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;
  79   sha1exp_str ::= sha1exp;
  80   tag_str     ::= 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         uint32_t 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         uint32_t 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 short 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                 uint16_t 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
 301
 302 static void alloc_objects(unsigned int cnt)
 303 {
 304         struct object_entry_pool *b;
 305
 306         b = xmalloc(sizeof(struct object_entry_pool)
 307                 + cnt * sizeof(struct object_entry));
 308         b->next_pool = blocks;
 309         b->next_free = b->entries;
 310         b->end = b->entries + cnt;
 311         blocks = b;
 312         alloc_count += cnt;
 313 }
 314
 315 static struct object_entry* new_object(unsigned char *sha1)
 316 {
 317         struct object_entry *e;
 318
 319         if (blocks->next_free == blocks->end)
 320                 alloc_objects(object_entry_alloc);
 321
 322         e = blocks->next_free++;
 323         hashcpy(e->sha1, sha1);
 324         return e;
 325 }
 326
 327 static struct object_entry* find_object(unsigned char *sha1)
 328 {
 329         unsigned int h = sha1[0] << 8 | sha1[1];
 330         struct object_entry *e;
 331         for (e = object_table[h]; e; e = e->next)
 332                 if (!hashcmp(sha1, e->sha1))
 333                         return e;
 334         return NULL;
 335 }
 336
 337 static struct object_entry* insert_object(unsigned char *sha1)
 338 {
 339         unsigned int h = sha1[0] << 8 | sha1[1];
 340         struct object_entry *e = object_table[h];
 341         struct object_entry *p = NULL;
 342
 343         while (e) {
 344                 if (!hashcmp(sha1, e->sha1))
 345                         return e;
 346                 p = e;
 347                 e = e->next;
 348         }
 349
 350         e = new_object(sha1);
 351         e->next = NULL;
 352         e->offset = 0;
 353         if (p)
 354                 p->next = e;
 355         else
 356                 object_table[h] = e;
 357         return e;
 358 }
 359
 360 static unsigned int hc_str(const char *s, size_t len)
 361 {
 362         unsigned int r = 0;
 363         while (len-- > 0)
 364                 r = r * 31 + *s++;
 365         return r;
 366 }
 367
 368 static void* pool_alloc(size_t len)
 369 {
 370         struct mem_pool *p;
 371         void *r;
 372
 373         for (p = mem_pool; p; p = p->next_pool)
 374                 if ((p->end - p->next_free >= len))
 375                         break;
 376
 377         if (!p) {
 378                 if (len >= (mem_pool_alloc/2)) {
 379                         total_allocd += len;
 380                         return xmalloc(len);
 381                 }
 382                 total_allocd += sizeof(struct mem_pool) + mem_pool_alloc;
 383                 p = xmalloc(sizeof(struct mem_pool) + mem_pool_alloc);
 384                 p->next_pool = mem_pool;
 385                 p->next_free = p->space;
 386                 p->end = p->next_free + mem_pool_alloc;
 387                 mem_pool = p;
 388         }
 389
 390         r = p->next_free;
 391         /* round out to a pointer alignment */
 392         if (len & (sizeof(void*) - 1))
 393                 len += sizeof(void*) - (len & (sizeof(void*) - 1));
 394         p->next_free += len;
 395         return r;
 396 }
 397
 398 static void* pool_calloc(size_t count, size_t size)
 399 {
 400         size_t len = count * size;
 401         void *r = pool_alloc(len);
 402         memset(r, 0, len);
 403         return r;
 404 }
 405
 406 static char* pool_strdup(const char *s)
 407 {
 408         char *r = pool_alloc(strlen(s) + 1);
 409         strcpy(r, s);
 410         return r;
 411 }
 412
 413 static void size_dbuf(struct dbuf *b, size_t maxlen)
 414 {
 415         if (b->buffer) {
 416                 if (b->capacity >= maxlen)
 417                         return;
 418                 free(b->buffer);
 419         }
 420         b->capacity = ((maxlen / 1024) + 1) * 1024;
 421         b->buffer = xmalloc(b->capacity);
 422 }
 423
 424 static void insert_mark(uintmax_t idnum, struct object_entry *oe)
 425 {
 426         struct mark_set *s = marks;
 427         while ((idnum >> s->shift) >= 1024) {
 428                 s = pool_calloc(1, sizeof(struct mark_set));
 429                 s->shift = marks->shift + 10;
 430                 s->data.sets[0] = marks;
 431                 marks = s;
 432         }
 433         while (s->shift) {
 434                 uintmax_t i = idnum >> s->shift;
 435                 idnum -= i << s->shift;
 436                 if (!s->data.sets[i]) {
 437                         s->data.sets[i] = pool_calloc(1, sizeof(struct mark_set));
 438                         s->data.sets[i]->shift = s->shift - 10;
 439                 }
 440                 s = s->data.sets[i];
 441         }
 442         if (!s->data.marked[idnum])
 443                 marks_set_count++;
 444         s->data.marked[idnum] = oe;
 445 }
 446
 447 static struct object_entry* find_mark(uintmax_t idnum)
 448 {
 449         uintmax_t orig_idnum = idnum;
 450         struct mark_set *s = marks;
 451         struct object_entry *oe = NULL;
 452         if ((idnum >> s->shift) < 1024) {
 453                 while (s && s->shift) {
 454                         uintmax_t i = idnum >> s->shift;
 455                         idnum -= i << s->shift;
 456                         s = s->data.sets[i];
 457                 }
 458                 if (s)
 459                         oe = s->data.marked[idnum];
 460         }
 461         if (!oe)
