sha1, S_IFREG | mode);
1411
1412         if (p_uq)
1413                 free(p_uq);
1414 }
1415
1416 static void file_change_d(struct branch *b)
1417 {
1418         const char *p = command_buf.buf + 2;
1419         char *p_uq;
1420         const char *endp;
1421
1422         p_uq = unquote_c_style(p, &endp);
1423         if (p_uq) {
1424                 if (*endp)
1425                         die("Garbage after path in: %s", command_buf.buf);
1426                 p = p_uq;
1427         }
1428         tree_content_remove(&b->branch_tree, p);
1429         if (p_uq)
1430                 free(p_uq);
1431 }
1432
1433 static void cmd_from(struct branch *b)
1434 {
1435         const char *from, *endp;
1436         char *str_uq;
1437         struct branch *s;
1438
1439         if (strncmp("from ", command_buf.buf, 5))
1440                 return;
1441
1442         if (b->last_commit)
1443                 die("Can't reinitailize branch %s", b->name);
1444
1445         from = strchr(command_buf.buf, ' ') + 1;
1446         str_uq = unquote_c_style(from, &endp);
1447         if (str_uq) {
1448                 if (*endp)
1449                         die("Garbage after string in: %s", command_buf.buf);
1450                 from = str_uq;
1451         }
1452
1453         s = lookup_branch(from);
1454         if (b == s)
1455                 die("Can't create a branch from itself: %s", b->name);
1456         else if (s) {
1457                 unsigned char *t = s->branch_tree.versions[1].sha1;
1458                 hashcpy(b->sha1, s->sha1);
1459                 hashcpy(b->branch_tree.versions[0].sha1, t);
1460                 hashcpy(b->branch_tree.versions[1].sha1, t);
1461         } else if (*from == ':') {
1462                 unsigned long idnum = strtoul(from + 1, NULL, 10);
1463                 struct object_entry *oe = find_mark(idnum);
1464                 unsigned long size;
1465                 unsigned int depth;
1466                 char *buf;
1467                 if (oe->type != OBJ_COMMIT)
1468                         die("Mark :%lu not a commit", idnum);
1469                 hashcpy(b->sha1, oe->sha1);
1470                 buf = unpack_entry(oe->offset, &size, &depth);
1471                 if (!buf || size < 46)
1472                         die("Not a valid commit: %s", from);
1473                 if (memcmp("tree ", buf, 5)
1474                         || get_sha1_hex(buf + 5, b->branch_tree.versions[1].sha1))
1475                         die("The commit %s is corrupt", sha1_to_hex(b->sha1));
1476                 free(buf);
1477                 hashcpy(b->branch_tree.versions[0].sha1,
1478                         b->branch_tree.versions[1].sha1);
1479         } else if (!get_sha1(from, b->sha1)) {
1480                 if (is_null_sha1(b->sha1)) {
1481                         hashclr(b->branch_tree.versions[0].sha1);
1482                         hashclr(b->branch_tree.versions[1].sha1);
1483                 } else {
1484                         unsigned long size;
1485                         char *buf;
1486
1487                         buf = read_object_with_reference(b->sha1,
1488                                 type_names[OBJ_COMMIT], &size, b->sha1);
1489                         if (!buf || size < 46)
1490                                 die("Not a valid commit: %s", from);
1491                         if (memcmp("tree ", buf, 5)
1492                                 || get_sha1_hex(buf + 5, b->branch_tree.versions[1].sha1))
1493                                 die("The commit %s is corrupt", sha1_to_hex(b->sha1));
1494                         free(buf);
1495                         hashcpy(b->branch_tree.versions[0].sha1,
1496                                 b->branch_tree.versions[1].sha1);
1497                 }
1498         } else
1499                 die("Invalid ref name or SHA1 expression: %s", from);
1500
1501         read_next_command();
1502 }
1503
1504 static void cmd_new_commit()
1505 {
1506         struct branch *b;
1507         void *msg;
1508         size_t msglen;
1509         char *str_uq;
1510         const char *endp;
1511         char *sp;
1512         char *author = NULL;
1513         char *committer = NULL;
1514
1515         /* Obtain the branch name from the rest of our command */
1516         sp = strchr(command_buf.buf, ' ') + 1;
1517         str_uq = unquote_c_style(sp, &endp);
1518         if (str_uq) {
1519                 if (*endp)
1520                         die("Garbage after ref in: %s", command_buf.buf);
1521                 sp = str_uq;
1522         }
1523         b = lookup_branch(sp);
