git.oblomov.eu Git - git/blob - fast-import.c

   1 /*
   2 Format of STDIN stream:
   3
   4   stream ::= cmd*;
   5
   6   cmd ::= new_blob
   7         | new_commit
   8         | new_branch
   9         | new_tag
  10         ;
  11
  12   new_blob ::= 'blob' blob_data;
  13
  14   new_commit ::= 'comt' ref_name author_committer_msg
  15     file_change*
  16     '0';
  17
  18   new_branch ::= 'brch' dst_ref_name src_ref_name;
  19   dst_ref_name ::= ref_name;
  20   src_ref_name ::= ref_name | sha1_exp;
  21
  22   new_tag ::= 'tagg' ref_name tag_name tagger_msg;
  23
  24   file_change ::= 'M' path_name hexsha1
  25                 | 'D' path_name
  26                 ;
  27
  28   author_committer_msg ::= len32
  29     'author' sp name '<' email '>' ts tz lf
  30     'committer' sp name '<' email '>' ts tz lf
  31     lf
  32     binary_data;
  33
  34   tagger_msg ::= len32
  35     'tagger' sp name '<' email '>' ts tz lf
  36     lf
  37     binary_data;
  38
  39   blob_data ::= len32 binary_data; # max len is 2^32-1
  40   path_name ::= len32 path;        # max len is PATH_MAX-1
  41   ref_name  ::= len32 ref;         # max len is PATH_MAX-1
  42   tag_name  ::= len32 tag;         # max len is PATH_MAX-1
  43   sha1_exp  ::= len32 sha1exp;     # max len is PATH_MAX-1
  44
  45   len32 ::= # unsigned 32 bit value, native format;
  46   binary_data ::= # file content, not interpreted;
  47   sp ::= # ASCII space character;
  48   lf ::= # ASCII newline (LF) character;
  49   path ::= # GIT style file path, e.g. "a/b/c";
  50   ref ::= # GIT ref name, e.g. "refs/heads/MOZ_GECKO_EXPERIMENT";
  51   tag ::= # GIT tag name, e.g. "FIREFOX_1_5";
  52   sha1exp ::= # Any valid GIT SHA1 expression;
  53   hexsha1 ::= # SHA1 in hexadecimal format;
  54   name ::= # valid GIT author/committer name;
  55   email ::= # valid GIT author/committer email;
  56   ts ::= # time since the epoch in seconds, ascii decimal;
  57   tz ::= # GIT style timezone;
  58 */
  59
  60 #include "builtin.h"
  61 #include "cache.h"
  62 #include "object.h"
  63 #include "blob.h"
  64 #include "tree.h"
  65 #include "delta.h"
  66 #include "pack.h"
  67 #include "refs.h"
  68 #include "csum-file.h"
  69
  70 struct object_entry
  71 {
  72         struct object_entry *next;
  73         unsigned long offset;
  74         unsigned char sha1[20];
  75 };
  76
  77 struct object_entry_pool
  78 {
  79         struct object_entry_pool *next_pool;
  80         struct object_entry *next_free;
  81         struct object_entry *end;
  82         struct object_entry entries[FLEX_ARRAY]; /* more */
  83 };
  84
  85 struct last_object
  86 {
  87         void *data;
  88         unsigned int len;
  89         unsigned int depth;
  90         unsigned char sha1[20];
  91 };
  92
  93 struct mem_pool
  94 {
  95         struct mem_pool *next_pool;
  96         char *next_free;
  97         char *end;
  98         char space[FLEX_ARRAY]; /* more */
  99 };
 100
 101 struct atom_str
 102 {
 103         struct atom_str *next_atom;
 104         int str_len;
 105         char str_dat[FLEX_ARRAY]; /* more */
 106 };
 107
 108 struct tree_content;
 109 struct tree_entry
 110 {
 111         struct tree_content *tree;
 112         struct atom_str* name;
 113         unsigned int mode;
 114         unsigned char sha1[20];
 115 };
 116
 117 struct tree_content
 118 {
 119         unsigned int entry_capacity; /* must match avail_tree_content */
 120         unsigned int entry_count;
 121         struct tree_entry *entries[FLEX_ARRAY]; /* more */
 122 };
 123
 124 struct avail_tree_content
 125 {
 126         unsigned int entry_capacity; /* must match tree_content */
 127         struct avail_tree_content *next_avail;
 128 };
 129
 130 struct branch
 131 {
 132         struct branch *table_next_branch;
 133         struct branch *active_next_branch;
 134         const char *name;
 135         unsigned long last_commit;
 136         struct tree_entry branch_tree;
 137         unsigned char sha1[20];
 138 };
 139
 140
 141 /* Stats and misc. counters */
 142 static int max_depth = 10;
 143 static unsigned long alloc_count;
 144 static unsigned long branch_count;
 145 static unsigned long object_count;
 146 static unsigned long duplicate_count;
 147 static unsigned long object_count_by_type[9];
 148 static unsigned long duplicate_count_by_type[9];
 149
 150 /* Memory pools */
 151 static size_t mem_pool_alloc = 2*1024*1024 - sizeof(struct mem_pool);
 152 static size_t total_allocd;
 153 static struct mem_pool *mem_pool;
 154
 155 /* atom management */
 156 static unsigned int atom_table_sz = 4451;
 157 static unsigned int atom_cnt;
 158 static struct atom_str **atom_table;
 159
 160 /* The .pack file being generated */
 161 static int pack_fd;
 162 static unsigned long pack_offset;
 163 static unsigned char pack_sha1[20];
 164
 165 /* Table of objects we've written. */
