fast-import.c

   1 /*
   2 Format of STDIN stream:
   3
   4   stream ::= cmd*;
   5
   6   cmd ::= new_blob
   7         | new_commit
   8         | new_tag
   9         | reset_branch
  10         ;
  11
  12   new_blob ::= 'blob' lf
  13         mark?
  14     file_content;
  15   file_content ::= data;
  16
  17   new_commit ::= 'commit' sp ref_str lf
  18     mark?
  19     ('author' sp name '<' email '>' ts tz lf)?
  20     'committer' sp name '<' email '>' ts tz lf
  21     commit_msg
  22     ('from' sp (ref_str | hexsha1 | sha1exp_str | idnum) lf)?
  23     ('merge' sp (ref_str | hexsha1 | sha1exp_str | idnum) lf)*
  24     file_change*
  25     lf;
  26   commit_msg ::= data;
  27
  28   file_change ::= 'M' sp mode sp (hexsha1 | idnum) sp path_str lf
  29                 | 'D' sp path_str lf
  30                 ;
  31   mode ::= '644' | '755';
  32
  33   new_tag ::= 'tag' sp tag_str lf
  34     'from' sp (ref_str | hexsha1 | sha1exp_str | idnum) lf
  35         'tagger' sp name '<' email '>' ts tz lf
  36     tag_msg;
  37   tag_msg ::= data;
  38
  39   reset_branch ::= 'reset' sp ref_str lf
  40     ('from' sp (ref_str | hexsha1 | sha1exp_str | idnum) lf)?
  41     lf;
  42
  43   checkpoint ::= 'checkpoint' lf
  44     lf;
  45
  46      # note: the first idnum in a stream should be 1 and subsequent
  47      # idnums should not have gaps between values as this will cause
  48      # the stream parser to reserve space for the gapped values.  An
  49          # idnum can be updated in the future to a new object by issuing
  50      # a new mark directive with the old idnum.
  51          #
  52   mark ::= 'mark' sp idnum lf;
  53
  54      # note: declen indicates the length of binary_data in bytes.
  55      # declen does not include the lf preceeding or trailing the
  56      # binary data.
  57      #
  58   data ::= 'data' sp declen lf
  59     binary_data
  60         lf;
  61
  62      # note: quoted strings are C-style quoting supporting \c for
  63      # common escapes of 'c' (e..g \n, \t, \\, \") or \nnn where nnn
  64          # is the signed byte value in octal.  Note that the only
  65      # characters which must actually be escaped to protect the
  66      # stream formatting is: \, " and LF.  Otherwise these values
  67          # are UTF8.
  68      #
  69   ref_str     ::= ref     | '"' quoted(ref)     '"' ;
  70   sha1exp_str ::= sha1exp | '"' quoted(sha1exp) '"' ;
  71   tag_str     ::= tag     | '"' quoted(tag)     '"' ;
  72   path_str    ::= path    | '"' quoted(path)    '"' ;
  73
  74   declen ::= # unsigned 32 bit value, ascii base10 notation;
  75   binary_data ::= # file content, not interpreted;
  76
  77   sp ::= # ASCII space character;
  78   lf ::= # ASCII newline (LF) character;
  79
  80      # note: a colon (':') must precede the numerical value assigned to
  81          # an idnum.  This is to distinguish it from a ref or tag name as
  82      # GIT does not permit ':' in ref or tag strings.
  83          #
  84   idnum   ::= ':' declen;
  85   path    ::= # GIT style file path, e.g. "a/b/c";
  86   ref     ::= # GIT ref name, e.g. "refs/heads/MOZ_GECKO_EXPERIMENT";
  87   tag     ::= # GIT tag name, e.g. "FIREFOX_1_5";
  88   sha1exp ::= # Any valid GIT SHA1 expression;
  89   hexsha1 ::= # SHA1 in hexadecimal format;
  90
  91      # note: name and email are UTF8 strings, however name must not
  92          # contain '<' or lf and email must not contain any of the
  93      # following: '<', '>', lf.
  94          #
  95   name  ::= # valid GIT author/committer name;
  96   email ::= # valid GIT author/committer email;
  97   ts    ::= # time since the epoch in seconds, ascii base10 notation;
  98   tz    ::= # GIT style timezone;
  99 */
 100
 101 #include "builtin.h"
 102 #include "cache.h"
 103 #include "object.h"
 104 #include "blob.h"
 105 #include "tree.h"
 106 #include "delta.h"
 107 #include "pack.h"
 108 #include "refs.h"
 109 #include "csum-file.h"
 110 #include "strbuf.h"
 111 #include "quote.h"
 112
 113 struct object_entry
 114 {
 115         struct object_entry *next;
 116         unsigned long offset;
 117         unsigned type : TYPE_BITS;
 118         unsigned pack_id : 16;
 119         unsigned char sha1[20];
 120 };
 121
 122 struct object_entry_pool
 123 {
 124         struct object_entry_pool *next_pool;
 125         struct object_entry *next_free;
 126         struct object_entry *end;
 127         struct object_entry entries[FLEX_ARRAY]; /* more */
 128 };
 129
 130 struct mark_set
 131 {
 132         int shift;
 133         union {
 134                 struct object_entry *marked[1024];
 135                 struct mark_set *sets[1024];
 136         } data;
 137 };
 138
 139 struct last_object
 140 {
 141         void *data;
 142         unsigned long len;
 143         unsigned long offset;
 144         unsigned int depth;
 145         unsigned no_free:1;
 146 };
 147
 148 struct mem_pool
 149 {
 150         struct mem_pool *next_pool;
 151         char *next_free;
 152         char *end;
 153         char space[FLEX_ARRAY]; /* more */
 154 };
 155
 156 struct atom_str
 157 {
 158         struct atom_str *next_atom;
 159         int str_len;
 160         char str_dat[FLEX_ARRAY]; /* more */
 161 };
 162
 163 struct tree_content;
 164 struct tree_entry
 165 {
 166         struct tree_content *tree;
 167         struct atom_str* name;
 168         struct tree_entry_ms
 169         {
 170                 unsigned int mode;
 171                 unsigned char sha1[20];
 172         } versions[2];
 173 };
 174
 175 struct tree_content
 176 {
 177         unsigned int entry_capacity; /* must match avail_tree_content */
 178         unsigned int entry_count;
 179         unsigned int delta_depth;
 180         struct tree_entry *entries[FLEX_ARRAY]; /* more */
 181 };
 182
 183 struct avail_tree_content
 184 {
 185         unsigned int entry_capacity; /* must match tree_content */
 186         struct avail_tree_content *next_avail;
 187 };
 188
 189 struct branch
 190 {
 191         struct branch *table_next_branch;
 192         struct branch *active_next_branch;
 193         const char *name;
 194         unsigned long last_commit;
 195         struct tree_entry branch_tree;
 196         unsigned char sha1[20];
 197 };
 198
 199 struct tag
 200 {
 201         struct tag *next_tag;
 202         const char *name;
 203         unsigned char sha1[20];
 204 };
 205
 206 struct dbuf
 207 {
 208         void *buffer;
 209         size_t capacity;
 210 };
 211
 212 struct hash_list
 213 {
 214         struct hash_list *next;
 215         unsigned char sha1[20];
 216 };
 217
 218 /* Stats and misc. counters */
 219 static unsigned long max_depth = 10;
 220 static unsigned long alloc_count;
 221 static unsigned long branch_count;
 222 static unsigned long branch_load_count;
 223 static unsigned long object_count;
 224 static unsigned long marks_set_count;
 225 static unsigned long object_count_by_type[1 << TYPE_BITS];
 226 static unsigned long duplicate_count_by_type[1 << TYPE_BITS];
 227 static unsigned long delta_count_by_type[1 << TYPE_BITS];
 228
 229 /* Memory pools */
 230 static size_t mem_pool_alloc = 2*1024*1024 - sizeof(struct mem_pool);
 231 static size_t total_allocd;
 232 static struct mem_pool *mem_pool;
 233
 234 /* Atom management */
 235 static unsigned int atom_table_sz = 4451;
 236 static unsigned int atom_cnt;
 237 static struct atom_str **atom_table;
 238
 239 /* The .pack file being generated */
 240 static const char *base_name;
 241 static unsigned int pack_id;
 242 static char *idx_name;
 243 static struct packed_git *pack_data;
 244 static struct packed_git **all_packs;
 245 static int pack_fd;
 246 static unsigned long pack_size;
 247 static unsigned char pack_sha1[20];
 248
 249 /* Table of objects we've written. */
 250 static unsigned int object_entry_alloc = 5000;
 251 static struct object_entry_pool *blocks;
 252 static struct object_entry *object_table[1 << 16];
 253 static struct mark_set *marks;
 254 static const char* mark_file;
 255
 256 /* Our last blob */
 257 static struct last_object last_blob;
 258
 259 /* Tree management */
 260 static unsigned int tree_entry_alloc = 1000;
 261 static void *avail_tree_entry;
 262 static unsigned int avail_tree_table_sz = 100;
 263 static struct avail_tree_content **avail_tree_table;
 264 static struct dbuf old_tree;
 265 static struct dbuf new_tree;
 266
 267 /* Branch data */
 268 static unsigned long max_active_branches = 5;
 269 static unsigned long cur_active_branches;