 462                 die("mark :%ju not declared", orig_idnum);
 463         return oe;
 464 }
 465
 466 static struct atom_str* to_atom(const char *s, unsigned short len)
 467 {
 468         unsigned int hc = hc_str(s, len) % atom_table_sz;
 469         struct atom_str *c;
 470
 471         for (c = atom_table[hc]; c; c = c->next_atom)
 472                 if (c->str_len == len && !strncmp(s, c->str_dat, len))
 473                         return c;
 474
 475         c = pool_alloc(sizeof(struct atom_str) + len + 1);
 476         c->str_len = len;
 477         strncpy(c->str_dat, s, len);
 478         c->str_dat[len] = 0;
 479         c->next_atom = atom_table[hc];
 480         atom_table[hc] = c;
 481         atom_cnt++;
 482         return c;
 483 }
 484
 485 static struct branch* lookup_branch(const char *name)
 486 {
 487         unsigned int hc = hc_str(name, strlen(name)) % branch_table_sz;
 488         struct branch *b;
 489
 490         for (b = branch_table[hc]; b; b = b->table_next_branch)
 491                 if (!strcmp(name, b->name))
 492                         return b;
 493         return NULL;
 494 }
 495
 496 static struct branch* new_branch(const char *name)
 497 {
 498         unsigned int hc = hc_str(name, strlen(name)) % branch_table_sz;
 499         struct branch* b = lookup_branch(name);
 500
 501         if (b)
 502                 die("Invalid attempt to create duplicate branch: %s", name);
 503         if (check_ref_format(name))
 504                 die("Branch name doesn't conform to GIT standards: %s", name);
 505
 506         b = pool_calloc(1, sizeof(struct branch));
 507         b->name = pool_strdup(name);
 508         b->table_next_branch = branch_table[hc];
 509         b->branch_tree.versions[0].mode = S_IFDIR;
 510         b->branch_tree.versions[1].mode = S_IFDIR;
 511         b->pack_id = MAX_PACK_ID;
 512         branch_table[hc] = b;
 513         branch_count++;
 514         return b;
 515 }
 516
 517 static unsigned int hc_entries(unsigned int cnt)
 518 {
 519         cnt = cnt & 7 ? (cnt / 8) + 1 : cnt / 8;
 520         return cnt < avail_tree_table_sz ? cnt : avail_tree_table_sz - 1;
 521 }
 522
 523 static struct tree_content* new_tree_content(unsigned int cnt)
 524 {
 525         struct avail_tree_content *f, *l = NULL;
 526         struct tree_content *t;
 527         unsigned int hc = hc_entries(cnt);
 528
 529         for (f = avail_tree_table[hc]; f; l = f, f = f->next_avail)
 530                 if (f->entry_capacity >= cnt)
 531                         break;
 532
 533         if (f) {
 534                 if (l)
 535                         l->next_avail = f->next_avail;
 536                 else
 537                         avail_tree_table[hc] = f->next_avail;
 538         } else {
 539                 cnt = cnt & 7 ? ((cnt / 8) + 1) * 8 : cnt;
 540                 f = pool_alloc(sizeof(*t) + sizeof(t->entries[0]) * cnt);
 541                 f->entry_capacity = cnt;
 542         }
 543
 544         t = (struct tree_content*)f;
 545         t->entry_count = 0;
 546         t->delta_depth = 0;
 547         return t;
 548 }
 549
 550 static void release_tree_entry(struct tree_entry *e);
 551 static void release_tree_content(struct tree_content *t)
 552 {
 553         struct avail_tree_content *f = (struct avail_tree_content*)t;
 554         unsigned int hc = hc_entries(f->entry_capacity);
 555         f->next_avail = avail_tree_table[hc];
 556         avail_tree_table[hc] = f;
 557 }
 558
 559 static void release_tree_content_recursive(struct tree_content *t)
 560 {
 561         unsigned int i;
 562         for (i = 0; i < t->entry_count; i++)
 563                 release_tree_entry(t->entries[i]);
 564         release_tree_content(t);
 565 }
 566
 567 static struct tree_content* grow_tree_content(
 568         struct tree_content *t,
 569         int amt)
 570 {
 571         struct tree_content *r = new_tree_content(t->entry_count + amt);
 572         r->entry_count = t->entry_count;
 573         r->delta_depth = t->delta_depth;
 574         memcpy(r->entries,t->entries,t->entry_count*sizeof(t->entries[0]));
 575         release_tree_content(t);
 576         return r;
 577 }
 578
 579 static struct tree_entry* new_tree_entry(void)
 580 {
 581         struct tree_entry *e;
 582
 583         if (!avail_tree_entry) {
 584                 unsigned int n = tree_entry_alloc;
 585                 total_allocd += n * sizeof(struct tree_entry);
 586                 avail_tree_entry = e = xmalloc(n * sizeof(struct tree_entry));
 587                 while (n-- > 1) {
 588                         *((void**)e) = e + 1;
 589                         e++;
 590                 }
 591                 *((void**)e) = NULL;
 592         }
 593
 594         e = avail_tree_entry;
 595         avail_tree_entry = *((void**)e);
 596         return e;
 597 }
 598
 599 static void release_tree_entry(struct tree_entry *e)
 600 {
 601         if (e->tree)
 602                 release_tree_content_recursive(e->tree);
 603         *((void**)e) = avail_tree_entry;
 604         avail_tree_entry = e;
 605 }
 606
 607 static void start_packfile(void)
 608 {
 609         static char tmpfile[PATH_MAX];
 610         struct packed_git *p;
 611         struct pack_header hdr;
 612         int pack_fd;
 613
 614         snprintf(tmpfile, sizeof(tmpfile),
 615                 "%s/pack_XXXXXX", get_object_directory());
 616         pack_fd = mkstemp(tmpfile);
 617         if (pack_fd < 0)
 618                 die("Can't create %s: %s", tmpfile, strerror(errno));
 619         p = xcalloc(1, sizeof(*p) + strlen(tmpfile) + 2);
 620         strcpy(p->pack_name, tmpfile);
 621         p->pack_fd = pack_fd;
 622
 623         hdr.hdr_signature = htonl(PACK_SIGNATURE);