 166 static unsigned int object_entry_alloc = 1000;
 167 static struct object_entry_pool *blocks;
 168 static struct object_entry *object_table[1 << 16];
 169
 170 /* Our last blob */
 171 static struct last_object last_blob;
 172
 173 /* Tree management */
 174 static unsigned int tree_entry_alloc = 1000;
 175 static void *avail_tree_entry;
 176 static unsigned int avail_tree_table_sz = 100;
 177 static struct avail_tree_content **avail_tree_table;
 178
 179 /* Branch data */
 180 static unsigned int max_active_branches = 5;
 181 static unsigned int cur_active_branches;
 182 static unsigned int branch_table_sz = 1039;
 183 static struct branch **branch_table;
 184 static struct branch *active_branches;
 185
 186
 187 static void alloc_objects(int cnt)
 188 {
 189         struct object_entry_pool *b;
 190
 191         b = xmalloc(sizeof(struct object_entry_pool)
 192                 + cnt * sizeof(struct object_entry));
 193         b->next_pool = blocks;
 194         b->next_free = b->entries;
 195         b->end = b->entries + cnt;
 196         blocks = b;
 197         alloc_count += cnt;
 198 }
 199
 200 static struct object_entry* new_object(unsigned char *sha1)
 201 {
 202         struct object_entry *e;
 203
 204         if (blocks->next_free == blocks->end)
 205                 alloc_objects(object_entry_alloc);
 206
 207         e = blocks->next_free++;
 208         memcpy(e->sha1, sha1, sizeof(e->sha1));
 209         return e;
 210 }
 211
 212 static struct object_entry* find_object(unsigned char *sha1)
 213 {
 214         unsigned int h = sha1[0] << 8 | sha1[1];
 215         struct object_entry *e;
 216         for (e = object_table[h]; e; e = e->next)
 217                 if (!memcmp(sha1, e->sha1, sizeof(e->sha1)))
 218                         return e;
 219         return NULL;
 220 }
 221
 222 static struct object_entry* insert_object(unsigned char *sha1)
 223 {
 224         unsigned int h = sha1[0] << 8 | sha1[1];
 225         struct object_entry *e = object_table[h];
 226         struct object_entry *p = NULL;
 227
 228         while (e) {
 229                 if (!memcmp(sha1, e->sha1, sizeof(e->sha1)))
 230                         return e;
 231                 p = e;
 232                 e = e->next;
 233         }
 234
 235         e = new_object(sha1);
 236         e->next = NULL;
 237         e->offset = 0;
 238         if (p)
 239                 p->next = e;
 240         else
 241                 object_table[h] = e;
 242         return e;
 243 }
 244
 245 static unsigned int hc_str(const char *s, size_t len)
 246 {
 247         unsigned int r = 0;
 248         while (len-- > 0)
 249                 r = r * 31 + *s++;
 250         return r;
 251 }
 252
 253 static void* pool_alloc(size_t len)
 254 {
 255         struct mem_pool *p;
 256         void *r;
 257
 258         for (p = mem_pool; p; p = p->next_pool)
 259                 if ((p->end - p->next_free >= len))
 260                         break;
 261
 262         if (!p) {
 263                 if (len >= (mem_pool_alloc/2)) {
 264                         total_allocd += len;
 265                         return xmalloc(len);
 266                 }
 267                 total_allocd += sizeof(struct mem_pool) + mem_pool_alloc;
 268                 p = xmalloc(sizeof(struct mem_pool) + mem_pool_alloc);
 269                 p->next_pool = mem_pool;
 270                 p->next_free = p->space;
 271                 p->end = p->next_free + mem_pool_alloc;
 272                 mem_pool = p;
 273         }
 274
 275         r = p->next_free;
 276         p->next_free += len;
 277         return r;
 278 }
 279
 280 static void* pool_calloc(size_t count, size_t size)
 281 {
 282         size_t len = count * size;
 283         void *r = pool_alloc(len);
 284         memset(r, 0, len);
 285         return r;
 286 }
 287
 288 static char* pool_strdup(const char *s)
 289 {
 290         char *r = pool_alloc(strlen(s) + 1);
 291         strcpy(r, s);
 292         return r;
 293 }
 294
 295 static struct atom_str* to_atom(const char *s, size_t len)
 296 {
 297         unsigned int hc = hc_str(s, len) % atom_table_sz;
 298         struct atom_str *c;
 299
 300         for (c = atom_table[hc]; c; c = c->next_atom)
 301                 if (c->str_len == len && !strncmp(s, c->str_dat, len))
 302                         return c;
 303
 304         c = pool_alloc(sizeof(struct atom_str) + len + 1);
 305         c->str_len = len;
 306         strncpy(c->str_dat, s, len);
 307         c->str_dat[len] = 0;
 308         c->next_atom = atom_table[hc];
 309         atom_table[hc] = c;
 310         atom_cnt++;
 311         return c;
 312 }
 313
 314 static struct branch* lookup_branch(const char *name)
 315 {
 316         unsigned int hc = hc_str(name, strlen(name)) % branch_table_sz;
 317         struct branch *b;
 318
 319         for (b = branch_table[hc]; b; b = b->table_next_branch)
