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_branch
   8         | new_commit
   9         | new_tag
  10         ;
  11
  12   new_blob ::= 'blob' lf
  13         mark?
  14     file_content;
  15   file_content ::= data;
  16
  17   new_branch ::= 'branch' sp ref_str lf
  18     ('from' sp (ref_str | hexsha1 | sha1exp_str | idnum) lf)?
  19     lf;
  20
  21   new_commit ::= 'commit' sp ref_str lf
  22         mark?
  23         ('author' sp name '<' email '>' ts tz lf)?
  24         'committer' sp name '<' email '>' ts tz lf
  25         commit_msg
  26     file_change*
  27     lf;
  28   commit_msg ::= data;
  29
  30   file_change ::= 'M' sp mode sp (hexsha1 | idnum) sp path_str lf
  31                 | 'D' sp path_str lf
  32                 ;
  33   mode ::= '644' | '755';
  34
  35   new_tag ::= 'tag' sp tag_str lf
  36     'from' sp (ref_str | hexsha1 | sha1exp_str | idnum) lf
  37         'tagger' sp name '<' email '>' ts tz lf
  38     tag_msg;
  39   tag_msg ::= data;
  40
  41      # note: the first idnum in a stream should be 1 and subsequent
  42      # idnums should not have gaps between values as this will cause
  43      # the stream parser to reserve space for the gapped values.  An
  44          # idnum can be updated in the future to a new object by issuing
  45      # a new mark directive with the old idnum.
  46          #
  47   mark ::= 'mark' sp idnum lf;
  48
  49      # note: declen indicates the length of binary_data in bytes.
  50      # declen does not include the lf preceeding or trailing the
  51      # binary data.
  52      #
  53   data ::= 'data' sp declen lf
  54     binary_data
  55         lf;
  56
  57      # note: quoted strings are C-style quoting supporting \c for
  58      # common escapes of 'c' (e..g \n, \t, \\, \") or \nnn where nnn
  59          # is the signed byte value in octal.  Note that the only
  60      # characters which must actually be escaped to protect the
  61      # stream formatting is: \, " and LF.  Otherwise these values
  62          # are UTF8.
  63      #
  64   ref_str     ::= ref     | '"' quoted(ref)     '"' ;
  65   sha1exp_str ::= sha1exp | '"' quoted(sha1exp) '"' ;
  66   tag_str     ::= tag     | '"' quoted(tag)     '"' ;
  67   path_str    ::= path    | '"' quoted(path)    '"' ;
  68
  69   declen ::= # unsigned 32 bit value, ascii base10 notation;
  70   binary_data ::= # file content, not interpreted;
  71
  72   sp ::= # ASCII space character;
  73   lf ::= # ASCII newline (LF) character;
  74
  75      # note: a colon (':') must precede the numerical value assigned to
  76          # an idnum.  This is to distinguish it from a ref or tag name as
  77      # GIT does not permit ':' in ref or tag strings.
  78          #
  79   idnum   ::= ':' declen;
  80   path    ::= # GIT style file path, e.g. "a/b/c";
  81   ref     ::= # GIT ref name, e.g. "refs/heads/MOZ_GECKO_EXPERIMENT";
  82   tag     ::= # GIT tag name, e.g. "FIREFOX_1_5";
  83   sha1exp ::= # Any valid GIT SHA1 expression;
  84   hexsha1 ::= # SHA1 in hexadecimal format;
  85
  86      # note: name and email are UTF8 strings, however name must not
  87          # contain '<' or lf and email must not contain any of the
  88      # following: '<', '>', lf.
  89          #
  90   name  ::= # valid GIT author/committer name;
  91   email ::= # valid GIT author/committer email;
  92   ts    ::= # time since the epoch in seconds, ascii base10 notation;
  93   tz    ::= # GIT style timezone;
  94 */
  95
  96 #include "builtin.h"
  97 #include "cache.h"
  98 #include "object.h"
  99 #include "blob.h"
 100 #include "tree.h"
 101 #include "delta.h"
 102 #include "pack.h"
 103 #include "refs.h"
 104 #include "csum-file.h"
 105 #include "strbuf.h"
 106 #include "quote.h"
 107
 108 struct object_entry
 109 {
 110         struct object_entry *next;
 111         enum object_type type;
 112         unsigned long offset;
 113         unsigned char sha1[20];
 114 };
 115
 116 struct object_entry_pool
 117 {
 118         struct object_entry_pool *next_pool;
 119         struct object_entry *next_free;
 120         struct object_entry *end;
 121         struct object_entry entries[FLEX_ARRAY]; /* more */
 122 };
 123
 124 struct mark_set
 125 {
 126         int shift;
 127         union {
 128                 struct object_entry *marked[1024];
 129                 struct mark_set *sets[1024];
 130         } data;
 131 };
 132
 133 struct last_object
 134 {
 135         void *data;
 136         unsigned int len;
 137         unsigned int depth;
 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         unsigned int mode;
 162         unsigned char sha1[20];
 163 };
 164
 165 struct tree_content
 166 {
 167         unsigned int entry_capacity; /* must match avail_tree_content */
 168         unsigned int entry_count;
 169         struct tree_entry *entries[FLEX_ARRAY]; /* more */
 170 };
 171
 172 struct avail_tree_content
 173 {
 174         unsigned int entry_capacity; /* must match tree_content */
 175         struct avail_tree_content *next_avail;
 176 };
 177
 178 struct branch
 179 {
 180         struct branch *table_next_branch;
 181         struct branch *active_next_branch;
 182         const char *name;
 183         unsigned long last_commit;
 184         struct tree_entry branch_tree;
 185         unsigned char sha1[20];
 186 };
 187
 188 struct tag
 189 {
 190         struct tag *next_tag;
 191         const char *name;
 192         unsigned char sha1[20];
 193 };
 194
 195
 196 /* Stats and misc. counters */
 197 static unsigned long max_depth = 10;
 198 static unsigned long alloc_count;
 199 static unsigned long branch_count;
 200 static unsigned long branch_load_count;
 201 static unsigned long remap_count;
 202 static unsigned long object_count;
 203 static unsigned long duplicate_count;
 204 static unsigned long marks_set_count;
 205 static unsigned long object_count_by_type[9];
 206 static unsigned long duplicate_count_by_type[9];
 207
 208 /* Memory pools */
 209 static size_t mem_pool_alloc = 2*1024*1024 - sizeof(struct mem_pool);
 210 static size_t total_allocd;
 211 static struct mem_pool *mem_pool;
 212
 213 /* Atom management */
 214 static unsigned int atom_table_sz = 4451;
 215 static unsigned int atom_cnt;
 216 static struct atom_str **atom_table;
 217
 218 /* The .pack file being generated */
 219 static int pack_fd;
 220 static unsigned long pack_size;
 221 static unsigned char pack_sha1[20];
 222 static void* pack_base;
 223 static size_t pack_mlen;
 224
 225 /* Table of objects we've written. */
 226 static unsigned int object_entry_alloc = 1000;
 227 static struct object_entry_pool *blocks;
 228 static struct object_entry *object_table[1 << 16];
 229 static struct mark_set *marks;
 230
 231 /* Our last blob */
 232 static struct last_object last_blob;
 233
 234 /* Tree management */
 235 static unsigned int tree_entry_alloc = 1000;
 236 static void *avail_tree_entry;
 237 static unsigned int avail_tree_table_sz = 100;
 238 static struct avail_tree_content **avail_tree_table;
 239
 240 /* Branch data */
 241 static unsigned long max_active_branches = 5;
 242 static unsigned long cur_active_branches;
 243 static unsigned long branch_table_sz = 1039;
 244 static struct branch **branch_table;
 245 static struct branch *active_branches;
 246
 247 /* Tag data */
 248 static struct tag *first_tag;
 249 static struct tag *last_tag;
 250
 251 /* Input stream parsing */
 252 static struct strbuf command_buf;
 253 static unsigned long next_mark;
