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

   1 /*
   2 Format of STDIN stream:
   3
   4   stream ::= cmd*;
   5
   6   cmd ::= new_blob
   7         | new_commit
   8         | new_tag
   9         | reset_branch
  10         | checkpoint
  11         ;
  12
  13   new_blob ::= 'blob' lf
  14         mark?
  15     file_content;
  16   file_content ::= data;
  17
  18   new_commit ::= 'commit' sp ref_str lf
  19     mark?
  20     ('author' sp name '<' email '>' ts tz lf)?
  21     'committer' sp name '<' email '>' ts tz lf
  22     commit_msg
  23     ('from' sp (ref_str | hexsha1 | sha1exp_str | idnum) lf)?
  24     ('merge' sp (ref_str | hexsha1 | sha1exp_str | idnum) lf)*
  25     file_change*
  26     lf;
  27   commit_msg ::= data;
  28
  29   file_change ::= file_del | file_obm | file_inm;
  30   file_del ::= 'D' sp path_str lf;
  31   file_obm ::= 'M' sp mode sp (hexsha1 | idnum) sp path_str lf;
  32   file_inm ::= 'M' sp mode sp 'inline' sp path_str lf
  33     data;
  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   reset_branch ::= 'reset' sp ref_str lf
  42     ('from' sp (ref_str | hexsha1 | sha1exp_str | idnum) lf)?
  43     lf;
  44
  45   checkpoint ::= 'checkpoint' lf
  46     lf;
  47
  48      # note: the first idnum in a stream should be 1 and subsequent
  49      # idnums should not have gaps between values as this will cause
  50      # the stream parser to reserve space for the gapped values.  An
  51          # idnum can be updated in the future to a new object by issuing
  52      # a new mark directive with the old idnum.
  53          #
  54   mark ::= 'mark' sp idnum lf;
  55   data ::= (delimited_data | exact_data)
  56     lf;
  57
  58     # note: delim may be any string but must not contain lf.
  59     # data_line may contain any data but must not be exactly
  60     # delim.
  61   delimited_data ::= 'data' sp '<<' delim lf
  62     (data_line lf)*
  63         delim lf;
  64
  65      # note: declen indicates the length of binary_data in bytes.
  66      # declen does not include the lf preceeding the binary data.
  67      #
  68   exact_data ::= 'data' sp declen lf
  69     binary_data;
  70
  71      # note: quoted strings are C-style quoting supporting \c for
  72      # common escapes of 'c' (e..g \n, \t, \\, \") or \nnn where nnn
  73          # is the signed byte value in octal.  Note that the only
  74      # characters which must actually be escaped to protect the
  75      # stream formatting is: \, " and LF.  Otherwise these values
  76          # are UTF8.
  77      #
  78   ref_str     ::= ref     | '"' quoted(ref)     '"' ;
  79   sha1exp_str ::= sha1exp | '"' quoted(sha1exp) '"' ;
  80   tag_str     ::= tag     | '"' quoted(tag)     '"' ;
  81   path_str    ::= path    | '"' quoted(path)    '"' ;
  82   mode        ::= '100644' | '644'
  83                 | '100755' | '755'
  84                 | '140000'
  85                 ;
  86
  87   declen ::= # unsigned 32 bit value, ascii base10 notation;
  88   bigint ::= # unsigned integer value, ascii base10 notation;
  89   binary_data ::= # file content, not interpreted;
  90
  91   sp ::= # ASCII space character;
  92   lf ::= # ASCII newline (LF) character;
  93
  94      # note: a colon (':') must precede the numerical value assigned to
  95          # an idnum.  This is to distinguish it from a ref or tag name as
  96      # GIT does not permit ':' in ref or tag strings.
  97          #
  98   idnum   ::= ':' bigint;
  99   path    ::= # GIT style file path, e.g. "a/b/c";
 100   ref     ::= # GIT ref name, e.g. "refs/heads/MOZ_GECKO_EXPERIMENT";
 101   tag     ::= # GIT tag name, e.g. "FIREFOX_1_5";
 102   sha1exp ::= # Any valid GIT SHA1 expression;
 103   hexsha1 ::= # SHA1 in hexadecimal format;
 104
 105      # note: name and email are UTF8 strings, however name must not
 106          # contain '<' or lf and email must not contain any of the
 107      # following: '<', '>', lf.
 108          #
 109   name  ::= # valid GIT author/committer name;
 110   email ::= # valid GIT author/committer email;
 111   ts    ::= # time since the epoch in seconds, ascii base10 notation;
 112   tz    ::= # GIT style timezone;
 113 */
 114
 115 #include "builtin.h"
 116 #include "cache.h"
 117 #include "object.h"
 118 #include "blob.h"
 119 #include "tree.h"
 120 #include "delta.h"
 121 #include "pack.h"
 122 #include "refs.h"
 123 #include "csum-file.h"
 124 #include "strbuf.h"
 125 #include "quote.h"
 126
 127 #define PACK_ID_BITS 16
 128 #define MAX_PACK_ID ((1<<PACK_ID_BITS)-1)
 129
 130 struct object_entry
 131 {
 132         struct object_entry *next;
 133         unsigned long offset;
 134         unsigned type : TYPE_BITS;
 135         unsigned pack_id : PACK_ID_BITS;
 136         unsigned char sha1[20];
 137 };
 138
 139 struct object_entry_pool
 140 {
 141         struct object_entry_pool *next_pool;
 142         struct object_entry *next_free;
 143         struct object_entry *end;
 144         struct object_entry entries[FLEX_ARRAY]; /* more */
 145 };
 146
 147 struct mark_set
 148 {
 149         union {
 150                 struct object_entry *marked[1024];
 151                 struct mark_set *sets[1024];
 152         } data;
 153         unsigned int shift;
 154 };
 155
 156 struct last_object
 157 {
 158         void *data;
 159         unsigned long len;
 160         unsigned long offset;
 161         unsigned int depth;
 162         unsigned no_free:1;
 163 };
 164
 165 struct mem_pool
 166 {
 167         struct mem_pool *next_pool;
 168         char *next_free;
 169         char *end;
 170         char space[FLEX_ARRAY]; /* more */
 171 };
 172
 173 struct atom_str
 174 {
 175         struct atom_str *next_atom;
 176         unsigned int str_len;
 177         char str_dat[FLEX_ARRAY]; /* more */
 178 };
 179
 180 struct tree_content;
 181 struct tree_entry
 182 {
 183         struct tree_content *tree;
 184         struct atom_str* name;
 185         struct tree_entry_ms
 186         {
 187                 unsigned int mode;
 188                 unsigned char sha1[20];
 189         } versions[2];
 190 };
 191
 192 struct tree_content
 193 {
 194         unsigned int entry_capacity; /* must match avail_tree_content */
 195         unsigned int entry_count;
 196         unsigned int delta_depth;
 197         struct tree_entry *entries[FLEX_ARRAY]; /* more */
 198 };
 199
 200 struct avail_tree_content
 201 {
 202         unsigned int entry_capacity; /* must match tree_content */
 203         struct avail_tree_content *next_avail;
 204 };
 205
 206 struct branch
 207 {
 208         struct branch *table_next_branch;
 209         struct branch *active_next_branch;
 210         const char *name;
 211         struct tree_entry branch_tree;
 212         uintmax_t last_commit;
 213         unsigned int pack_id;
 214         unsigned char sha1[20];
 215 };
 216
 217 struct tag
 218 {
 219         struct tag *next_tag;
 220         const char *name;
 221         unsigned int pack_id;
 222         unsigned char sha1[20];
 223 };
 224
 225 struct dbuf
 226 {
 227         void *buffer;
 228         size_t capacity;
 229 };
 230
 231 struct hash_list
 232 {
 233         struct hash_list *next;
 234         unsigned char sha1[20];
 235 };
 236
 237 /* Configured limits on output */
 238 static unsigned long max_depth = 10;
 239 static unsigned long max_packsize = (1LL << 32) - 1;
 240
 241 /* Stats and misc. counters */
 242 static uintmax_t alloc_count;
 243 static uintmax_t marks_set_count;
 244 static uintmax_t object_count_by_type[1 << TYPE_BITS];
 245 static uintmax_t duplicate_count_by_type[1 << TYPE_BITS];
 246 static uintmax_t delta_count_by_type[1 << TYPE_BITS];
 247 static unsigned long object_count;
 248 static unsigned long branch_count;
 249 static unsigned long branch_load_count;
 250
 251 /* Memory pools */
 252 static size_t mem_pool_alloc = 2*1024*1024 - sizeof(struct mem_pool);
 253 static size_t total_allocd;
 254 static struct mem_pool *mem_pool;
 255
 256 /* Atom management */
 257 static unsigned int atom_table_sz = 4451;
 258 static unsigned int atom_cnt;
 259 static struct atom_str **atom_table;
 260
 261 /* The .pack file being generated */
 262 static unsigned int pack_id;
 263 static struct packed_git *pack_data;
 264 static struct packed_git **all_packs;
 265 static unsigned long pack_size;
 266
 267 /* Table of objects we've written. */
 268 static unsigned int object_entry_alloc = 5000;
 269 static struct object_entry_pool *blocks;
 270 static struct object_entry *object_table[1 << 16];
 271 static struct mark_set *marks;
 272 static const char* mark_file;
 273
 274 /* Our last blob */
 275 static struct last_object last_blob;
 276
 277 /* Tree management */
 278 static unsigned int tree_entry_alloc = 1000;
 279 static void *avail_tree_entry;
 280 static unsigned int avail_tree_table_sz = 100;
 281 static struct avail_tree_content **avail_tree_table;
 282 static struct dbuf old_tree;
 283 static struct dbuf new_tree;
 284
 285 /* Branch data */
 286 static unsigned long max_active_branches = 5;
 287 static unsigned long cur_active_branches;
 288 static unsigned long branch_table_sz = 1039;
 289 static struct branch **branch_table;
 290 static struct branch *active_branches;
 291
 292 /* Tag data */
 293 static struct tag *first_tag;
 294 static struct tag *last_tag;
 295
 296 /* Input stream parsing */
 297 static struct strbuf command_buf;