 270 static unsigned long branch_table_sz = 1039;
 271 static struct branch **branch_table;
 272 static struct branch *active_branches;
 273
 274 /* Tag data */
 275 static struct tag *first_tag;
 276 static struct tag *last_tag;
 277
 278 /* Input stream parsing */
 279 static struct strbuf command_buf;
 280 static unsigned long next_mark;
 281 static struct dbuf new_data;
 282 static FILE* branch_log;
 283
 284
 285 static void alloc_objects(unsigned int cnt)
 286 {
 287         struct object_entry_pool *b;
 288
 289         b = xmalloc(sizeof(struct object_entry_pool)
 290                 + cnt * sizeof(struct object_entry));
 291         b->next_pool = blocks;
 292         b->next_free = b->entries;
 293         b->end = b->entries + cnt;
 294         blocks = b;
 295         alloc_count += cnt;
 296 }
 297
 298 static struct object_entry* new_object(unsigned char *sha1)
 299 {
 300         struct object_entry *e;
 301
 302         if (blocks->next_free == blocks->end)
 303                 alloc_objects(object_entry_alloc);
 304
 305         e = blocks->next_free++;
 306         e->pack_id = pack_id;
 307         hashcpy(e->sha1, sha1);
 308         return e;
 309 }
 310
 311 static struct object_entry* find_object(unsigned char *sha1)
 312 {
 313         unsigned int h = sha1[0] << 8 | sha1[1];
 314         struct object_entry *e;
 315         for (e = object_table[h]; e; e = e->next)
 316                 if (!hashcmp(sha1, e->sha1))
 317                         return e;
 318         return NULL;
 319 }
 320
 321 static struct object_entry* insert_object(unsigned char *sha1)
 322 {
 323         unsigned int h = sha1[0] << 8 | sha1[1];
 324         struct object_entry *e = object_table[h];
 325         struct object_entry *p = NULL;
 326
 327         while (e) {
 328                 if (!hashcmp(sha1, e->sha1))
 329                         return e;
 330                 p = e;
 331                 e = e->next;
 332         }
 333
 334         e = new_object(sha1);
 335         e->next = NULL;
 336         e->offset = 0;
 337         if (p)
 338                 p->next = e;
 339         else
 340                 object_table[h] = e;
 341         return e;
 342 }
 343
 344 static unsigned int hc_str(const char *s, size_t len)
 345 {
 346         unsigned int r = 0;
 347         while (len-- > 0)
 348                 r = r * 31 + *s++;
 349         return r;
 350 }
 351
 352 static void* pool_alloc(size_t len)
 353 {
 354         struct mem_pool *p;
 355         void *r;
 356
 357         for (p = mem_pool; p; p = p->next_pool)
 358                 if ((p->end - p->next_free >= len))
 359                         break;
 360
 361         if (!p) {
 362                 if (len >= (mem_pool_alloc/2)) {
 363                         total_allocd += len;
 364                         return xmalloc(len);
 365                 }
 366                 total_allocd += sizeof(struct mem_pool) + mem_pool_alloc;
 367                 p = xmalloc(sizeof(struct mem_pool) + mem_pool_alloc);
 368                 p->next_pool = mem_pool;
 369                 p->next_free = p->space;
 370                 p->end = p->next_free + mem_pool_alloc;
 371                 mem_pool = p;
 372         }
 373
 374         r = p->next_free;
 375         /* round out to a pointer alignment */
 376         if (len & (sizeof(void*) - 1))
 377                 len += sizeof(void*) - (len & (sizeof(void*) - 1));
 378         p->next_free += len;
 379         return r;
 380 }
 381
 382 static void* pool_calloc(size_t count, size_t size)
 383 {
 384         size_t len = count * size;
 385         void *r = pool_alloc(len);
 386         memset(r, 0, len);
 387         return r;
 388 }
 389
 390 static char* pool_strdup(const char *s)
 391 {
 392         char *r = pool_alloc(strlen(s) + 1);
 393         strcpy(r, s);
 394         return r;
 395 }
 396
 397 static void size_dbuf(struct dbuf *b, size_t maxlen)
 398 {
 399         if (b->buffer) {
 400                 if (b->capacity >= maxlen)
 401                         return;
 402                 free(b->buffer);
 403         }
 404         b->capacity = ((maxlen / 1024) + 1) * 1024;
 405         b->buffer = xmalloc(b->capacity);
 406 }
 407
 408 static void insert_mark(unsigned long idnum, struct object_entry *oe)
 409 {
 410         struct mark_set *s = marks;
 411         while ((idnum >> s->shift) >= 1024) {
 412                 s = pool_calloc(1, sizeof(struct mark_set));
 413                 s->shift = marks->shift + 10;
 414                 s->data.sets[0] = marks;
 415                 marks = s;
 416         }
 417         while (s->shift) {
 418                 unsigned long i = idnum >> s->shift;
 419                 idnum -= i << s->shift;
 420                 if (!s->data.sets[i]) {
 421                         s->data.sets[i] = pool_calloc(1, sizeof(struct mark_set));
 422                         s->data.sets[i]->shift = s->shift - 10;
 423                 }
 424                 s = s->data.sets[i];
 425         }
 426         if (!s->data.marked[idnum])
 427                 marks_set_count++;
 428         s->data.marked[idnum] = oe;
 429 }
 430
 431 static struct object_entry* find_mark(unsigned long idnum)
 432 {
 433         unsigned long orig_idnum = idnum;
 434         struct mark_set *s = marks;
 435         struct object_entry *oe = NULL;
 436         if ((idnum >> s->shift) < 1024) {
 437                 while (s && s->shift) {
 438                         unsigned long i = idnum >> s->shift;
 439                         idnum -= i << s->shift;
 440                         s = s->data.sets[i];
 441                 }
 442                 if (s)
 443                         oe = s->data.marked[idnum];
 444         }
 445         if (!oe)
 446                 die("mark :%lu not declared", orig_idnum);
 447         return oe;
 448 }
 449
 450 static struct atom_str* to_atom(const char *s, size_t len)
 451 {
 452         unsigned int hc = hc_str(s, len) % atom_table_sz;
 453         struct atom_str *c;
 454
 455         for (c = atom_table[hc]; c; c = c->next_atom)
 456                 if (c->str_len == len && !strncmp(s, c->str_dat, len))
 457                         return c;
 458
 459         c = pool_alloc(sizeof(struct atom_str) + len + 1);
 460         c->str_len = len;
 461         strncpy(c->str_dat, s, len);
 462         c->str_dat[len] = 0;
 463         c->next_atom = atom_table[hc];
 464         atom_table[hc] = c;
 465         atom_cnt++;
 466         return c;
 467 }
 468
 469 static struct branch* lookup_branch(const char *name)
 470 {
 471         unsigned int hc = hc_str(name, strlen(name)) % branch_table_sz;
 472         struct branch *b;
 473
 474         for (b = branch_table[hc]; b; b = b->table_next_branch)
 475                 if (!strcmp(name, b->name))
 476                         return b;
 477         return NULL;
 478 }
 479
 480 static struct branch* new_branch(const char *name)
 481 {
 482         unsigned int hc = hc_str(name, strlen(name)) % branch_table_sz;
 483         struct branch* b = lookup_branch(name);
 484
 485         if (b)
 486                 die("Invalid attempt to create duplicate branch: %s", name);
 487         if (check_ref_format(name))
 488                 die("Branch name doesn't conform to GIT standards: %s", name);
 489
 490         b = pool_calloc(1, sizeof(struct branch));
 491         b->name = pool_strdup(name);
 492         b->table_next_branch = branch_table[hc];
 493         b->branch_tree.versions[0].mode = S_IFDIR;
 494         b->branch_tree.versions[1].mode = S_IFDIR;
 495         branch_table[hc] = b;
 496         branch_count++;
 497         return b;
 498 }
 499
 500 static unsigned int hc_entries(unsigned int cnt)
 501 {
 502         cnt = cnt & 7 ? (cnt / 8) + 1 : cnt / 8;
 503         return cnt < avail_tree_table_sz ? cnt : avail_tree_table_sz - 1;
 504 }
 505
 506 static struct tree_content* new_tree_content(unsigned int cnt)
 507 {
 508         struct avail_tree_content *f, *l = NULL;
 509         struct tree_content *t;
 510         unsigned int hc = hc_entries(cnt);
 511
 512         for (f = avail_tree_table[hc]; f; l = f, f = f->next_avail)
 513                 if (f->entry_capacity >= cnt)
 514                         break;
 515
 516         if (f) {
 517                 if (l)
 518                         l->next_avail = f->next_avail;
 519                 else
 520                         avail_tree_table[hc] = f->next_avail;
 521         } else {