 320                 if (!strcmp(name, b->name))
 321                         return b;
 322         return NULL;
 323 }
 324
 325 static struct branch* new_branch(const char *name)
 326 {
 327         unsigned int hc = hc_str(name, strlen(name)) % branch_table_sz;
 328         struct branch* b = lookup_branch(name);
 329
 330         if (b)
 331                 die("Invalid attempt to create duplicate branch: %s", name);
 332
 333         b = pool_calloc(1, sizeof(struct branch));
 334         b->name = pool_strdup(name);
 335         b->table_next_branch = branch_table[hc];
 336         branch_table[hc] = b;
 337         branch_count++;
 338         return b;
 339 }
 340
 341 static unsigned int hc_entries(unsigned int cnt)
 342 {
 343         cnt = cnt & 7 ? (cnt / 8) + 1 : cnt / 8;
 344         return cnt < avail_tree_table_sz ? cnt : avail_tree_table_sz - 1;
 345 }
 346
 347 static struct tree_content* new_tree_content(unsigned int cnt)
 348 {
 349         struct avail_tree_content *f, *l = NULL;
 350         struct tree_content *t;
 351         unsigned int hc = hc_entries(cnt);
 352
 353         for (f = avail_tree_table[hc]; f; l = f, f = f->next_avail)
 354                 if (f->entry_capacity >= cnt)
 355                         break;
 356
 357         if (f) {
 358                 if (l)
 359                         l->next_avail = f->next_avail;
 360                 else
 361                         avail_tree_table[hc] = f->next_avail;
 362         } else {
 363                 cnt = cnt & 7 ? ((cnt / 8) + 1) * 8 : cnt;
 364                 f = pool_alloc(sizeof(*t) + sizeof(t->entries[0]) * cnt);
 365                 f->entry_capacity = cnt;
 366         }
 367
 368         t = (struct tree_content*)f;
 369         t->entry_count = 0;
 370         return t;
 371 }
 372
 373 static void release_tree_entry(struct tree_entry *e);
 374 static void release_tree_content(struct tree_content *t)
 375 {
 376         struct avail_tree_content *f = (struct avail_tree_content*)t;
 377         unsigned int hc = hc_entries(f->entry_capacity);
 378         unsigned int i;
 379         for (i = 0; i < t->entry_count; i++)
 380                 release_tree_entry(t->entries[i]);
 381         f->next_avail = avail_tree_table[hc];
 382         avail_tree_table[hc] = f;
 383 }
 384
 385 static struct tree_content* grow_tree_content(
 386         struct tree_content *t,
 387         int amt)
 388 {
 389         struct tree_content *r = new_tree_content(t->entry_count + amt);
 390         r->entry_count = t->entry_count;
 391         memcpy(r->entries,t->entries,t->entry_count*sizeof(t->entries[0]));
 392         release_tree_content(t);
 393         return r;
 394 }
 395
 396 static struct tree_entry* new_tree_entry()
 397 {
 398         struct tree_entry *e;
 399
 400         if (!avail_tree_entry) {
 401                 unsigned int n = tree_entry_alloc;
 402                 avail_tree_entry = e = xmalloc(n * sizeof(struct tree_entry));
 403                 while (n--) {
 404                         *((void**)e) = e + 1;
 405                         e++;
 406                 }
 407         }
 408
 409         e = avail_tree_entry;
 410         avail_tree_entry = *((void**)e);
 411         return e;
 412 }
 413
 414 static void release_tree_entry(struct tree_entry *e)
 415 {
 416         if (e->tree)
 417                 release_tree_content(e->tree);
 418         *((void**)e) = avail_tree_entry;
 419         avail_tree_entry = e;
 420 }
 421
 422 static void yread(int fd, void *buffer, size_t length)
 423 {
 424         ssize_t ret = 0;
 425         while (ret < length) {
 426                 ssize_t size = xread(fd, (char *) buffer + ret, length - ret);
 427                 if (!size)
 428                         die("Read from descriptor %i: end of stream", fd);
 429                 if (size < 0)
 430                         die("Read from descriptor %i: %s", fd, strerror(errno));
 431                 ret += size;
 432         }
 433 }
 434
 435 static int optional_read(int fd, void *buffer, size_t length)
 436 {
 437         ssize_t ret = 0;
 438         while (ret < length) {
 439                 ssize_t size = xread(fd, (char *) buffer + ret, length - ret);
 440                 if (!size && !ret)
 441                         return 1;
 442                 if (!size)
 443                         die("Read from descriptor %i: end of stream", fd);
 444                 if (size < 0)
 445                         die("Read from descriptor %i: %s", fd, strerror(errno));
 446                 ret += size;
 447         }
 448         return 0;
 449 }
 450
 451 static void ywrite(int fd, void *buffer, size_t length)
 452 {
 453         ssize_t ret = 0;
 454         while (ret < length) {
 455                 ssize_t size = xwrite(fd, (char *) buffer + ret, length - ret);
 456                 if (!size)
 457                         die("Write to descriptor %i: end of file", fd);
