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
 189 /* Stats and misc. counters */
 190 static unsigned long max_depth = 10;
 191 static unsigned long alloc_count;
 192 static unsigned long branch_count;
 193 static unsigned long branch_load_count;
 194 static unsigned long object_count;
 195 static unsigned long duplicate_count;
 196 static unsigned long marks_set_count;
 197 static unsigned long object_count_by_type[9];
 198 static unsigned long duplicate_count_by_type[9];
 199
 200 /* Memory pools */
 201 static size_t mem_pool_alloc = 2*1024*1024 - sizeof(struct mem_pool);
 202 static size_t total_allocd;
 203 static struct mem_pool *mem_pool;
 204
 205 /* Atom management */
 206 static unsigned int atom_table_sz = 4451;
 207 static unsigned int atom_cnt;
 208 static struct atom_str **atom_table;
 209
 210 /* The .pack file being generated */
 211 static int pack_fd;
 212 static unsigned long pack_offset;
 213 static unsigned char pack_sha1[20];
 214
 215 /* Table of objects we've written. */
 216 static unsigned int object_entry_alloc = 1000;
 217 static struct object_entry_pool *blocks;
 218 static struct object_entry *object_table[1 << 16];
 219 static struct mark_set *marks;
 220
 221 /* Our last blob */
 222 static struct last_object last_blob;
 223
 224 /* Tree management */
 225 static unsigned int tree_entry_alloc = 1000;
 226 static void *avail_tree_entry;
 227 static unsigned int avail_tree_table_sz = 100;
 228 static struct avail_tree_content **avail_tree_table;
 229
 230 /* Branch data */
 231 static unsigned long max_active_branches = 5;
 232 static unsigned long cur_active_branches;
 233 static unsigned long branch_table_sz = 1039;
 234 static struct branch **branch_table;
 235 static struct branch *active_branches;
 236
 237 /* Input stream parsing */
 238 static struct strbuf command_buf;
 239 static unsigned long next_mark;
 240
 241
 242 static void alloc_objects(int cnt)
 243 {
 244         struct object_entry_pool *b;
 245
 246         b = xmalloc(sizeof(struct object_entry_pool)
 247                 + cnt * sizeof(struct object_entry));
 248         b->next_pool = blocks;
 249         b->next_free = b->entries;
 250         b->end = b->entries + cnt;
 251         blocks = b;
 252         alloc_count += cnt;
 253 }
 254
 255 static struct object_entry* new_object(unsigned char *sha1)
 256 {
 257         struct object_entry *e;
 258
 259         if (blocks->next_free == blocks->end)
 260                 alloc_objects(object_entry_alloc);
 261
 262         e = blocks->next_free++;
 263         memcpy(e->sha1, sha1, sizeof(e->sha1));
 264         return e;
 265 }
 266
 267 static struct object_entry* find_object(unsigned char *sha1)
 268 {
 269         unsigned int h = sha1[0] << 8 | sha1[1];
 270         struct object_entry *e;
 271         for (e = object_table[h]; e; e = e->next)
 272                 if (!memcmp(sha1, e->sha1, sizeof(e->sha1)))
 273                         return e;
 274         return NULL;
 275 }
 276
 277 static struct object_entry* insert_object(unsigned char *sha1)
 278 {
 279         unsigned int h = sha1[0] << 8 | sha1[1];
 280         struct object_entry *e = object_table[h];
 281         struct object_entry *p = NULL;
 282
 283         while (e) {
 284                 if (!memcmp(sha1, e->sha1, sizeof(e->sha1)))
 285                         return e;
 286                 p = e;
 287                 e = e->next;
 288         }
 289
 290         e = new_object(sha1);
 291         e->next = NULL;
 292         e->offset = 0;
 293         if (p)
 294                 p->next = e;
 295         else
 296                 object_table[h] = e;
 297         return e;
 298 }
 299
 300 static unsigned int hc_str(const char *s, size_t len)
 301 {
 302         unsigned int r = 0;
 303         while (len-- > 0)
 304                 r = r * 31 + *s++;
 305         return r;
 306 }
 307
 308 static void* pool_alloc(size_t len)
 309 {
 310         struct mem_pool *p;
 311         void *r;
 312
 313         for (p = mem_pool; p; p = p->next_pool)
 314                 if ((p->end - p->next_free >= len))
 315                         break;
 316
 317         if (!p) {
 318                 if (len >= (mem_pool_alloc/2)) {
 319                         total_allocd += len;
 320                         return xmalloc(len);
 321                 }
 322                 total_allocd += sizeof(struct mem_pool) + mem_pool_alloc;
 323                 p = xmalloc(sizeof(struct mem_pool) + mem_pool_alloc);
 324                 p->next_pool = mem_pool;
 325                 p->next_free = p->space;
 326                 p->end = p->next_free + mem_pool_alloc;
 327                 mem_pool = p;
 328         }
 329
 330         r = p->next_free;
 331         p->next_free += len;
 332         return r;
 333 }
 334
 335 static void* pool_calloc(size_t count, size_t size)
 336 {
 337         size_t len = count * size;
 338         void *r = pool_alloc(len);
 339         memset(r, 0, len);
 340         return r;
 341 }
 342
 343 static char* pool_strdup(const char *s)
 344 {
 345         char *r = pool_alloc(strlen(s) + 1);
 346         strcpy(r, s);
 347         return r;
 348 }
 349
 350 static void insert_mark(unsigned long idnum, struct object_entry *oe)
 351 {
 352         struct mark_set *s = marks;
 353         while ((idnum >> s->shift) >= 1024) {
 354                 s = pool_calloc(1, sizeof(struct mark_set));
 355                 s->shift = marks->shift + 10;
 356                 s->data.sets[0] = marks;
 357                 marks = s;
 358         }
 359         while (s->shift) {