 522                 cnt = cnt & 7 ? ((cnt / 8) + 1) * 8 : cnt;
 523                 f = pool_alloc(sizeof(*t) + sizeof(t->entries[0]) * cnt);
 524                 f->entry_capacity = cnt;
 525         }
 526
 527         t = (struct tree_content*)f;
 528         t->entry_count = 0;
 529         t->delta_depth = 0;
 530         return t;
 531 }
 532
 533 static void release_tree_entry(struct tree_entry *e);
 534 static void release_tree_content(struct tree_content *t)
 535 {
 536         struct avail_tree_content *f = (struct avail_tree_content*)t;
 537         unsigned int hc = hc_entries(f->entry_capacity);
 538         f->next_avail = avail_tree_table[hc];
 539         avail_tree_table[hc] = f;
 540 }
 541
 542 static void release_tree_content_recursive(struct tree_content *t)
 543 {
 544         unsigned int i;
 545         for (i = 0; i < t->entry_count; i++)
 546                 release_tree_entry(t->entries[i]);
 547         release_tree_content(t);
 548 }
 549
 550 static struct tree_content* grow_tree_content(
 551         struct tree_content *t,
 552         int amt)
 553 {
 554         struct tree_content *r = new_tree_content(t->entry_count + amt);
 555         r->entry_count = t->entry_count;
 556         r->delta_depth = t->delta_depth;
 557         memcpy(r->entries,t->entries,t->entry_count*sizeof(t->entries[0]));
 558         release_tree_content(t);
 559         return r;
 560 }
 561
 562 static struct tree_entry* new_tree_entry()
 563 {
 564         struct tree_entry *e;
 565
 566         if (!avail_tree_entry) {
 567                 unsigned int n = tree_entry_alloc;
 568                 total_allocd += n * sizeof(struct tree_entry);
 569                 avail_tree_entry = e = xmalloc(n * sizeof(struct tree_entry));
 570                 while (n-- > 1) {
 571                         *((void**)e) = e + 1;
 572                         e++;
 573                 }
 574                 *((void**)e) = NULL;
 575         }
 576
 577         e = avail_tree_entry;
 578         avail_tree_entry = *((void**)e);
 579         return e;
 580 }
 581
 582 static void release_tree_entry(struct tree_entry *e)
 583 {
 584         if (e->tree)
 585                 release_tree_content_recursive(e->tree);
 586         *((void**)e) = avail_tree_entry;
 587         avail_tree_entry = e;
 588 }
 589
 590 static void yread(int fd, void *buffer, size_t length)
 591 {
 592         ssize_t ret = 0;
 593         while (ret < length) {
 594                 ssize_t size = xread(fd, (char *) buffer + ret, length - ret);
 595                 if (!size)
 596                         die("Read from descriptor %i: end of stream", fd);
 597                 if (size < 0)
 598                         die("Read from descriptor %i: %s", fd, strerror(errno));
 599                 ret += size;
 600         }
 601 }
 602
 603 static void start_packfile()
 604 {
 605         struct packed_git *p;
 606         struct pack_header hdr;
 607
 608         idx_name = xmalloc(strlen(base_name) + 11);
 609         p = xcalloc(1, sizeof(*p) + strlen(base_name) + 13);
 610         sprintf(p->pack_name, "%s%5.5i.pack", base_name, pack_id + 1);
 611         sprintf(idx_name, "%s%5.5i.idx", base_name, pack_id + 1);
 612
 613         pack_fd = open(p->pack_name, O_RDWR|O_CREAT|O_EXCL, 0666);
 614         if (pack_fd < 0)
 615                 die("Can't create %s: %s", p->pack_name, strerror(errno));
 616         p->pack_fd = pack_fd;
 617
 618         hdr.hdr_signature = htonl(PACK_SIGNATURE);
 619         hdr.hdr_version = htonl(2);
 620         hdr.hdr_entries = 0;
 621         write_or_die(pack_fd, &hdr, sizeof(hdr));
 622
 623         pack_data = p;
 624         pack_size = sizeof(hdr);
 625         object_count = 0;
 626
 627         all_packs = xrealloc(all_packs, sizeof(*all_packs) * (pack_id + 1));
 628         all_packs[pack_id] = p;
 629 }
 630
 631 static void fixup_header_footer()
 632 {
 633         SHA_CTX c;
 634         char hdr[8];
 635         unsigned long cnt;
 636         char *buf;
 637
 638         if (lseek(pack_fd, 0, SEEK_SET) != 0)
 639                 die("Failed seeking to start: %s", strerror(errno));
 640
 641         SHA1_Init(&c);
 642         yread(pack_fd, hdr, 8);
 643         SHA1_Update(&c, hdr, 8);
 644
 645         cnt = htonl(object_count);
 646         SHA1_Update(&c, &cnt, 4);
 647         write_or_die(pack_fd, &cnt, 4);
 648
 649         buf = xmalloc(128 * 1024);
 650         for (;;) {
 651                 size_t n = xread(pack_fd, buf, 128 * 1024);
 652                 if (n <= 0)
 653                         break;
 654                 SHA1_Update(&c, buf, n);
 655         }
 656         free(buf);
 657
 658         SHA1_Final(pack_sha1, &c);
 659         write_or_die(pack_fd, pack_sha1, sizeof(pack_sha1));
 660 }
 661
 662 static int oecmp (const void *a_, const void *b_)
 663 {
 664         struct object_entry *a = *((struct object_entry**)a_);
 665         struct object_entry *b = *((struct object_entry**)b_);
 666         return hashcmp(a->sha1, b->sha1);
 667 }
 668
 669 static void write_index(const char *idx_name)
 670 {
 671         struct sha1file *f;
 672         struct object_entry **idx, **c, **last, *e;
 673         struct object_entry_pool *o;
 674         unsigned int array[256];
 675         int i;
 676
 677         /* Build the sorted table of object IDs. */
 678         idx = xmalloc(object_count * sizeof(struct object_entry*));
 679         c = idx;
 680         for (o = blocks; o; o = o->next_pool)
 681                 for (e = o->entries; e != o->next_free; e++)
 682                         if (pack_id == e->pack_id)
 683                                 *c++ = e;
 684         last = idx + object_count;
 685         qsort(idx, object_count, sizeof(struct object_entry*), oecmp);
 686
 687         /* Generate the fan-out array. */
 688         c = idx;
 689         for (i = 0; i < 256; i++) {
 690                 struct object_entry **next = c;;
 691                 while (next < last) {
 692                         if ((*next)->sha1[0] != i)
 693                                 break;
 694                         next++;
 695                 }
 696                 array[i] = htonl(next - idx);
 697                 c = next;
 698         }
 699
 700         f = sha1create("%s", idx_name);
 701         sha1write(f, array, 256 * sizeof(int));
 702         for (c = idx; c != last; c++) {
 703                 unsigned int offset = htonl((*c)->offset);
 704                 sha1write(f, &offset, 4);
 705                 sha1write(f, (*c)->sha1, sizeof((*c)->sha1));
 706         }
 707         sha1write(f, pack_sha1, sizeof(pack_sha1));
 708         sha1close(f, NULL, 1);
 709         free(idx);
 710 }
 711
 712 static void end_packfile()
 713 {
 714         struct packed_git *old_p = pack_data, *new_p;
 715
 716         if (object_count) {
 717                 fixup_header_footer();
 718                 write_index(idx_name);
 719
 720                 /* Register the packfile with core git's machinary. */
 721                 new_p = add_packed_git(idx_name, strlen(idx_name), 1);
 722                 if (!new_p)
 723                         die("core git rejected index %s", idx_name);
 724                 new_p->windows = old_p->windows;
 725                 new_p->pack_fd = old_p->pack_fd;
 726                 all_packs[pack_id++] = new_p;
 727                 install_packed_git(new_p);
 728         }
 729         else {
 730                 close(pack_fd);
 731                 unlink(old_p->pack_name);
 732         }
 733         free(old_p);
 734         free(idx_name);
 735
 736         /* We can't carry a delta across packfiles. */
 737         free(last_blob.data);
 738         last_blob.data = NULL;
 739         last_blob.len = 0;
 740         last_blob.offset = 0;
 741         last_blob.depth = 0;
 742 }
 743
 744 static size_t encode_header(
 745         enum object_type type,
 746         size_t size,
 747         unsigned char *hdr)
 748 {
 749         int n = 1;
 750         unsigned char c;
 751
 752         if (type < OBJ_COMMIT || type > OBJ_REF_DELTA)
 753                 die("bad type %d", type);
 754
 755         c = (type << 4) | (size & 15);
 756         size >>= 4;
 757         while (size) {
 758                 *hdr++ = c | 0x80;
 759                 c = size & 0x7f;
 760                 size >>= 7;
 761                 n++;
 762         }
 763         *hdr = c;
 764         return n;
 765 }
 766
 767 static int store_object(
 768         enum object_type type,