1524         if (!b)
1525                 b = new_branch(sp);
1526         if (str_uq)
1527                 free(str_uq);
1528
1529         read_next_command();
1530         cmd_mark();
1531         if (!strncmp("author ", command_buf.buf, 7)) {
1532                 author = strdup(command_buf.buf);
1533                 read_next_command();
1534         }
1535         if (!strncmp("committer ", command_buf.buf, 10)) {
1536                 committer = strdup(command_buf.buf);
1537                 read_next_command();
1538         }
1539         if (!committer)
1540                 die("Expected committer but didn't get one");
1541         msg = cmd_data(&msglen);
1542         read_next_command();
1543         cmd_from(b);
1544
1545         /* ensure the branch is active/loaded */
1546         if (!b->branch_tree.tree || !max_active_branches) {
1547                 unload_one_branch();
1548                 load_branch(b);
1549         }
1550
1551         /* file_change* */
1552         for (;;) {
1553                 if (1 == command_buf.len)
1554                         break;
1555                 else if (!strncmp("M ", command_buf.buf, 2))
1556                         file_change_m(b);
1557                 else if (!strncmp("D ", command_buf.buf, 2))
1558                         file_change_d(b);
1559                 else
1560                         die("Unsupported file_change: %s", command_buf.buf);
1561                 read_next_command();
1562         }
1563
1564         /* build the tree and the commit */
1565         store_tree(&b->branch_tree);
1566         hashcpy(b->branch_tree.versions[0].sha1,
1567                 b->branch_tree.versions[1].sha1);
1568         size_dbuf(&new_data, 97 + msglen
1569                 + (author
1570                         ? strlen(author) + strlen(committer)
1571                         : 2 * strlen(committer)));
1572         sp = new_data.buffer;
1573         sp += sprintf(sp, "tree %s\n",
1574                 sha1_to_hex(b->branch_tree.versions[1].sha1));
1575         if (!is_null_sha1(b->sha1))
1576                 sp += sprintf(sp, "parent %s\n", sha1_to_hex(b->sha1));
1577         if (author)
1578                 sp += sprintf(sp, "%s\n", author);
1579         else
1580                 sp += sprintf(sp, "author %s\n", committer + 10);
1581         sp += sprintf(sp, "%s\n\n", committer);
1582         memcpy(sp, msg, msglen);
1583         sp += msglen;
1584         if (author)
1585                 free(author);
1586         free(committer);
1587         free(msg);
1588
1589         store_object(OBJ_COMMIT,
1590                 new_data.buffer, sp - (char*)new_data.buffer,
1591                 NULL, b->sha1, next_mark);
1592         b->last_commit = object_count_by_type[OBJ_COMMIT];
1593
1594         if (branch_log) {
1595                 int need_dq = quote_c_style(b->name, NULL, NULL, 0);
1596                 fprintf(branch_log, "commit ");
1597                 if (need_dq) {
1598                         fputc('"', branch_log);
1599                         quote_c_style(b->name, NULL, branch_log, 0);
1600                         fputc('"', branch_log);
1601                 } else
1602                         fprintf(branch_log, "%s", b->name);
1603                 fprintf(branch_log," :%lu %s\n",next_mark,sha1_to_hex(b->sha1));
1604         }
1605 }
1606
1607 static void cmd_new_tag()
1608 {
1609         char *str_uq;
1610         const char *endp;
1611         char *sp;
1612         const char *from;
1613         char *tagger;
1614         struct branch *s;
1615         void *msg;
1616         size_t msglen;
1617         struct tag *t;
1618         unsigned long from_mark = 0;
1619         unsigned char sha1[20];
1620
1621         /* Obtain the new tag name from the rest of our command */
1622         sp = strchr(command_buf.buf, ' ') + 1;
1623         str_uq = unquote_c_style(sp, &endp);
1624         if (str_uq) {
1625                 if (*endp)
1626                         die("Garbage after tag name in: %s", command_buf.buf);
1627                 sp = str_uq;
1628         }
1629         t = pool_alloc(sizeof(struct tag));
1630         t->next_tag = NULL;
1631         t->name = pool_strdup(sp);
1632         if (last_tag)
1633                 last_tag->next_tag = t;
1634         else
1635                 first_tag = t;
1636         last_tag = t;
1637         if (str_uq)
1638                 free(str_uq);
1639         read_next_command();
1640
1641         /* from ... */
1642         if (strncmp("from ", command_buf.buf, 5))
1643                 die("Expected from command, got %s", command_buf.buf);