 298 static uintmax_t next_mark;
 299 static struct dbuf new_data;
 300 static FILE* branch_log;
 301
 302
 303 static void alloc_objects(unsigned int cnt)
 304 {
 305         struct object_entry_pool *b;
 306
 307         b = xmalloc(sizeof(struct object_entry_pool)
 308                 + cnt * sizeof(struct object_entry));
 309         b->next_pool = blocks;
 310         b->next_free = b->entries;
 311         b->end = b->entries + cnt;
 312         blocks = b;
 313         alloc_count += cnt;
 314 }
 315
 316 static struct object_entry* new_object(unsigned char *sha1)
 317 {
 318         struct object_entry *e;
 319
 320         if (blocks->next_free == blocks->end)
 321                 alloc_objects(object_entry_alloc);
 322
 323         e = blocks->next_free++;
 324         hashcpy(e->sha1, sha1);
 325         return e;
 326 }
 327
 328 static struct object_entry* find_object(unsigned char *sha1)
 329 {
 330         unsigned int h = sha1[0] << 8 | sha1[1];
 331         struct object_entry *e;
 332         for (e = object_table[h]; e; e = e->next)
 333                 if (!hashcmp(sha1, e->sha1))
 334                         return e;
 335         return NULL;
 336 }
 337
 338 static struct object_entry* insert_object(unsigned char *sha1)
 339 {
 340         unsigned int h = sha1[0] << 8 | sha1[1];
 341         struct object_entry *e = object_table[h];
 342         struct object_entry *p = NULL;
 343
 344         while (e) {
 345                 if (!hashcmp(sha1, e->sha1))
 346                         return e;
 347                 p = e;
 348                 e = e->next;
 349         }
 350
 351         e = new_object(sha1);
 352         e->next = NULL;
 353         e->offset = 0;
 354         if (p)
 355                 p->next = e;
 356         else
 357                 object_table[h] = e;
 358         return e;
 359 }
 360
 361 static unsigned int hc_str(const char *s, size_t len)
 362 {
 363         unsigned int r = 0;
 364         while (len-- > 0)
 365                 r = r * 31 + *s++;
 366         return r;
 367 }
 368
 369 static void* pool_alloc(size_t len)
 370 {
 371         struct mem_pool *p;
 372         void *r;
 373
 374         for (p = mem_pool; p; p = p->next_pool)
 375                 if ((p->end - p->next_free >= len))
 376                         break;
 377
 378         if (!p) {
 379                 if (len >= (mem_pool_alloc/2)) {
 380                         total_allocd += len;
 381                         return xmalloc(len);
 382                 }
 383                 total_allocd += sizeof(struct mem_pool) + mem_pool_alloc;
 384                 p = xmalloc(sizeof(struct mem_pool) + mem_pool_alloc);
 385                 p->next_pool = mem_pool;
 386                 p->next_free = p->space;
 387                 p->end = p->next_free + mem_pool_alloc;
 388                 mem_pool = p;
 389         }
 390
 391         r = p->next_free;
 392         /* round out to a pointer alignment */
 393         if (len & (sizeof(void*) - 1))
 394                 len += sizeof(void*) - (len & (sizeof(void*) - 1));
 395         p->next_free += len;
 396         return r;
 397 }
 398
 399 static void* pool_calloc(size_t count, size_t size)
 400 {
 401         size_t len = count * size;
 402         void *r = pool_alloc(len);
 403         memset(r, 0, len);
 404         return r;
 405 }
 406
 407 static char* pool_strdup(const char *s)
 408 {
 409         char *r = pool_alloc(strlen(s) + 1);
 410         strcpy(r, s);
 411         return r;
 412 }
 413
 414 static void size_dbuf(struct dbuf *b, size_t maxlen)
 415 {
 416         if (b->buffer) {
 417                 if (b->capacity >= maxlen)
 418                         return;
 419                 free(b->buffer);
 420         }
 421         b->capacity = ((maxlen / 1024) + 1) * 1024;
 422         b->buffer = xmalloc(b->capacity);
 423 }
 424
 425 static void insert_mark(uintmax_t idnum, struct object_entry *oe)
 426 {
 427         struct mark_set *s = marks;
 428         while ((idnum >> s->shift) >= 1024) {
 429                 s = pool_calloc(1, sizeof(struct mark_set));
 430                 s->shift = marks->shift + 10;
 431                 s->data.sets[0] = marks;
 432                 marks = s;
 433         }
 434         while (s->shift) {
 435                 uintmax_t i = idnum >> s->shift;
 436                 idnum -= i << s->shift;
 437                 if (!s->data.sets[i]) {
 438                         s->data.sets[i] = pool_calloc(1, sizeof(struct mark_set));
 439                         s->data.sets[i]->shift = s->shift - 10;
 440                 }
 441                 s = s->data.sets[i];
 442         }
 443         if (!s->data.marked[idnum])
 444                 marks_set_count++;
 445         s->data.marked[idnum] = oe;
 446 }
 447
 448 static struct object_entry* find_mark(uintmax_t idnum)
 449 {
 450         uintmax_t orig_idnum = idnum;
 451         struct mark_set *s = marks;
 452         struct object_entry *oe = NULL;
 453         if ((idnum >> s->shift) < 1024) {
 454                 while (s && s->shift) {
 455                         uintmax_t i = idnum >> s->shift;
 456                         idnum -= i << s->shift;
 457                         s = s->data.sets[i];
 458                 }
 459                 if (s)
 460                         oe = s->data.marked[idnum];
 461         }
 462         if (!oe)
 463                 die("mark :%ju not declared", orig_idnum);
 464         return oe;
 465 }
 466
 467 static struct atom_str* to_atom(const char *s, size_t len)
 468 {
 469         unsigned int hc = hc_str(s, len) % atom_table_sz;
 470         struct atom_str *c;
 471
 472         for (c = atom_table[hc]; c; c = c->next_atom)
 473                 if (c->str_len == len && !strncmp(s, c->str_dat, len))
 474                         return c;
 475
 476         c = pool_alloc(sizeof(struct atom_str) + len + 1);
 477         c->str_len = len;
 478         strncpy(c->str_dat, s, len);
 479         c->str_dat[len] = 0;
 480         c->next_atom = atom_table[hc];
 481         atom_table[hc] = c;
 482         atom_cnt++;
 483         return c;
 484 }
 485
 486 static struct branch* lookup_branch(const char *name)
 487 {
 488         unsigned int hc = hc_str(name, strlen(name)) % branch_table_sz;
 489         struct branch *b;
 490
 491         for (b = branch_table[hc]; b; b = b->table_next_branch)
 492                 if (!strcmp(name, b->name))
 493                         return b;
 494         return NULL;
 495 }
 496
 497 static struct branch* new_branch(const char *name)
 498 {
 499         unsigned int hc = hc_str(name, strlen(name)) % branch_table_sz;
 500         struct branch* b = lookup_branch(name);
 501
 502         if (b)
 503                 die("Invalid attempt to create duplicate branch: %s", name);
 504         if (check_ref_format(name))
 505                 die("Branch name doesn't conform to GIT standards: %s", name);
 506
 507         b = pool_calloc(1, sizeof(struct branch));
 508         b->name = pool_strdup(name);
 509         b->table_next_branch = branch_table[hc];
 510         b->branch_tree.versions[0].mode = S_IFDIR;
 511         b->branch_tree.versions[1].mode = S_IFDIR;
 512         b->pack_id = MAX_PACK_ID;
 513         branch_table[hc] = b;
 514         branch_count++;
 515         return b;
 516 }
 517
 518 static unsigned int hc_entries(unsigned int cnt)
 519 {
 520         cnt = cnt & 7 ? (cnt / 8) + 1 : cnt / 8;
 521         return cnt < avail_tree_table_sz ? cnt : avail_tree_table_sz - 1;
 522 }
 523
 524 static struct tree_content* new_tree_content(unsigned int cnt)
 525 {
 526         struct avail_tree_content *f, *l = NULL;
 527         struct tree_content *t;
 528         unsigned int hc = hc_entries(cnt);
 529
 530         for (f = avail_tree_table[hc]; f; l = f, f = f->next_avail)
 531                 if (f->entry_capacity >= cnt)
 532                         break;
 533
 534         if (f) {
 535                 if (l)
 536                         l->next_avail = f->next_avail;
 537                 else
 538                         avail_tree_table[hc] = f->next_avail;
 539         } else {
 540                 cnt = cnt & 7 ? ((cnt / 8) + 1) * 8 : cnt;
 541                 f = pool_alloc(sizeof(*t) + sizeof(t->entries[0]) * cnt);
 542                 f->entry_capacity = cnt;
 543         }
 544
 545         t = (struct tree_content*)f;
 546         t->entry_count = 0;
 547         t->delta_depth = 0;
 548         return t;
 549 }
 550
 551 static void release_tree_entry(struct tree_entry *e);
 552 static void release_tree_content(struct tree_content *t)
 553 {
 554         struct avail_tree_content *f = (struct avail_tree_content*)t;
 555         unsigned int hc = hc_entries(f->entry_capacity);
 556         f->next_avail = avail_tree_table[hc];
 557         avail_tree_table[hc] = f;
 558 }
 559
 560 static void release_tree_content_recursive(struct tree_content *t)
 561 {
 562         unsigned int i;
 563         for (i = 0; i < t->entry_count; i++)
 564                 release_tree_entry(t->entries[i]);
 565         release_tree_content(t);
 566 }
 567
 568 static struct tree_content* grow_tree_content(
 569         struct tree_content *t,
 570         int amt)
 571 {
 572         struct tree_content *r = new_tree_content(t->entry_count + amt);
 573         r->entry_count = t->entry_count;
 574         r->delta_depth = t->delta_depth;