 458                 if (size < 0)
 459                         die("Write to descriptor %i: %s", fd, strerror(errno));
 460                 ret += size;
 461         }
 462 }
 463
 464 static const char* read_path()
 465 {
 466         static char sn[PATH_MAX];
 467         unsigned long slen;
 468
 469         yread(0, &slen, 4);
 470         if (!slen)
 471                 die("Expected string command parameter, didn't find one");
 472         if (slen > (PATH_MAX - 1))
 473                 die("Can't handle excessive string length %lu", slen);
 474         yread(0, sn, slen);
 475         sn[slen] = 0;
 476         return sn;
 477 }
 478
 479 static unsigned long encode_header(
 480         enum object_type type,
 481         unsigned long size,
 482         unsigned char *hdr)
 483 {
 484         int n = 1;
 485         unsigned char c;
 486
 487         if (type < OBJ_COMMIT || type > OBJ_DELTA)
 488                 die("bad type %d", type);
 489
 490         c = (type << 4) | (size & 15);
 491         size >>= 4;
 492         while (size) {
 493                 *hdr++ = c | 0x80;
 494                 c = size & 0x7f;
 495                 size >>= 7;
 496                 n++;
 497         }
 498         *hdr = c;
 499         return n;
 500 }
 501
 502 static int store_object(
 503         enum object_type type,
 504         void *dat,
 505         unsigned long datlen,
 506         struct last_object *last,
 507         unsigned char *sha1out)
 508 {
 509         void *out, *delta;
 510         struct object_entry *e;
 511         unsigned char hdr[96];
 512         unsigned char sha1[20];
 513         unsigned long hdrlen, deltalen;
 514         SHA_CTX c;
 515         z_stream s;
 516
 517         hdrlen = sprintf((char*)hdr,"%s %lu",type_names[type],datlen) + 1;
 518         SHA1_Init(&c);
 519         SHA1_Update(&c, hdr, hdrlen);
 520         SHA1_Update(&c, dat, datlen);
 521         SHA1_Final(sha1, &c);
 522         if (sha1out)
 523                 memcpy(sha1out, sha1, sizeof(sha1));
 524
 525         e = insert_object(sha1);
 526         if (e->offset) {
 527                 duplicate_count++;
 528                 duplicate_count_by_type[type]++;
 529                 return 1;
 530         }
 531         e->offset = pack_offset;
 532         object_count++;
 533         object_count_by_type[type]++;
 534
 535         if (last && last->data && last->depth < max_depth)
 536                 delta = diff_delta(last->data, last->len,
 537                         dat, datlen,
 538                         &deltalen, 0);
 539         else
 540                 delta = 0;
 541
 542         memset(&s, 0, sizeof(s));
 543         deflateInit(&s, zlib_compression_level);
 544
 545         if (delta) {
 546                 last->depth++;
 547                 s.next_in = delta;
 548                 s.avail_in = deltalen;
 549                 hdrlen = encode_header(OBJ_DELTA, deltalen, hdr);
 550                 ywrite(pack_fd, hdr, hdrlen);
 551                 ywrite(pack_fd, last->sha1, sizeof(sha1));
 552                 pack_offset += hdrlen + sizeof(sha1);
 553         } else {
 554                 if (last)
 555                         last->depth = 0;
 556                 s.next_in = dat;
 557                 s.avail_in = datlen;
 558                 hdrlen = encode_header(type, datlen, hdr);
 559                 ywrite(pack_fd, hdr, hdrlen);
 560                 pack_offset += hdrlen;
 561         }
 562
 563         s.avail_out = deflateBound(&s, s.avail_in);
 564         s.next_out = out = xmalloc(s.avail_out);
 565         while (deflate(&s, Z_FINISH) == Z_OK)
 566                 /* nothing */;
 567         deflateEnd(&s);
 568
 569         ywrite(pack_fd, out, s.total_out);
 570         pack_offset += s.total_out;
 571
 572         free(out);
 573         if (delta)
 574                 free(delta);
 575         if (last) {
 576                 if (last->data)
 577                         free(last->data);
 578                 last->data = dat;
 579                 last->len = datlen;
 580                 memcpy(last->sha1, sha1, sizeof(sha1));
 581         }
 582         return 0;
 583 }
 584
 585 static const char *get_mode(const char *str, unsigned int *modep)
 586 {
 587         unsigned char c;
 588         unsigned int mode = 0;
 589
 590         while ((c = *str++) != ' ') {
 591                 if (c < '0' || c > '7')
 592                         return NULL;
 593                 mode = (mode << 3) + (c - '0');
 594         }
 595         *modep = mode;
 596         return str;
 597 }
 598
 599 static void load_tree(struct tree_entry *root)
 600 {
 601         struct object_entry *myoe;
 602         struct tree_content *t;
 603         unsigned long size;
 604         char *buf;
 605         const char *c;
 606         char type[20];
 607
 608         root->tree = t = new_tree_content(8);