 624         hdr.hdr_version = htonl(2);
 625         hdr.hdr_entries = 0;
 626         write_or_die(p->pack_fd, &hdr, sizeof(hdr));
 627
 628         pack_data = p;
 629         pack_size = sizeof(hdr);
 630         object_count = 0;
 631
 632         all_packs = xrealloc(all_packs, sizeof(*all_packs) * (pack_id + 1));
 633         all_packs[pack_id] = p;
 634 }
 635
 636 static void fixup_header_footer(void)
 637 {
 638         static const int buf_sz = 128 * 1024;
 639         int pack_fd = pack_data->pack_fd;
 640         SHA_CTX c;
 641         struct pack_header hdr;
 642         char *buf;
 643
 644         if (lseek(pack_fd, 0, SEEK_SET) != 0)
 645                 die("Failed seeking to start: %s", strerror(errno));
 646         if (read_in_full(pack_fd, &hdr, sizeof(hdr)) != sizeof(hdr))
 647                 die("Unable to reread header of %s", pack_data->pack_name);
 648         if (lseek(pack_fd, 0, SEEK_SET) != 0)
 649                 die("Failed seeking to start: %s", strerror(errno));
 650         hdr.hdr_entries = htonl(object_count);
 651         write_or_die(pack_fd, &hdr, sizeof(hdr));
 652
 653         SHA1_Init(&c);
 654         SHA1_Update(&c, &hdr, sizeof(hdr));
 655
 656         buf = xmalloc(buf_sz);
 657         for (;;) {
 658                 size_t n = xread(pack_fd, buf, buf_sz);
 659                 if (!n)
 660                         break;
 661                 if (n < 0)
 662                         die("Failed to checksum %s", pack_data->pack_name);
 663                 SHA1_Update(&c, buf, n);
 664         }
 665         free(buf);
 666
 667         SHA1_Final(pack_data->sha1, &c);
 668         write_or_die(pack_fd, pack_data->sha1, sizeof(pack_data->sha1));
 669         close(pack_fd);
 670 }
 671
 672 static int oecmp (const void *a_, const void *b_)
 673 {
 674         struct object_entry *a = *((struct object_entry**)a_);
 675         struct object_entry *b = *((struct object_entry**)b_);
 676         return hashcmp(a->sha1, b->sha1);
 677 }
 678
 679 static char* create_index(void)
 680 {
 681         static char tmpfile[PATH_MAX];
 682         SHA_CTX ctx;
 683         struct sha1file *f;
 684         struct object_entry **idx, **c, **last, *e;
 685         struct object_entry_pool *o;
 686         uint32_t array[256];
 687         int i, idx_fd;
 688
 689         /* Build the sorted table of object IDs. */
 690         idx = xmalloc(object_count * sizeof(struct object_entry*));
 691         c = idx;
 692         for (o = blocks; o; o = o->next_pool)
 693                 for (e = o->next_free; e-- != o->entries;)
 694                         if (pack_id == e->pack_id)
 695                                 *c++ = e;
 696         last = idx + object_count;
 697         if (c != last)
 698                 die("internal consistency error creating the index");
 699         qsort(idx, object_count, sizeof(struct object_entry*), oecmp);
 700
 701         /* Generate the fan-out array. */
 702         c = idx;
 703         for (i = 0; i < 256; i++) {
 704                 struct object_entry **next = c;;
 705                 while (next < last) {
 706                         if ((*next)->sha1[0] != i)
 707                                 break;
 708                         next++;
 709                 }
 710                 array[i] = htonl(next - idx);
 711                 c = next;
 712         }
 713
 714         snprintf(tmpfile, sizeof(tmpfile),
 715                 "%s/index_XXXXXX", get_object_directory());
 716         idx_fd = mkstemp(tmpfile);
 717         if (idx_fd < 0)
 718                 die("Can't create %s: %s", tmpfile, strerror(errno));
 719         f = sha1fd(idx_fd, tmpfile);
 720         sha1write(f, array, 256 * sizeof(int));
 721         SHA1_Init(&ctx);
 722         for (c = idx; c != last; c++) {
 723                 uint32_t offset = htonl((*c)->offset);
 724                 sha1write(f, &offset, 4);
 725                 sha1write(f, (*c)->sha1, sizeof((*c)->sha1));
 726                 SHA1_Update(&ctx, (*c)->sha1, 20);
 727         }
 728         sha1write(f, pack_data->sha1, sizeof(pack_data->sha1));
 729         sha1close(f, NULL, 1);
 730         free(idx);
 731         SHA1_Final(pack_data->sha1, &ctx);
 732         return tmpfile;
 733 }
 734
 735 static char* keep_pack(char *curr_index_name)
 736 {
 737         static char name[PATH_MAX];
 738         static char *keep_msg = "fast-import";
 739         int keep_fd;
 740
 741         chmod(pack_data->pack_name, 0444);
 742         chmod(curr_index_name, 0444);
 743
 744         snprintf(name, sizeof(name), "%s/pack/pack-%s.keep",
 745                  get_object_directory(), sha1_to_hex(pack_data->sha1));
 746         keep_fd = open(name, O_RDWR|O_CREAT|O_EXCL, 0600);
 747         if (keep_fd < 0)
 748                 die("cannot create keep file");
 749         write(keep_fd, keep_msg, strlen(keep_msg));
 750         close(keep_fd);
 751
 752         snprintf(name, sizeof(name), "%s/pack/pack-%s.pack",
 753                  get_object_directory(), sha1_to_hex(pack_data->sha1));
 754         if (move_temp_to_file(pack_data->pack_name, name))
 755                 die("cannot store pack file");
 756
 757         snprintf(name, sizeof(name), "%s/pack/pack-%s.idx",
 758                  get_object_directory(), sha1_to_hex(pack_data->sha1));
 759         if (move_temp_to_file(curr_index_name, name))
 760                 die("cannot store index file");
 761         return name;
 762 }
 763
 764 static void unkeep_all_packs(void)
 765 {
 766         static char name[PATH_MAX];
 767         int k;
 768
 769         for (k = 0; k < pack_id; k++) {
 770                 struct packed_git *p = all_packs[k];
 771                 snprintf(name, sizeof(name), "%s/pack/pack-%s.keep",