 769         void *dat,
 770         size_t datlen,
 771         struct last_object *last,
 772         unsigned char *sha1out,
 773         unsigned long mark)
 774 {
 775         void *out, *delta;
 776         struct object_entry *e;
 777         unsigned char hdr[96];
 778         unsigned char sha1[20];
 779         unsigned long hdrlen, deltalen;
 780         SHA_CTX c;
 781         z_stream s;
 782
 783         hdrlen = sprintf((char*)hdr,"%s %lu",type_names[type],datlen) + 1;
 784         SHA1_Init(&c);
 785         SHA1_Update(&c, hdr, hdrlen);
 786         SHA1_Update(&c, dat, datlen);
 787         SHA1_Final(sha1, &c);
 788         if (sha1out)
 789                 hashcpy(sha1out, sha1);
 790
 791         e = insert_object(sha1);
 792         if (mark)
 793                 insert_mark(mark, e);
 794         if (e->offset) {
 795                 duplicate_count_by_type[type]++;
 796                 return 1;
 797         }
 798         e->type = type;
 799         e->offset = pack_size;
 800         object_count++;
 801         object_count_by_type[type]++;
 802
 803         if (last && last->data && last->depth < max_depth)
 804                 delta = diff_delta(last->data, last->len,
 805                         dat, datlen,
 806                         &deltalen, 0);
 807         else
 808                 delta = 0;
 809
 810         memset(&s, 0, sizeof(s));
 811         deflateInit(&s, zlib_compression_level);
 812
 813         if (delta) {
 814                 unsigned long ofs = e->offset - last->offset;
 815                 unsigned pos = sizeof(hdr) - 1;
 816
 817                 delta_count_by_type[type]++;
 818                 last->depth++;
 819                 s.next_in = delta;
 820                 s.avail_in = deltalen;
 821
 822                 hdrlen = encode_header(OBJ_OFS_DELTA, deltalen, hdr);
 823                 write_or_die(pack_fd, hdr, hdrlen);
 824                 pack_size += hdrlen;
 825
 826                 hdr[pos] = ofs & 127;
 827                 while (ofs >>= 7)
 828                         hdr[--pos] = 128 | (--ofs & 127);
 829                 write_or_die(pack_fd, hdr + pos, sizeof(hdr) - pos);
 830                 pack_size += sizeof(hdr) - pos;
 831         } else {
 832                 if (last)
 833                         last->depth = 0;
 834                 s.next_in = dat;
 835                 s.avail_in = datlen;
 836                 hdrlen = encode_header(type, datlen, hdr);
 837                 write_or_die(pack_fd, hdr, hdrlen);
 838                 pack_size += hdrlen;
 839         }
 840
 841         s.avail_out = deflateBound(&s, s.avail_in);
 842         s.next_out = out = xmalloc(s.avail_out);
 843         while (deflate(&s, Z_FINISH) == Z_OK)
 844                 /* nothing */;
 845         deflateEnd(&s);
 846
 847         write_or_die(pack_fd, out, s.total_out);
 848         pack_size += s.total_out;
 849
 850         free(out);
 851         if (delta)
 852                 free(delta);
 853         if (last) {
 854                 if (last->data && !last->no_free)
 855                         free(last->data);
 856                 last->data = dat;
 857                 last->offset = e->offset;
 858                 last->len = datlen;
 859         }
 860         return 0;
 861 }
 862
 863 static void *gfi_unpack_entry(
 864         struct object_entry *oe,
 865         unsigned long *sizep)
 866 {
 867         static char type[20];
 868         struct packed_git *p = all_packs[oe->pack_id];
 869         if (p == pack_data)
 870                 p->pack_size = pack_size + 20;
 871         return unpack_entry(p, oe->offset, type, sizep);
 872 }
 873
 874 static const char *get_mode(const char *str, unsigned int *modep)
 875 {
 876         unsigned char c;
 877         unsigned int mode = 0;
 878
 879         while ((c = *str++) != ' ') {
 880                 if (c < '0' || c > '7')
 881                         return NULL;
 882                 mode = (mode << 3) + (c - '0');
 883         }
 884         *modep = mode;
 885         return str;
 886 }
 887
 888 static void load_tree(struct tree_entry *root)
 889 {
 890         unsigned char* sha1 = root->versions[1].sha1;
 891         struct object_entry *myoe;
 892         struct tree_content *t;
 893         unsigned long size;
 894         char *buf;
 895         const char *c;
 896
 897         root->tree = t = new_tree_content(8);
 898         if (is_null_sha1(sha1))
 899                 return;
 900
 901         myoe = find_object(sha1);
 902         if (myoe) {
 903                 if (myoe->type != OBJ_TREE)
 904                         die("Not a tree: %s", sha1_to_hex(sha1));
 905                 t->delta_depth = 0;
 906                 buf = gfi_unpack_entry(myoe, &size);
 907         } else {
 908                 char type[20];
 909                 buf = read_sha1_file(sha1, type, &size);
 910                 if (!buf || strcmp(type, tree_type))
 911                         die("Can't load tree %s", sha1_to_hex(sha1));
 912         }
 913
 914         c = buf;
 915         while (c != (buf + size)) {
 916                 struct tree_entry *e = new_tree_entry();
 917
 918                 if (t->entry_count == t->entry_capacity)
 919                         root->tree = t = grow_tree_content(t, 8);
 920                 t->entries[t->entry_count++] = e;
 921
 922                 e->tree = NULL;
 923                 c = get_mode(c, &e->versions[1].mode);
 924                 if (!c)
 925                         die("Corrupt mode in %s", sha1_to_hex(sha1));
 926                 e->versions[0].mode = e->versions[1].mode;
 927                 e->name = to_atom(c, strlen(c));
 928                 c += e->name->str_len + 1;
 929                 hashcpy(e->versions[0].sha1, (unsigned char*)c);
 930                 hashcpy(e->versions[1].sha1, (unsigned char*)c);
 931                 c += 20;
 932         }
 933         free(buf);
 934 }
 935
 936 static int tecmp0 (const void *_a, const void *_b)
 937 {
 938         struct tree_entry *a = *((struct tree_entry**)_a);
 939         struct tree_entry *b = *((struct tree_entry**)_b);
 940         return base_name_compare(
 941                 a->name->str_dat, a->name->str_len, a->versions[0].mode,
 942                 b->name->str_dat, b->name->str_len, b->versions[0].mode);
 943 }
 944
 945 static int tecmp1 (const void *_a, const void *_b)
 946 {
 947         struct tree_entry *a = *((struct tree_entry**)_a);
 948         struct tree_entry *b = *((struct tree_entry**)_b);
 949         return base_name_compare(
 950                 a->name->str_dat, a->name->str_len, a->versions[1].mode,
 951                 b->name->str_dat, b->name->str_len, b->versions[1].mode);
 952 }
 953
 954 static void mktree(struct tree_content *t,
 955         int v,
 956         unsigned long *szp,
 957         struct dbuf *b)
 958 {
 959         size_t maxlen = 0;
 960         unsigned int i;
 961         char *c;
 962
 963         if (!v)
 964                 qsort(t->entries,t->entry_count,sizeof(t->entries[0]),tecmp0);
 965         else
 966                 qsort(t->entries,t->entry_count,sizeof(t->entries[0]),tecmp1);
 967
 968         for (i = 0; i < t->entry_count; i++) {
 969                 if (t->entries[i]->versions[v].mode)
 970                         maxlen += t->entries[i]->name->str_len + 34;
 971         }
 972
 973         size_dbuf(b, maxlen);
 974         c = b->buffer;
 975         for (i = 0; i < t->entry_count; i++) {
 976                 struct tree_entry *e = t->entries[i];
 977                 if (!e->versions[v].mode)
 978                         continue;
 979                 c += sprintf(c, "%o", e->versions[v].mode);
 980                 *c++ = ' ';
 981                 strcpy(c, e->name->str_dat);
 982                 c += e->name->str_len + 1;
 983                 hashcpy((unsigned char*)c, e->versions[v].sha1);
 984                 c += 20;
 985         }
 986         *szp = c - (char*)b->buffer;
 987 }
 988
 989 static void store_tree(struct tree_entry *root)
 990 {
 991         struct tree_content *t = root->tree;
 992         unsigned int i, j, del;
 993         unsigned long new_len;
 994         struct last_object lo;
 995         struct object_entry *le;
 996
 997         if (!is_null_sha1(root->versions[1].sha1))
 998                 return;
 999
1000         for (i = 0; i < t->entry_count; i++) {
1001                 if (t->entries[i]->tree)
1002                         store_tree(t->entries[i]);
1003         }
1004