 609         if (!memcmp(root->sha1, null_sha1, 20))
 610                 return;
 611
 612         myoe = find_object(root->sha1);
 613         if (myoe) {
 614                 die("FIXME");
 615         } else {
 616                 buf = read_sha1_file(root->sha1, type, &size);
 617                 if (!buf || strcmp(type, tree_type))
 618                         die("Can't load existing tree %s", sha1_to_hex(root->sha1));
 619         }
 620
 621         c = buf;
 622         while (c != (buf + size)) {
 623                 struct tree_entry *e = new_tree_entry();
 624
 625                 if (t->entry_count == t->entry_capacity)
 626                         root->tree = t = grow_tree_content(t, 8);
 627                 t->entries[t->entry_count++] = e;
 628
 629                 e->tree = NULL;
 630                 c = get_mode(c, &e->mode);
 631                 if (!c)
 632                         die("Corrupt mode in %s", sha1_to_hex(root->sha1));
 633                 e->name = to_atom(c, strlen(c));
 634                 c += e->name->str_len + 1;
 635                 memcpy(e->sha1, c, sizeof(e->sha1));
 636                 c += 20;
 637         }
 638         free(buf);
 639 }
 640
 641 static int tecmp (const void *_a, const void *_b)
 642 {
 643         struct tree_entry *a = *((struct tree_entry**)_a);
 644         struct tree_entry *b = *((struct tree_entry**)_b);
 645         return base_name_compare(
 646                 a->name->str_dat, a->name->str_len, a->mode,
 647                 b->name->str_dat, b->name->str_len, b->mode);
 648 }
 649
 650 static void store_tree(struct tree_entry *root)
 651 {
 652         struct tree_content *t = root->tree;
 653         unsigned int i;
 654         size_t maxlen;
 655         char *buf, *c;
 656
 657         if (memcmp(root->sha1, null_sha1, 20))
 658                 return;
 659
 660         maxlen = 0;
 661         for (i = 0; i < t->entry_count; i++) {
 662                 maxlen += t->entries[i]->name->str_len + 34;
 663                 if (t->entries[i]->tree)
 664                         store_tree(t->entries[i]);
 665         }
 666
 667         qsort(t->entries, t->entry_count, sizeof(t->entries[0]), tecmp);
 668         buf = c = xmalloc(maxlen);
 669         for (i = 0; i < t->entry_count; i++) {
 670                 struct tree_entry *e = t->entries[i];
 671                 c += sprintf(c, "%o", e->mode);
 672                 *c++ = ' ';
 673                 strcpy(c, e->name->str_dat);
 674                 c += e->name->str_len + 1;
 675                 memcpy(c, e->sha1, 20);
 676                 c += 20;
 677         }
 678         store_object(OBJ_TREE, buf, c - buf, NULL, root->sha1);
 679         free(buf);
 680 }
 681
 682 static int tree_content_set(
 683         struct tree_entry *root,
 684         const char *p,
 685         const unsigned char *sha1,
 686         const unsigned int mode)
 687 {
 688         struct tree_content *t = root->tree;
 689         const char *slash1;
 690         unsigned int i, n;
 691         struct tree_entry *e;
 692
 693         slash1 = strchr(p, '/');
 694         if (slash1)
 695                 n = slash1 - p;
 696         else
 697                 n = strlen(p);
 698
 699         for (i = 0; i < t->entry_count; i++) {
 700                 e = t->entries[i];
 701                 if (e->name->str_len == n && !strncmp(p, e->name->str_dat, n)) {
 702                         if (!slash1) {
 703                                 if (e->mode == mode && !memcmp(e->sha1, sha1, 20))
 704                                         return 0;
 705                                 e->mode = mode;
 706                                 memcpy(e->sha1, sha1, 20);
 707                                 if (e->tree) {
 708                                         release_tree_content(e->tree);
 709                                         e->tree = NULL;
 710                                 }
 711                                 memcpy(root->sha1, null_sha1, 20);
 712                                 return 1;
 713                         }
 714                         if (!S_ISDIR(e->mode)) {
 715                                 e->tree = new_tree_content(8);
 716                                 e->mode = 040000;
 717                         }
 718                         if (!e->tree)
 719                                 load_tree(e);
 720                         if (tree_content_set(e, slash1 + 1, sha1, mode)) {
 721                                 memcpy(root->sha1, null_sha1, 20);
 722                                 return 1;
 723                         }
 724                         return 0;
 725                 }
 726         }
 727
 728         if (t->entry_count == t->entry_capacity)
 729                 root->tree = t = grow_tree_content(t, 8);
 730         e = new_tree_entry();
 731         e->name = to_atom(p, n);