 254
 255
 256 static void alloc_objects(int cnt)
 257 {
 258         struct object_entry_pool *b;
 259
 260         b = xmalloc(sizeof(struct object_entry_pool)
 261                 + cnt * sizeof(struct object_entry));
 262         b->next_pool = blocks;
 263         b->next_free = b->entries;
 264         b->end = b->entries + cnt;
 265         blocks = b;
 266         alloc_count += cnt;
 267 }
 268
 269 static struct object_entry* new_object(unsigned char *sha1)
 270 {
 271         struct object_entry *e;
 272
 273         if (blocks->next_free == blocks->end)
 274                 alloc_objects(object_entry_alloc);
 275
 276         e = blocks->next_free++;
 277         memcpy(e->sha1, sha1, sizeof(e->sha1));
 278         return e;
 279 }
 280
 281 static struct object_entry* find_object(unsigned char *sha1)
 282 {
 283         unsigned int h = sha1[0] << 8 | sha1[1];
 284         struct object_entry *e;
 285         for (e = object_table[h]; e; e = e->next)
 286                 if (!memcmp(sha1, e->sha1, sizeof(e->sha1)))
 287                         return e;
 288         return NULL;
 289 }
 290
 291 static struct object_entry* insert_object(unsigned char *sha1)
 292 {
 293         unsigned int h = sha1[0] << 8 | sha1[1];
 294         struct object_entry *e = object_table[h];
 295         struct object_entry *p = NULL;
 296
 297         while (e) {
 298                 if (!memcmp(sha1, e->sha1, sizeof(e->sha1)))
 299                         return e;
 300                 p = e;
 301                 e = e->next;
 302         }
 303
 304         e = new_object(sha1);
 305         e->next = NULL;
 306         e->offset = 0;
 307         if (p)
 308                 p->next = e;
 309         else
 310                 object_table[h] = e;
 311         return e;
 312 }
 313
 314 static unsigned int hc_str(const char *s, size_t len)
 315 {
 316         unsigned int r = 0;
 317         while (len-- > 0)
 318                 r = r * 31 + *s++;
 319         return r;
 320 }
 321
 322 static void* pool_alloc(size_t len)
 323 {
 324         struct mem_pool *p;
 325         void *r;
 326
 327         for (p = mem_pool; p; p = p->next_pool)
 328                 if ((p->end - p->next_free >= len))
 329                         break;
 330
 331         if (!p) {
 332                 if (len >= (mem_pool_alloc/2)) {
 333                         total_allocd += len;
 334                         return xmalloc(len);
 335                 }
 336                 total_allocd += sizeof(struct mem_pool) + mem_pool_alloc;
 337                 p = xmalloc(sizeof(struct mem_pool) + mem_pool_alloc);
 338                 p->next_pool = mem_pool;
 339                 p->next_free = p->space;
 340                 p->end = p->next_free + mem_pool_alloc;
 341                 mem_pool = p;
 342         }
 343
 344         r = p->next_free;
 345         /* round out to a pointer alignment */
 346         if (len & (sizeof(void*) - 1))
 347                 len += sizeof(void*) - (len & (sizeof(void*) - 1));
 348         p->next_free += len;
 349         return r;
 350 }
 351
 352 static void* pool_calloc(size_t count, size_t size)
 353 {
 354         size_t len = count * size;
 355         void *r = pool_alloc(len);
 356         memset(r, 0, len);
 357         return r;
 358 }
 359
 360 static char* pool_strdup(const char *s)
 361 {
 362         char *r = pool_alloc(strlen(s) + 1);
 363         strcpy(r, s);
 364         return r;
 365 }
 366
 367 static void insert_mark(unsigned long idnum, struct object_entry *oe)
 368 {
 369         struct mark_set *s = marks;
 370         while ((idnum >> s->shift) >= 1024) {
 371                 s = pool_calloc(1, sizeof(struct mark_set));
 372                 s->shift = marks->shift + 10;
 373                 s->data.sets[0] = marks;
 374                 marks = s;
 375         }
 376         while (s->shift) {
 377                 unsigned long i = idnum >> s->shift;
 378                 idnum -= i << s->shift;
 379                 if (!s->data.sets[i]) {
 380                         s->data.sets[i] = pool_calloc(1, sizeof(struct mark_set));
 381                         s->data.sets[i]->shift = s->shift - 10;
 382                 }
 383                 s = s->data.sets[i];
 384         }
 385         if (!s->data.marked[idnum])
 386                 marks_set_count++;
 387         s->data.marked[idnum] = oe;
 388 }
 389
 390 static struct object_entry* find_mark(unsigned long idnum)
 391 {
 392         unsigned long orig_idnum = idnum;
 393         struct mark_set *s = marks;
 394         struct object_entry *oe = NULL;
 395         if ((idnum >> s->shift) < 1024) {
 396                 while (s && s->shift) {
 397                         unsigned long i = idnum >> s->shift;
 398                         idnum -= i << s->shift;
 399                         s = s->data.sets[i];
 400                 }
 401                 if (s)
 402                         oe = s->data.marked[idnum];
 403         }
 404         if (!oe)
 405                 die("mark :%lu not declared", orig_idnum);
 406         return oe;
 407 }
 408
 409 static struct atom_str* to_atom(const char *s, size_t len)
 410 {
 411         unsigned int hc = hc_str(s, len) % atom_table_sz;
 412         struct atom_str *c;
 413
 414         for (c = atom_table[hc]; c; c = c->next_atom)
 415                 if (c->str_len == len && !strncmp(s, c->str_dat, len))
 416                         return c;
 417
 418         c = pool_alloc(sizeof(struct atom_str) + len + 1);
 419         c->str_len = len;
 420         strncpy(c->str_dat, s, len);
 421         c->str_dat[len] = 0;
 422         c->next_atom = atom_table[hc];
 423         atom_table[hc] = c;
 424         atom_cnt++;
 425         return c;
 426 }
 427
 428 static struct branch* lookup_branch(const char *name)
 429 {
 430         unsigned int hc = hc_str(name, strlen(name)) % branch_table_sz;
 431         struct branch *b;
 432
 433         for (b = branch_table[hc]; b; b = b->table_next_branch)
 434                 if (!strcmp(name, b->name))
 435                         return b;
 436         return NULL;
 437 }
 438
 439 static struct branch* new_branch(const char *name)
 440 {
 441         unsigned int hc = hc_str(name, strlen(name)) % branch_table_sz;
 442         struct branch* b = lookup_branch(name);
 443
 444         if (b)
 445                 die("Invalid attempt to create duplicate branch: %s", name);
 446         if (check_ref_format(name))
 447                 die("Branch name doesn't conform to GIT standards: %s", name);
 448
 449         b = pool_calloc(1, sizeof(struct branch));
 450         b->name = pool_strdup(name);
 451         b->table_next_branch = branch_table[hc];
 452         branch_table[hc] = b;
 453         branch_count++;
 454         return b;
 455 }
 456
 457 static unsigned int hc_entries(unsigned int cnt)
 458 {
 459         cnt = cnt & 7 ? (cnt / 8) + 1 : cnt / 8;
 460         return cnt < avail_tree_table_sz ? cnt : avail_tree_table_sz - 1;
 461 }
 462
 463 static struct tree_content* new_tree_content(unsigned int cnt)
 464 {
 465         struct avail_tree_content *f, *l = NULL;
 466         struct tree_content *t;
 467         unsigned int hc = hc_entries(cnt);
 468
 469         for (f = avail_tree_table[hc]; f; l = f, f = f->next_avail)
 470                 if (f->entry_capacity >= cnt)
 471                         break;
 472
 473         if (f) {
 474                 if (l)
 475                         l->next_avail = f->next_avail;