1644
1645         from = strchr(command_buf.buf, ' ') + 1;
1646         str_uq = unquote_c_style(from, &endp);
1647         if (str_uq) {
1648                 if (*endp)
1649                         die("Garbage after string in: %s", command_buf.buf);
1650                 from = str_uq;
1651         }
1652
1653         s = lookup_branch(from);
1654         if (s) {
1655                 hashcpy(sha1, s->sha1);
1656         } else if (*from == ':') {
1657                 from_mark = strtoul(from + 1, NULL, 10);
1658                 struct object_entry *oe = find_mark(from_mark);
1659                 if (oe->type != OBJ_COMMIT)
1660                         die("Mark :%lu not a commit", from_mark);
1661                 hashcpy(sha1, oe->sha1);
1662         } else if (!get_sha1(from, sha1)) {
1663                 unsigned long size;
1664                 char *buf;
1665
1666                 buf = read_object_with_reference(sha1,
1667                         type_names[OBJ_COMMIT], &size, sha1);
1668                 if (!buf || size < 46)
1669                         die("Not a valid commit: %s", from);
1670                 free(buf);
1671         } else
1672                 die("Invalid ref name or SHA1 expression: %s", from);
1673
1674         if (str_uq)
1675                 free(str_uq);
1676         read_next_command();
1677
1678         /* tagger ... */
1679         if (strncmp("tagger ", command_buf.buf, 7))
1680                 die("Expected tagger command, got %s", command_buf.buf);
1681         tagger = strdup(command_buf.buf);
1682
1683         /* tag payload/message */
1684         read_next_command();
1685         msg = cmd_data(&msglen);
1686
1687         /* build the tag object */
1688         size_dbuf(&new_data, 67+strlen(t->name)+strlen(tagger)+msglen);
1689         sp = new_data.buffer;
1690         sp += sprintf(sp, "object %s\n", sha1_to_hex(sha1));
1691         sp += sprintf(sp, "type %s\n", type_names[OBJ_COMMIT]);
1692         sp += sprintf(sp, "tag %s\n", t->name);
1693         sp += sprintf(sp, "%s\n\n", tagger);
1694         memcpy(sp, msg, msglen);
1695         sp += msglen;
1696         free(tagger);
1697         free(msg);
1698
1699         store_object(OBJ_TAG, new_data.buffer, sp - (char*)new_data.buffer,
1700                 NULL, t->sha1, 0);
1701
1702         if (branch_log) {
1703                 int need_dq = quote_c_style(t->name, NULL, NULL, 0);
1704                 fprintf(branch_log, "tag ");
1705                 if (need_dq) {
1706                         fputc('"', branch_log);
1707                         quote_c_style(t->name, NULL, branch_log, 0);
1708                         fputc('"', branch_log);
1709                 } else
1710                         fprintf(branch_log, "%s", t->name);
1711                 fprintf(branch_log," :%lu %s\n",from_mark,sha1_to_hex(t->sha1));
1712         }
1713 }
1714
1715 static void cmd_reset_branch()
1716 {
1717         struct branch *b;
1718         char *str_uq;
1719         const char *endp;
1720         char *sp;
1721
1722         /* Obtain the branch name from the rest of our command */
1723         sp = strchr(command_buf.buf, ' ') + 1;
1724         str_uq = unquote_c_style(sp, &endp);
1725         if (str_uq) {
1726                 if (*endp)
1727                         die("Garbage after ref in: %s", command_buf.buf);
1728                 sp = str_uq;
1729         }
1730         b = lookup_branch(sp);
1731         if (b) {
1732                 b->last_commit = 0;
1733                 if (b->branch_tree.tree) {
1734                         release_tree_content_recursive(b->branch_tree.tree);
1735                         b->branch_tree.tree = NULL;
1736                 }
1737         }
1738         if (str_uq)
1739                 free(str_uq);
1740 }
1741
1742 static const char fast_import_usage[] =
1743 "git-fast-import [--objects=n] [--depth=n] [--active-branches=n] [--export-marks=marks.file] [--branch-log=log] temp.pack";
1744
1745 int main(int argc, const char **argv)
1746 {
1747         const char *base_name;
1748         int i;
1749         unsigned long est_obj_cnt = object_entry_alloc;
1750         char *pack_name;
1751         char *idx_name;
1752         struct stat sb;
1753
1754         setup_ident();
1755         git_config(git_default_config);
1756         page_size = getpagesize();
1757
1758         for (i = 1; i < argc; i++) {
1759                 const char *a = argv[i];
1760
1761                 if (*a != '-' || !strcmp(a, "--"))
1762                         break;
1763                 else if (!strncmp(a, "--objects=", 10))