 732         t->entries[t->entry_count++] = e;
 733         if (slash1) {
 734                 e->tree = new_tree_content(8);
 735                 e->mode = 040000;
 736                 tree_content_set(e, slash1 + 1, sha1, mode);
 737         } else {
 738                 e->tree = NULL;
 739                 e->mode = mode;
 740                 memcpy(e->sha1, sha1, 20);
 741         }
 742         memcpy(root->sha1, null_sha1, 20);
 743         return 1;
 744 }
 745
 746 static int tree_content_remove(struct tree_entry *root, const char *p)
 747 {
 748         struct tree_content *t = root->tree;
 749         const char *slash1;
 750         unsigned int i, n;
 751         struct tree_entry *e;
 752
 753         slash1 = strchr(p, '/');
 754         if (slash1)
 755                 n = slash1 - p;
 756         else
 757                 n = strlen(p);
 758
 759         for (i = 0; i < t->entry_count; i++) {
 760                 e = t->entries[i];
 761                 if (e->name->str_len == n && !strncmp(p, e->name->str_dat, n)) {
 762                         if (!slash1 || !S_ISDIR(e->mode))
 763                                 goto del_entry;
 764                         if (!e->tree)
 765                                 load_tree(e);
 766                         if (tree_content_remove(e, slash1 + 1)) {
 767                                 if (!e->tree->entry_count)
 768                                         goto del_entry;
 769                                 memcpy(root->sha1, null_sha1, 20);
 770                                 return 1;
 771                         }
 772                         return 0;
 773                 }
 774         }
 775         return 0;
 776
 777 del_entry:
 778         for (i++; i < t->entry_count; i++)
 779                 t->entries[i-1] = t->entries[i];
 780         t->entry_count--;
 781         release_tree_entry(e);
 782         memcpy(root->sha1, null_sha1, 20);
 783         return 1;
 784 }
 785
 786 static void init_pack_header()
 787 {
 788         const char* magic = "PACK";
 789         unsigned long version = 3;
 790         unsigned long zero = 0;
 791
 792         version = htonl(version);
 793         ywrite(pack_fd, (char*)magic, 4);
 794         ywrite(pack_fd, &version, 4);
 795         ywrite(pack_fd, &zero, 4);
 796         pack_offset = 4 * 3;
 797 }
 798
 799 static void fixup_header_footer()
 800 {
 801         SHA_CTX c;
 802         char hdr[8];
 803         unsigned long cnt;
 804         char *buf;
 805         size_t n;
 806
 807         if (lseek(pack_fd, 0, SEEK_SET) != 0)
 808                 die("Failed seeking to start: %s", strerror(errno));
 809
 810         SHA1_Init(&c);
 811         yread(pack_fd, hdr, 8);
 812         SHA1_Update(&c, hdr, 8);
 813
 814         cnt = htonl(object_count);
 815         SHA1_Update(&c, &cnt, 4);
 816         ywrite(pack_fd, &cnt, 4);
 817
 818         buf = xmalloc(128 * 1024);
 819         for (;;) {
 820                 n = xread(pack_fd, buf, 128 * 1024);
 821                 if (n <= 0)
 822                         break;
 823                 SHA1_Update(&c, buf, n);
 824         }
 825         free(buf);
 826
 827         SHA1_Final(pack_sha1, &c);
 828         ywrite(pack_fd, pack_sha1, sizeof(pack_sha1));
 829 }
 830
 831 static int oecmp (const void *_a, const void *_b)
 832 {
 833         struct object_entry *a = *((struct object_entry**)_a);
 834         struct object_entry *b = *((struct object_entry**)_b);
 835         return memcmp(a->sha1, b->sha1, sizeof(a->sha1));
 836 }
 837
 838 static void write_index(const char *idx_name)
 839 {
 840         struct sha1file *f;
 841         struct object_entry **idx, **c, **last;
 842         struct object_entry *e;
 843         struct object_entry_pool *o;
 844         unsigned int array[256];
 845         int i;
 846
 847         /* Build the sorted table of object IDs. */
 848         idx = xmalloc(object_count * sizeof(struct object_entry*));
 849         c = idx;
 850         for (o = blocks; o; o = o->next_pool)
 851                 for (e = o->entries; e != o->next_free; e++)
 852                         *c++ = e;
 853         last = idx + object_count;
 854         qsort(idx, object_count, sizeof(struct object_entry*), oecmp);
 855
 856         /* Generate the fan-out array. */
 857         c = idx;
 858         for (i = 0; i < 256; i++) {
 859                 struct object_entry **next = c;;
 860                 while (next < last) {
 861                         if ((*next)->sha1[0] != i)
 862                                 break;
 863                         next++;
 864                 }
 865                 array[i] = htonl(next - idx);
 866                 c = next;
 867         }
 868
 869         f = sha1create("%s", idx_name);
 870         sha1write(f, array, 256 * sizeof(int));
 871         for (c = idx; c != last; c++) {
 872                 unsigned int offset = htonl((*c)->offset);