 476                 else
 477                         avail_tree_table[hc] = f->next_avail;
 478         } else {
 479                 cnt = cnt & 7 ? ((cnt / 8) + 1) * 8 : cnt;
 480                 f = pool_alloc(sizeof(*t) + sizeof(t->entries[0]) * cnt);
 481                 f->entry_capacity = cnt;
 482         }
 483
 484         t = (struct tree_content*)f;
 485         t->entry_count = 0;
 486         return t;
 487 }
 488
 489 static void release_tree_entry(struct tree_entry *e);
 490 static void release_tree_content(struct tree_content *t)
 491 {
 492         struct avail_tree_content *f = (struct avail_tree_content*)t;
 493         unsigned int hc = hc_entries(f->entry_capacity);
 494         f->next_avail = avail_tree_table[hc];
 495         avail_tree_table[hc] = f;
 496 }
 497
 498 static void release_tree_content_recursive(struct tree_content *t)
 499 {
 500         unsigned int i;
 501         for (i = 0; i < t->entry_count; i++)
 502                 release_tree_entry(t->entries[i]);
 503         release_tree_content(t);
 504 }
 505
 506 static struct tree_content* grow_tree_content(
 507         struct tree_content *t,
 508         int amt)
 509 {
 510         struct tree_content *r = new_tree_content(t->entry_count + amt);
 511         r->entry_count = t->entry_count;
 512         memcpy(r->entries,t->entries,t->entry_count*sizeof(t->entries[0]));
 513         release_tree_content(t);
 514         return r;
 515 }
 516
 517 static struct tree_entry* new_tree_entry()
 518 {
 519         struct tree_entry *e;
 520
 521         if (!avail_tree_entry) {
 522                 unsigned int n = tree_entry_alloc;
 523                 avail_tree_entry = e = xmalloc(n * sizeof(struct tree_entry));
 524                 while (n--) {
 525                         *((void**)e) = e + 1;
 526                         e++;
 527                 }
 528         }
 529
 530         e = avail_tree_entry;
 531         avail_tree_entry = *((void**)e);
 532         return e;
 533 }
 534
 535 static void release_tree_entry(struct tree_entry *e)
 536 {
 537         if (e->tree)
 538                 release_tree_content_recursive(e->tree);
 539         *((void**)e) = avail_tree_entry;
 540         avail_tree_entry = e;
 541 }
 542
 543 static void yread(int fd, void *buffer, size_t length)
 544 {
 545         ssize_t ret = 0;
 546         while (ret < length) {
 547                 ssize_t size = xread(fd, (char *) buffer + ret, length - ret);
 548                 if (!size)
 549                         die("Read from descriptor %i: end of stream", fd);
 550                 if (size < 0)
 551                         die("Read from descriptor %i: %s", fd, strerror(errno));
 552                 ret += size;
 553         }
 554 }
 555
 556 static void ywrite(int fd, void *buffer, size_t length)
 557 {
 558         ssize_t ret = 0;
 559         while (ret < length) {
 560                 ssize_t size = xwrite(fd, (char *) buffer + ret, length - ret);
 561                 if (!size)
 562                         die("Write to descriptor %i: end of file", fd);
 563                 if (size < 0)
 564                         die("Write to descriptor %i: %s", fd, strerror(errno));
 565                 ret += size;
 566         }
 567 }
 568
 569 static size_t encode_header(
 570         enum object_type type,
 571         size_t size,
 572         unsigned char *hdr)
 573 {
 574         int n = 1;
 575         unsigned char c;
 576
 577         if (type < OBJ_COMMIT || type > OBJ_DELTA)
 578                 die("bad type %d", type);
 579
 580         c = (type << 4) | (size & 15);
 581         size >>= 4;
 582         while (size) {
 583                 *hdr++ = c | 0x80;
 584                 c = size & 0x7f;
 585                 size >>= 7;
 586                 n++;
 587         }
 588         *hdr = c;
 589         return n;
 590 }
 591
 592 static int store_object(
 593         enum object_type type,
 594         void *dat,
 595         size_t datlen,
 596         struct last_object *last,
 597         unsigned char *sha1out,
 598         unsigned long mark)
 599 {
 600         void *out, *delta;
 601         struct object_entry *e;
 602         unsigned char hdr[96];
 603         unsigned char sha1[20];
 604         unsigned long hdrlen, deltalen;
 605         SHA_CTX c;
 606         z_stream s;
 607
 608         hdrlen = sprintf((char*)hdr,"%s %lu",type_names[type],datlen) + 1;
 609         SHA1_Init(&c);
 610         SHA1_Update(&c, hdr, hdrlen);
 611         SHA1_Update(&c, dat, datlen);
 612         SHA1_Final(sha1, &c);
 613         if (sha1out)
 614                 memcpy(sha1out, sha1, sizeof(sha1));
 615
 616         e = insert_object(sha1);
 617         if (mark)
 618                 insert_mark(mark, e);
 619         if (e->offset) {
 620                 duplicate_count++;
 621                 duplicate_count_by_type[type]++;
 622                 return 1;
 623         }
 624         e->type = type;
 625         e->offset = pack_size;
 626         object_count++;
 627         object_count_by_type[type]++;
 628
 629         if (last && last->data && last->depth < max_depth)
 630                 delta = diff_delta(last->data, last->len,
 631                         dat, datlen,
 632                         &deltalen, 0);
 633         else
 634                 delta = 0;
 635
 636         memset(&s, 0, sizeof(s));
 637         deflateInit(&s, zlib_compression_level);
 638
 639         if (delta) {
 640                 last->depth++;
 641                 s.next_in = delta;
 642                 s.avail_in = deltalen;
 643                 hdrlen = encode_header(OBJ_DELTA, deltalen, hdr);
 644                 ywrite(pack_fd, hdr, hdrlen);
 645                 ywrite(pack_fd, last->sha1, sizeof(sha1));
 646                 pack_size += hdrlen + sizeof(sha1);
 647         } else {
 648                 if (last)
 649                         last->depth = 0;
 650                 s.next_in = dat;
 651                 s.avail_in = datlen;
 652                 hdrlen = encode_header(type, datlen, hdr);
 653                 ywrite(pack_fd, hdr, hdrlen);
 654                 pack_size += hdrlen;
 655         }
 656
 657         s.avail_out = deflateBound(&s, s.avail_in);
 658         s.next_out = out = xmalloc(s.avail_out);
 659         while (deflate(&s, Z_FINISH) == Z_OK)
 660                 /* nothing */;
 661         deflateEnd(&s);
 662
 663         ywrite(pack_fd, out, s.total_out);
 664         pack_size += s.total_out;
 665
 666         free(out);
 667         if (delta)
 668                 free(delta);
 669         if (last) {
 670                 if (last->data)
 671                         free(last->data);
 672                 last->data = dat;
 673                 last->len = datlen;
 674                 memcpy(last->sha1, sha1, sizeof(sha1));
 675         }
 676         return 0;
 677 }
 678
 679 static void* map_pack(unsigned long offset)
 680 {
 681         if (offset >= pack_size)
 682                 die("object offset outside of pack file");
 683         if (offset >= pack_mlen) {
 684                 if (pack_base)
 685                         munmap(pack_base, pack_mlen);
 686                 /* round out how much we map to 16 MB units */
 687                 pack_mlen = pack_size;
 688                 if (pack_mlen & ((1 << 24) - 1))
 689                         pack_mlen = ((pack_mlen >> 24) + 1) << 24;
 690                 pack_base = mmap(NULL,pack_mlen,PROT_READ,MAP_SHARED,pack_fd,0);