 575         memcpy(r->entries,t->entries,t->entry_count*sizeof(t->entries[0]));
 576         release_tree_content(t);
 577         return r;
 578 }
 579
 580 static struct tree_entry* new_tree_entry(void)
 581 {
 582         struct tree_entry *e;
 583
 584         if (!avail_tree_entry) {
 585                 unsigned int n = tree_entry_alloc;
 586                 total_allocd += n * sizeof(struct tree_entry);
 587                 avail_tree_entry = e = xmalloc(n * sizeof(struct tree_entry));
 588                 while (n-- > 1) {
 589                         *((void**)e) = e + 1;
 590                         e++;
 591                 }
 592                 *((void**)e) = NULL;
 593         }
 594
 595         e = avail_tree_entry;
 596         avail_tree_entry = *((void**)e);
 597         return e;
 598 }
 599
 600 static void release_tree_entry(struct tree_entry *e)
 601 {
 602         if (e->tree)
 603                 release_tree_content_recursive(e->tree);
 604         *((void**)e) = avail_tree_entry;
 605         avail_tree_entry = e;
 606 }
 607
 608 static void start_packfile(void)
 609 {
 610         static char tmpfile[PATH_MAX];
 611         struct packed_git *p;
 612         struct pack_header hdr;
 613         int pack_fd;
 614
 615         snprintf(tmpfile, sizeof(tmpfile),
 616                 "%s/pack_XXXXXX", get_object_directory());
 617         pack_fd = mkstemp(tmpfile);
 618         if (pack_fd < 0)
 619                 die("Can't create %s: %s", tmpfile, strerror(errno));
 620         p = xcalloc(1, sizeof(*p) + strlen(tmpfile) + 2);
 621         strcpy(p->pack_name, tmpfile);
 622         p->pack_fd = pack_fd;
 623
 624         hdr.hdr_signature = htonl(PACK_SIGNATURE);
 625         hdr.hdr_version = htonl(2);
 626         hdr.hdr_entries = 0;
 627         write_or_die(p->pack_fd, &hdr, sizeof(hdr));
 628
 629         pack_data = p;
 630         pack_size = sizeof(hdr);
 631         object_count = 0;
 632
 633         all_packs = xrealloc(all_packs, sizeof(*all_packs) * (pack_id + 1));
 634         all_packs[pack_id] = p;
 635 }
 636
 637 static void fixup_header_footer(void)
 638 {
 639         static const int buf_sz = 128 * 1024;
 640         int pack_fd = pack_data->pack_fd;
 641         SHA_CTX c;
 642         struct pack_header hdr;
 643         char *buf;
 644
 645         if (lseek(pack_fd, 0, SEEK_SET) != 0)
 646                 die("Failed seeking to start: %s", strerror(errno));
 647         if (read_in_full(pack_fd, &hdr, sizeof(hdr)) != sizeof(hdr))
 648                 die("Unable to reread header of %s", pack_data->pack_name);
 649         if (lseek(pack_fd, 0, SEEK_SET) != 0)
 650                 die("Failed seeking to start: %s", strerror(errno));
 651         hdr.hdr_entries = htonl(object_count);
 652         write_or_die(pack_fd, &hdr, sizeof(hdr));
 653
 654         SHA1_Init(&c);
 655         SHA1_Update(&c, &hdr, sizeof(hdr));
 656
 657         buf = xmalloc(buf_sz);
 658         for (;;) {
 659                 size_t n = xread(pack_fd, buf, buf_sz);
 660                 if (!n)
 661                         break;
 662                 if (n < 0)
 663                         die("Failed to checksum %s", pack_data->pack_name);
 664                 SHA1_Update(&c, buf, n);
 665         }
 666         free(buf);
 667
 668         SHA1_Final(pack_data->sha1, &c);
 669         write_or_die(pack_fd, pack_data->sha1, sizeof(pack_data->sha1));
 670         close(pack_fd);
 671 }
 672
 673 static int oecmp (const void *a_, const void *b_)
 674 {
 675         struct object_entry *a = *((struct object_entry**)a_);
 676         struct object_entry *b = *((struct object_entry**)b_);
 677         return hashcmp(a->sha1, b->sha1);
 678 }
 679
 680 static char* create_index(void)
 681 {
 682         static char tmpfile[PATH_MAX];
 683         SHA_CTX ctx;
 684         struct sha1file *f;
 685         struct object_entry **idx, **c, **last, *e;
 686         struct object_entry_pool *o;
 687         uint32_t array[256];
 688         int i, idx_fd;
 689
 690         /* Build the sorted table of object IDs. */
 691         idx = xmalloc(object_count * sizeof(struct object_entry*));
 692         c = idx;
 693         for (o = blocks; o; o = o->next_pool)
 694                 for (e = o->next_free; e-- != o->entries;)
 695                         if (pack_id == e->pack_id)
 696                                 *c++ = e;
 697         last = idx + object_count;
 698         if (c != last)
 699                 die("internal consistency error creating the index");
 700         qsort(idx, object_count, sizeof(struct object_entry*), oecmp);
 701
 702         /* Generate the fan-out array. */
 703         c = idx;
 704         for (i = 0; i < 256; i++) {
 705                 struct object_entry **next = c;;
 706                 while (next < last) {
 707                         if ((*next)->sha1[0] != i)
 708                                 break;
 709                         next++;
 710                 }
 711                 array[i] = htonl(next - idx);
 712                 c = next;
 713         }
 714
 715         snprintf(tmpfile, sizeof(tmpfile),
 716                 "%s/index_XXXXXX", get_object_directory());
 717         idx_fd = mkstemp(tmpfile);
 718         if (idx_fd < 0)
 719                 die("Can't create %s: %s", tmpfile, strerror(errno));
 720         f = sha1fd(idx_fd, tmpfile);
 721         sha1write(f, array, 256 * sizeof(int));
 722         SHA1_Init(&ctx);
 723         for (c = idx; c != last; c++) {
 724                 uint32_t offset = htonl((*c)->offset);
 725                 sha1write(f, &offset, 4);
 726                 sha1write(f, (*c)->sha1, sizeof((*c)->sha1));
 727                 SHA1_Update(&ctx, (*c)->sha1, 20);
 728         }
 729         sha1write(f, pack_data->sha1, sizeof(pack_data->sha1));
 730         sha1close(f, NULL, 1);
 731         free(idx);
 732         SHA1_Final(pack_data->sha1, &ctx);
 733         return tmpfile;
 734 }
 735
 736 static char* keep_pack(char *curr_index_name)
 737 {
 738         static char name[PATH_MAX];
 739         static char *keep_msg = "fast-import";
 740         int keep_fd;
 741
 742         chmod(pack_data->pack_name, 0444);
 743         chmod(curr_index_name, 0444);
 744
 745         snprintf(name, sizeof(name), "%s/pack/pack-%s.keep",
 746                  get_object_directory(), sha1_to_hex(pack_data->sha1));
 747         keep_fd = open(name, O_RDWR|O_CREAT|O_EXCL, 0600);
 748         if (keep_fd < 0)
 749                 die("cannot create keep file");
 750         write(keep_fd, keep_msg, strlen(keep_msg));
 751         close(keep_fd);
 752
 753         snprintf(name, sizeof(name), "%s/pack/pack-%s.pack",
 754                  get_object_directory(), sha1_to_hex(pack_data->sha1));
 755         if (move_temp_to_file(pack_data->pack_name, name))
 756                 die("cannot store pack file");
 757
 758         snprintf(name, sizeof(name), "%s/pack/pack-%s.idx",
 759                  get_object_directory(), sha1_to_hex(pack_data->sha1));
 760         if (move_temp_to_file(curr_index_name, name))
 761                 die("cannot store index file");
 762         return name;
 763 }
 764
 765 static void unkeep_all_packs(void)
 766 {
 767         static char name[PATH_MAX];
 768         int k;
 769
 770         for (k = 0; k < pack_id; k++) {
 771                 struct packed_git *p = all_packs[k];
 772                 snprintf(name, sizeof(name), "%s/pack/pack-%s.keep",
 773                          get_object_directory(), sha1_to_hex(p->sha1));
 774                 unlink(name);
 775         }
 776 }
 777
 778 static void end_packfile(void)
 779 {
 780         struct packed_git *old_p = pack_data, *new_p;
 781
 782         if (object_count) {
 783                 char *idx_name;
 784                 int i;
 785                 struct branch *b;
 786                 struct tag *t;
 787
 788                 fixup_header_footer();
 789                 idx_name = keep_pack(create_index());
 790
 791                 /* Register the packfile with core git's machinary. */
 792                 new_p = add_packed_git(idx_name, strlen(idx_name), 1);
 793                 if (!new_p)
 794                         die("core git rejected index %s", idx_name);
 795                 new_p->windows = old_p->windows;
 796                 all_packs[pack_id] = new_p;
 797                 install_packed_git(new_p);
 798
 799                 /* Print the boundary */
 800                 fprintf(stdout, "%s:", new_p->pack_name);
 801                 for (i = 0; i < branch_table_sz; i++) {
 802                         for (b = branch_table[i]; b; b = b->table_next_branch) {
 803                                 if (b->pack_id == pack_id)
 804                                         fprintf(stdout, " %s", sha1_to_hex(b->sha1));
 805                         }
 806                 }
 807                 for (t = first_tag; t; t = t->next_tag) {
 808                         if (t->pack_id == pack_id)
 809                                 fprintf(stdout, " %s", sha1_to_hex(t->sha1));
 810                 }
 811                 fputc('\n', stdout);
 812
 813                 pack_id++;
 814         }
 815         else
 816                 unlink(old_p->pack_name);
 817         free(old_p);
 818
 819         /* We can't carry a delta across packfiles. */
 820         free(last_blob.data);
 821         last_blob.data = NULL;
 822         last_blob.len = 0;