1005         le = find_object(root->versions[0].sha1);
1006         if (!S_ISDIR(root->versions[0].mode)
1007                 || !le
1008                 || le->pack_id != pack_id) {
1009                 lo.data = NULL;
1010                 lo.depth = 0;
1011         } else {
1012                 mktree(t, 0, &lo.len, &old_tree);
1013                 lo.data = old_tree.buffer;
1014                 lo.offset = le->offset;
1015                 lo.depth = t->delta_depth;
1016                 lo.no_free = 1;
1017         }
1018
1019         mktree(t, 1, &new_len, &new_tree);
1020         store_object(OBJ_TREE, new_tree.buffer, new_len,
1021                 &lo, root->versions[1].sha1, 0);
1022
1023         t->delta_depth = lo.depth;
1024         for (i = 0, j = 0, del = 0; i < t->entry_count; i++) {
1025                 struct tree_entry *e = t->entries[i];
1026                 if (e->versions[1].mode) {
1027                         e->versions[0].mode = e->versions[1].mode;
1028                         hashcpy(e->versions[0].sha1, e->versions[1].sha1);
1029                         t->entries[j++] = e;
1030                 } else {
1031                         release_tree_entry(e);
1032                         del++;
1033                 }
1034         }
1035         t->entry_count -= del;
1036 }
1037
1038 static int tree_content_set(
1039         struct tree_entry *root,
1040         const char *p,
1041         const unsigned char *sha1,
1042         const unsigned int mode)
1043 {
1044         struct tree_content *t = root->tree;
1045         const char *slash1;
1046         unsigned int i, n;
1047         struct tree_entry *e;
1048
1049         slash1 = strchr(p, '/');
1050         if (slash1)
1051                 n = slash1 - p;
1052         else
1053                 n = strlen(p);
1054
1055         for (i = 0; i < t->entry_count; i++) {
1056                 e = t->entries[i];
1057                 if (e->name->str_len == n && !strncmp(p, e->name->str_dat, n)) {
1058                         if (!slash1) {
1059                                 if (e->versions[1].mode == mode
1060                                                 && !hashcmp(e->versions[1].sha1, sha1))
1061                                         return 0;
1062                                 e->versions[1].mode = mode;
1063                                 hashcpy(e->versions[1].sha1, sha1);
1064                                 if (e->tree) {
1065                                         release_tree_content_recursive(e->tree);
1066                                         e->tree = NULL;
1067                                 }
1068                                 hashclr(root->versions[1].sha1);
1069                                 return 1;
1070                         }
1071                         if (!S_ISDIR(e->versions[1].mode)) {
1072                                 e->tree = new_tree_content(8);
1073                                 e->versions[1].mode = S_IFDIR;
1074                         }
1075                         if (!e->tree)
1076                                 load_tree(e);
1077                         if (tree_content_set(e, slash1 + 1, sha1, mode)) {
1078                                 hashclr(root->versions[1].sha1);
1079                                 return 1;
1080                         }
1081                         return 0;
1082                 }
1083         }
1084
1085         if (t->entry_count == t->entry_capacity)
1086                 root->tree = t = grow_tree_content(t, 8);
1087         e = new_tree_entry();
1088         e->name = to_atom(p, n);
1089         e->versions[0].mode = 0;
1090         hashclr(e->versions[0].sha1);
1091         t->entries[t->entry_count++] = e;
1092         if (slash1) {
1093                 e->tree = new_tree_content(8);
1094                 e->versions[1].mode = S_IFDIR;
1095                 tree_content_set(e, slash1 + 1, sha1, mode);
1096         } else {
1097                 e->tree = NULL;
1098                 e->versions[1].mode = mode;
1099                 hashcpy(e->versions[1].sha1, sha1);
1100         }
1101         hashclr(root->versions[1].sha1);
1102         return 1;
1103 }
1104
1105 static int tree_content_remove(struct tree_entry *root, const char *p)
1106 {
1107         struct tree_content *t = root->tree;
1108         const char *slash1;
1109         unsigned int i, n;
1110         struct tree_entry *e;
1111
1112         slash1 = strchr(p, '/');
1113         if (slash1)
1114                 n = slash1 - p;
1115         else
1116                 n = strlen(p);
1117
1118         for (i = 0; i < t->entry_count; i++) {
1119                 e = t->entries[i];
1120                 if (e->name->str_len == n && !strncmp(p, e->name->str_dat, n)) {
1121                         if (!slash1 || !S_ISDIR(e->versions[1].mode))
1122                                 goto del_entry;
1123                         if (!e->tree)
1124                                 load_tree(e);
1125                         if (tree_content_remove(e, slash1 + 1)) {
1126                                 for (n = 0; n < e->tree->entry_count; n++) {
1127                                         if (e->tree->entries[n]->versions[1].mode) {
1128                                                 hashclr(root->versions[1].sha1);
1129                                                 return 1;
1130                                         }
1131                                 }
1132                                 goto del_entry;
1133                         }
1134                         return 0;
1135                 }
1136         }
1137         return 0;
1138
1139 del_entry:
1140         if (e->tree) {
1141                 release_tree_content_recursive(e->tree);
1142                 e->tree = NULL;
1143         }
1144         e->versions[1].mode = 0;
1145         hashclr(e->versions[1].sha1);
1146         hashclr(root->versions[1].sha1);
1147         return 1;
1148 }
1149
1150 static void dump_branches()
1151 {
1152         static const char *msg = "fast-import";
1153         unsigned int i;
1154         struct branch *b;
1155         struct ref_lock *lock;
1156
1157         for (i = 0; i < branch_table_sz; i++) {
1158                 for (b = branch_table[i]; b; b = b->table_next_branch) {
1159                         lock = lock_any_ref_for_update(b->name, NULL);
1160                         if (!lock || write_ref_sha1(lock, b->sha1, msg) < 0)
1161                                 die("Can't write %s", b->name);
1162                 }
1163         }
1164 }
1165
1166 static void dump_tags()
1167 {
1168         static const char *msg = "fast-import";
1169         struct tag *t;
1170         struct ref_lock *lock;
1171         char path[PATH_MAX];
1172
1173         for (t = first_tag; t; t = t->next_tag) {
1174                 sprintf(path, "refs/tags/%s", t->name);
1175                 lock = lock_any_ref_for_update(path, NULL);
1176                 if (!lock || write_ref_sha1(lock, t->sha1, msg) < 0)
1177                         die("Can't write %s", path);
1178         }
1179 }
1180
1181 static void dump_marks_helper(FILE *f,
1182         unsigned long base,
1183         struct mark_set *m)
1184 {
1185         int k;
1186         if (m->shift) {
1187                 for (k = 0; k < 1024; k++) {
1188                         if (m->data.sets[k])
1189                                 dump_marks_helper(f, (base + k) << m->shift,
1190                                         m->data.sets[k]);
1191                 }
1192         } else {
1193                 for (k = 0; k < 1024; k++) {
1194                         if (m->data.marked[k])
1195                                 fprintf(f, ":%lu %s\n", base + k,
1196                                         sha1_to_hex(m->data.marked[k]->sha1));
1197                 }
1198         }
1199 }
1200
1201 static void dump_marks()
1202 {
1203         if (mark_file)
1204         {
1205                 FILE *f = fopen(mark_file, "w");
1206                 dump_marks_helper(f, 0, marks);
1207                 fclose(f);
1208         }
1209 }
1210
1211 static void read_next_command()
1212 {
1213         read_line(&command_buf, stdin, '\n');
1214 }
1215
1216 static void cmd_mark()
1217 {
1218         if (!strncmp("mark :", command_buf.buf, 6)) {
1219                 next_mark = strtoul(command_buf.buf + 6, NULL, 10);
1220                 read_next_command();
1221         }
1222         else
1223                 next_mark = 0;
1224 }
1225
1226 static void* cmd_data (size_t *size)