 772                          get_object_directory(), sha1_to_hex(p->sha1));
 773                 unlink(name);
 774         }
 775 }
 776
 777 static void end_packfile(void)
 778 {
 779         struct packed_git *old_p = pack_data, *new_p;
 780
 781         if (object_count) {
 782                 char *idx_name;
 783                 int i;
 784                 struct branch *b;
 785                 struct tag *t;
 786
 787                 fixup_header_footer();
 788                 idx_name = keep_pack(create_index());
 789
 790                 /* Register the packfile with core git's machinary. */
 791                 new_p = add_packed_git(idx_name, strlen(idx_name), 1);
 792                 if (!new_p)
 793                         die("core git rejected index %s", idx_name);
 794                 new_p->windows = old_p->windows;
 795                 all_packs[pack_id] = new_p;
 796                 install_packed_git(new_p);
 797
 798                 /* Print the boundary */
 799                 fprintf(stdout, "%s:", new_p->pack_name);
 800                 for (i = 0; i < branch_table_sz; i++) {
 801                         for (b = branch_table[i]; b; b = b->table_next_branch) {
 802                                 if (b->pack_id == pack_id)
 803                                         fprintf(stdout, " %s", sha1_to_hex(b->sha1));
 804                         }
 805                 }
 806                 for (t = first_tag; t; t = t->next_tag) {
 807                         if (t->pack_id == pack_id)
 808                                 fprintf(stdout, " %s", sha1_to_hex(t->sha1));
 809                 }
 810                 fputc('\n', stdout);
 811
 812                 pack_id++;
 813         }
 814         else
 815                 unlink(old_p->pack_name);
 816         free(old_p);
 817
 818         /* We can't carry a delta across packfiles. */
 819         free(last_blob.data);
 820         last_blob.data = NULL;
 821         last_blob.len = 0;
 822         last_blob.offset = 0;
 823         last_blob.depth = 0;
 824 }
 825
 826 static void checkpoint(void)
 827 {
 828         end_packfile();
 829         start_packfile();
 830 }
 831
 832 static size_t encode_header(
 833         enum object_type type,
 834         size_t size,
 835         unsigned char *hdr)
 836 {
 837         int n = 1;
 838         unsigned char c;
 839
 840         if (type < OBJ_COMMIT || type > OBJ_REF_DELTA)
 841                 die("bad type %d", type);
 842
 843         c = (type << 4) | (size & 15);
 844         size >>= 4;
 845         while (size) {
 846                 *hdr++ = c | 0x80;
 847                 c = size & 0x7f;
 848                 size >>= 7;
 849                 n++;
 850         }
 851         *hdr = c;
 852         return n;
 853 }
 854
 855 static int store_object(
 856         enum object_type type,
 857         void *dat,
 858         size_t datlen,
 859         struct last_object *last,
 860         unsigned char *sha1out,
 861         uintmax_t mark)
 862 {
 863         void *out, *delta;
 864         struct object_entry *e;
 865         unsigned char hdr[96];
 866         unsigned char sha1[20];
 867         unsigned long hdrlen, deltalen;
 868         SHA_CTX c;
 869         z_stream s;
 870
 871         hdrlen = sprintf((char*)hdr,"%s %lu",type_names[type],datlen) + 1;
 872         SHA1_Init(&c);
 873         SHA1_Update(&c, hdr, hdrlen);
 874         SHA1_Update(&c, dat, datlen);
 875         SHA1_Final(sha1, &c);
 876         if (sha1out)
 877                 hashcpy(sha1out, sha1);
 878
 879         e = insert_object(sha1);
 880         if (mark)
 881                 insert_mark(mark, e);
 882         if (e->offset) {
 883                 duplicate_count_by_type[type]++;
 884                 return 1;
 885         }
 886
 887         if (last && last->data && last->depth < max_depth) {
 888                 delta = diff_delta(last->data, last->len,
 889                         dat, datlen,
 890                         &deltalen, 0);
 891                 if (delta && deltalen >= datlen) {
 892                         free(delta);
 893                         delta = NULL;
 894                 }
 895         } else
 896                 delta = NULL;
 897
 898         memset(&s, 0, sizeof(s));
 899         deflateInit(&s, zlib_compression_level);
 900         if (delta) {
 901                 s.next_in = delta;
 902                 s.avail_in = deltalen;
 903         } else {
 904                 s.next_in = dat;
 905                 s.avail_in = datlen;
 906         }
 907         s.avail_out = deflateBound(&s, s.avail_in);
 908         s.next_out = out = xmalloc(s.avail_out);
 909         while (deflate(&s, Z_FINISH) == Z_OK)
 910                 /* nothing */;
 911         deflateEnd(&s);
 912
 913         /* Determine if we should auto-checkpoint. */
 914         if ((pack_size + 60 + s.total_out) > max_packsize
 915                 || (pack_size + 60 + s.total_out) < pack_size) {
 916
 917                 /* This new object needs to *not* have the current pack_id. */
 918                 e->pack_id = pack_id + 1;
 919                 checkpoint();
 920
 921                 /* We cannot carry a delta into the new pack. */
 922                 if (delta) {
 923                         free(delta);
 924                         delta = NULL;
 925
 926                         memset(&s, 0, sizeof(s));
 927                         deflateInit(&s, zlib_compression_level);
 928                         s.next_in = dat;
 929                         s.avail_in = datlen;
 930                         s.avail_out = deflateBound(&s, s.avail_in);
 931                         s.next_out = out = xrealloc(out, s.avail_out);
 932                         while (deflate(&s, Z_FINISH) == Z_OK)