 823         last_blob.offset = 0;
 824         last_blob.depth = 0;
 825 }
 826
 827 static void checkpoint(void)
 828 {
 829         end_packfile();
 830         start_packfile();
 831 }
 832
 833 static size_t encode_header(
 834         enum object_type type,
 835         size_t size,
 836         unsigned char *hdr)
 837 {
 838         int n = 1;
 839         unsigned char c;
 840
 841         if (type < OBJ_COMMIT || type > OBJ_REF_DELTA)
 842                 die("bad type %d", type);
 843
 844         c = (type << 4) | (size & 15);
 845         size >>= 4;
 846         while (size) {
 847                 *hdr++ = c | 0x80;
 848                 c = size & 0x7f;
 849                 size >>= 7;
 850                 n++;
 851         }
 852         *hdr = c;
 853         return n;
 854 }
 855
 856 static int store_object(
 857         enum object_type type,
 858         void *dat,
 859         size_t datlen,
 860         struct last_object *last,
 861         unsigned char *sha1out,
 862         uintmax_t mark)
 863 {
 864         void *out, *delta;
 865         struct object_entry *e;
 866         unsigned char hdr[96];
 867         unsigned char sha1[20];
 868         unsigned long hdrlen, deltalen;
 869         SHA_CTX c;
 870         z_stream s;
 871
 872         hdrlen = sprintf((char*)hdr,"%s %lu",type_names[type],datlen) + 1;
 873         SHA1_Init(&c);
 874         SHA1_Update(&c, hdr, hdrlen);
 875         SHA1_Update(&c, dat, datlen);
 876         SHA1_Final(sha1, &c);
 877         if (sha1out)
 878                 hashcpy(sha1out, sha1);
 879
 880         e = insert_object(sha1);
 881         if (mark)
 882                 insert_mark(mark, e);
 883         if (e->offset) {
 884                 duplicate_count_by_type[type]++;
 885                 return 1;
 886         }
 887
 888         if (last && last->data && last->depth < max_depth) {
 889                 delta = diff_delta(last->data, last->len,
 890                         dat, datlen,
 891                         &deltalen, 0);
 892                 if (delta && deltalen >= datlen) {
 893                         free(delta);
 894                         delta = NULL;
 895                 }
 896         } else
 897                 delta = NULL;
 898
 899         memset(&s, 0, sizeof(s));
 900         deflateInit(&s, zlib_compression_level);
 901         if (delta) {
 902                 s.next_in = delta;
 903                 s.avail_in = deltalen;
 904         } else {
 905                 s.next_in = dat;
 906                 s.avail_in = datlen;
 907         }
 908         s.avail_out = deflateBound(&s, s.avail_in);
 909         s.next_out = out = xmalloc(s.avail_out);
 910         while (deflate(&s, Z_FINISH) == Z_OK)
 911                 /* nothing */;
 912         deflateEnd(&s);
 913
 914         /* Determine if we should auto-checkpoint. */
 915         if ((pack_size + 60 + s.total_out) > max_packsize
 916                 || (pack_size + 60 + s.total_out) < pack_size) {
 917
 918                 /* This new object needs to *not* have the current pack_id. */
 919                 e->pack_id = pack_id + 1;
 920                 checkpoint();
 921
 922                 /* We cannot carry a delta into the new pack. */
 923                 if (delta) {
 924                         free(delta);
 925                         delta = NULL;
 926
 927                         memset(&s, 0, sizeof(s));
 928                         deflateInit(&s, zlib_compression_level);
 929                         s.next_in = dat;
 930                         s.avail_in = datlen;
 931                         s.avail_out = deflateBound(&s, s.avail_in);
 932                         s.next_out = out = xrealloc(out, s.avail_out);
 933                         while (deflate(&s, Z_FINISH) == Z_OK)
 934                                 /* nothing */;
 935                         deflateEnd(&s);
 936                 }
 937         }
 938
 939         e->type = type;
 940         e->pack_id = pack_id;
 941         e->offset = pack_size;
 942         object_count++;
 943         object_count_by_type[type]++;
 944
 945         if (delta) {
 946                 unsigned long ofs = e->offset - last->offset;
 947                 unsigned pos = sizeof(hdr) - 1;
 948
 949                 delta_count_by_type[type]++;
 950                 last->depth++;
 951
 952                 hdrlen = encode_header(OBJ_OFS_DELTA, deltalen, hdr);
 953                 write_or_die(pack_data->pack_fd, hdr, hdrlen);
 954                 pack_size += hdrlen;
 955
 956                 hdr[pos] = ofs & 127;
 957                 while (ofs >>= 7)
 958                         hdr[--pos] = 128 | (--ofs & 127);
 959                 write_or_die(pack_data->pack_fd, hdr + pos, sizeof(hdr) - pos);
 960                 pack_size += sizeof(hdr) - pos;
 961         } else {
 962                 if (last)
 963                         last->depth = 0;
 964                 hdrlen = encode_header(type, datlen, hdr);
 965                 write_or_die(pack_data->pack_fd, hdr, hdrlen);
 966                 pack_size += hdrlen;
 967         }
 968
 969         write_or_die(pack_data->pack_fd, out, s.total_out);
 970         pack_size += s.total_out;
 971
 972         free(out);
 973         if (delta)
 974                 free(delta);
 975         if (last) {
 976                 if (last->data && !last->no_free)
 977                         free(last->data);
 978                 last->data = dat;
 979                 last->offset = e->offset;
 980                 last->len = datlen;
 981         }
 982         return 0;
 983 }
 984
 985 static void *gfi_unpack_entry(
 986         struct object_entry *oe,
 987         unsigned long *sizep)
 988 {
 989         static char type[20];
 990         struct packed_git *p = all_packs[oe->pack_id];
 991         if (p == pack_data)
 992                 p->pack_size = pack_size + 20;
 993         return unpack_entry(p, oe->offset, type, sizep);
 994 }
 995
 996 static const char *get_mode(const char *str, unsigned int *modep)
 997 {
 998         unsigned char c;
 999         unsigned int mode = 0;
1000
1001         while ((c = *str++) != ' ') {
1002                 if (c < '0' || c > '7')
1003                         return NULL;
1004                 mode = (mode << 3) + (c - '0');
1005         }
1006         *modep = mode;
1007         return str;
1008 }
1009
1010 static void load_tree(struct tree_entry *root)
1011 {
1012         unsigned char* sha1 = root->versions[1].sha1;
1013         struct object_entry *myoe;
1014         struct tree_content *t;
1015         unsigned long size;
1016         char *buf;
1017         const char *c;
1018
1019         root->tree = t = new_tree_content(8);
1020         if (is_null_sha1(sha1))
1021                 return;
1022
1023         myoe = find_object(sha1);
1024         if (myoe) {
1025                 if (myoe->type != OBJ_TREE)
1026                         die("Not a tree: %s", sha1_to_hex(sha1));
1027                 t->delta_depth = 0;
1028                 buf = gfi_unpack_entry(myoe, &size);
1029         } else {
1030                 char type[20];
1031                 buf = read_sha1_file(sha1, type, &size);
1032                 if (!buf || strcmp(type, tree_type))
1033                         die("Can't load tree %s", sha1_to_hex(sha1));
1034         }
1035
1036         c = buf;
1037         while (c != (buf + size)) {
1038                 struct tree_entry *e = new_tree_entry();
1039
1040                 if (t->entry_count == t->entry_capacity)
1041                         root->tree = t = grow_tree_content(t, 8);
1042                 t->entries[t->entry_count++] = e;
1043
1044                 e->tree = NULL;
1045                 c = get_mode(c, &e->versions[1].mode);
1046                 if (!c)
1047                         die("Corrupt mode in %s", sha1_to_hex(sha1));
1048                 e->versions[0].mode = e->versions[1].mode;
1049                 e->name = to_atom(c, strlen(c));
1050                 c += e->name->str_len + 1;
1051                 hashcpy(e->versions[0].sha1, (unsigned char*)c);
1052                 hashcpy(e->versions[1].sha1, (unsigned char*)c);
1053                 c += 20;
1054         }
1055         free(buf);
1056 }
1057
1058 static int tecmp0 (const void *_a, const void *_b)
1059 {
1060         struct tree_entry *a = *((struct tree_entry**)_a);
1061         struct tree_entry *b = *((struct tree_entry**)_b);
1062         return base_name_compare(
1063                 a->name->str_dat, a->name->str_len, a->versions[0].mode,
1064                 b->name->str_dat, b->name->str_len, b->versions[0].mode);
1065 }
1066
1067 static int tecmp1 (const void *_a, const void *_b)
1068 {
1069         struct tree_entry *a = *((struct tree_entry**)_a);
1070         struct tree_entry *b = *((struct tree_entry**)_b);
1071         return base_name_compare(
1072                 a->name->str_dat, a->name->str_len, a->versions[1].mode,
1073                 b->name->str_dat, b->name->str_len, b->versions[1].mode);
1074 }
1075
1076 static void mktree(struct tree_content *t,
1077         int v,
1078         unsigned long *szp,
1079         struct dbuf *b)
1080 {
1081         size_t maxlen = 0;
1082         unsigned int i;
1083         char *c;
1084
1085         if (!v)
1086                 qsort(t->entries,t->entry_count,sizeof(t->entries[0]),tecmp0);
1087         else