 360                 unsigned long i = idnum >> s->shift;
 361                 idnum -= i << s->shift;
 362                 if (!s->data.sets[i]) {
 363                         s->data.sets[i] = pool_calloc(1, sizeof(struct mark_set));
 364                         s->data.sets[i]->shift = s->shift - 10;
 365                 }
 366                 s = s->data.sets[i];
 367         }
 368         if (!s->data.marked[idnum])
 369                 marks_set_count++;
 370         s->data.marked[idnum] = oe;
 371 }
 372
 373 static struct object_entry* find_mark(unsigned long idnum)
 374 {
 375         unsigned long orig_idnum = idnum;
 376         struct mark_set *s = marks;
 377         struct object_entry *oe = NULL;
 378         if ((idnum >> s->shift) < 1024) {
 379                 while (s && s->shift) {
 380                         unsigned long i = idnum >> s->shift;
 381                         idnum -= i << s->shift;
 382                         s = s->data.sets[i];
 383                 }
 384                 if (s)
 385                         oe = s->data.marked[idnum];
 386         }
 387         if (!oe)
 388                 die("mark :%lu not declared", orig_idnum);
 389         return oe;
 390 }
 391
 392 static struct atom_str* to_atom(const char *s, size_t len)
 393 {
 394         unsigned int hc = hc_str(s, len) % atom_table_sz;
 395         struct atom_str *c;
 396
 397         for (c = atom_table[hc]; c; c = c->next_atom)
 398                 if (c->str_len == len && !strncmp(s, c->str_dat, len))
 399                         return c;
 400
 401         c = pool_alloc(sizeof(struct atom_str) + len + 1);
 402         c->str_len = len;
 403         strncpy(c->str_dat, s, len);
 404         c->str_dat[len] = 0;
 405         c->next_atom = atom_table[hc];
 406         atom_table[hc] = c;
 407         atom_cnt++;
 408         return c;
 409 }
 410
 411 static struct branch* lookup_branch(const char *name)
 412 {
 413         unsigned int hc = hc_str(name, strlen(name)) % branch_table_sz;
 414         struct branch *b;
 415
 416         for (b = branch_table[hc]; b; b = b->table_next_branch)
 417                 if (!strcmp(name, b->name))
 418                         return b;
 419         return NULL;
 420 }
 421
 422 static struct branch* new_branch(const char *name)
 423 {
 424         unsigned int hc = hc_str(name, strlen(name)) % branch_table_sz;
 425         struct branch* b = lookup_branch(name);
 426
 427         if (b)
 428                 die("Invalid attempt to create duplicate branch: %s", name);
 429         if (check_ref_format(name))
 430                 die("Branch name doesn't conform to GIT standards: %s", name);
 431
 432         b = pool_calloc(1, sizeof(struct branch));
 433         b->name = pool_strdup(name);
 434         b->table_next_branch = branch_table[hc];
 435         branch_table[hc] = b;
 436         branch_count++;
 437         return b;
 438 }
 439
 440 static unsigned int hc_entries(unsigned int cnt)
 441 {
 442         cnt = cnt & 7 ? (cnt / 8) + 1 : cnt / 8;
 443         return cnt < avail_tree_table_sz ? cnt : avail_tree_table_sz - 1;
 444 }
 445
 446 static struct tree_content* new_tree_content(unsigned int cnt)
 447 {
 448         struct avail_tree_content *f, *l = NULL;
 449         struct tree_content *t;
 450         unsigned int hc = hc_entries(cnt);
 451
 452         for (f = avail_tree_table[hc]; f; l = f, f = f->next_avail)
 453                 if (f->entry_capacity >= cnt)
 454                         break;
 455
 456         if (f) {
 457                 if (l)
 458                         l->next_avail = f->next_avail;
 459                 else
 460                         avail_tree_table[hc] = f->next_avail;
 461         } else {
 462                 cnt = cnt & 7 ? ((cnt / 8) + 1) * 8 : cnt;
 463                 f = pool_alloc(sizeof(*t) + sizeof(t->entries[0]) * cnt);
 464                 f->entry_capacity = cnt;
 465         }
 466
 467         t = (struct tree_content*)f;
 468         t->entry_count = 0;
 469         return t;
 470 }
 471
 472 static void release_tree_entry(struct tree_entry *e);
 473 static void release_tree_content(struct tree_content *t)
 474 {
 475         struct avail_tree_content *f = (struct avail_tree_content*)t;
 476         unsigned int hc = hc_entries(f->entry_capacity);
 477         f->next_avail = avail_tree_table[hc];
 478         avail_tree_table[hc] = f;
 479 }
 480
 481 static void release_tree_content_recursive(struct tree_content *t)
 482 {
 483         unsigned int i;
 484         for (i = 0; i < t->entry_count; i++)
 485                 release_tree_entry(t->entries[i]);
 486         release_tree_content(t);
 487 }
 488
 489 static struct tree_content* grow_tree_content(
 490         struct tree_content *t,
 491         int amt)
 492 {
 493         struct tree_content *r = new_tree_content(t->entry_count + amt);
 494         r->entry_count = t->entry_count;
 495         memcpy(r->entries,t->entries,t->entry_count*sizeof(t->entries[0]));
 496         release_tree_content(t);
 497         return r;
 498 }
 499
 500 static struct tree_entry* new_tree_entry()
 501 {
 502         struct tree_entry *e;
 503
 504         if (!avail_tree_entry) {
 505                 unsigned int n = tree_entry_alloc;
 506                 avail_tree_entry = e = xmalloc(n * sizeof(struct tree_entry));
 507                 while (n--) {
 508                         *((void**)e) = e + 1;
 509                         e++;
 510                 }
 511         }
 512
 513         e = avail_tree_entry;
 514         avail_tree_entry = *((void**)e);
 515         return e;