 933                                 /* nothing */;
 934                         deflateEnd(&s);
 935                 }
 936         }
 937
 938         e->type = type;
 939         e->pack_id = pack_id;
 940         e->offset = pack_size;
 941         object_count++;
 942         object_count_by_type[type]++;
 943
 944         if (delta) {
 945                 unsigned long ofs = e->offset - last->offset;
 946                 unsigned pos = sizeof(hdr) - 1;
 947
 948                 delta_count_by_type[type]++;
 949                 last->depth++;
 950
 951                 hdrlen = encode_header(OBJ_OFS_DELTA, deltalen, hdr);
 952                 write_or_die(pack_data->pack_fd, hdr, hdrlen);
 953                 pack_size += hdrlen;
 954
 955                 hdr[pos] = ofs & 127;
 956                 while (ofs >>= 7)
 957                         hdr[--pos] = 128 | (--ofs & 127);
 958                 write_or_die(pack_data->pack_fd, hdr + pos, sizeof(hdr) - pos);
 959                 pack_size += sizeof(hdr) - pos;
 960         } else {
 961                 if (last)
 962                         last->depth = 0;
 963                 hdrlen = encode_header(type, datlen, hdr);
 964                 write_or_die(pack_data->pack_fd, hdr, hdrlen);
 965                 pack_size += hdrlen;
 966         }
 967
 968         write_or_die(pack_data->pack_fd, out, s.total_out);
 969         pack_size += s.total_out;
 970
 971         free(out);
 972         if (delta)
 973                 free(delta);
 974         if (last) {
 975                 if (last->data && !last->no_free)
 976                         free(last->data);
 977                 last->data = dat;
 978                 last->offset = e->offset;
 979                 last->len = datlen;
 980         }
 981         return 0;
 982 }
 983
 984 static void *gfi_unpack_entry(
 985         struct object_entry *oe,
 986         unsigned long *sizep)
 987 {
 988         static char type[20];
 989         struct packed_git *p = all_packs[oe->pack_id];
 990         if (p == pack_data)
 991                 p->pack_size = pack_size + 20;
 992         return unpack_entry(p, oe->offset, type, sizep);
 993 }
 994
 995 static const char *get_mode(const char *str, uint16_t *modep)
 996 {
 997         unsigned char c;
 998         uint16_t mode = 0;
 999
1000         while ((c = *str++) != ' ') {
1001                 if (c < '0' || c > '7')
1002                         return NULL;
1003                 mode = (mode << 3) + (c - '0');
1004         }
1005         *modep = mode;
1006         return str;
1007 }
1008
1009 static void load_tree(struct tree_entry *root)
1010 {
1011         unsigned char* sha1 = root->versions[1].sha1;
1012         struct object_entry *myoe;
1013         struct tree_content *t;
1014         unsigned long size;
1015         char *buf;
1016         const char *c;
1017
1018         root->tree = t = new_tree_content(8);
1019         if (is_null_sha1(sha1))
1020                 return;
1021
1022         myoe = find_object(sha1);
1023         if (myoe) {
1024                 if (myoe->type != OBJ_TREE)
1025                         die("Not a tree: %s", sha1_to_hex(sha1));
1026                 t->delta_depth = 0;
1027                 buf = gfi_unpack_entry(myoe, &size);
1028         } else {
1029                 char type[20];
1030                 buf = read_sha1_file(sha1, type, &size);
1031                 if (!buf || strcmp(type, tree_type))
1032                         die("Can't load tree %s", sha1_to_hex(sha1));
1033         }
1034
1035         c = buf;
1036         while (c != (buf + size)) {
1037                 struct tree_entry *e = new_tree_entry();
1038
1039                 if (t->entry_count == t->entry_capacity)
1040                         root->tree = t = grow_tree_content(t, 8);
1041                 t->entries[t->entry_count++] = e;
1042
1043                 e->tree = NULL;
1044                 c = get_mode(c, &e->versions[1].mode);
1045                 if (!c)
1046                         die("Corrupt mode in %s", sha1_to_hex(sha1));
1047                 e->versions[0].mode = e->versions[1].mode;
1048                 e->name = to_atom(c, (unsigned short)strlen(c));
1049                 c += e->name->str_len + 1;
1050                 hashcpy(e->versions[0].sha1, (unsigned char*)c);
1051                 hashcpy(e->versions[1].sha1, (unsigned char*)c);
1052                 c += 20;
1053         }
1054         free(buf);
1055 }
1056
1057 static int tecmp0 (const void *_a, const void *_b)
1058 {
1059         struct tree_entry *a = *((struct tree_entry**)_a);
1060         struct tree_entry *b = *((struct tree_entry**)_b);
1061         return base_name_compare(
1062                 a->name->str_dat, a->name->str_len, a->versions[0].mode,
1063                 b->name->str_dat, b->name->str_len, b->versions[0].mode);
1064 }
1065
1066 static int tecmp1 (const void *_a, const void *_b)
1067 {
1068         struct tree_entry *a = *((struct tree_entry**)_a);
1069         struct tree_entry *b = *((struct tree_entry**)_b);
1070         return base_name_compare(
1071                 a->name->str_dat, a->name->str_len, a->versions[1].mode,
1072                 b->name->str_dat, b->name->str_len, b->versions[1].mode);
1073 }
1074
1075 static void mktree(struct tree_content *t,
1076         int v,
1077         unsigned long *szp,
1078         struct dbuf *b)
1079 {
1080         size_t maxlen = 0;
1081         unsigned int i;
1082         char *c;
1083
1084         if (!v)
1085                 qsort(t->entries,t->entry_count,sizeof(t->entries[0]),tecmp0);
1086         else
1087                 qsort(t->entries,t->entry_count,sizeof(t->entries[0]),tecmp1);
1088
1089         for (i = 0; i < t->entry_count; i++) {