 873                 sha1write(f, &offset, 4);
 874                 sha1write(f, (*c)->sha1, sizeof((*c)->sha1));
 875         }
 876         sha1write(f, pack_sha1, sizeof(pack_sha1));
 877         sha1close(f, NULL, 1);
 878         free(idx);
 879 }
 880
 881 static void dump_branches()
 882 {
 883         static const char *msg = "fast-import";
 884         unsigned int i;
 885         struct branch *b;
 886         struct ref_lock *lock;
 887
 888         for (i = 0; i < branch_table_sz; i++) {
 889                 for (b = branch_table[i]; b; b = b->table_next_branch) {
 890                         lock = lock_any_ref_for_update(b->name, NULL, 0);
 891                         if (!lock || write_ref_sha1(lock, b->sha1, msg) < 0)
 892                                 die("Can't write %s", b->name);
 893                 }
 894         }
 895 }
 896
 897 static void cmd_new_blob()
 898 {
 899         unsigned long datlen;
 900         unsigned char sha1[20];
 901         void *dat;
 902
 903         yread(0, &datlen, 4);
 904         dat = xmalloc(datlen);
 905         yread(0, dat, datlen);
 906         if (store_object(OBJ_BLOB, dat, datlen, &last_blob, sha1))
 907                 free(dat);
 908 }
 909
 910 static void unload_one_branch()
 911 {
 912         while (cur_active_branches >= max_active_branches) {
 913                 unsigned long min_commit = ULONG_MAX;
 914                 struct branch *e, *l = NULL, *p = NULL;
 915
 916                 for (e = active_branches; e; e = e->active_next_branch) {
 917                         if (e->last_commit < min_commit) {
 918                                 p = l;
 919                                 min_commit = e->last_commit;
 920                         }
 921                         l = e;
 922                 }
 923
 924                 if (p) {
 925                         e = p->active_next_branch;
 926                         p->active_next_branch = e->active_next_branch;
 927                 } else {
 928                         e = active_branches;
 929                         active_branches = e->active_next_branch;
 930                 }
 931                 e->active_next_branch = NULL;
 932                 if (e->branch_tree.tree) {
 933                         release_tree_content(e->branch_tree.tree);
 934                         e->branch_tree.tree = NULL;
 935                 }
 936                 cur_active_branches--;
 937         }
 938 }
 939
 940 static void load_branch(struct branch *b)
 941 {
 942         load_tree(&b->branch_tree);
 943         b->active_next_branch = active_branches;
 944         active_branches = b;
 945         cur_active_branches++;
 946 }
 947
 948 static void file_change_m(struct branch *b)
 949 {
 950         const char *path = read_path();
 951         char hexsha1[41];
 952         unsigned char sha1[20];
 953
 954         yread(0, hexsha1, 40);
 955         hexsha1[40] = 0;
 956
 957         if (get_sha1_hex(hexsha1, sha1))
 958                 die("Invalid sha1 %s for %s", hexsha1, path);
 959
 960         tree_content_set(&b->branch_tree, path, sha1, 0100644);
 961 }
 962
 963 static void file_change_d(struct branch *b)
 964 {
 965         tree_content_remove(&b->branch_tree, read_path());
 966 }
 967
 968 static void cmd_new_commit()
 969 {
 970         static const unsigned int max_hdr_len = 94;
 971         const char *name = read_path();
 972         struct branch *b = lookup_branch(name);
 973         unsigned int acmsglen;
 974         char *body, *c;
 975
 976         if (!b)
 977                 die("Branch not declared: %s", name);
 978         if (!b->branch_tree.tree) {
 979                 unload_one_branch();
 980                 load_branch(b);
 981         }
 982
 983         /* author_committer_msg */
 984         yread(0, &acmsglen, 4);
 985         body = xmalloc(acmsglen + max_hdr_len);
 986         c = body + max_hdr_len;
 987         yread(0, c, acmsglen);
 988
 989         /* file_change* */
 990         for (;;) {
 991                 unsigned char cmd;
 992                 yread(0, &cmd, 1);
 993                 if (cmd == '0')
 994                         break;
 995                 else if (cmd == 'M')
 996                         file_change_m(b);
 997                 else if (cmd == 'D')
 998                         file_change_d(b);
 999                 else
1000                         die("Unsupported file_change: %c", cmd);
1001         }
1002
1003         if (memcmp(b->sha1, null_sha1, 20)) {
1004                 sprintf(c - 48, "parent %s", sha1_to_hex(b->sha1));
1005                 *(c - 1) = '\n';
1006                 c -= 48;
1007         }
1008         store_tree(&b->branch_tree);
1009         sprintf(c - 46, "tree %s", sha1_to_hex(b->branch_tree.sha1));
1010         *(c - 1) = '\n';
1011         c -= 46;
1012
1013         store_object(OBJ_COMMIT,