 516 }
 517
 518 static void release_tree_entry(struct tree_entry *e)
 519 {
 520         if (e->tree)
 521                 release_tree_content_recursive(e->tree);
 522         *((void**)e) = avail_tree_entry;
 523         avail_tree_entry = e;
 524 }
 525
 526 static void yread(int fd, void *buffer, size_t length)
 527 {
 528         ssize_t ret = 0;
 529         while (ret < length) {
 530                 ssize_t size = xread(fd, (char *) buffer + ret, length - ret);
 531                 if (!size)
 532                         die("Read from descriptor %i: end of stream", fd);
 533                 if (size < 0)
 534                         die("Read from descriptor %i: %s", fd, strerror(errno));
 535                 ret += size;
 536         }
 537 }
 538
 539 static void ywrite(int fd, void *buffer, size_t length)
 540 {
 541         ssize_t ret = 0;
 542         while (ret < length) {
 543                 ssize_t size = xwrite(fd, (char *) buffer + ret, length - ret);
 544                 if (!size)
 545                         die("Write to descriptor %i: end of file", fd);
 546                 if (size < 0)
 547                         die("Write to descriptor %i: %s", fd, strerror(errno));
 548                 ret += size;
 549         }
 550 }
 551
 552 static size_t encode_header(
 553         enum object_type type,
 554         size_t size,
 555         unsigned char *hdr)
 556 {
 557         int n = 1;
 558         unsigned char c;
 559
 560         if (type < OBJ_COMMIT || type > OBJ_DELTA)
 561                 die("bad type %d", type);
 562
 563         c = (type << 4) | (size & 15);
 564         size >>= 4;
 565         while (size) {
 566                 *hdr++ = c | 0x80;
 567                 c = size & 0x7f;
 568                 size >>= 7;
 569                 n++;
 570         }
 571         *hdr = c;
 572         return n;
 573 }
 574
 575 static int store_object(
 576         enum object_type type,
 577         void *dat,
 578         size_t datlen,
 579         struct last_object *last,
 580         unsigned char *sha1out,
 581         unsigned long mark)
 582 {
 583         void *out, *delta;
 584         struct object_entry *e;
 585         unsigned char hdr[96];
 586         unsigned char sha1[20];
 587         unsigned long hdrlen, deltalen;
 588         SHA_CTX c;
 589         z_stream s;
 590
 591         hdrlen = sprintf((char*)hdr,"%s %lu",type_names[type],datlen) + 1;
 592         SHA1_Init(&c);
 593         SHA1_Update(&c, hdr, hdrlen);
 594         SHA1_Update(&c, dat, datlen);
 595         SHA1_Final(sha1, &c);
 596         if (sha1out)
 597                 memcpy(sha1out, sha1, sizeof(sha1));
 598
 599         e = insert_object(sha1);
 600         if (mark)
 601                 insert_mark(mark, e);
 602         if (e->offset) {
 603                 duplicate_count++;
 604                 duplicate_count_by_type[type]++;
 605                 return 1;
 606         }
 607         e->type = type;
 608         e->offset = pack_offset;
 609         object_count++;
 610         object_count_by_type[type]++;
 611
 612         if (last && last->data && last->depth < max_depth)
 613                 delta = diff_delta(last->data, last->len,
 614                         dat, datlen,
 615                         &deltalen, 0);
 616         else
 617                 delta = 0;
 618
 619         memset(&s, 0, sizeof(s));
 620         deflateInit(&s, zlib_compression_level);
 621
 622         if (delta) {
 623                 last->depth++;
 624                 s.next_in = delta;
 625                 s.avail_in = deltalen;
 626                 hdrlen = encode_header(OBJ_DELTA, deltalen, hdr);
 627                 ywrite(pack_fd, hdr, hdrlen);
 628                 ywrite(pack_fd, last->sha1, sizeof(sha1));
 629                 pack_offset += hdrlen + sizeof(sha1);
 630         } else {
 631                 if (last)
 632                         last->depth = 0;
 633                 s.next_in = dat;
 634                 s.avail_in = datlen;
 635                 hdrlen = encode_header(type, datlen, hdr);
 636                 ywrite(pack_fd, hdr, hdrlen);
 637                 pack_offset += hdrlen;
 638         }
 639
 640         s.avail_out = deflateBound(&s, s.avail_in);
 641         s.next_out = out = xmalloc(s.avail_out);
 642         while (deflate(&s, Z_FINISH) == Z_OK)
 643                 /* nothing */;
 644         deflateEnd(&s);
 645
 646         ywrite(pack_fd, out, s.total_out);
 647         pack_offset += s.total_out;
 648
 649         free(out);
 650         if (delta)
 651                 free(delta);
 652         if (last) {
 653                 if (last->data)
 654                         free(last->data);
 655                 last->data = dat;
 656                 last->len = datlen;
 657                 memcpy(last->sha1, sha1, sizeof(sha1));
 658         }
 659         return 0;
 660 }
 661
 662 static const char *get_mode(const char *str, unsigned int *modep)
 663 {
 664         unsigned char c;
 665         unsigned int mode = 0;
 666
 667         while ((c = *str++) != ' ') {
 668                 if (c < '0' || c > '7')
 669                         return NULL;
 670                 mode = (mode << 3) + (c - '0');
 671         }
 672         *modep = mode;
 673         return str;
 674 }
 675
 676 static void load_tree(struct tree_entry *root)
 677 {
 678         struct object_entry *myoe;
 679         struct tree_content *t;
 680         unsigned long size;
 681         char *buf;
 682         const char *c;
 683         char type[20];
 684