1090                 if (t->entries[i]->versions[v].mode)
1091                         maxlen += t->entries[i]->name->str_len + 34;
1092         }
1093
1094         size_dbuf(b, maxlen);
1095         c = b->buffer;
1096         for (i = 0; i < t->entry_count; i++) {
1097                 struct tree_entry *e = t->entries[i];
1098                 if (!e->versions[v].mode)
1099                         continue;
1100                 c += sprintf(c, "%o", (unsigned int)e->versions[v].mode);
1101                 *c++ = ' ';
1102                 strcpy(c, e->name->str_dat);
1103                 c += e->name->str_len + 1;
1104                 hashcpy((unsigned char*)c, e->versions[v].sha1);
1105                 c += 20;
1106         }
1107         *szp = c - (char*)b->buffer;
1108 }
1109
1110 static void store_tree(struct tree_entry *root)
1111 {
1112         struct tree_content *t = root->tree;
1113         unsigned int i, j, del;
1114         unsigned long new_len;
1115         struct last_object lo;
1116         struct object_entry *le;
1117
1118         if (!is_null_sha1(root->versions[1].sha1))
1119                 return;
1120
1121         for (i = 0; i < t->entry_count; i++) {
1122                 if (t->entries[i]->tree)
1123                         store_tree(t->entries[i]);
1124         }
1125
1126         le = find_object(root->versions[0].sha1);
1127         if (!S_ISDIR(root->versions[0].mode)
1128                 || !le
1129                 || le->pack_id != pack_id) {
1130                 lo.data = NULL;
1131                 lo.depth = 0;
1132         } else {
1133                 mktree(t, 0, &lo.len, &old_tree);
1134                 lo.data = old_tree.buffer;
1135                 lo.offset = le->offset;
1136                 lo.depth = t->delta_depth;
1137                 lo.no_free = 1;
1138         }
1139
1140         mktree(t, 1, &new_len, &new_tree);
1141         store_object(OBJ_TREE, new_tree.buffer, new_len,
1142                 &lo, root->versions[1].sha1, 0);
1143
1144         t->delta_depth = lo.depth;
1145         for (i = 0, j = 0, del = 0; i < t->entry_count; i++) {
1146                 struct tree_entry *e = t->entries[i];
1147                 if (e->versions[1].mode) {
1148                         e->versions[0].mode = e->versions[1].mode;
1149                         hashcpy(e->versions[0].sha1, e->versions[1].sha1);
1150                         t->entries[j++] = e;
1151                 } else {
1152                         release_tree_entry(e);
1153                         del++;
1154                 }
1155         }
1156         t->entry_count -= del;
1157 }
1158
1159 static int tree_content_set(
1160         struct tree_entry *root,
1161         const char *p,
1162         const unsigned char *sha1,
1163         const uint16_t mode)
1164 {
1165         struct tree_content *t = root->tree;
1166         const char *slash1;
1167         unsigned int i, n;
1168         struct tree_entry *e;
1169
1170         slash1 = strchr(p, '/');
1171         if (slash1)
1172                 n = slash1 - p;
1173         else
1174                 n = strlen(p);
1175
1176         for (i = 0; i < t->entry_count; i++) {
1177                 e = t->entries[i];
1178                 if (e->name->str_len == n && !strncmp(p, e->name->str_dat, n)) {
1179                         if (!slash1) {
1180                                 if (e->versions[1].mode == mode
1181                                                 && !hashcmp(e->versions[1].sha1, sha1))
1182                                         return 0;
1183                                 e->versions[1].mode = mode;
1184                                 hashcpy(e->versions[1].sha1, sha1);
1185                                 if (e->tree) {
1186                                         release_tree_content_recursive(e->tree);
1187                                         e->tree = NULL;
1188                                 }
1189                                 hashclr(root->versions[1].sha1);
1190                                 return 1;
1191                         }
1192                         if (!S_ISDIR(e->versions[1].mode)) {
1193                                 e->tree = new_tree_content(8);
1194                                 e->versions[1].mode = S_IFDIR;
1195                         }
1196                         if (!e->tree)
1197                                 load_tree(e);
1198                         if (tree_content_set(e, slash1 + 1, sha1, mode)) {
1199                                 hashclr(root->versions[1].sha1);
1200                                 return 1;
1201                         }
1202                         return 0;
1203                 }
1204         }
1205
1206         if (t->entry_count == t->entry_capacity)
1207                 root->tree = t = grow_tree_content(t, 8);
1208         e = new_tree_entry();
1209         e->name = to_atom(p, (unsigned short)n);
1210         e->versions[0].mode = 0;
1211         hashclr(e->versions[0].sha1);
1212         t->entries[t->entry_count++] = e;
1213         if (slash1) {
1214                 e->tree = new_tree_content(8);
1215                 e->versions[1].mode = S_IFDIR;
1216                 tree_content_set(e, slash1 + 1, sha1, mode);
1217         } else {
1218                 e->tree = NULL;
1219                 e->versions[1].mode = mode;
1220                 hashcpy(e->versions[1].sha1, sha1);
1221         }
1222         hashclr(root->versions[1].sha1);
1223         return 1;
1224 }
1225
1226 static int tree_content_remove(struct tree_entry *root, const char *p)
1227 {
1228         struct tree_content *t = root->tree;
1229         const char *slash1;
1230         unsigned int i, n;
1231         struct tree_entry *e;