1227 {
1228         size_t n = 0;
1229         void *buffer;
1230         size_t length;
1231
1232         if (strncmp("data ", command_buf.buf, 5))
1233                 die("Expected 'data n' command, found: %s", command_buf.buf);
1234
1235         length = strtoul(command_buf.buf + 5, NULL, 10);
1236         buffer = xmalloc(length);
1237
1238         while (n < length) {
1239                 size_t s = fread((char*)buffer + n, 1, length - n, stdin);
1240                 if (!s && feof(stdin))
1241                         die("EOF in data (%lu bytes remaining)", length - n);
1242                 n += s;
1243         }
1244
1245         if (fgetc(stdin) != '\n')
1246                 die("An lf did not trail the binary data as expected.");
1247
1248         *size = length;
1249         return buffer;
1250 }
1251
1252 static void cmd_new_blob()
1253 {
1254         size_t l;
1255         void *d;
1256
1257         read_next_command();
1258         cmd_mark();
1259         d = cmd_data(&l);
1260
1261         if (store_object(OBJ_BLOB, d, l, &last_blob, NULL, next_mark))
1262                 free(d);
1263 }
1264
1265 static void unload_one_branch()
1266 {
1267         while (cur_active_branches
1268                 && cur_active_branches >= max_active_branches) {
1269                 unsigned long min_commit = ULONG_MAX;
1270                 struct branch *e, *l = NULL, *p = NULL;
1271
1272                 for (e = active_branches; e; e = e->active_next_branch) {
1273                         if (e->last_commit < min_commit) {
1274                                 p = l;
1275                                 min_commit = e->last_commit;
1276                         }
1277                         l = e;
1278                 }
1279
1280                 if (p) {
1281                         e = p->active_next_branch;
1282                         p->active_next_branch = e->active_next_branch;
1283                 } else {
1284                         e = active_branches;
1285                         active_branches = e->active_next_branch;
1286                 }
1287                 e->active_next_branch = NULL;
1288                 if (e->branch_tree.tree) {
1289                         release_tree_content_recursive(e->branch_tree.tree);
1290                         e->branch_tree.tree = NULL;
1291                 }
1292                 cur_active_branches--;
1293         }
1294 }
1295
1296 static void load_branch(struct branch *b)
1297 {
1298         load_tree(&b->branch_tree);
1299         b->active_next_branch = active_branches;
1300         active_branches = b;
1301         cur_active_branches++;
1302         branch_load_count++;
1303 }
1304
1305 static void file_change_m(struct branch *b)
1306 {
1307         const char *p = command_buf.buf + 2;
1308         char *p_uq;
1309         const char *endp;
1310         struct object_entry *oe;
1311         unsigned char sha1[20];
1312         unsigned int mode;
1313         char type[20];
1314
1315         p = get_mode(p, &mode);
1316         if (!p)
1317                 die("Corrupt mode: %s", command_buf.buf);
1318         switch (mode) {
1319         case S_IFREG | 0644:
1320         case S_IFREG | 0755:
1321         case S_IFLNK:
1322         case 0644:
1323         case 0755:
1324                 /* ok */
1325                 break;
1326         default:
1327                 die("Corrupt mode: %s", command_buf.buf);
1328         }
1329
1330         if (*p == ':') {
1331                 char *x;
1332                 oe = find_mark(strtoul(p + 1, &x, 10));
1333                 hashcpy(sha1, oe->sha1);
1334                 p = x;
1335         } else {
1336                 if (get_sha1_hex(p, sha1))
1337                         die("Invalid SHA1: %s", command_buf.buf);
1338                 oe = find_object(sha1);
1339                 p += 40;
1340         }
1341         if (*p++ != ' ')
1342                 die("Missing space after SHA1: %s", command_buf.buf);
1343
1344         p_uq = unquote_c_style(p, &endp);
1345         if (p_uq) {
1346                 if (*endp)
1347                         die("Garbage after path in: %s", command_buf.buf);
1348                 p = p_uq;
1349         }
1350
1351         if (oe) {
1352                 if (oe->type != OBJ_BLOB)
1353                         die("Not a blob (actually a %s): %s",
1354                                 command_buf.buf, type_names[oe->type]);
1355         } else {
1356                 if (sha1_object_info(sha1, type, NULL))
1357                         die("Blob not found: %s", command_buf.buf);
1358                 if (strcmp(blob_type, type))
1359                         die("Not a blob (actually a %s): %s",
1360                                 command_buf.buf, type);
1361         }
1362
1363         tree_content_set(&b->branch_tree, p, sha1, S_IFREG | mode);
1364
1365         if (p_uq)
1366                 free(p_uq);
1367 }
1368
1369 static void file_change_d(struct branch *b)
1370 {
1371         const char *p = command_buf.buf + 2;
1372         char *p_uq;
1373         const char *endp;
1374
1375         p_uq = unquote_c_style(p, &endp);
1376         if (p_uq) {
1377                 if (*endp)
1378                         die("Garbage after path in: %s", command_buf.buf);
1379                 p = p_uq;
1380         }
1381         tree_content_remove(&b->branch_tree, p);
1382         if (p_uq)
1383                 free(p_uq);
1384 }
1385
1386 static void cmd_from(struct branch *b)
1387 {
1388         const char *from, *endp;
1389         char *str_uq;
1390         struct branch *s;
1391
1392         if (strncmp("from ", command_buf.buf, 5))
1393                 return;
1394
1395         if (b->last_commit)
1396                 die("Can't reinitailize branch %s", b->name);
1397
1398         from = strchr(command_buf.buf, ' ') + 1;
1399         str_uq = unquote_c_style(from, &endp);
1400         if (str_uq) {
1401                 if (*endp)
1402                         die("Garbage after string in: %s", command_buf.buf);
1403                 from = str_uq;
1404         }
1405
1406         s = lookup_branch(from);
1407         if (b == s)
1408                 die("Can't create a branch from itself: %s", b->name);
1409         else if (s) {
1410                 unsigned char *t = s->branch_tree.versions[1].sha1;
1411                 hashcpy(b->sha1, s->sha1);
1412                 hashcpy(b->branch_tree.versions[0].sha1, t);
1413                 hashcpy(b->branch_tree.versions[1].sha1, t);
1414         } else if (*from == ':') {
1415                 unsigned long idnum = strtoul(from + 1, NULL, 10);
1416                 struct object_entry *oe = find_mark(idnum);
1417                 unsigned long size;
1418                 char *buf;
1419                 if (oe->type != OBJ_COMMIT)
1420                         die("Mark :%lu not a commit", idnum);
1421                 hashcpy(b->sha1, oe->sha1);
1422                 buf = gfi_unpack_entry(oe, &size);
1423                 if (!buf || size < 46)
1424                         die("Not a valid commit: %s", from);
1425                 if (memcmp("tree ", buf, 5)
1426                         || get_sha1_hex(buf + 5, b->branch_tree.versions[1].sha1))
1427                         die("The commit %s is corrupt", sha1_to_hex(b->sha1));
1428                 free(buf);
1429                 hashcpy(b->branch_tree.versions[0].sha1,
1430                         b->branch_tree.versions[1].sha1);
1431         } else if (!get_sha1(from, b->sha1)) {
1432                 if (is_null_sha1(b->sha1)) {
1433                         hashclr(b->branch_tree.versions[0].sha1);
1434                         hashclr(b->branch_tree.versions[1].sha1);
1435                 } else {
1436                         unsigned long size;
1437                         char *buf;
1438
1439                         buf = read_object_with_reference(b->sha1,
1440                                 type_names[OBJ_COMMIT], &size, b->sha1);
1441                         if (!buf || size < 46)
1442                                 die("Not a valid commit: %s", from);
1443                         if (memcmp("tree ", buf, 5)
1444                                 || get_sha1_hex(buf + 5, b->branch_tree.versions[1].sha1))
1445                                 die("The commit %s is corrupt", sha1_to_hex(b->sha1));
1446                         free(buf);
1447                         hashcpy(b->branch_tree.versions[0].sha1,
1448                                 b->branch_tree.versions[1].sha1);