 685         root->tree = t = new_tree_content(8);
 686         if (!memcmp(root->sha1, null_sha1, 20))
 687                 return;
 688
 689         myoe = find_object(root->sha1);
 690         if (myoe) {
 691                 die("FIXME");
 692         } else {
 693                 buf = read_sha1_file(root->sha1, type, &size);
 694                 if (!buf || strcmp(type, tree_type))
 695                         die("Can't load existing tree %s", sha1_to_hex(root->sha1));
 696         }
 697
 698         c = buf;
 699         while (c != (buf + size)) {
 700                 struct tree_entry *e = new_tree_entry();
 701
 702                 if (t->entry_count == t->entry_capacity)
 703                         root->tree = t = grow_tree_content(t, 8);
 704                 t->entries[t->entry_count++] = e;
 705
 706                 e->tree = NULL;
 707                 c = get_mode(c, &e->mode);
 708                 if (!c)
 709                         die("Corrupt mode in %s", sha1_to_hex(root->sha1));
 710                 e->name = to_atom(c, strlen(c));
 711                 c += e->name->str_len + 1;
 712                 memcpy(e->sha1, c, sizeof(e->sha1));
 713                 c += 20;
 714         }
 715         free(buf);
 716 }
 717
 718 static int tecmp (const void *_a, const void *_b)
 719 {
 720         struct tree_entry *a = *((struct tree_entry**)_a);
 721         struct tree_entry *b = *((struct tree_entry**)_b);
 722         return base_name_compare(
 723                 a->name->str_dat, a->name->str_len, a->mode,
 724                 b->name->str_dat, b->name->str_len, b->mode);
 725 }
 726
 727 static void store_tree(struct tree_entry *root)
 728 {
 729         struct tree_content *t = root->tree;
 730         unsigned int i;
 731         size_t maxlen;
 732         char *buf, *c;
 733
 734         if (memcmp(root->sha1, null_sha1, 20))
 735                 return;
 736
 737         maxlen = 0;
 738         for (i = 0; i < t->entry_count; i++) {
 739                 maxlen += t->entries[i]->name->str_len + 34;
 740                 if (t->entries[i]->tree)
 741                         store_tree(t->entries[i]);
 742         }
 743
 744         qsort(t->entries, t->entry_count, sizeof(t->entries[0]), tecmp);
 745         buf = c = xmalloc(maxlen);
 746         for (i = 0; i < t->entry_count; i++) {
 747                 struct tree_entry *e = t->entries[i];
 748                 c += sprintf(c, "%o", e->mode);
 749                 *c++ = ' ';
 750                 strcpy(c, e->name->str_dat);
 751                 c += e->name->str_len + 1;
 752                 memcpy(c, e->sha1, 20);
 753                 c += 20;
 754         }
 755         store_object(OBJ_TREE, buf, c - buf, NULL, root->sha1, 0);
 756         free(buf);
 757 }
 758
 759 static int tree_content_set(
 760         struct tree_entry *root,
 761         const char *p,
 762         const unsigned char *sha1,
 763         const unsigned int mode)
 764 {
 765         struct tree_content *t = root->tree;
 766         const char *slash1;
 767         unsigned int i, n;
 768         struct tree_entry *e;
 769
 770         slash1 = strchr(p, '/');
 771         if (slash1)
 772                 n = slash1 - p;
 773         else
 774                 n = strlen(p);
 775
 776         for (i = 0; i < t->entry_count; i++) {
 777                 e = t->entries[i];
 778                 if (e->name->str_len == n && !strncmp(p, e->name->str_dat, n)) {
 779                         if (!slash1) {
 780                                 if (e->mode == mode && !memcmp(e->sha1, sha1, 20))
 781                                         return 0;
 782                                 e->mode = mode;
 783                                 memcpy(e->sha1, sha1, 20);
 784                                 if (e->tree) {
 785                                         release_tree_content_recursive(e->tree);
 786                                         e->tree = NULL;
 787                                 }
 788                                 memcpy(root->sha1, null_sha1, 20);
 789                                 return 1;
 790                         }
 791                         if (!S_ISDIR(e->mode)) {
 792                                 e->tree = new_tree_content(8);
 793                                 e->mode = S_IFDIR;
 794                         }
 795                         if (!e->tree)
 796                                 load_tree(e);
 797                         if (tree_content_set(e, slash1 + 1, sha1, mode)) {
 798                                 memcpy(root->sha1, null_sha1, 20);
 799                                 return 1;
 800                         }
 801                         return 0;
 802                 }
 803         }
 804
 805         if (t->entry_count == t->entry_capacity)
 806                 root->tree = t = grow_tree_content(t, 8);
 807         e = new_tree_entry();
 808         e->name = to_atom(p, n);
 809         t->entries[t->entry_count++] = e;
 810         if (slash1) {
 811                 e->tree = new_tree_content(8);
 812                 e->mode = S_IFDIR;
 813                 tree_content_set(e, slash1 + 1, sha1, mode);
 814         } else {
 815                 e->tree = NULL;
 816                 e->mode = mode;
 817                 memcpy(e->sha1, sha1, 20);
 818         }
 819         memcpy(root->sha1, null_sha1, 20);
 820         return 1;
 821 }
 822
 823 static int tree_content_remove(struct tree_entry *root, const char *p)
 824 {
 825         struct tree_content *t = root->tree;
 826         const char *slash1;
 827         unsigned int i, n;