1088                 qsort(t->entries,t->entry_count,sizeof(t->entries[0]),tecmp1);
1089
1090         for (i = 0; i < t->entry_count; i++) {
1091                 if (t->entries[i]->versions[v].mode)
1092                         maxlen += t->entries[i]->name->str_len + 34;
1093         }
1094
1095         size_dbuf(b, maxlen);
1096         c = b->buffer;
1097         for (i = 0; i < t->entry_count; i++) {
1098                 struct tree_entry *e = t->entries[i];
1099                 if (!e->versions[v].mode)
1100                         continue;
1101                 c += sprintf(c, "%o", e->versions[v].mode);
1102                 *c++ = ' ';
1103                 strcpy(c, e->name->str_dat);
1104                 c += e->name->str_len + 1;
1105                 hashcpy((unsigned char*)c, e->versions[v].sha1);
1106                 c += 20;
1107         }
1108         *szp = c - (char*)b->buffer;
1109 }
1110
1111 static void store_tree(struct tree_entry *root)
1112 {
1113         struct tree_content *t = root->tree;
1114         unsigned int i, j, del;
1115         unsigned long new_len;
1116         struct last_object lo;
1117         struct object_entry *le;
1118
1119         if (!is_null_sha1(root->versions[1].sha1))
1120                 return;
1121
1122         for (i = 0; i < t->entry_count; i++) {
1123                 if (t->entries[i]->tree)
1124                         store_tree(t->entries[i]);
1125         }
1126
1127         le = find_object(root->versions[0].sha1);
1128         if (!S_ISDIR(root->versions[0].mode)
1129                 || !le
1130                 || le->pack_id != pack_id) {
1131                 lo.data = NULL;
1132                 lo.depth = 0;
1133         } else {
1134                 mktree(t, 0, &lo.len, &old_tree);
1135                 lo.data = old_tree.buffer;
1136                 lo.offset = le->offset;
1137                 lo.depth = t->delta_depth;
1138                 lo.no_free = 1;
1139         }
1140
1141         mktree(t, 1, &new_len, &new_tree);
1142         store_object(OBJ_TREE, new_tree.buffer, new_len,
1143                 &lo, root->versions[1].sha1, 0);
1144
1145         t->delta_depth = lo.depth;
1146         for (i = 0, j = 0, del = 0; i < t->entry_count; i++) {
1147                 struct tree_entry *e = t->entries[i];
1148                 if (e->versions[1].mode) {
1149                         e->versions[0].mode = e->versions[1].mode;
1150                         hashcpy(e->versions[0].sha1, e->versions[1].sha1);
1151                         t->entries[j++] = e;
1152                 } else {
1153                         release_tree_entry(e);
1154                         del++;
1155                 }
1156         }
1157         t->entry_count -= del;
1158 }
1159
1160 static int tree_content_set(
1161         struct tree_entry *root,
1162         const char *p,
1163         const unsigned char *sha1,
1164         const unsigned int mode)
1165 {
1166         struct tree_content *t = root->tree;
1167         const char *slash1;
1168         unsigned int i, n;
1169         struct tree_entry *e;
1170
1171         slash1 = strchr(p, '/');
1172         if (slash1)
1173                 n = slash1 - p;
1174         else
1175                 n = strlen(p);
1176
1177         for (i = 0; i < t->entry_count; i++) {
1178                 e = t->entries[i];
1179                 if (e->name->str_len == n && !strncmp(p, e->name->str_dat, n)) {
1180                         if (!slash1) {
1181                                 if (e->versions[1].mode == mode
1182                                                 && !hashcmp(e->versions[1].sha1, sha1))
1183                                         return 0;
1184                                 e->versions[1].mode = mode;
1185                                 hashcpy(e->versions[1].sha1, sha1);
1186                                 if (e->tree) {
1187                                         release_tree_content_recursive(e->tree);
1188                                         e->tree = NULL;
1189                                 }
1190                                 hashclr(root->versions[1].sha1);
1191                                 return 1;
1192                         }
1193                         if (!S_ISDIR(e->versions[1].mode)) {
1194                                 e->tree = new_tree_content(8);
1195                                 e->versions[1].mode = S_IFDIR;
1196                         }
1197                         if (!e->tree)
1198                                 load_tree(e);
1199                         if (tree_content_set(e, slash1 + 1, sha1, mode)) {
1200                                 hashclr(root->versions[1].sha1);
1201                                 return 1;
1202                         }
1203                         return 0;
1204                 }
1205         }
1206
1207         if (t->entry_count == t->entry_capacity)
1208                 root->tree = t = grow_tree_content(t, 8);
1209         e = new_tree_entry();
1210         e->name = to_atom(p, n);
1211         e->versions[0].mode = 0;
1212         hashclr(e->versions[0].sha1);
1213         t->entries[t->entry_count++] = e;
1214         if (slash1) {
1215                 e->tree = new_tree_content(8);
1216                 e->versions[1].mode = S_IFDIR;
1217                 tree_content_set(e, slash1 + 1, sha1, mode);
1218         } else {
1219                 e->tree = NULL;
1220                 e->versions[1].mode = mode;
1221                 hashcpy(e->versions[1].sha1, sha1);
1222         }
1223         hashclr(root->versions[1].sha1);
1224         return 1;
1225 }
1226
1227 static int tree_content_remove(struct tree_entry *root, const char *p)
1228 {
1229         struct tree_content *t = root->tree;
1230         const char *slash1;
1231         unsigned int i, n;
1232         struct tree_entry *e;
1233
1234         slash1 = strchr(p, '/');
1235         if (slash1)
1236                 n = slash1 - p;
1237         else
1238                 n = strlen(p);
1239
1240         for (i = 0; i < t->entry_count; i++) {
1241                 e = t->entries[i];
1242                 if (e->name->str_len == n && !strncmp(p, e->name->str_dat, n)) {
1243                         if (!slash1 || !S_ISDIR(e->versions[1].mode))
1244                                 goto del_entry;
1245                         if (!e->tree)
1246                                 load_tree(e);
1247                         if (tree_content_remove(e, slash1 + 1)) {
1248                                 for (n = 0; n < e->tree->entry_count; n++) {
1249                                         if (e->tree->entries[n]->versions[1].mode) {
1250                                                 hashclr(root->versions[1].sha1);
1251                                                 return 1;
1252                                         }
1253                                 }
1254                                 goto del_entry;
1255                         }
1256                         return 0;
1257                 }
1258         }
1259         return 0;
1260
1261 del_entry:
1262         if (e->tree) {
1263                 release_tree_content_recursive(e->tree);
1264                 e->tree = NULL;
1265         }
1266         e->versions[1].mode = 0;
1267         hashclr(e->versions[1].sha1);
1268         hashclr(root->versions[1].sha1);
1269         return 1;
1270 }
1271
1272 static void dump_branches(void)
1273 {
1274         static const char *msg = "fast-import";
1275         unsigned int i;
1276         struct branch *b;
1277         struct ref_lock *lock;
1278
1279         for (i = 0; i < branch_table_sz; i++) {
1280                 for (b = branch_table[i]; b; b = b->table_next_branch) {
1281                         lock = lock_any_ref_for_update(b->name, NULL);
1282                         if (!lock || write_ref_sha1(lock, b->sha1, msg) < 0)
1283                                 die("Can't write %s", b->name);
1284                 }
1285         }
1286 }
1287
1288 static void dump_tags(void)
1289 {
1290         static const char *msg = "fast-import";
1291         struct tag *t;
1292         struct ref_lock *lock;
1293         char path[PATH_MAX];
1294
1295         for (t = first_tag; t; t = t->next_tag) {
1296                 sprintf(path, "refs/tags/%s", t->name);
1297                 lock = lock_any_ref_for_update(path, NULL);
1298                 if (!lock || write_ref_sha1(lock, t->sha1, msg) < 0)
1299                         die("Can't write %s", path);
1300         }
1301 }
1302
1303 static void dump_marks_helper(FILE *f,
1304         uintmax_t base,
1305         struct mark_set *m)
1306 {
1307         uintmax_t k;
1308         if (m->shift) {
1309                 for (k = 0; k < 1024; k++) {
1310                         if (m->data.sets[k])
1311                                 dump_marks_helper(f, (base + k) << m->shift,
1312                                         m->data.sets[k]);
1313                 }
1314         } else {
1315                 for (k = 0; k < 1024; k++) {
1316                         if (m->data.marked[k])
1317                                 fprintf(f, ":%ju %s\n", base + k,
1318                                         sha1_to_hex(m->data.marked[k]->sha1));
1319                 }
1320         }
1321 }
1322
1323 static void dump_marks(void)
1324 {
1325         if (mark_file)
1326         {
1327                 FILE *f = fopen(mark_file, "w");
1328                 dump_marks_helper(f, 0, marks);