1764                         est_obj_cnt = strtoul(a + 10, NULL, 0);
1765                 else if (!strncmp(a, "--depth=", 8))
1766                         max_depth = strtoul(a + 8, NULL, 0);
1767                 else if (!strncmp(a, "--active-branches=", 18))
1768                         max_active_branches = strtoul(a + 18, NULL, 0);
1769                 else if (!strncmp(a, "--export-marks=", 15))
1770                         mark_file = a + 15;
1771                 else if (!strncmp(a, "--branch-log=", 13)) {
1772                         branch_log = fopen(a + 13, "w");
1773                         if (!branch_log)
1774                                 die("Can't create %s: %s", a + 13, strerror(errno));
1775                 }
1776                 else
1777                         die("unknown option %s", a);
1778         }
1779         if ((i+1) != argc)
1780                 usage(fast_import_usage);
1781         base_name = argv[i];
1782
1783         pack_name = xmalloc(strlen(base_name) + 6);
1784         sprintf(pack_name, "%s.pack", base_name);
1785         idx_name = xmalloc(strlen(base_name) + 5);
1786         sprintf(idx_name, "%s.idx", base_name);
1787
1788         pack_fd = open(pack_name, O_RDWR|O_CREAT|O_EXCL, 0666);
1789         if (pack_fd < 0)
1790                 die("Can't create %s: %s", pack_name, strerror(errno));
1791
1792         init_pack_header();
1793         alloc_objects(est_obj_cnt);
1794         strbuf_init(&command_buf);
1795
1796         atom_table = xcalloc(atom_table_sz, sizeof(struct atom_str*));
1797         branch_table = xcalloc(branch_table_sz, sizeof(struct branch*));
1798         avail_tree_table = xcalloc(avail_tree_table_sz, sizeof(struct avail_tree_content*));
1799         marks = pool_calloc(1, sizeof(struct mark_set));
1800
1801         for (;;) {
1802                 read_next_command();
1803                 if (command_buf.eof)
1804                         break;
1805                 else if (!strcmp("blob", command_buf.buf))
1806                         cmd_new_blob();
1807                 else if (!strncmp("commit ", command_buf.buf, 7))
1808                         cmd_new_commit();
1809                 else if (!strncmp("tag ", command_buf.buf, 4))
1810                         cmd_new_tag();
1811                 else if (!strncmp("reset ", command_buf.buf, 6))
1812                         cmd_reset_branch();
1813                 else
1814                         die("Unsupported command: %s", command_buf.buf);
1815         }
1816
1817         fixup_header_footer();
1818         close(pack_fd);
1819         write_index(idx_name);
1820         dump_branches();
1821         dump_tags();
1822         dump_marks();
1823         if (branch_log)
1824                 fclose(branch_log);
1825
1826         fprintf(stderr, "%s statistics:\n", argv[0]);
1827         fprintf(stderr, "---------------------------------------------------------------------\n");
1828         fprintf(stderr, "Alloc'd objects: %10lu (%10lu overflow  )\n", alloc_count, alloc_count - est_obj_cnt);
1829         fprintf(stderr, "Total objects:   %10lu (%10lu duplicates                  )\n", object_count, duplicate_count);
1830         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]);
1831         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]);
1832         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]);
1833         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]);
1834         fprintf(stderr, "Total branches:  %10lu (%10lu loads     )\n", branch_count, branch_load_count);
1835         fprintf(stderr, "      marks:     %10u (%10lu unique    )\n", (1 << marks->shift) * 1024, marks_set_count);
1836         fprintf(stderr, "      atoms:     %10u\n", atom_cnt);
1837         fprintf(stderr, "Memory total:    %10lu KiB\n", (total_allocd + alloc_count*sizeof(struct object_entry))/1024);
1838         fprintf(stderr, "       pools:    %10lu KiB\n", total_allocd/1024);
1839         fprintf(stderr, "     objects:    %10lu KiB\n", (alloc_count*sizeof(struct object_entry))/1024);
1840         fprintf(stderr, "Pack remaps:     %10lu\n", remap_count);
1841         stat(pack_name, &sb);
1842         fprintf(stderr, "Pack size:       %10lu KiB\n", (unsigned long)(sb.st_size/1024));
1843         stat(idx_name, &sb);
1844         fprintf(stderr, "Index size:      %10lu KiB\n", (unsigned long)(sb.st_size/1024));
1845         fprintf(stderr, "---------------------------------------------------------------------\n");
1846
1847         fprintf(stderr, "\n");
1848
1849         return 0;
1850 }