 828         struct tree_entry *e;
 829
 830         slash1 = strchr(p, '/');
 831         if (slash1)
 832                 n = slash1 - p;
 833         else
 834                 n = strlen(p);
 835
 836         for (i = 0; i < t->entry_count; i++) {
 837                 e = t->entries[i];
 838                 if (e->name->str_len == n && !strncmp(p, e->name->str_dat, n)) {
 839                         if (!slash1 || !S_ISDIR(e->mode))
 840                                 goto del_entry;
 841                         if (!e->tree)
 842                                 load_tree(e);
 843                         if (tree_content_remove(e, slash1 + 1)) {
 844                                 if (!e->tree->entry_count)
 845                                         goto del_entry;
 846                                 memcpy(root->sha1, null_sha1, 20);
 847                                 return 1;
 848                         }
 849                         return 0;
 850                 }
 851         }
 852         return 0;
 853
 854 del_entry:
 855         for (i++; i < t->entry_count; i++)
 856                 t->entries[i-1] = t->entries[i];
 857         t->entry_count--;
 858         release_tree_entry(e);
 859         memcpy(root->sha1, null_sha1, 20);
 860         return 1;
 861 }
 862
 863 static void init_pack_header()
 864 {
 865         struct pack_header hdr;
 866
 867         hdr.hdr_signature = htonl(PACK_SIGNATURE);
 868         hdr.hdr_version = htonl(2);
 869         hdr.hdr_entries = 0;
 870
 871         ywrite(pack_fd, &hdr, sizeof(hdr));
 872         pack_offset = sizeof(hdr);
 873 }
 874
 875 static void fixup_header_footer()
 876 {
 877         SHA_CTX c;
 878         char hdr[8];
 879         unsigned long cnt;
 880         char *buf;
 881         size_t n;
 882
 883         if (lseek(pack_fd, 0, SEEK_SET) != 0)
 884                 die("Failed seeking to start: %s", strerror(errno));
 885
 886         SHA1_Init(&c);
 887         yread(pack_fd, hdr, 8);
 888         SHA1_Update(&c, hdr, 8);
 889
 890         cnt = htonl(object_count);
 891         SHA1_Update(&c, &cnt, 4);
 892         ywrite(pack_fd, &cnt, 4);
 893
 894         buf = xmalloc(128 * 1024);
 895         for (;;) {
 896                 n = xread(pack_fd, buf, 128 * 1024);
 897                 if (n <= 0)
 898                         break;
 899                 SHA1_Update(&c, buf, n);
 900         }
 901         free(buf);
 902
 903         SHA1_Final(pack_sha1, &c);
 904         ywrite(pack_fd, pack_sha1, sizeof(pack_sha1));
 905 }
 906
 907 static int oecmp (const void *_a, const void *_b)
 908 {
 909         struct object_entry *a = *((struct object_entry**)_a);
 910         struct object_entry *b = *((struct object_entry**)_b);
 911         return memcmp(a->sha1, b->sha1, sizeof(a->sha1));
 912 }
 913
 914 static void write_index(const char *idx_name)
 915 {
 916         struct sha1file *f;
 917         struct object_entry **idx, **c, **last;
 918         struct object_entry *e;
 919         struct object_entry_pool *o;
 920         unsigned int array[256];
 921         int i;
 922
 923         /* Build the sorted table of object IDs. */
 924         idx = xmalloc(object_count * sizeof(struct object_entry*));
 925         c = idx;
 926         for (o = blocks; o; o = o->next_pool)
 927                 for (e = o->entries; e != o->next_free; e++)
 928                         *c++ = e;
 929         last = idx + object_count;
 930         qsort(idx, object_count, sizeof(struct object_entry*), oecmp);
 931
 932         /* Generate the fan-out array. */
 933         c = idx;
 934         for (i = 0; i < 256; i++) {
 935                 struct object_entry **next = c;;
 936                 while (next < last) {
 937                         if ((*next)->sha1[0] != i)
 938                                 break;
 939                         next++;
 940                 }
 941                 array[i] = htonl(next - idx);
 942                 c = next;
 943         }
 944
 945         f = sha1create("%s", idx_name);
 946         sha1write(f, array, 256 * sizeof(int));
 947         for (c = idx; c != last; c++) {
 948                 unsigned int offset = htonl((*c)->offset);
 949                 sha1write(f, &offset, 4);
 950                 sha1write(f, (*c)->sha1, sizeof((*c)->sha1));
 951         }
 952         sha1write(f, pack_sha1, sizeof(pack_sha1));
 953         sha1close(f, NULL, 1);
 954         free(idx);
 955 }
 956
 957 static void dump_branches()
 958 {
 959         static const char *msg = "fast-import";
 960         unsigned int i;
 961         struct branch *b;
 962         struct ref_lock *lock;
 963
 964         for (i = 0; i < branch_table_sz; i++) {
 965                 for (b = branch_table[i]; b; b = b->table_next_branch) {
 966                         lock = lock_any_ref_for_update(b->name, NULL, 0);
 967                         if (!lock || write_ref_sha1(lock, b->sha1, msg) < 0)
 968                                 die("Can't write %s", b->name);
 969                 }
 970         }
 971 }
 972
 973 static void read_next_command()
 974 {
 975         read_line(&command_buf, stdin, '\n');
 976 }
 977
 978 static void cmd_mark()
 979 {
 980         if (!strncmp("mark :", command_buf.buf, 6)) {
 981                 next_mark = strtoul(command_buf.buf + 6, NULL, 10);
 982                 read_next_command();
 983         }
 984         else
 985                 next_mark = 0;
 986 }
 987
 988 static void* cmd_data (size_t *size)
 989 {
 990         size_t n = 0;
 991         void *buffer;
 992         size_t length;
 993