1329                 fclose(f);
1330         }
1331 }
1332
1333 static void read_next_command(void)
1334 {
1335         read_line(&command_buf, stdin, '\n');
1336 }
1337
1338 static void cmd_mark(void)
1339 {
1340         if (!strncmp("mark :", command_buf.buf, 6)) {
1341                 next_mark = strtoumax(command_buf.buf + 6, NULL, 10);
1342                 read_next_command();
1343         }
1344         else
1345                 next_mark = 0;
1346 }
1347
1348 static void* cmd_data (size_t *size)
1349 {
1350         size_t length;
1351         char *buffer;
1352
1353         if (strncmp("data ", command_buf.buf, 5))
1354                 die("Expected 'data n' command, found: %s", command_buf.buf);
1355
1356         if (!strncmp("<<", command_buf.buf + 5, 2)) {
1357                 char *term = xstrdup(command_buf.buf + 5 + 2);
1358                 size_t sz = 8192, term_len = command_buf.len - 5 - 2;
1359                 length = 0;
1360                 buffer = xmalloc(sz);
1361                 for (;;) {
1362                         read_next_command();
1363                         if (command_buf.eof)
1364                                 die("EOF in data (terminator '%s' not found)", term);
1365                         if (term_len == command_buf.len
1366                                 && !strcmp(term, command_buf.buf))
1367                                 break;
1368                         if (sz < (length + command_buf.len)) {
1369                                 sz = sz * 3 / 2 + 16;
1370                                 if (sz < (length + command_buf.len))
1371                                         sz = length + command_buf.len;
1372                                 buffer = xrealloc(buffer, sz);
1373                         }
1374                         memcpy(buffer + length,
1375                                 command_buf.buf,
1376                                 command_buf.len - 1);
1377                         length += command_buf.len - 1;
1378                         buffer[length++] = '\n';
1379                 }
1380                 free(term);
1381         }
1382         else {
1383                 size_t n = 0;
1384                 length = strtoul(command_buf.buf + 5, NULL, 10);
1385                 buffer = xmalloc(length);
1386                 while (n < length) {
1387                         size_t s = fread(buffer + n, 1, length - n, stdin);
1388                         if (!s && feof(stdin))
1389                                 die("EOF in data (%lu bytes remaining)", length - n);
1390                         n += s;
1391                 }
1392         }
1393
1394         if (fgetc(stdin) != '\n')
1395                 die("An lf did not trail the binary data as expected.");
1396
1397         *size = length;
1398         return buffer;
1399 }
1400
1401 static void cmd_new_blob(void)
1402 {
1403         size_t l;
1404         void *d;
1405
1406         read_next_command();
1407         cmd_mark();
1408         d = cmd_data(&l);
1409
1410         if (store_object(OBJ_BLOB, d, l, &last_blob, NULL, next_mark))
1411                 free(d);
1412 }
1413
1414 static void unload_one_branch(void)
1415 {
1416         while (cur_active_branches
1417                 && cur_active_branches >= max_active_branches) {
1418                 unsigned long min_commit = ULONG_MAX;
1419                 struct branch *e, *l = NULL, *p = NULL;
1420
1421                 for (e = active_branches; e; e = e->active_next_branch) {
1422                         if (e->last_commit < min_commit) {
1423                                 p = l;
1424                                 min_commit = e->last_commit;
1425                         }
1426                         l = e;
1427                 }
1428
1429                 if (p) {
1430                         e = p->active_next_branch;
1431                         p->active_next_branch = e->active_next_branch;
1432                 } else {
1433                         e = active_branches;
1434                         active_branches = e->active_next_branch;
1435                 }
1436                 e->active_next_branch = NULL;
1437                 if (e->branch_tree.tree) {
1438                         release_tree_content_recursive(e->branch_tree.tree);
1439                         e->branch_tree.tree = NULL;
1440                 }
1441                 cur_active_branches--;
1442         }
1443 }
1444
1445 static void load_branch(struct branch *b)
1446 {
1447         load_tree(&b->branch_tree);
1448         b->active_next_branch = active_branches;
1449         active_branches = b;
1450         cur_active_branches++;
1451         branch_load_count++;
1452 }
1453
1454 static void file_change_m(struct branch *b)
1455 {
1456         const char *p = command_buf.buf + 2;
1457         char *p_uq;
1458         const char *endp;
1459         struct object_entry *oe;
1460         unsigned char sha1[20];
1461         unsigned int mode, inline_data = 0;
1462         char type[20];
1463
1464         p = get_mode(p, &mode);
1465         if (!p)
1466                 die("Corrupt mode: %s", command_buf.buf);
1467         switch (mode) {
1468         case S_IFREG | 0644:
1469         case S_IFREG | 0755:
1470         case S_IFLNK:
1471         case 0644:
1472         case 0755:
1473                 /* ok */
1474                 break;
1475         default:
1476                 die("Corrupt mode: %s", command_buf.buf);
1477         }
1478
1479         if (*p == ':') {
1480                 char *x;
1481                 oe = find_mark(strtoumax(p + 1, &x, 10));
1482                 hashcpy(sha1, oe->sha1);
1483                 p = x;
1484         } else if (!strncmp("inline", p, 6)) {
1485                 inline_data = 1;
1486                 p += 6;
1487         } else {
1488                 if (get_sha1_hex(p, sha1))
1489                         die("Invalid SHA1: %s", command_buf.buf);
1490                 oe = find_object(sha1);
1491                 p += 40;
1492         }
1493         if (*p++ != ' ')
1494                 die("Missing space after SHA1: %s", command_buf.buf);
1495
1496         p_uq = unquote_c_style(p, &endp);
1497         if (p_uq) {
1498                 if (*endp)
1499                         die("Garbage after path in: %s", command_buf.buf);
1500                 p = p_uq;
1501         }
1502
1503         if (inline_data) {
1504                 size_t l;
1505                 void *d;
1506                 if (!p_uq)
1507                         p = p_uq = xstrdup(p);
1508                 read_next_command();
1509                 d = cmd_data(&l);
1510                 if (store_object(OBJ_BLOB, d, l, &last_blob, sha1, 0))
1511                         free(d);
1512         } else if (oe) {
1513                 if (oe->type != OBJ_BLOB)
1514                         die("Not a blob (actually a %s): %s",
1515                                 command_buf.buf, type_names[oe->type]);
1516         } else {
1517                 if (sha1_object_info(sha1, type, NULL))
1518                         die("Blob not found: %s", command_buf.buf);
1519                 if (strcmp(blob_type, type))
1520                         die("Not a blob (actually a %s): %s",
1521                                 command_buf.buf, type);
1522         }
1523
1524         tree_content_set(&b->branch_tree, p, sha1, S_IFREG | mode);
1525
1526         if (p_uq)
1527                 free(p_uq);
1528 }
1529
1530 static void file_change_d(struct branch *b)
1531 {
1532         const char *p = command_buf.buf + 2;
1533         char *p_uq;
1534         const char *endp;
1535
1536         p_uq = unquote_c_style(p, &endp);
1537         if (p_uq) {
1538                 if (*endp)
1539                         die("Garbage after path in: %s", command_buf.buf);
1540                 p = p_uq;
1541         }
1542         tree_content_remove(&b->branch_tree, p);
1543         if (p_uq)
1544                 free(p_uq);
1545 }
1546
1547 static void cmd_from(struct branch *b)
1548 {
1549         const char *from, *endp;
1550         char *str_uq;
1551         struct branch *s;
1552
1553         if (strncmp("from ", command_buf.buf, 5))
1554                 return;
1555
1556         if (b->last_commit)
1557                 die("Can't reinitailize branch %s", b->name);
1558
1559         from = strchr(command_buf.buf, ' ') + 1;
1560         str_uq = unquote_c_style(from, &endp);
1561         if (str_uq) {
1562                 if (*endp)
1563                         die("Garbage after string in: %s", command_buf.buf);
1564                 from = str_uq;
1565         }
1566
1567         s = lookup_branch(from);
1568         if (b == s)
1569                 die("Can't create a branch from itself: %s", b->name);
1570         else if (s) {
1571                 unsigned char *t = s->branch_tree.versions[1].sha1;
1572                 hashcpy(b->sha1, s->sha1);
1573                 hashcpy(b->branch_tree.versions[0].sha1, t);
1574                 hashcpy(b->branch_tree.versions[1].sha1, t);
1575         } else if (*from == ':') {
1576                 uintmax_t idnum = strtoumax(from + 1, NULL, 10);
1577                 struct object_entry *oe = find_mark(idnum);
1578                 unsigned long size;
1579                 char *buf;
1580                 if (oe->type != OBJ_COMMIT)
1581                         die("Mark :%ju not a commit", idnum);
1582                 hashcpy(b->sha1, oe->sha1);