 691                 if (pack_base == MAP_FAILED)
 692                         die("Failed to map generated pack: %s", strerror(errno));
 693                 remap_count++;
 694         }
 695         return (char*)pack_base + offset;
 696 }
 697
 698 static unsigned long unpack_object_header(unsigned long offset,
 699         enum object_type *type,
 700         unsigned long *sizep)
 701 {
 702         unsigned shift;
 703         unsigned char c;
 704         unsigned long size;
 705
 706         c = *(unsigned char*)map_pack(offset++);
 707         *type = (c >> 4) & 7;
 708         size = c & 15;
 709         shift = 4;
 710         while (c & 0x80) {
 711                 c = *(unsigned char*)map_pack(offset++);
 712                 size += (c & 0x7f) << shift;
 713                 shift += 7;
 714         }
 715         *sizep = size;
 716         return offset;
 717 }
 718
 719 static void *unpack_non_delta_entry(unsigned long o, unsigned long sz)
 720 {
 721         z_stream stream;
 722         unsigned char *result;
 723
 724         result = xmalloc(sz + 1);
 725         result[sz] = 0;
 726
 727         memset(&stream, 0, sizeof(stream));
 728         stream.next_in = map_pack(o);
 729         stream.avail_in = pack_mlen - o;
 730         stream.next_out = result;
 731         stream.avail_out = sz;
 732
 733         inflateInit(&stream);
 734         for (;;) {
 735                 int st = inflate(&stream, Z_FINISH);
 736                 if (st == Z_STREAM_END)
 737                         break;
 738                 if (st == Z_OK) {
 739                         o = stream.next_in - (unsigned char*)pack_base;
 740                         stream.next_in = map_pack(o);
 741                         stream.avail_in = pack_mlen - o;
 742                         continue;
 743                 }
 744                 die("Error from zlib during inflate.");
 745         }
 746         inflateEnd(&stream);
 747         if (stream.total_out != sz)
 748                 die("Error after inflate: sizes mismatch");
 749         return result;
 750 }
 751
 752 static void *unpack_entry(unsigned long offset, unsigned long *sizep);
 753
 754 static void *unpack_delta_entry(unsigned long offset,
 755         unsigned long delta_size,
 756         unsigned long *sizep)
 757 {
 758         struct object_entry *base_oe;
 759         unsigned char *base_sha1;
 760         void *delta_data, *base, *result;
 761         unsigned long base_size, result_size;
 762
 763         base_sha1 = (unsigned char*)map_pack(offset + 20) - 20;
 764         base_oe = find_object(base_sha1);
 765         if (!base_oe)
 766                 die("I'm broken; I can't find a base I know must be here.");
 767         base = unpack_entry(base_oe->offset, &base_size);
 768         delta_data = unpack_non_delta_entry(offset + 20, delta_size);
 769         result = patch_delta(base, base_size,
 770                              delta_data, delta_size,
 771                              &result_size);
 772         if (!result)
 773                 die("failed to apply delta");
 774         free(delta_data);
 775         free(base);
 776         *sizep = result_size;
 777         return result;
 778 }
 779
 780 static void *unpack_entry(unsigned long offset, unsigned long *sizep)
 781 {
 782         unsigned long size;
 783         enum object_type kind;
 784
 785         offset = unpack_object_header(offset, &kind, &size);
 786         switch (kind) {
 787         case OBJ_DELTA:
 788                 return unpack_delta_entry(offset, size, sizep);
 789         case OBJ_COMMIT:
 790         case OBJ_TREE:
 791         case OBJ_BLOB:
 792         case OBJ_TAG:
 793                 *sizep = size;
 794                 return unpack_non_delta_entry(offset, size);
 795         default:
 796                 die("I created an object I can't read!");
 797         }
 798 }
 799
 800 static const char *get_mode(const char *str, unsigned int *modep)
 801 {
 802         unsigned char c;
 803         unsigned int mode = 0;
 804
 805         while ((c = *str++) != ' ') {
 806                 if (c < '0' || c > '7')
 807                         return NULL;
 808                 mode = (mode << 3) + (c - '0');
 809         }
 810         *modep = mode;
 811         return str;
 812 }
 813
 814 static void load_tree(struct tree_entry *root)
 815 {
 816         struct object_entry *myoe;
 817         struct tree_content *t;
 818         unsigned long size;
 819         char *buf;
 820         const char *c;
 821
 822         root->tree = t = new_tree_content(8);
 823         if (!memcmp(root->sha1, null_sha1, 20))
 824                 return;
 825
 826         myoe = find_object(root->sha1);
 827         if (myoe) {
 828                 if (myoe->type != OBJ_TREE)
 829                         die("Not a tree: %s", sha1_to_hex(root->sha1));
 830                 buf = unpack_entry(myoe->offset, &size);
 831         } else {
 832                 char type[20];
 833                 buf = read_sha1_file(root->sha1, type, &size);
 834                 if (!buf || !strcmp(type, tree_type))
 835                         die("Can't load tree %s", sha1_to_hex(root->sha1));
 836         }
 837
 838         c = buf;
 839         while (c != (buf + size)) {
 840                 struct tree_entry *e = new_tree_entry();
 841
 842                 if (t->entry_count == t->entry_capacity)
 843                         root->tree = t = grow_tree_content(t, 8);
 844                 t->entries[t->entry_count++] = e;
 845
 846                 e->tree = NULL;
 847                 c = get_mode(c, &e->mode);
 848                 if (!c)
 849                         die("Corrupt mode in %s", sha1_to_hex(root->sha1));
 850                 e->name = to_atom(c, strlen(c));
 851                 c += e->name->str_len + 1;
 852                 memcpy(e->sha1, c, sizeof(e->sha1));
 853                 c += 20;
 854         }
 855         free(buf);
 856 }
 857
 858 static int tecmp (const void *_a, const void *_b)
 859 {
 860         struct tree_entry *a = *((struct tree_entry**)_a);
 861         struct tree_entry *b = *((struct tree_entry**)_b);
 862         return base_name_compare(
 863                 a->name->str_dat, a->name->str_len, a->mode,
 864                 b->name->str_dat, b->name->str_len, b->mode);
 865 }
 866
 867 static void store_tree(struct tree_entry *root)
 868 {
 869         struct tree_content *t = root->tree;
 870         unsigned int i;
 871         size_t maxlen;
 872         char *buf, *c;
 873
 874         if (memcmp(root->sha1, null_sha1, 20))
 875                 return;
 876
 877         maxlen = 0;
 878         for (i = 0; i < t->entry_count; i++) {
 879                 maxlen += t->entries[i]->name->str_len + 34;
 880                 if (t->entries[i]->tree)
 881                         store_tree(t->entries[i]);
 882         }
 883
 884         qsort(t->entries, t->entry_count, sizeof(t->entries[0]), tecmp);
 885         buf = c = xmalloc(maxlen);
 886         for (i = 0; i < t->entry_count; i++) {
 887                 struct tree_entry *e = t->entries[i];
 888                 c += sprintf(c, "%o", e->mode);
 889                 *c++ = ' ';
 890                 strcpy(c, e->name->str_dat);
 891                 c += e->name->str_len + 1;
 892                 memcpy(c, e->sha1, 20);
 893                 c += 20;
 894         }
 895         store_object(OBJ_TREE, buf, c - buf, NULL, root->sha1, 0);
 896         free(buf);
 897 }
 898
 899 static int tree_content_set(