 994         if (strncmp("data ", command_buf.buf, 5))
 995                 die("Expected 'data n' command, found: %s", command_buf.buf);
 996
 997         length = strtoul(command_buf.buf + 5, NULL, 10);
 998         buffer = xmalloc(length);
 999
1000         while (n < length) {
1001                 size_t s = fread((char*)buffer + n, 1, length - n, stdin);
1002                 if (!s && feof(stdin))
1003                         die("EOF in data (%lu bytes remaining)", length - n);
1004                 n += s;
1005         }
1006
1007         if (fgetc(stdin) != '\n')
1008                 die("An lf did not trail the binary data as expected.");
1009
1010         *size = length;
1011         return buffer;
1012 }
1013
1014 static void cmd_new_blob()
1015 {
1016         size_t l;
1017         void *d;
1018
1019         read_next_command();
1020         cmd_mark();
1021         d = cmd_data(&l);
1022
1023         if (store_object(OBJ_BLOB, d, l, &last_blob, NULL, next_mark))
1024                 free(d);
1025 }
1026
1027 static void unload_one_branch()
1028 {
1029         while (cur_active_branches >= max_active_branches) {
1030                 unsigned long min_commit = ULONG_MAX;
1031                 struct branch *e, *l = NULL, *p = NULL;
1032
1033                 for (e = active_branches; e; e = e->active_next_branch) {
1034                         if (e->last_commit < min_commit) {
1035                                 p = l;
1036                                 min_commit = e->last_commit;
1037                         }
1038                         l = e;
1039                 }
1040
1041                 if (p) {
1042                         e = p->active_next_branch;
1043                         p->active_next_branch = e->active_next_branch;
1044                 } else {
1045                         e = active_branches;
1046                         active_branches = e->active_next_branch;
1047                 }
1048                 e->active_next_branch = NULL;
1049                 if (e->branch_tree.tree) {
1050                         release_tree_content_recursive(e->branch_tree.tree);
1051                         e->branch_tree.tree = NULL;
1052                 }
1053                 cur_active_branches--;
1054         }
1055 }
1056
1057 static void load_branch(struct branch *b)
1058 {
1059         load_tree(&b->branch_tree);
1060         b->active_next_branch = active_branches;
1061         active_branches = b;
1062         cur_active_branches++;
1063         branch_load_count++;
1064 }
1065
1066 static void file_change_m(struct branch *b)
1067 {
1068         const char *p = command_buf.buf + 2;
1069         char *p_uq;
1070         const char *endp;
1071         struct object_entry *oe;
1072         unsigned char sha1[20];
1073         unsigned int mode;
1074         char type[20];
1075
1076         p = get_mode(p, &mode);
1077         if (!p)
1078                 die("Corrupt mode: %s", command_buf.buf);
1079         switch (mode) {
1080         case S_IFREG | 0644:
1081         case S_IFREG | 0755:
1082         case S_IFLNK:
1083         case 0644:
1084         case 0755:
1085                 /* ok */
1086                 break;
1087         default:
1088                 die("Corrupt mode: %s", command_buf.buf);
1089         }
1090
1091         if (*p == ':') {
1092                 char *x;
1093                 oe = find_mark(strtoul(p + 1, &x, 10));
1094                 p = x;
1095         } else {
1096                 if (get_sha1_hex(p, sha1))
1097                         die("Invalid SHA1: %s", command_buf.buf);
1098                 oe = find_object(sha1);
1099                 p += 40;
1100         }
1101         if (*p++ != ' ')
1102                 die("Missing space after SHA1: %s", command_buf.buf);
1103
1104         p_uq = unquote_c_style(p, &endp);
1105         if (p_uq) {
1106                 if (*endp)
1107                         die("Garbage after path in: %s", command_buf.buf);
1108                 p = p_uq;
1109         }
1110
1111         if (oe) {
1112                 if (oe->type != OBJ_BLOB)
1113                         die("Not a blob (actually a %s): %s",
1114                                 command_buf.buf, type_names[oe->type]);
1115         } else {
1116                 if (sha1_object_info(sha1, type, NULL))
1117                         die("Blob not found: %s", command_buf.buf);
1118                 if (strcmp(blob_type, type))
1119                         die("Not a blob (actually a %s): %s",
1120                                 command_buf.buf, type);
1121         }
1122
1123         tree_content_set(&b->branch_tree, p, sha1, S_IFREG | mode);
1124
1125         if (p_uq)
1126                 free(p_uq);
1127 }
1128
1129 static void file_change_d(struct branch *b)
1130 {
1131         const char *p = command_buf.buf + 2;
1132         char *p_uq;
1133         const char *endp;
1134
1135         p_uq = unquote_c_style(p, &endp);
1136         if (p_uq) {
1137                 if (*endp)
1138                         die("Garbage after path in: %s", command_buf.buf);
1139                 p = p_uq;
1140         }
1141         tree_content_remove(&b->branch_tree, p);
1142         if (p_uq)
1143                 free(p_uq);
1144 }
1145
1146 static void cmd_new_commit()
1147 {
1148         struct branch *b;
1149         void *msg;
1150         size_t msglen;
1151         char *str_uq;
1152         const char *endp;
1153         char *sp;
1154         char *author = NULL;
1155         char *committer = NULL;
1156         char *body;
1157
1158         /* Obtain the branch name from the rest of our command */