1583                 buf = gfi_unpack_entry(oe, &size);
1584                 if (!buf || size < 46)
1585                         die("Not a valid commit: %s", from);
1586                 if (memcmp("tree ", buf, 5)
1587                         || get_sha1_hex(buf + 5, b->branch_tree.versions[1].sha1))
1588                         die("The commit %s is corrupt", sha1_to_hex(b->sha1));
1589                 free(buf);
1590                 hashcpy(b->branch_tree.versions[0].sha1,
1591                         b->branch_tree.versions[1].sha1);
1592         } else if (!get_sha1(from, b->sha1)) {
1593                 if (is_null_sha1(b->sha1)) {
1594                         hashclr(b->branch_tree.versions[0].sha1);
1595                         hashclr(b->branch_tree.versions[1].sha1);
1596                 } else {
1597                         unsigned long size;
1598                         char *buf;
1599
1600                         buf = read_object_with_reference(b->sha1,
1601                                 type_names[OBJ_COMMIT], &size, b->sha1);
1602                         if (!buf || size < 46)
1603                                 die("Not a valid commit: %s", from);
1604                         if (memcmp("tree ", buf, 5)
1605                                 || get_sha1_hex(buf + 5, b->branch_tree.versions[1].sha1))
1606                                 die("The commit %s is corrupt", sha1_to_hex(b->sha1));
1607                         free(buf);
1608                         hashcpy(b->branch_tree.versions[0].sha1,
1609                                 b->branch_tree.versions[1].sha1);
1610                 }
1611         } else
1612                 die("Invalid ref name or SHA1 expression: %s", from);
1613
1614         read_next_command();
1615 }
1616
1617 static struct hash_list* cmd_merge(unsigned int *count)
1618 {
1619         struct hash_list *list = NULL, *n, *e;
1620         const char *from, *endp;
1621         char *str_uq;
1622         struct branch *s;
1623
1624         *count = 0;
1625         while (!strncmp("merge ", command_buf.buf, 6)) {
1626                 from = strchr(command_buf.buf, ' ') + 1;
1627                 str_uq = unquote_c_style(from, &endp);
1628                 if (str_uq) {
1629                         if (*endp)
1630                                 die("Garbage after string in: %s", command_buf.buf);
1631                         from = str_uq;
1632                 }
1633
1634                 n = xmalloc(sizeof(*n));
1635                 s = lookup_branch(from);
1636                 if (s)
1637                         hashcpy(n->sha1, s->sha1);
1638                 else if (*from == ':') {
1639                         uintmax_t idnum = strtoumax(from + 1, NULL, 10);
1640                         struct object_entry *oe = find_mark(idnum);
1641                         if (oe->type != OBJ_COMMIT)
1642                                 die("Mark :%ju not a commit", idnum);
1643                         hashcpy(n->sha1, oe->sha1);
1644                 } else if (get_sha1(from, n->sha1))
1645                         die("Invalid ref name or SHA1 expression: %s", from);
1646
1647                 n->next = NULL;
1648                 if (list)
1649                         e->next = n;
1650                 else
1651                         list = n;
1652                 e = n;
1653                 *count++;
1654                 read_next_command();
1655         }
1656         return list;
1657 }
1658
1659 static void cmd_new_commit(void)
1660 {
1661         struct branch *b;
1662         void *msg;
1663         size_t msglen;
1664         char *str_uq;
1665         const char *endp;
1666         char *sp;
1667         char *author = NULL;
1668         char *committer = NULL;
1669         struct hash_list *merge_list = NULL;
1670         unsigned int merge_count;
1671
1672         /* Obtain the branch name from the rest of our command */
1673         sp = strchr(command_buf.buf, ' ') + 1;
1674         str_uq = unquote_c_style(sp, &endp);
1675         if (str_uq) {
1676                 if (*endp)
1677                         die("Garbage after ref in: %s", command_buf.buf);
1678                 sp = str_uq;
1679         }
1680         b = lookup_branch(sp);
1681         if (!b)
1682                 b = new_branch(sp);
1683         if (str_uq)
1684                 free(str_uq);
1685
1686         read_next_command();
1687         cmd_mark();
1688         if (!strncmp("author ", command_buf.buf, 7)) {
1689                 author = strdup(command_buf.buf);
1690                 read_next_command();
1691         }
1692         if (!strncmp("committer ", command_buf.buf, 10)) {
1693                 committer = strdup(command_buf.buf);
1694                 read_next_command();
1695         }
1696         if (!committer)
1697                 die("Expected committer but didn't get one");
1698         msg = cmd_data(&msglen);
1699         read_next_command();
1700         cmd_from(b);
1701         merge_list = cmd_merge(&merge_count);
1702
1703         /* ensure the branch is active/loaded */
1704         if (!b->branch_tree.tree || !max_active_branches) {
1705                 unload_one_branch();
1706                 load_branch(b);
1707         }
1708
1709         /* file_change* */
1710         for (;;) {
1711                 if (1 == command_buf.len)
1712                         break;
1713                 else if (!strncmp("M ", command_buf.buf, 2))
1714                         file_change_m(b);
1715                 else if (!strncmp("D ", command_buf.buf, 2))
1716                         file_change_d(b);
1717                 else
1718                         die("Unsupported file_change: %s", command_buf.buf);
1719                 read_next_command();
1720         }
1721
1722         /* build the tree and the commit */
1723         store_tree(&b->branch_tree);
1724         hashcpy(b->branch_tree.versions[0].sha1,
1725                 b->branch_tree.versions[1].sha1);
1726         size_dbuf(&new_data, 97 + msglen
1727                 + merge_count * 49
1728                 + (author
1729                         ? strlen(author) + strlen(committer)
1730                         : 2 * strlen(committer)));
1731         sp = new_data.buffer;
1732         sp += sprintf(sp, "tree %s\n",
1733                 sha1_to_hex(b->branch_tree.versions[1].sha1));
1734         if (!is_null_sha1(b->sha1))
1735                 sp += sprintf(sp, "parent %s\n", sha1_to_hex(b->sha1));
1736         while (merge_list) {
1737                 struct hash_list *next = merge_list->next;
1738                 sp += sprintf(sp, "parent %s\n", sha1_to_hex(merge_list->sha1));
1739                 free(merge_list);
1740                 merge_list = next;
1741         }
1742         if (author)
1743                 sp += sprintf(sp, "%s\n", author);
1744         else
1745                 sp += sprintf(sp, "author %s\n", committer + 10);
1746         sp += sprintf(sp, "%s\n\n", committer);
1747         memcpy(sp, msg, msglen);
1748         sp += msglen;
1749         if (author)
1750                 free(author);
1751         free(committer);
1752         free(msg);
1753
1754         if (!store_object(OBJ_COMMIT,
1755                 new_data.buffer, sp - (char*)new_data.buffer,
1756                 NULL, b->sha1, next_mark))
1757                 b->pack_id = pack_id;
1758         b->last_commit = object_count_by_type[OBJ_COMMIT];
1759
1760         if (branch_log) {
1761                 int need_dq = quote_c_style(b->name, NULL, NULL, 0);
1762                 fprintf(branch_log, "commit ");
1763                 if (need_dq) {
1764                         fputc('"', branch_log);
1765                         quote_c_style(b->name, NULL, branch_log, 0);
1766                         fputc('"', branch_log);
1767                 } else
1768                         fprintf(branch_log, "%s", b->name);
1769                 fprintf(branch_log," :%ju %s\n",next_mark,sha1_to_hex(b->sha1));
1770         }
1771 }
1772
1773 static void cmd_new_tag(void)
1774 {
1775         char *str_uq;
1776         const char *endp;
1777         char *sp;
1778         const char *from;
1779         char *tagger;
1780         struct branch *s;
1781         void *msg;
1782         size_t msglen;
1783         struct tag *t;
1784         uintmax_t from_mark = 0;
1785         unsigned char sha1[20];
1786
1787         /* Obtain the new tag name from the rest of our command */
1788         sp = strchr(command_buf.buf, ' ') + 1;
1789         str_uq = unquote_c_style(sp, &endp);
1790         if (str_uq) {
1791                 if (*endp)
1792                         die("Garbage after tag name in: %s", command_buf.buf);
1793                 sp = str_uq;
1794         }
1795         t = pool_alloc(sizeof(struct tag));
1796         t->next_tag = NULL;
1797         t->name = pool_strdup(sp);
1798         if (last_tag)
1799                 last_tag->next_tag = t;
1800         else
1801                 first_tag = t;
1802         last_tag = t;
1803         if (str_uq)
1804                 free(str_uq);
1805         read_next_command();
1806
1807         /* from ... */
1808         if (strncmp("from ", command_buf.buf, 5))
1809                 die("Expected from command, got %s", command_buf.buf);
1810
1811         from = strchr(command_buf.buf, ' ') + 1;
1812         str_uq = unquote_c_style(from, &endp);
1813         if (str_uq) {
1814                 if (*endp)