 900         struct tree_entry *root,
 901         const char *p,
 902         const unsigned char *sha1,
 903         const unsigned int mode)
 904 {
 905         struct tree_content *t = root->tree;
 906         const char *slash1;
 907         unsigned int i, n;
 908         struct tree_entry *e;
 909
 910         slash1 = strchr(p, '/');
 911         if (slash1)
 912                 n = slash1 - p;
 913         else
 914                 n = strlen(p);
 915
 916         for (i = 0; i < t->entry_count; i++) {
 917                 e = t->entries[i];
 918                 if (e->name->str_len == n && !strncmp(p, e->name->str_dat, n)) {
 919                         if (!slash1) {
 920                                 if (e->mode == mode && !memcmp(e->sha1, sha1, 20))
 921                                         return 0;
 922                                 e->mode = mode;
 923                                 memcpy(e->sha1, sha1, 20);
 924                                 if (e->tree) {
 925                                         release_tree_content_recursive(e->tree);
 926                                         e->tree = NULL;
 927                                 }
 928                                 memcpy(root->sha1, null_sha1, 20);
 929                                 return 1;
 930                         }
 931                         if (!S_ISDIR(e->mode)) {
 932                                 e->tree = new_tree_content(8);
 933                                 e->mode = S_IFDIR;
 934                         }
 935                         if (!e->tree)
 936                                 load_tree(e);
 937                         if (tree_content_set(e, slash1 + 1, sha1, mode)) {
 938                                 memcpy(root->sha1, null_sha1, 20);
 939                                 return 1;
 940                         }
 941                         return 0;
 942                 }
 943         }
 944
 945         if (t->entry_count == t->entry_capacity)
 946                 root->tree = t = grow_tree_content(t, 8);
 947         e = new_tree_entry();
 948         e->name = to_atom(p, n);
 949         t->entries[t->entry_count++] = e;
 950         if (slash1) {
 951                 e->tree = new_tree_content(8);
 952                 e->mode = S_IFDIR;
 953                 tree_content_set(e, slash1 + 1, sha1, mode);
 954         } else {
 955                 e->tree = NULL;
 956                 e->mode = mode;
 957                 memcpy(e->sha1, sha1, 20);
 958         }
 959         memcpy(root->sha1, null_sha1, 20);
 960         return 1;
 961 }
 962
 963 static int tree_content_remove(struct tree_entry *root, const char *p)
 964 {
 965         struct tree_content *t = root->tree;
 966         const char *slash1;
 967         unsigned int i, n;
 968         struct tree_entry *e;
 969
 970         slash1 = strchr(p, '/');
 971         if (slash1)
 972                 n = slash1 - p;
 973         else
 974                 n = strlen(p);
 975
 976         for (i = 0; i < t->entry_count; i++) {
 977                 e = t->entries[i];
 978                 if (e->name->str_len == n && !strncmp(p, e->name->str_dat, n)) {
 979                         if (!slash1 || !S_ISDIR(e->mode))
 980                                 goto del_entry;
 981                         if (!e->tree)
 982                                 load_tree(e);
 983                         if (tree_content_remove(e, slash1 + 1)) {
 984                                 if (!e->tree->entry_count)
 985                                         goto del_entry;
 986                                 memcpy(root->sha1, null_sha1, 20);
 987                                 return 1;
 988                         }
 989                         return 0;
 990                 }
 991         }
 992         return 0;
 993
 994 del_entry:
 995         for (i++; i < t->entry_count; i++)
 996                 t->entries[i-1] = t->entries[i];
 997         t->entry_count--;
 998         release_tree_entry(e);
 999         memcpy(root->sha1, null_sha1, 20);
1000         return 1;
1001 }
1002
1003 static void init_pack_header()
1004 {
1005         struct pack_header hdr;
1006
1007         hdr.hdr_signature = htonl(PACK_SIGNATURE);
1008         hdr.hdr_version = htonl(2);
1009         hdr.hdr_entries = 0;
1010
1011         ywrite(pack_fd, &hdr, sizeof(hdr));
1012         pack_size = sizeof(hdr);
1013 }
1014
1015 static void fixup_header_footer()
1016 {
1017         SHA_CTX c;
1018         char hdr[8];
1019         unsigned long cnt;
1020         char *buf;
1021         size_t n;
1022
1023         if (lseek(pack_fd, 0, SEEK_SET) != 0)
1024                 die("Failed seeking to start: %s", strerror(errno));
1025
1026         SHA1_Init(&c);
1027         yread(pack_fd, hdr, 8);
1028         SHA1_Update(&c, hdr, 8);
1029
1030         cnt = htonl(object_count);
1031         SHA1_Update(&c, &cnt, 4);
1032         ywrite(pack_fd, &cnt, 4);
1033
1034         buf = xmalloc(128 * 1024);
1035         for (;;) {
1036                 n = xread(pack_fd, buf, 128 * 1024);
1037                 if (n <= 0)
1038                         break;
1039                 SHA1_Update(&c, buf, n);
1040         }
1041         free(buf);
1042
1043         SHA1_Final(pack_sha1, &c);
1044         ywrite(pack_fd, pack_sha1, sizeof(pack_sha1));
1045 }
1046
1047 static int oecmp (const void *_a, const void *_b)
1048 {
1049         struct object_entry *a = *((struct object_entry**)_a);
1050         struct object_entry *b = *((struct object_entry**)_b);
1051         return memcmp(a->sha1, b->sha1, sizeof(a->sha1));
1052 }
1053
1054 static void write_index(const char *idx_name)
1055 {
1056         struct sha1file *f;
1057         struct object_entry **idx, **c, **last;
1058         struct object_entry *e;
1059         struct object_entry_pool *o;
1060         unsigned int array[256];
1061         int i;
1062
1063         /* Build the sorted table of object IDs. */
1064         idx = xmalloc(object_count * sizeof(struct object_entry*));
1065         c = idx;
1066         for (o = blocks; o; o = o->next_pool)
1067                 for (e = o->entries; e != o->next_free; e++)
1068                         *c++ = e;
1069         last = idx + object_count;
1070         qsort(idx, object_count, sizeof(struct object_entry*), oecmp);
1071
1072         /* Generate the fan-out array. */
1073         c = idx;
1074         for (i = 0; i < 256; i++) {
1075                 struct object_entry **next = c;;
1076                 while (next < last) {
1077                         if ((*next)->sha1[0] != i)
1078                                 break;
1079                         next++;
1080                 }
1081                 array[i] = htonl(next - idx);
1082                 c = next;
1083         }
1084
1085         f = sha1create("%s", idx_name);
1086         sha1write(f, array, 256 * sizeof(int));
1087         for (c = idx; c != last; c++) {
1088                 unsigned int offset = htonl((*c)->offset);
1089                 sha1write(f, &offset, 4);
1090                 sha1write(f, (*c)->sha1, sizeof((*c)->sha1));
1091         }
1092         sha1write(f, pack_sha1, sizeof(pack_sha1));
1093         sha1close(f, NULL, 1);
1094         free(idx);
1095 }
1096
1097 static void dump_branches()
1098 {
1099         static const char *msg = "fast-import";
1100         unsigned int i;
1101         struct branch *b;
1102         struct ref_lock *lock;
1103
1104         for (i = 0; i < branch_table_sz; i++) {