1014                 c, (body + max_hdr_len + acmsglen) - c,
1015                 NULL, b->sha1);
1016         free(body);
1017         b->last_commit = object_count_by_type[OBJ_COMMIT];
1018 }
1019
1020 static void cmd_new_branch()
1021 {
1022         struct branch *b = new_branch(read_path());
1023         const char *base = read_path();
1024         struct branch *s = lookup_branch(base);
1025
1026         if (!strcmp(b->name, base))
1027                 die("Can't create a branch from itself: %s", base);
1028         else if (s) {
1029                 memcpy(b->sha1, s->sha1, 20);
1030                 memcpy(b->branch_tree.sha1, s->branch_tree.sha1, 20);
1031         }
1032         else if (!get_sha1(base, b->sha1)) {
1033                 if (!memcmp(b->sha1, null_sha1, 20))
1034                         memcpy(b->branch_tree.sha1, null_sha1, 20);
1035                 else {
1036                         unsigned long size;
1037                         char *buf;
1038
1039                         buf = read_object_with_reference(b->sha1,
1040                                 type_names[OBJ_COMMIT], &size, b->sha1);
1041                         if (!buf || size < 46)
1042                                 die("Not a valid commit: %s", base);
1043                         if (memcmp("tree ", buf, 5)
1044                                 || get_sha1_hex(buf + 5, b->branch_tree.sha1))
1045                                 die("The commit %s is corrupt", sha1_to_hex(b->sha1));
1046                         free(buf);
1047                 }
1048         } else
1049                 die("Not a SHA1 or branch: %s", base);
1050 }
1051
1052 int main(int argc, const char **argv)
1053 {
1054         const char *base_name = argv[1];
1055         int est_obj_cnt = atoi(argv[2]);
1056         char *pack_name;
1057         char *idx_name;
1058         struct stat sb;
1059
1060         setup_ident();
1061         git_config(git_default_config);
1062
1063         pack_name = xmalloc(strlen(base_name) + 6);
1064         sprintf(pack_name, "%s.pack", base_name);
1065         idx_name = xmalloc(strlen(base_name) + 5);
1066         sprintf(idx_name, "%s.idx", base_name);
1067
1068         pack_fd = open(pack_name, O_RDWR|O_CREAT|O_EXCL, 0666);
1069         if (pack_fd < 0)
1070                 die("Can't create %s: %s", pack_name, strerror(errno));
1071
1072         alloc_objects(est_obj_cnt);
1073
1074         atom_table = xcalloc(atom_table_sz, sizeof(struct atom_str*));
1075         branch_table = xcalloc(branch_table_sz, sizeof(struct branch*));
1076         avail_tree_table = xcalloc(avail_tree_table_sz, sizeof(struct avail_tree_content*));
1077
1078         init_pack_header();
1079         for (;;) {
1080                 unsigned long cmd;
1081                 if (optional_read(0, &cmd, 4))
1082                         break;
1083
1084                 switch (ntohl(cmd)) {
1085                 case 'blob': cmd_new_blob();   break;
1086                 case 'comt': cmd_new_commit(); break;
1087                 case 'brch': cmd_new_branch(); break;
1088                 default:
1089                         die("Invalid command %lu", cmd);
1090                 }
1091         }
1092         fixup_header_footer();
1093         close(pack_fd);
1094         write_index(idx_name);
1095         dump_branches();
1096
1097         fprintf(stderr, "%s statistics:\n", argv[0]);
1098         fprintf(stderr, "---------------------------------------------------\n");
1099         fprintf(stderr, "Alloc'd objects: %10lu (%10lu overflow  )\n", alloc_count, alloc_count - est_obj_cnt);
1100         fprintf(stderr, "Total objects:   %10lu (%10lu duplicates)\n", object_count, duplicate_count);
1101         fprintf(stderr, "      blobs  :   %10lu (%10lu duplicates)\n", object_count_by_type[OBJ_BLOB], duplicate_count_by_type[OBJ_BLOB]);
1102         fprintf(stderr, "      trees  :   %10lu (%10lu duplicates)\n", object_count_by_type[OBJ_TREE], duplicate_count_by_type[OBJ_TREE]);
1103         fprintf(stderr, "      commits:   %10lu (%10lu duplicates)\n", object_count_by_type[OBJ_COMMIT], duplicate_count_by_type[OBJ_COMMIT]);
1104         fprintf(stderr, "      tags   :   %10lu (%10lu duplicates)\n", object_count_by_type[OBJ_TAG], duplicate_count_by_type[OBJ_TAG]);
1105         fprintf(stderr, "Total branches:  %10lu\n", branch_count);
1106         fprintf(stderr, "Total atoms:     %10u\n", atom_cnt);
1107         fprintf(stderr, "Memory pools:    %10lu MiB\n", total_allocd/(1024*1024));
1108         fprintf(stderr, "---------------------------------------------------\n");
1109
1110         stat(pack_name, &sb);
1111         fprintf(stderr, "Pack size:       %10lu KiB\n", (unsigned long)(sb.st_size/1024));
1112         stat(idx_name, &sb);
1113         fprintf(stderr, "Index size:      %10lu KiB\n", (unsigned long)(sb.st_size/1024));
1114
1115         fprintf(stderr, "\n");
1116
1117         return 0;
1118 }