1449                 }
1450         } else
1451                 die("Invalid ref name or SHA1 expression: %s", from);
1452
1453         read_next_command();
1454 }
1455
1456 static struct hash_list* cmd_merge(unsigned int *count)
1457 {
1458         struct hash_list *list = NULL, *n, *e;
1459         const char *from, *endp;
1460         char *str_uq;
1461         struct branch *s;
1462
1463         *count = 0;
1464         while (!strncmp("merge ", command_buf.buf, 6)) {
1465                 from = strchr(command_buf.buf, ' ') + 1;
1466                 str_uq = unquote_c_style(from, &endp);
1467                 if (str_uq) {
1468                         if (*endp)
1469                                 die("Garbage after string in: %s", command_buf.buf);
1470                         from = str_uq;
1471                 }
1472
1473                 n = xmalloc(sizeof(*n));
1474                 s = lookup_branch(from);
1475                 if (s)
1476                         hashcpy(n->sha1, s->sha1);
1477                 else if (*from == ':') {
1478                         unsigned long idnum = strtoul(from + 1, NULL, 10);
1479                         struct object_entry *oe = find_mark(idnum);
1480                         if (oe->type != OBJ_COMMIT)
1481                                 die("Mark :%lu not a commit", idnum);
1482                         hashcpy(n->sha1, oe->sha1);
1483                 } else if (get_sha1(from, n->sha1))
1484                         die("Invalid ref name or SHA1 expression: %s", from);
1485
1486                 n->next = NULL;
1487                 if (list)
1488                         e->next = n;
1489                 else
1490                         list = n;
1491                 e = n;
1492                 *count++;
1493                 read_next_command();
1494         }
1495         return list;
1496 }
1497
1498 static void cmd_new_commit()
1499 {
1500         struct branch *b;
1501         void *msg;
1502         size_t msglen;
1503         char *str_uq;
1504         const char *endp;
1505         char *sp;
1506         char *author = NULL;
1507         char *committer = NULL;
1508         struct hash_list *merge_list = NULL;
1509         unsigned int merge_count;
1510
1511         /* Obtain the branch name from the rest of our command */
1512         sp = strchr(command_buf.buf, ' ') + 1;
1513         str_uq = unquote_c_style(sp, &endp);
1514         if (str_uq) {
1515                 if (*endp)
1516                         die("Garbage after ref in: %s", command_buf.buf);
1517                 sp = str_uq;
1518         }
1519         b = lookup_branch(sp);
1520         if (!b)
1521                 b = new_branch(sp);
1522         if (str_uq)
1523                 free(str_uq);
1524
1525         read_next_command();
1526         cmd_mark();
1527         if (!strncmp("author ", command_buf.buf, 7)) {
1528                 author = strdup(command_buf.buf);
1529                 read_next_command();
1530         }
1531         if (!strncmp("committer ", command_buf.buf, 10)) {
1532                 committer = strdup(command_buf.buf);
1533                 read_next_command();
1534         }
1535         if (!committer)
1536                 die("Expected committer but didn't get one");
1537         msg = cmd_data(&msglen);
1538         read_next_command();
1539         cmd_from(b);
1540         merge_list = cmd_merge(&merge_count);
1541
1542         /* ensure the branch is active/loaded */
1543         if (!b->branch_tree.tree || !max_active_branches) {
1544                 unload_one_branch();
1545                 load_branch(b);
1546         }
1547
1548         /* file_change* */
1549         for (;;) {
1550                 if (1 == command_buf.len)
1551                         break;
1552                 else if (!strncmp("M ", command_buf.buf, 2))
1553                         file_change_m(b);
1554                 else if (!strncmp("D ", command_buf.buf, 2))
1555                         file_change_d(b);
1556                 else
1557                         die("Unsupported file_change: %s", command_buf.buf);
1558                 read_next_command();
1559         }
1560
1561         /* build the tree and the commit */
1562         store_tree(&b->branch_tree);
1563         hashcpy(b->branch_tree.versions[0].sha1,
1564                 b->branch_tree.versions[1].sha1);
1565         size_dbuf(&new_data, 97 + msglen
1566                 + merge_count * 49
1567                 + (author
1568                         ? strlen(author) + strlen(committer)
1569                         : 2 * strlen(committer)));
1570         sp = new_data.buffer;
1571         sp += sprintf(sp, "tree %s\n",
1572                 sha1_to_hex(b->branch_tree.versions[1].sha1));
1573         if (!is_null_sha1(b->sha1))
1574                 sp += sprintf(sp, "parent %s\n", sha1_to_hex(b->sha1));
1575         while (merge_list) {
1576                 struct hash_list *next = merge_list->next;
1577                 sp += sprintf(sp, "parent %s\n", sha1_to_hex(merge_list->sha1));
1578                 free(merge_list);
1579                 merge_list = next;
1580         }
1581         if (author)
1582                 sp += sprintf(sp, "%s\n", author);
1583         else
1584                 sp += sprintf(sp, "author %s\n", committer + 10);
1585         sp += sprintf(sp, "%s\n\n", committer);
1586         memcpy(sp, msg, msglen);
1587         sp += msglen;
1588         if (author)
1589                 free(author);
1590         free(committer);
1591         free(msg);
1592
1593         store_object(OBJ_COMMIT,
1594                 new_data.buffer, sp - (char*)new_data.buffer,
1595                 NULL, b->sha1, next_mark);
1596         b->last_commit = object_count_by_type[OBJ_COMMIT];
1597
1598         if (branch_log) {
1599                 int need_dq = quote_c_style(b->name, NULL, NULL, 0);
1600                 fprintf(branch_log, "commit ");
1601                 if (need_dq) {
1602                         fputc('"', branch_log);
1603                         quote_c_style(b->name, NULL, branch_log, 0);
1604                         fputc('"', branch_log);
1605                 } else
1606                         fprintf(branch_log, "%s", b->name);
1607                 fprintf(branch_log," :%lu %s\n",next_mark,sha1_to_hex(b->sha1));
1608         }
1609 }
1610
1611 static void cmd_new_tag()
1612 {
1613         char *str_uq;
1614         const char *endp;
1615         char *sp;
1616         const char *from;
1617         char *tagger;
1618         struct branch *s;
1619         void *msg;
1620         size_t msglen;
1621         struct tag *t;
1622         unsigned long from_mark = 0;
1623         unsigned char sha1[20];
1624
1625         /* Obtain the new tag name from the rest of our command */
1626         sp = strchr(command_buf.buf, ' ') + 1;
1627         str_uq = unquote_c_style(sp, &endp);
1628         if (str_uq) {
1629                 if (*endp)
1630                         die("Garbage after tag name in: %s", command_buf.buf);
1631                 sp = str_uq;
1632         }
1633         t = pool_alloc(sizeof(struct tag));
1634         t->next_tag = NULL;
1635         t->name = pool_strdup(sp);
1636         if (last_tag)
1637                 last_tag->next_tag = t;
1638         else
1639                 first_tag = t;
1640         last_tag = t;
1641         if (str_uq)
1642                 free(str_uq);
1643         read_next_command();
1644
1645         /* from ... */
1646         if (strncmp("from ", command_buf.buf, 5))
1647                 die("Expected from command, got %s", command_buf.buf);
1648
1649         from = strchr(command_buf.buf, ' ') + 1;
1650         str_uq = unquote_c_style(from, &endp);
1651         if (str_uq) {
1652                 if (*endp)
1653                         die("Garbage after string in: %s", command_buf.buf);
1654                 from = str_uq;
1655         }
1656
1657         s = lookup_branch(from);
1658         if (s) {
1659                 hashcpy(sha1, s->sha1);
1660         } else if (*from == ':') {
1661                 from_mark = strtoul(from + 1, NULL, 10);
1662                 struct object_entry *oe = find_mark(from_mark);
1663                 if (oe->type != OBJ_COMMIT)
1664                         die("Mark :%lu not a commit", from_mark);
1665                 hashcpy(sha1, oe->sha1);
1666         } else if (!get_sha1(from, sha1)) {
1667                 unsigned long size;
1668                 char *buf;
1669
1670                 buf = read_object_with_reference(sha1,