1815                         die("Garbage after string in: %s", command_buf.buf);
1816                 from = str_uq;
1817         }
1818
1819         s = lookup_branch(from);
1820         if (s) {
1821                 hashcpy(sha1, s->sha1);
1822         } else if (*from == ':') {
1823                 from_mark = strtoumax(from + 1, NULL, 10);
1824                 struct object_entry *oe = find_mark(from_mark);
1825                 if (oe->type != OBJ_COMMIT)
1826                         die("Mark :%ju not a commit", from_mark);
1827                 hashcpy(sha1, oe->sha1);
1828         } else if (!get_sha1(from, sha1)) {
1829                 unsigned long size;
1830                 char *buf;
1831
1832                 buf = read_object_with_reference(sha1,
1833                         type_names[OBJ_COMMIT], &size, sha1);
1834                 if (!buf || size < 46)
1835                         die("Not a valid commit: %s", from);
1836                 free(buf);
1837         } else
1838                 die("Invalid ref name or SHA1 expression: %s", from);
1839
1840         if (str_uq)
1841                 free(str_uq);
1842         read_next_command();
1843
1844         /* tagger ... */
1845         if (strncmp("tagger ", command_buf.buf, 7))
1846                 die("Expected tagger command, got %s", command_buf.buf);
1847         tagger = strdup(command_buf.buf);
1848
1849         /* tag payload/message */
1850         read_next_command();
1851         msg = cmd_data(&msglen);
1852
1853         /* build the tag object */
1854         size_dbuf(&new_data, 67+strlen(t->name)+strlen(tagger)+msglen);
1855         sp = new_data.buffer;
1856         sp += sprintf(sp, "object %s\n", sha1_to_hex(sha1));
1857         sp += sprintf(sp, "type %s\n", type_names[OBJ_COMMIT]);
1858         sp += sprintf(sp, "tag %s\n", t->name);
1859         sp += sprintf(sp, "%s\n\n", tagger);
1860         memcpy(sp, msg, msglen);
1861         sp += msglen;
1862         free(tagger);
1863         free(msg);
1864
1865         if (store_object(OBJ_TAG, new_data.buffer,
1866                 sp - (char*)new_data.buffer,
1867                 NULL, t->sha1, 0))
1868                 t->pack_id = MAX_PACK_ID;
1869         else
1870                 t->pack_id = pack_id;
1871
1872         if (branch_log) {
1873                 int need_dq = quote_c_style(t->name, NULL, NULL, 0);
1874                 fprintf(branch_log, "tag ");
1875                 if (need_dq) {
1876                         fputc('"', branch_log);
1877                         quote_c_style(t->name, NULL, branch_log, 0);
1878                         fputc('"', branch_log);
1879                 } else
1880                         fprintf(branch_log, "%s", t->name);
1881                 fprintf(branch_log," :%ju %s\n",from_mark,sha1_to_hex(t->sha1));
1882         }
1883 }
1884
1885 static void cmd_reset_branch(void)
1886 {
1887         struct branch *b;
1888         char *str_uq;
1889         const char *endp;
1890         char *sp;
1891
1892         /* Obtain the branch name from the rest of our command */
1893         sp = strchr(command_buf.buf, ' ') + 1;
1894         str_uq = unquote_c_style(sp, &endp);
1895         if (str_uq) {
1896                 if (*endp)
1897                         die("Garbage after ref in: %s", command_buf.buf);
1898                 sp = str_uq;
1899         }
1900         b = lookup_branch(sp);
1901         if (b) {
1902                 b->last_commit = 0;
1903                 if (b->branch_tree.tree) {
1904                         release_tree_content_recursive(b->branch_tree.tree);
1905                         b->branch_tree.tree = NULL;
1906                 }
1907         }
1908         else
1909                 b = new_branch(sp);
1910         if (str_uq)
1911                 free(str_uq);
1912         read_next_command();
1913         cmd_from(b);
1914 }
1915
1916 static void cmd_checkpoint(void)
1917 {
1918         if (object_count)
1919                 checkpoint();
1920         read_next_command();
1921 }
1922
1923 static const char fast_import_usage[] =
1924 "git-fast-import [--depth=n] [--active-branches=n] [--export-marks=marks.file] [--branch-log=log]";
1925
1926 int main(int argc, const char **argv)
1927 {
1928         int i;
1929         uintmax_t total_count, duplicate_count;
1930
1931         git_config(git_default_config);
1932
1933         for (i = 1; i < argc; i++) {
1934                 const char *a = argv[i];
1935
1936                 if (*a != '-' || !strcmp(a, "--"))
1937                         break;
1938                 else if (!strncmp(a, "--max-pack-size=", 16))
1939                         max_packsize = strtoumax(a + 16, NULL, 0) * 1024 * 1024;
1940                 else if (!strncmp(a, "--depth=", 8))
1941                         max_depth = strtoul(a + 8, NULL, 0);
1942                 else if (!strncmp(a, "--active-branches=", 18))
1943                         max_active_branches = strtoul(a + 18, NULL, 0);
1944                 else if (!strncmp(a, "--export-marks=", 15))
1945                         mark_file = a + 15;
1946                 else if (!strncmp(a, "--branch-log=", 13)) {
1947                         branch_log = fopen(a + 13, "w");
1948                         if (!branch_log)
1949                                 die("Can't create %s: %s", a + 13, strerror(errno));
1950                 }
1951                 else
1952                         die("unknown option %s", a);
1953         }
1954         if (i != argc)
1955                 usage(fast_import_usage);
1956
1957         alloc_objects(object_entry_alloc);
1958         strbuf_init(&command_buf);
1959
1960         atom_table = xcalloc(atom_table_sz, sizeof(struct atom_str*));
1961         branch_table = xcalloc(branch_table_sz, sizeof(struct branch*));
1962         avail_tree_table = xcalloc(avail_tree_table_sz, sizeof(struct avail_tree_content*));
1963         marks = pool_calloc(1, sizeof(struct mark_set));
1964
1965         start_packfile();
1966         for (;;) {
1967                 read_next_command();
1968                 if (command_buf.eof)
1969                         break;
1970                 else if (!strcmp("blob", command_buf.buf))
1971                         cmd_new_blob();
1972                 else if (!strncmp("commit ", command_buf.buf, 7))
1973                         cmd_new_commit();
1974                 else if (!strncmp("tag ", command_buf.buf, 4))
1975                         cmd_new_tag();
1976                 else if (!strncmp("reset ", command_buf.buf, 6))
1977                         cmd_reset_branch();
1978                 else if (!strcmp("checkpoint", command_buf.buf))
1979                         cmd_checkpoint();
1980                 else
1981                         die("Unsupported command: %s", command_buf.buf);
1982         }
1983         end_packfile();
1984
1985         dump_branches();
1986         dump_tags();
1987         unkeep_all_packs();
1988         dump_marks();
1989         if (branch_log)
1990                 fclose(branch_log);
1991
1992         total_count = 0;
1993         for (i = 0; i < ARRAY_SIZE(object_count_by_type); i++)
1994                 total_count += object_count_by_type[i];
1995         duplicate_count = 0;
1996         for (i = 0; i < ARRAY_SIZE(duplicate_count_by_type); i++)
1997                 duplicate_count += duplicate_count_by_type[i];
1998
1999         fprintf(stderr, "%s statistics:\n", argv[0]);
2000         fprintf(stderr, "---------------------------------------------------------------------\n");
2001         fprintf(stderr, "Alloc'd objects: %10ju\n", alloc_count);
2002         fprintf(stderr, "Total objects:   %10ju (%10ju duplicates                  )\n", total_count, duplicate_count);
2003         fprintf(stderr, "      blobs  :   %10ju (%10ju duplicates %10ju deltas)\n", object_count_by_type[OBJ_BLOB], duplicate_count_by_type[OBJ_BLOB], delta_count_by_type[OBJ_BLOB]);
2004         fprintf(stderr, "      trees  :   %10ju (%10ju duplicates %10ju deltas)\n", object_count_by_type[OBJ_TREE], duplicate_count_by_type[OBJ_TREE], delta_count_by_type[OBJ_TREE]);
2005         fprintf(stderr, "      commits:   %10ju (%10ju duplicates %10ju deltas)\n", object_count_by_type[OBJ_COMMIT], duplicate_count_by_type[OBJ_COMMIT], delta_count_by_type[OBJ_COMMIT]);
2006         fprintf(stderr, "      tags   :   %10ju (%10ju duplicates %10ju deltas)\n", object_count_by_type[OBJ_TAG], duplicate_count_by_type[OBJ_TAG], delta_count_by_type[OBJ_TAG]);
2007         fprintf(stderr, "Total branches:  %10lu (%10lu loads     )\n", branch_count, branch_load_count);
2008         fprintf(stderr, "      marks:     %10ju (%10ju unique    )\n", (((uintmax_t)1) << marks->shift) * 1024, marks_set_count);
2009         fprintf(stderr, "      atoms:     %10u\n", atom_cnt);
2010         fprintf(stderr, "Memory total:    %10ju KiB\n", (total_allocd + alloc_count*sizeof(struct object_entry))/1024);
2011         fprintf(stderr, "       pools:    %10lu KiB\n", total_allocd/1024);
2012         fprintf(stderr, "     objects:    %10ju KiB\n", (alloc_count*sizeof(struct object_entry))/1024);
2013         fprintf(stderr, "---------------------------------------------------------------------\n");
2014         pack_report();
2015         fprintf(stderr, "---------------------------------------------------------------------\n");
2016         fprintf(stderr, "\n");
2017
2018         return 0;
2019 }