1232
1233         slash1 = strchr(p, '/');
1234         if (slash1)
1235                 n = slash1 - p;
1236         else
1237                 n = strlen(p);
1238
1239         for (i = 0; i < t->entry_count; i++) {
1240                 e = t->entries[i];
1241                 if (e->name->str_len == n && !strncmp(p, e->name->str_dat, n)) {
1242                         if (!slash1 || !S_ISDIR(e->versions[1].mode))
1243                                 goto del_entry;
1244                         if (!e->tree)
1245                                 load_tree(e);
1246                         if (tree_content_remove(e, slash1 + 1)) {
1247                                 for (n = 0; n < e->tree->entry_count; n++) {
1248                                         if (e->tree->entries[n]->versions[1].mode) {
1249                                                 hashclr(root->versions[1].sha1);
1250                                                 return 1;
1251                                         }
1252                                 }
1253                                 goto del_entry;
1254                         }
1255                         return 0;
1256                 }
1257         }
1258         return 0;
1259
1260 del_entry:
1261         if (e->tree) {
1262                 release_tree_content_recursive(e->tree);
1263                 e->tree = NULL;
1264         }
1265         e->versions[1].mode = 0;
1266         hashclr(e->versions[1].sha1);
1267         hashclr(root->versions[1].sha1);
1268         return 1;
1269 }
1270
1271 static void dump_branches(void)
1272 {
1273         static const char *msg = "fast-import";
1274         unsigned int i;
1275         struct branch *b;
1276         struct ref_lock *lock;
1277
1278         for (i = 0; i < branch_table_sz; i++) {
1279                 for (b = branch_table[i]; b; b = b->table_next_branch) {
1280                         lock = lock_any_ref_for_update(b->name, NULL);
1281                         if (!lock || write_ref_sha1(lock, b->sha1, msg) < 0)
1282                                 die("Can't write %s", b->name);
1283                 }
1284         }
1285 }
1286
1287 static void dump_tags(void)
1288 {
1289         static const char *msg = "fast-import";
1290         struct tag *t;
1291         struct ref_lock *lock;
1292         char path[PATH_MAX];
1293
1294         for (t = first_tag; t; t = t->next_tag) {
1295                 sprintf(path, "refs/tags/%s", t->name);
1296                 lock = lock_any_ref_for_update(path, NULL);
1297                 if (!lock || write_ref_sha1(lock, t->sha1, msg) < 0)
1298                         die("Can't write %s", path);
1299         }
1300 }
1301
1302 static void dump_marks_helper(FILE *f,
1303         uintmax_t base,
1304         struct mark_set *m)
1305 {
1306         uintmax_t k;
1307         if (m->shift) {
1308                 for (k = 0; k < 1024; k++) {
1309                         if (m->data.sets[k])
1310                                 dump_marks_helper(f, (base + k) << m->shift,
1311                                         m->data.sets[k]);
1312                 }
1313         } else {
1314                 for (k = 0; k < 1024; k++) {
1315                         if (m->data.marked[k])
1316                                 fprintf(f, ":%ju %s\n", base + k,
1317                                         sha1_to_hex(m->data.marked[k]->sha1));
1318                 }
1319         }
1320 }
1321
1322 static void dump_marks(void)
1323 {
1324         if (mark_file)
1325         {
1326                 FILE *f = fopen(mark_file, "w");
1327                 dump_marks_helper(f, 0, marks);
1328                 fclose(f);
1329         }
1330 }
1331
1332 static void read_next_command(void)
1333 {
1334         read_line(&command_buf, stdin, '\n');
1335 }
1336
1337 static void cmd_mark(void)
1338 {
1339         if (!strncmp("mark :", command_buf.buf, 6)) {
1340                 next_mark = strtoumax(command_buf.buf + 6, NULL, 10);
1341                 read_next_command();
1342         }
1343         else
1344                 next_mark = 0;
1345 }
1346
1347 static void* cmd_data (size_t *size)
1348 {
1349         size_t length;
1350         char *buffer;
1351
1352         if (strncmp("data ", command_buf.buf, 5))
1353                 die("Expected 'data n' command, found: %s", command_buf.buf);
1354
1355         if (!strncmp("<<", command_buf.buf + 5, 2)) {
1356                 char *term = xstrdup(command_buf.buf + 5 + 2);
1357                 size_t sz = 8192, term_len = command_buf.len - 5 - 2;
1358                 length = 0;
1359                 buffer = xmalloc(sz);
1360                 for (;;) {
1361                         read_next_command();
1362                         if (command_buf.eof)
1363                                 die("EOF in data (terminator '%s' not found)", term);
1364                         if (term_len == command_buf.len
1365                                 && !strcmp(term, command_buf.buf))
1366                                 break;
1367                         if (sz < (length + command_buf.len)) {
1368                                 sz = sz * 3 / 2 + 16;
1369                                 if (sz < (length + command_buf.len))
1370                                         sz = length + command_buf.len;
1371                                 buffer = xrealloc(buffer, sz);
1372                         }
1373                         memcpy(buffer + length,
1374                                 command_buf.buf,
1375                                 command_buf.len - 1);
1376                         length += command_buf.len - 1;
1377                         buffer[length++] = '\n';
1378                 }
1379                 free(term);
1380         }