1105                 for (b = branch_table[i]; b; b = b->table_next_branch) {
1106                         lock = lock_any_ref_for_update(b->name, NULL, 0);
1107                         if (!lock || write_ref_sha1(lock, b->sha1, msg) < 0)
1108                                 die("Can't write %s", b->name);
1109                 }
1110         }
1111 }
1112
1113 static void dump_tags()
1114 {
1115         static const char *msg = "fast-import";
1116         struct tag *t;
1117         struct ref_lock *lock;
1118         char path[PATH_MAX];
1119
1120         for (t = first_tag; t; t = t->next_tag) {
1121                 sprintf(path, "refs/tags/%s", t->name);
1122                 lock = lock_any_ref_for_update(path, NULL, 0);
1123                 if (!lock || write_ref_sha1(lock, t->sha1, msg) < 0)
1124                         die("Can't write %s", path);
1125         }
1126 }
1127
1128 static void read_next_command()
1129 {
1130         read_line(&command_buf, stdin, '\n');
1131 }
1132
1133 static void cmd_mark()
1134 {
1135         if (!strncmp("mark :", command_buf.buf, 6)) {
1136                 next_mark = strtoul(command_buf.buf + 6, NULL, 10);
1137                 read_next_command();
1138         }
1139         else
1140                 next_mark = 0;
1141 }
1142
1143 static void* cmd_data (size_t *size)
1144 {
1145         size_t n = 0;
1146         void *buffer;
1147         size_t length;
1148
1149         if (strncmp("data ", command_buf.buf, 5))
1150                 die("Expected 'data n' command, found: %s", command_buf.buf);
1151
1152         length = strtoul(command_buf.buf + 5, NULL, 10);
1153         buffer = xmalloc(length);
1154
1155         while (n < length) {
1156                 size_t s = fread((char*)buffer + n, 1, length - n, stdin);
1157                 if (!s && feof(stdin))
1158                         die("EOF in data (%lu bytes remaining)", length - n);
1159                 n += s;
1160         }
1161
1162         if (fgetc(stdin) != '\n')
1163                 die("An lf did not trail the binary data as expected.");
1164
1165         *size = length;
1166         return buffer;
1167 }
1168
1169 static void cmd_new_blob()
1170 {
1171         size_t l;
1172         void *d;
1173
1174         read_next_command();
1175         cmd_mark();
1176         d = cmd_data(&l);
1177
1178         if (store_object(OBJ_BLOB, d, l, &last_blob, NULL, next_mark))
1179                 free(d);
1180 }
1181
1182 static void unload_one_branch()
1183 {
1184         while (cur_active_branches
1185                 && cur_active_branches >= max_active_branches) {
1186                 unsigned long min_commit = ULONG_MAX;
1187                 struct branch *e, *l = NULL, *p = NULL;
1188
1189                 for (e = active_branches; e; e = e->active_next_branch) {
1190                         if (e->last_commit < min_commit) {
1191                                 p = l;
1192                                 min_commit = e->last_commit;
1193                         }
1194                         l = e;
1195                 }
1196
1197                 if (p) {
1198                         e = p->active_next_branch;
1199                         p->active_next_branch = e->active_next_branch;
1200                 } else {
1201                         e = active_branches;
1202                         active_branches = e->active_next_branch;
1203                 }
1204                 e->active_next_branch = NULL;
1205                 if (e->branch_tree.tree) {
1206                         release_tree_content_recursive(e->branch_tree.tree);
1207                         e->branch_tree.tree = NULL;
1208                 }
1209                 cur_active_branches--;
1210         }
1211 }
1212
1213 static void load_branch(struct branch *b)
1214 {
1215         load_tree(&b->branch_tree);
1216         b->active_next_branch = active_branches;
1217         active_branches = b;
1218         cur_active_branches++;
1219         branch_load_count++;
1220 }
1221
1222 static void file_change_m(struct branch *b)
1223 {
1224         const char *p = command_buf.buf + 2;
1225         char *p_uq;
1226         const char *endp;
1227         struct object_entry *oe;
1228         unsigned char sha1[20];
1229         unsigned int mode;
1230         char type[20];
1231
1232         p = get_mode(p, &mode);
1233         if (!p)
1234                 die("Corrupt mode: %s", command_buf.buf);
1235         switch (mode) {
1236         case S_IFREG | 0644:
1237         case S_IFREG | 0755:
1238         case S_IFLNK:
1239         case 0644:
1240         case 0755:
1241                 /* ok */
1242                 break;
1243         default:
1244                 die("Corrupt mode: %s", command_buf.buf);
1245         }
1246
1247         if (*p == ':') {
1248                 char *x;
1249                 oe = find_mark(strtoul(p + 1, &x, 10));
1250                 p = x;
1251         } else {
1252                 if (get_sha1_hex(p, sha1))
1253                         die("Invalid SHA1: %s", command_buf.buf);
1254                 oe = find_object(sha1);
1255                 p += 40;
1256         }
1257         if (*p++ != ' ')
1258                 die("Missing space after SHA1: %s", command_buf.buf);
1259
1260         p_uq = unquote_c_style(p, &endp);
1261         if (p_uq) {
1262                 if (*endp)
1263                         die("Garbage after path in: %s", command_buf.buf);
1264                 p = p_uq;
1265         }
1266
1267         if (oe) {
1268                 if (oe->type != OBJ_BLOB)
1269                         die("Not a blob (actually a %s): %s",
1270                                 command_buf.buf, type_names[oe->type]);
1271         } else {
1272                 if (sha1_object_info(sha1, type, NULL))
1273                         die("Blob not found: %s", command_buf.buf);
1274                 if (strcmp(blob_type, type))
1275                         die("Not a blob (actually a %s): %s",
1276                                 command_buf.buf, type);
1277         }
1278
1279         tree_content_set(&b->branch_tree, p, sha1, S_IFREG | mode);
1280
1281         if (p_uq)
1282                 free(p_uq);
1283 }
1284
1285 static void file_change_d(struct branch *b)
1286 {
1287         const char *p = command_buf.buf + 2;
1288         char *p_uq;
1289         const char *endp;
1290
1291         p_uq = unquote_c_style(p, &endp);
1292         if (p_uq) {
1293                 if (*endp)
1294                         die("Garbage after path in: %s", command_buf.buf);
1295                 p = p_uq;
1296         }
1297         tree_content_remove(&b->branch_tree, p);
1298         if (p_uq)
1299                 free(p_uq);
1300 }
1301
1302 static void cmd_new_commit()
1303 {
1304         struct branch *b;
1305         void *msg;
1306         size_t msglen;
1307         char *str_uq;
1308         const char *endp;
1309         char *sp;
1310         char *author = NULL;
1311         char *committer = NULL;
1312         char *body;
1313
1314         /* Obtain the branch name from the rest of our command */
1315         sp = strchr(command_buf.buf, ' ') + 1;
1316         str_uq = unquote_c_style(sp, &endp);
1317         if (str_uq) {
1318                 if (*endp)
1319                         die("Garbage after ref in: %s", command_buf.buf);
1320                 sp = str_uq;
1321         }
1322         b = lookup_branch(sp);
1323         if (!b)
1324                 die("Branch not declared: %s", sp);