1159         sp = strchr(command_buf.buf, ' ') + 1;
1160         str_uq = unquote_c_style(sp, &endp);
1161         if (str_uq) {
1162                 if (*endp)
1163                         die("Garbage after ref in: %s", command_buf.buf);
1164                 sp = str_uq;
1165         }
1166         b = lookup_branch(sp);
1167         if (!b)
1168                 die("Branch not declared: %s", sp);
1169         if (str_uq)
1170                 free(str_uq);
1171
1172         read_next_command();
1173         cmd_mark();
1174         if (!strncmp("author ", command_buf.buf, 7)) {
1175                 author = strdup(command_buf.buf);
1176                 read_next_command();
1177         }
1178         if (!strncmp("committer ", command_buf.buf, 10)) {
1179                 committer = strdup(command_buf.buf);
1180                 read_next_command();
1181         }
1182         if (!committer)
1183                 die("Expected committer but didn't get one");
1184         msg = cmd_data(&msglen);
1185
1186         /* ensure the branch is active/loaded */
1187         if (!b->branch_tree.tree) {
1188                 unload_one_branch();
1189                 load_branch(b);
1190         }
1191
1192         /* file_change* */
1193         for (;;) {
1194                 read_next_command();
1195                 if (1 == command_buf.len)
1196                         break;
1197                 else if (!strncmp("M ", command_buf.buf, 2))
1198                         file_change_m(b);
1199                 else if (!strncmp("D ", command_buf.buf, 2))
1200                         file_change_d(b);
1201                 else
1202                         die("Unsupported file_change: %s", command_buf.buf);
1203         }
1204
1205         /* build the tree and the commit */
1206         store_tree(&b->branch_tree);
1207         body = xmalloc(97 + msglen
1208                 + (author
1209                         ? strlen(author) + strlen(committer)
1210                         : 2 * strlen(committer)));
1211         sp = body;
1212         sp += sprintf(sp, "tree %s\n", sha1_to_hex(b->branch_tree.sha1));
1213         if (memcmp(b->sha1, null_sha1, 20))
1214                 sp += sprintf(sp, "parent %s\n", sha1_to_hex(b->sha1));
1215         if (author)
1216                 sp += sprintf(sp, "%s\n", author);
1217         else
1218                 sp += sprintf(sp, "author %s\n", committer + 10);
1219         sp += sprintf(sp, "%s\n\n", committer);
1220         memcpy(sp, msg, msglen);
1221         sp += msglen;
1222         if (author)
1223                 free(author);
1224         free(committer);
1225         free(msg);
1226
1227         store_object(OBJ_COMMIT, body, sp - body, NULL, b->sha1, next_mark);
1228         free(body);
1229         b->last_commit = object_count_by_type[OBJ_COMMIT];
1230 }
1231
1232 static void cmd_new_branch()
1233 {
1234         struct branch *b;
1235         char *str_uq;
1236         const char *endp;
1237         char *sp;
1238
1239         /* Obtain the new branch name from the rest of our command */
1240         sp = strchr(command_buf.buf, ' ') + 1;
1241         str_uq = unquote_c_style(sp, &endp);
1242         if (str_uq) {
1243                 if (*endp)
1244                         die("Garbage after ref in: %s", command_buf.buf);
1245                 sp = str_uq;
1246         }
1247         b = new_branch(sp);
1248         if (str_uq)
1249                 free(str_uq);
1250         read_next_command();
1251
1252         /* from ... */
1253         if (!strncmp("from ", command_buf.buf, 5)) {
1254                 const char *from;
1255                 struct branch *s;
1256
1257                 from = strchr(command_buf.buf, ' ') + 1;
1258                 str_uq = unquote_c_style(from, &endp);
1259                 if (str_uq) {
1260                         if (*endp)
1261                                 die("Garbage after string in: %s", command_buf.buf);
1262                         from = str_uq;
1263                 }
1264
1265                 s = lookup_branch(from);
1266                 if (b == s)
1267                         die("Can't create a branch from itself: %s", b->name);
1268                 else if (s) {
1269                         memcpy(b->sha1, s->sha1, 20);
1270                         memcpy(b->branch_tree.sha1, s->branch_tree.sha1, 20);
1271                 } else if (*from == ':') {
1272                         unsigned long idnum = strtoul(from + 1, NULL, 10);
1273                         struct object_entry *oe = find_mark(idnum);
1274                         if (oe->type != OBJ_COMMIT)
1275                                 die("Mark :%lu not a commit", idnum);
1276                         memcpy(b->sha1, oe->sha1, 20);
1277                         memcpy(b->branch_tree.sha1, null_sha1, 20);
1278                 } else if (!get_sha1(from, b->sha1)) {
1279                         if (!memcmp(b->sha1, null_sha1, 20))
1280                                 memcpy(b->branch_tree.sha1, null_sha1, 20);
1281                         else {
1282                                 unsigned long size;
1283                                 char *buf;
1284
1285                                 buf = read_object_with_reference(b->sha1,
1286                                         type_names[OBJ_COMMIT], &size, b->sha1);
1287                                 if (!buf || size < 46)
1288                                         die("Not a valid commit: %s", from);
1289                                 if (memcmp("tree ", buf, 5)