1671                         type_names[OBJ_COMMIT], &size, sha1);
1672                 if (!buf || size < 46)
1673                         die("Not a valid commit: %s", from);
1674                 free(buf);
1675         } else
1676                 die("Invalid ref name or SHA1 expression: %s", from);
1677
1678         if (str_uq)
1679                 free(str_uq);
1680         read_next_command();
1681
1682         /* tagger ... */
1683         if (strncmp("tagger ", command_buf.buf, 7))
1684                 die("Expected tagger command, got %s", command_buf.buf);
1685         tagger = strdup(command_buf.buf);
1686
1687         /* tag payload/message */
1688         read_next_command();
1689         msg = cmd_data(&msglen);
1690
1691         /* build the tag object */
1692         size_dbuf(&new_data, 67+strlen(t->name)+strlen(tagger)+msglen);
1693         sp = new_data.buffer;
1694         sp += sprintf(sp, "object %s\n", sha1_to_hex(sha1));
1695         sp += sprintf(sp, "type %s\n", type_names[OBJ_COMMIT]);
1696         sp += sprintf(sp, "tag %s\n", t->name);
1697         sp += sprintf(sp, "%s\n\n", tagger);
1698         memcpy(sp, msg, msglen);
1699         sp += msglen;
1700         free(tagger);
1701         free(msg);
1702
1703         store_object(OBJ_TAG, new_data.buffer, sp - (char*)new_data.buffer,
1704                 NULL, t->sha1, 0);
1705
1706         if (branch_log) {
1707                 int need_dq = quote_c_style(t->name, NULL, NULL, 0);
1708                 fprintf(branch_log, "tag ");
1709                 if (need_dq) {
1710                         fputc('"', branch_log);
1711                         quote_c_style(t->name, NULL, branch_log, 0);
1712                         fputc('"', branch_log);
1713                 } else
1714                         fprintf(branch_log, "%s", t->name);
1715                 fprintf(branch_log," :%lu %s\n",from_mark,sha1_to_hex(t->sha1));
1716         }
1717 }
1718
1719 static void cmd_reset_branch()
1720 {
1721         struct branch *b;
1722         char *str_uq;
1723         const char *endp;
1724         char *sp;
1725
1726         /* Obtain the branch name from the rest of our command */
1727         sp = strchr(command_buf.buf, ' ') + 1;
1728         str_uq = unquote_c_style(sp, &endp);
1729         if (str_uq) {
1730                 if (*endp)
1731                         die("Garbage after ref in: %s", command_buf.buf);
1732                 sp = str_uq;
1733         }
1734         b = lookup_branch(sp);
1735         if (b) {
1736                 b->last_commit = 0;
1737                 if (b->branch_tree.tree) {
1738                         release_tree_content_recursive(b->branch_tree.tree);
1739                         b->branch_tree.tree = NULL;
1740                 }
1741         }
1742         else
1743                 b = new_branch(sp);
1744         if (str_uq)
1745                 free(str_uq);
1746         read_next_command();
1747         cmd_from(b);
1748 }
1749
1750 static void cmd_checkpoint()
1751 {
1752         if (object_count) {
1753                 end_packfile();
1754                 start_packfile();
1755         }
1756         read_next_command();
1757 }
1758
1759 static const char fast_import_usage[] =
1760 "git-fast-import [--objects=n] [--depth=n] [--active-branches=n] [--export-marks=marks.file] [--branch-log=log] temp.pack";
1761
1762 int main(int argc, const char **argv)
1763 {
1764         int i;
1765         unsigned long est_obj_cnt = object_entry_alloc;
1766         unsigned long duplicate_count;
1767
1768         setup_ident();
1769         git_config(git_default_config);
1770
1771         for (i = 1; i < argc; i++) {
1772                 const char *a = argv[i];
1773
1774                 if (*a != '-' || !strcmp(a, "--"))
1775                         break;
1776                 else if (!strncmp(a, "--objects=", 10))
1777                         est_obj_cnt = strtoul(a + 10, NULL, 0);
1778                 else if (!strncmp(a, "--depth=", 8))
1779                         max_depth = strtoul(a + 8, NULL, 0);
1780                 else if (!strncmp(a, "--active-branches=", 18))
1781                         max_active_branches = strtoul(a + 18, NULL, 0);
1782                 else if (!strncmp(a, "--export-marks=", 15))
1783                         mark_file = a + 15;
1784                 else if (!strncmp(a, "--branch-log=", 13)) {
1785                         branch_log = fopen(a + 13, "w");
1786                         if (!branch_log)
1787                                 die("Can't create %s: %s", a + 13, strerror(errno));
1788                 }
1789                 else
1790                         die("unknown option %s", a);
1791         }
1792         if ((i+1) != argc)
1793                 usage(fast_import_usage);
1794         base_name = argv[i];
1795
1796         alloc_objects(est_obj_cnt);
1797         strbuf_init(&command_buf);
1798
1799         atom_table = xcalloc(atom_table_sz, sizeof(struct atom_str*));
1800         branch_table = xcalloc(branch_table_sz, sizeof(struct branch*));
1801         avail_tree_table = xcalloc(avail_tree_table_sz, sizeof(struct avail_tree_content*));
1802         marks = pool_calloc(1, sizeof(struct mark_set));
1803
1804         start_packfile();
1805         for (;;) {
1806                 read_next_command();
1807                 if (command_buf.eof)
1808                         break;
1809                 else if (!strcmp("blob", command_buf.buf))
1810                         cmd_new_blob();
1811                 else if (!strncmp("commit ", command_buf.buf, 7))
1812                         cmd_new_commit();
1813                 else if (!strncmp("tag ", command_buf.buf, 4))
1814                         cmd_new_tag();
1815                 else if (!strncmp("reset ", command_buf.buf, 6))
1816                         cmd_reset_branch();
1817                 else if (!strcmp("checkpoint", command_buf.buf))
1818                         cmd_checkpoint();
1819                 else
1820                         die("Unsupported command: %s", command_buf.buf);
1821         }
1822         end_packfile();
1823
1824         dump_branches();
1825         dump_tags();
1826         dump_marks();
1827         if (branch_log)
1828                 fclose(branch_log);
1829
1830         for (i = 0; i < ARRAY_SIZE(duplicate_count_by_type); i++)
1831                 duplicate_count += duplicate_count_by_type[i];
1832
1833         fprintf(stderr, "%s statistics:\n", argv[0]);
1834         fprintf(stderr, "---------------------------------------------------------------------\n");
1835         fprintf(stderr, "Alloc'd objects: %10lu (%10lu overflow  )\n", alloc_count, alloc_count - est_obj_cnt);
1836         fprintf(stderr, "Total objects:   %10lu (%10lu duplicates                  )\n", object_count, duplicate_count);
1837         fprintf(stderr, "      blobs  :   %10lu (%10lu duplicates %10lu deltas)\n", object_count_by_type[OBJ_BLOB], duplicate_count_by_type[OBJ_BLOB], delta_count_by_type[OBJ_BLOB]);
1838         fprintf(stderr, "      trees  :   %10lu (%10lu duplicates %10lu deltas)\n", object_count_by_type[OBJ_TREE], duplicate_count_by_type[OBJ_TREE], delta_count_by_type[OBJ_TREE]);
1839         fprintf(stderr, "      commits:   %10lu (%10lu duplicates %10lu deltas)\n", object_count_by_type[OBJ_COMMIT], duplicate_count_by_type[OBJ_COMMIT], delta_count_by_type[OBJ_COMMIT]);
1840         fprintf(stderr, "      tags   :   %10lu (%10lu duplicates %10lu deltas)\n", object_count_by_type[OBJ_TAG], duplicate_count_by_type[OBJ_TAG], delta_count_by_type[OBJ_TAG]);
1841         fprintf(stderr, "Total branches:  %10lu (%10lu loads     )\n", branch_count, branch_load_count);
1842         fprintf(stderr, "      marks:     %10u (%10lu unique    )\n", (1 << marks->shift) * 1024, marks_set_count);
1843         fprintf(stderr, "      atoms:     %10u\n", atom_cnt);
1844         fprintf(stderr, "Memory total:    %10lu KiB\n", (total_allocd + alloc_count*sizeof(struct object_entry))/1024);
1845         fprintf(stderr, "       pools:    %10lu KiB\n", total_allocd/1024);
1846         fprintf(stderr, "     objects:    %10lu KiB\n", (alloc_count*sizeof(struct object_entry))/1024);
1847         fprintf(stderr, "---------------------------------------------------------------------\n");
1848         pack_report();
1849         fprintf(stderr, "---------------------------------------------------------------------\n");
1850         fprintf(stderr, "\n");
1851
1852         return 0;
1853 }