1325         if (str_uq)
1326                 free(str_uq);
1327
1328         read_next_command();
1329         cmd_mark();
1330         if (!strncmp("author ", command_buf.buf, 7)) {
1331                 author = strdup(command_buf.buf);
1332                 read_next_command();
1333         }
1334         if (!strncmp("committer ", command_buf.buf, 10)) {
1335                 committer = strdup(command_buf.buf);
1336                 read_next_command();
1337         }
1338         if (!committer)
1339                 die("Expected committer but didn't get one");
1340         msg = cmd_data(&msglen);
1341
1342         /* ensure the branch is active/loaded */
1343         if (!b->branch_tree.tree || !max_active_branches) {
1344                 unload_one_branch();
1345                 load_branch(b);
1346         }
1347
1348         /* file_change* */
1349         for (;;) {
1350                 read_next_command();
1351                 if (1 == command_buf.len)
1352                         break;
1353                 else if (!strncmp("M ", command_buf.buf, 2))
1354                         file_change_m(b);
1355                 else if (!strncmp("D ", command_buf.buf, 2))
1356                         file_change_d(b);
1357                 else
1358                         die("Unsupported file_change: %s", command_buf.buf);
1359         }
1360
1361         /* build the tree and the commit */
1362         store_tree(&b->branch_tree);
1363         body = xmalloc(97 + msglen
1364                 + (author
1365                         ? strlen(author) + strlen(committer)
1366                         : 2 * strlen(committer)));
1367         sp = body;
1368         sp += sprintf(sp, "tree %s\n", sha1_to_hex(b->branch_tree.sha1));
1369         if (memcmp(b->sha1, null_sha1, 20))
1370                 sp += sprintf(sp, "parent %s\n", sha1_to_hex(b->sha1));
1371         if (author)
1372                 sp += sprintf(sp, "%s\n", author);
1373         else
1374                 sp += sprintf(sp, "author %s\n", committer + 10);
1375         sp += sprintf(sp, "%s\n\n", committer);
1376         memcpy(sp, msg, msglen);
1377         sp += msglen;
1378         if (author)
1379                 free(author);
1380         free(committer);
1381         free(msg);
1382
1383         store_object(OBJ_COMMIT, body, sp - body, NULL, b->sha1, next_mark);
1384         free(body);
1385         b->last_commit = object_count_by_type[OBJ_COMMIT];
1386 }
1387
1388 static void cmd_new_branch()
1389 {
1390         struct branch *b;
1391         char *str_uq;
1392         const char *endp;
1393         char *sp;
1394
1395         /* Obtain the new branch name from the rest of our command */
1396         sp = strchr(command_buf.buf, ' ') + 1;
1397         str_uq = unquote_c_style(sp, &endp);
1398         if (str_uq) {
1399                 if (*endp)
1400                         die("Garbage after ref in: %s", command_buf.buf);
1401                 sp = str_uq;
1402         }
1403         b = new_branch(sp);
1404         if (str_uq)
1405                 free(str_uq);
1406         read_next_command();
1407
1408         /* from ... */
1409         if (!strncmp("from ", command_buf.buf, 5)) {
1410                 const char *from;
1411                 struct branch *s;
1412
1413                 from = strchr(command_buf.buf, ' ') + 1;
1414                 str_uq = unquote_c_style(from, &endp);
1415                 if (str_uq) {
1416                         if (*endp)
1417                                 die("Garbage after string in: %s", command_buf.buf);
1418                         from = str_uq;
1419                 }
1420
1421                 s = lookup_branch(from);
1422                 if (b == s)
1423                         die("Can't create a branch from itself: %s", b->name);
1424                 else if (s) {
1425                         memcpy(b->sha1, s->sha1, 20);
1426                         memcpy(b->branch_tree.sha1, s->branch_tree.sha1, 20);
1427                 } else if (*from == ':') {
1428                         unsigned long idnum = strtoul(from + 1, NULL, 10);
1429                         struct object_entry *oe = find_mark(idnum);
1430                         unsigned long size;
1431                         char *buf;
1432                         if (oe->type != OBJ_COMMIT)
1433                                 die("Mark :%lu not a commit", idnum);
1434                         memcpy(b->sha1, oe->sha1, 20);
1435                         buf = unpack_entry(oe->offset, &size);
1436                         if (!buf || size < 46)
1437                                 die("Not a valid commit: %s", from);
1438                         if (memcmp("tree ", buf, 5)
1439                                 || get_sha1_hex(buf + 5, b->branch_tree.sha1))
1440                                 die("The commit %s is corrupt", sha1_to_hex(b->sha1));
1441                         free(buf);
1442                 } else if (!get_sha1(from, b->sha1)) {
1443                         if (!memcmp(b->sha1, null_sha1, 20))
1444                                 memcpy(b->branch_tree.sha1, null_sha1, 20);
1445                         else {
1446                                 unsigned long size;
1447                                 char *buf;
1448
1449                                 buf = read_object_with_reference(b->sha1,
1450                                         type_names[OBJ_COMMIT], &size, b->sha1);
1451                                 if (!buf || size < 46)
1452                                         die("Not a valid commit: %s", from);
1453                                 if (memcmp("tree ", buf, 5)
1454                                         || get_sha1_hex(buf + 5, b->branch_tree.sha1))
1455                                         die("The commit %s is corrupt", sha1_to_hex(b->sha1));
1456                                 free(buf);
1457                         }
1458                 } else
1459                         die("Invalid ref name or SHA1 expression: %s", from);
1460
1461                 if (str_uq)
1462                         free(str_uq);
1463                 read_next_command();
1464         } else {
1465                 memcpy(b->sha1, null_sha1, 20);
1466                 memcpy(b->branch_tree.sha1, null_sha1, 20);
1467         }
1468
1469         if (command_buf.eof || command_buf.len > 1)
1470                 die("An lf did not terminate the branch command as expected.");
1471 }
1472
1473 static void cmd_new_tag()
1474 {
1475         char *str_uq;
1476         const char *endp;
1477         char *sp;
1478         const char *from;
1479         char *tagger;
1480         struct branch *s;
1481         void *msg;
1482         size_t msglen;
1483         char *body;
1484         struct tag *t;
1485         unsigned char sha1[20];
1486
1487         /* Obtain the new tag name from the rest of our command */
1488         sp = strchr(command_buf.buf, ' ') + 1;
1489         str_uq = unquote_c_style(sp, &endp);
1490         if (str_uq) {
1491                 if (*endp)
1492                         die("Garbage after tag name in: %s", command_buf.buf);
1493                 sp = str_uq;
1494         }
1495         t = pool_alloc(sizeof(struct tag));
1496         t->next_tag = NULL;
1497         t->name = pool_strdup(sp);
1498         if (last_tag)
1499                 last_tag->next_tag = t;
1500         else
1501                 first_tag = t;
1502         last_tag = t;
1503         if (str_uq)
1504                 free(str_uq);