1290                                         || get_sha1_hex(buf + 5, b->branch_tree.sha1))
1291                                         die("The commit %s is corrupt", sha1_to_hex(b->sha1));
1292                                 free(buf);
1293                         }
1294                 } else
1295                         die("Invalid ref name or SHA1 expression: %s", from);
1296
1297                 if (str_uq)
1298                         free(str_uq);
1299                 read_next_command();
1300         } else {
1301                 memcpy(b->sha1, null_sha1, 20);
1302                 memcpy(b->branch_tree.sha1, null_sha1, 20);
1303         }
1304
1305         if (command_buf.eof || command_buf.len > 1)
1306                 die("An lf did not terminate the branch command as expected.");
1307 }
1308
1309 static const char fast_import_usage[] =
1310 "git-fast-import [--objects=n] [--depth=n] [--active-branches=n] temp.pack";
1311
1312 int main(int argc, const char **argv)
1313 {
1314         const char *base_name;
1315         int i;
1316         unsigned long est_obj_cnt = 1000;
1317         char *pack_name;
1318         char *idx_name;
1319         struct stat sb;
1320
1321         setup_ident();
1322         git_config(git_default_config);
1323
1324         for (i = 1; i < argc; i++) {
1325                 const char *a = argv[i];
1326
1327                 if (*a != '-' || !strcmp(a, "--"))
1328                         break;
1329                 else if (!strncmp(a, "--objects=", 10))
1330                         est_obj_cnt = strtoul(a + 10, NULL, 0);
1331                 else if (!strncmp(a, "--depth=", 8))
1332                         max_depth = strtoul(a + 8, NULL, 0);
1333                 else if (!strncmp(a, "--active-branches=", 18))
1334                         max_active_branches = strtoul(a + 18, NULL, 0);
1335                 else
1336                         die("unknown option %s", a);
1337         }
1338         if ((i+1) != argc)
1339                 usage(fast_import_usage);
1340         base_name = argv[i];
1341
1342         pack_name = xmalloc(strlen(base_name) + 6);
1343         sprintf(pack_name, "%s.pack", base_name);
1344         idx_name = xmalloc(strlen(base_name) + 5);
1345         sprintf(idx_name, "%s.idx", base_name);
1346
1347         pack_fd = open(pack_name, O_RDWR|O_CREAT|O_EXCL, 0666);
1348         if (pack_fd < 0)
1349                 die("Can't create %s: %s", pack_name, strerror(errno));
1350
1351         init_pack_header();
1352         alloc_objects(est_obj_cnt);
1353         strbuf_init(&command_buf);
1354
1355         atom_table = xcalloc(atom_table_sz, sizeof(struct atom_str*));
1356         branch_table = xcalloc(branch_table_sz, sizeof(struct branch*));
1357         avail_tree_table = xcalloc(avail_tree_table_sz, sizeof(struct avail_tree_content*));
1358         marks = pool_calloc(1, sizeof(struct mark_set));
1359
1360         for (;;) {
1361                 read_next_command();
1362                 if (command_buf.eof)
1363                         break;
1364                 else if (!strcmp("blob", command_buf.buf))
1365                         cmd_new_blob();
1366                 else if (!strncmp("branch ", command_buf.buf, 7))
1367                         cmd_new_branch();
1368                 else if (!strncmp("commit ", command_buf.buf, 7))
1369                         cmd_new_commit();
1370                 else
1371                         die("Unsupported command: %s", command_buf.buf);
1372         }
1373
1374         fixup_header_footer();
1375         close(pack_fd);
1376         write_index(idx_name);
1377         dump_branches();
1378
1379         fprintf(stderr, "%s statistics:\n", argv[0]);
1380         fprintf(stderr, "---------------------------------------------------\n");
1381         fprintf(stderr, "Alloc'd objects: %10lu (%10lu overflow  )\n", alloc_count, alloc_count - est_obj_cnt);
1382         fprintf(stderr, "Total objects:   %10lu (%10lu duplicates)\n", object_count, duplicate_count);
1383         fprintf(stderr, "      blobs  :   %10lu (%10lu duplicates)\n", object_count_by_type[OBJ_BLOB], duplicate_count_by_type[OBJ_BLOB]);
1384         fprintf(stderr, "      trees  :   %10lu (%10lu duplicates)\n", object_count_by_type[OBJ_TREE], duplicate_count_by_type[OBJ_TREE]);
1385         fprintf(stderr, "      commits:   %10lu (%10lu duplicates)\n", object_count_by_type[OBJ_COMMIT], duplicate_count_by_type[OBJ_COMMIT]);
1386         fprintf(stderr, "      tags   :   %10lu (%10lu duplicates)\n", object_count_by_type[OBJ_TAG], duplicate_count_by_type[OBJ_TAG]);
1387         fprintf(stderr, "Total branches:  %10lu (%10lu loads     )\n", branch_count, branch_load_count);
1388         fprintf(stderr, "      marks:     %10u (%10lu unique    )\n", (1 << marks->shift) * 1024, marks_set_count);
1389         fprintf(stderr, "      atoms:     %10u\n", atom_cnt);
1390         fprintf(stderr, "Memory total:    %10lu KiB\n", (total_allocd + alloc_count*sizeof(struct object_entry))/1024);
1391         fprintf(stderr, "       pools:    %10lu KiB\n", total_allocd/1024);
1392         fprintf(stderr, "     objects:    %10lu KiB\n", (alloc_count*sizeof(struct object_entry))/1024);
1393         fprintf(stderr, "---------------------------------------------------\n");
1394
1395         stat(pack_name, &sb);
1396         fprintf(stderr, "Pack size:       %10lu KiB\n", (unsigned long)(sb.st_size/1024));
1397         stat(idx_name, &sb);
1398         fprintf(stderr, "Index size:      %10lu KiB\n", (unsigned long)(sb.st_size/1024));
1399
1400         fprintf(stderr, "\n");
1401
1402         return 0;
1403 }