1505         read_next_command();
1506
1507         /* from ... */
1508         if (strncmp("from ", command_buf.buf, 5))
1509                 die("Expected from command, got %s", command_buf.buf);
1510
1511         from = strchr(command_buf.buf, ' ') + 1;
1512         str_uq = unquote_c_style(from, &endp);
1513         if (str_uq) {
1514                 if (*endp)
1515                         die("Garbage after string in: %s", command_buf.buf);
1516                 from = str_uq;
1517         }
1518
1519         s = lookup_branch(from);
1520         if (s) {
1521                 memcpy(sha1, s->sha1, 20);
1522         } else if (*from == ':') {
1523                 unsigned long idnum = strtoul(from + 1, NULL, 10);
1524                 struct object_entry *oe = find_mark(idnum);
1525                 if (oe->type != OBJ_COMMIT)
1526                         die("Mark :%lu not a commit", idnum);
1527                 memcpy(sha1, oe->sha1, 20);
1528         } else if (!get_sha1(from, sha1)) {
1529                 unsigned long size;
1530                 char *buf;
1531
1532                 buf = read_object_with_reference(sha1,
1533                         type_names[OBJ_COMMIT], &size, sha1);
1534                 if (!buf || size < 46)
1535                         die("Not a valid commit: %s", from);
1536                 free(buf);
1537         } else
1538                 die("Invalid ref name or SHA1 expression: %s", from);
1539
1540         if (str_uq)
1541                 free(str_uq);
1542         read_next_command();
1543
1544         /* tagger ... */
1545         if (strncmp("tagger ", command_buf.buf, 7))
1546                 die("Expected tagger command, got %s", command_buf.buf);
1547         tagger = strdup(command_buf.buf);
1548
1549         /* tag payload/message */
1550         read_next_command();
1551         msg = cmd_data(&msglen);
1552
1553         /* build the tag object */
1554         body = xmalloc(67 + strlen(t->name) + strlen(tagger) + msglen);
1555         sp = body;
1556         sp += sprintf(sp, "object %s\n", sha1_to_hex(sha1));
1557         sp += sprintf(sp, "type %s\n", type_names[OBJ_COMMIT]);
1558         sp += sprintf(sp, "tag %s\n", t->name);
1559         sp += sprintf(sp, "%s\n\n", tagger);
1560         memcpy(sp, msg, msglen);
1561         sp += msglen;
1562         free(tagger);
1563         free(msg);
1564
1565         store_object(OBJ_TAG, body, sp - body, NULL, t->sha1, 0);
1566         free(body);
1567 }
1568
1569 static const char fast_import_usage[] =
1570 "git-fast-import [--objects=n] [--depth=n] [--active-branches=n] temp.pack";
1571
1572 int main(int argc, const char **argv)
1573 {
1574         const char *base_name;
1575         int i;
1576         unsigned long est_obj_cnt = 1000;
1577         char *pack_name;
1578         char *idx_name;
1579         struct stat sb;
1580
1581         setup_ident();
1582         git_config(git_default_config);
1583
1584         for (i = 1; i < argc; i++) {
1585                 const char *a = argv[i];
1586
1587                 if (*a != '-' || !strcmp(a, "--"))
1588                         break;
1589                 else if (!strncmp(a, "--objects=", 10))
1590                         est_obj_cnt = strtoul(a + 10, NULL, 0);
1591                 else if (!strncmp(a, "--depth=", 8))
1592                         max_depth = strtoul(a + 8, NULL, 0);
1593                 else if (!strncmp(a, "--active-branches=", 18))
1594                         max_active_branches = strtoul(a + 18, NULL, 0);
1595                 else
1596                         die("unknown option %s", a);
1597         }
1598         if ((i+1) != argc)
1599                 usage(fast_import_usage);
1600         base_name = argv[i];
1601
1602         pack_name = xmalloc(strlen(base_name) + 6);
1603         sprintf(pack_name, "%s.pack", base_name);
1604         idx_name = xmalloc(strlen(base_name) + 5);
1605         sprintf(idx_name, "%s.idx", base_name);
1606
1607         pack_fd = open(pack_name, O_RDWR|O_CREAT|O_EXCL, 0666);
1608         if (pack_fd < 0)
1609                 die("Can't create %s: %s", pack_name, strerror(errno));
1610
1611         init_pack_header();
1612         alloc_objects(est_obj_cnt);
1613         strbuf_init(&command_buf);
1614
1615         atom_table = xcalloc(atom_table_sz, sizeof(struct atom_str*));
1616         branch_table = xcalloc(branch_table_sz, sizeof(struct branch*));
1617         avail_tree_table = xcalloc(avail_tree_table_sz, sizeof(struct avail_tree_content*));
1618         marks = pool_calloc(1, sizeof(struct mark_set));
1619
1620         for (;;) {
1621                 read_next_command();
1622                 if (command_buf.eof)
1623                         break;
1624                 else if (!strcmp("blob", command_buf.buf))
1625                         cmd_new_blob();
1626                 else if (!strncmp("branch ", command_buf.buf, 7))
1627                         cmd_new_branch();
1628                 else if (!strncmp("commit ", command_buf.buf, 7))
1629                         cmd_new_commit();
1630                 else if (!strncmp("tag ", command_buf.buf, 4))
1631                         cmd_new_tag();
1632                 else
1633                         die("Unsupported command: %s", command_buf.buf);
1634         }
1635
1636         fixup_header_footer();
1637         close(pack_fd);
1638         write_index(idx_name);
1639         dump_branches();
1640         dump_tags();
1641
1642         fprintf(stderr, "%s statistics:\n", argv[0]);
1643         fprintf(stderr, "---------------------------------------------------\n");
1644         fprintf(stderr, "Alloc'd objects: %10lu (%10lu overflow  )\n", alloc_count, alloc_count - est_obj_cnt);
1645         fprintf(stderr, "Total objects:   %10lu (%10lu duplicates)\n", object_count, duplicate_count);
1646         fprintf(stderr, "      blobs  :   %10lu (%10lu duplicates)\n", object_count_by_type[OBJ_BLOB], duplicate_count_by_type[OBJ_BLOB]);
1647         fprintf(stderr, "      trees  :   %10lu (%10lu duplicates)\n", object_count_by_type[OBJ_TREE], duplicate_count_by_type[OBJ_TREE]);
1648         fprintf(stderr, "      commits:   %10lu (%10lu duplicates)\n", object_count_by_type[OBJ_COMMIT], duplicate_count_by_type[OBJ_COMMIT]);
1649         fprintf(stderr, "      tags   :   %10lu (%10lu duplicates)\n", object_count_by_type[OBJ_TAG], duplicate_count_by_type[OBJ_TAG]);
1650         fprintf(stderr, "Total branches:  %10lu (%10lu loads     )\n", branch_count, branch_load_count);
1651         fprintf(stderr, "      marks:     %10u (%10lu unique    )\n", (1 << marks->shift) * 1024, marks_set_count);
1652         fprintf(stderr, "      atoms:     %10u\n", atom_cnt);
1653         fprintf(stderr, "Memory total:    %10lu KiB\n", (total_allocd + alloc_count*sizeof(struct object_entry))/1024);
1654         fprintf(stderr, "       pools:    %10lu KiB\n", total_allocd/1024);
1655         fprintf(stderr, "     objects:    %10lu KiB\n", (alloc_count*sizeof(struct object_entry))/1024);
1656         fprintf(stderr, "Pack remaps:     %10lu\n", remap_count);
1657         fprintf(stderr, "---------------------------------------------------\n");
1658
1659         stat(pack_name, &sb);
1660         fprintf(stderr, "Pack size:       %10lu KiB\n", (unsigned long)(sb.st_size/1024));
1661         stat(idx_name, &sb);
1662         fprintf(stderr, "Index size:      %10lu KiB\n", (unsigned long)(sb.st_size/1024));
1663
1664         fprintf(stderr, "\n");
1665
1666         return 0;
1667 }