1381         else {
1382                 size_t n = 0;
1383                 length = strtoul(command_buf.buf + 5, NULL, 10);
1384                 buffer = xmalloc(length);
1385                 while (n < length) {
1386                         size_t s = fread(buffer + n, 1, length - n, stdin);
1387                         if (!s && feof(stdin))
1388                                 die("EOF in data (%lu bytes remaining)", length - n);
1389                         n += s;
1390                 }
1391         }
1392
1393         if (fgetc(stdin) != '\n')
1394                 die("An lf did not trail the binary data as expected.");
1395
1396         *size = length;
1397         return buffer;
1398 }
1399
1400 static void cmd_new_blob(void)
1401 {
1402         size_t l;
1403         void *d;
1404
1405         read_next_command();
1406         cmd_mark();
1407         d = cmd_data(&l);
1408
1409         if (store_object(OBJ_BLOB, d, l, &last_blob, NULL, next_mark))
1410                 free(d);
1411 }
1412
1413 static void unload_one_branch(void)
1414 {
1415         while (cur_active_branches
1416                 && cur_active_branches >= max_active_branches) {
1417                 unsigned long min_commit = ULONG_MAX;
1418                 struct branch *e, *l = NULL, *p = NULL;
1419
1420                 for (e = active_branches; e; e = e->active_next_branch) {
1421                         if (e->last_commit < min_commit) {
1422                                 p = l;
1423                                 min_commit = e->last_commit;
1424                         }
1425                         l = e;
1426                 }
1427
1428                 if (p) {
1429                         e = p->active_next_branch;
1430                         p->active_next_branch = e->active_next_branch;
1431                 } else {
1432                         e = active_branches;
1433                         active_branches = e->active_next_branch;
1434                 }
1435                 e->active_next_branch = NULL;
1436                 if (e->branch_tree.tree) {
1437                         release_tree_content_recursive(e->branch_tree.tree);
1438                         e->branch_tree.tree = NULL;
1439                 }
1440                 cur_active_branches--;
1441         }
1442 }
1443
1444 static void load_branch(struct branch *b)
1445 {
1446         load_tree(&b->branch_tree);
1447         b->active_next_branch = active_branches;
1448         active_branches = b;
1449         cur_active_branches++;
1450         branch_load_count++;
1451 }
1452
1453 static void file_change_m(struct branch *b)
1454 {
1455         const char *p = command_buf.buf + 2;
1456         char *p_uq;
1457         const char *endp;
1458         struct object_entry *oe = oe;
1459         unsigned char sha1[20];
1460         uint16_t mode, inline_data = 0;
1461         char type[20];
1462
1463         p = get_mode(p, &mode);
1464         if (!p)
1465                 die("Corrupt mode: %s", command_buf.buf);
1466         switch (mode) {
1467         case S_IFREG | 0644:
1468         case S_IFREG | 0755:
1469         case S_IFLNK:
1470         case 0644:
1471         case 0755:
1472                 /* ok */
1473                 break;
1474         default:
1475                 die("Corrupt mode: %s", command_buf.buf);
1476         }
1477
1478         if (*p == ':') {
1479                 char *x;
1480                 oe = find_mark(strtoumax(p + 1, &x, 10));
1481                 hashcpy(sha1, oe->sha1);
1482                 p = x;
1483         } else if (!strncmp("inline", p, 6)) {
1484                 inline_data = 1;
1485                 p += 6;
1486         } else {
1487                 if (get_sha1_hex(p, sha1))
1488                         die("Invalid SHA1: %s", command_buf.buf);
1489                 oe = find_object(sha1);
1490                 p += 40;
1491         }
1492         if (*p++ != ' ')
1493                 die("Missing space after SHA1: %s", command_buf.buf);
1494
1495         p_uq = unquote_c_style(p, &endp);
1496         if (p_uq) {
1497                 if (*endp)
1498                         die("Garbage after path in: %s", command_buf.buf);
1499                 p = p_uq;
1500         }
1501
1502         if (inline_data) {
1503                 size_t l;
1504                 void *d;
1505                 if (!p_uq)
1506                         p = p_uq = xstrdup(p);
1507                 read_next_command();
1508                 d = cmd_data(&l);
1509                 if (store_object(OBJ_BLOB, d, l, &last_blob, sha1, 0))
1510                         free(d);
1511         } else if (oe) {
1512                 if (oe->type != OBJ_BLOB)
1513                         die("Not a blob (actually a %s): %s",
1514                                 command_buf.buf, type_names[oe->type]);
1515         } else {
1516                 if (sha1_object_info(sha1, type, NULL))
1517                         die("Blob not found: %s", command_buf.buf);
1518                 if (strcmp(blob_type, type))
1519                         die("Not a blob (actually a %s): %s",
1520                                 command_buf.buf, type);
1521         }
1522
1523         tree_content_set(&b->branch_tree, p, sha1, S_IFREG | mode);
1524
1525         if (p_uq)
1526                 free(p_uq);
1527 }
1528
1529 static void file_change_d(struct branch *b)
1530 {
1531         const char *p = command_buf.buf + 2;
1532         char *p_uq;
1533         const char *endp;
1534
1535         p_uq = unquote_c_style(p, &endp);
1536         if (p_uq) {
1537                 if (*endp)
1538                         die("Garbage after path in: %s", command_buf.buf);
1539                 p = p_uq;
1540         }
1541         tree_content_remove(&b->branch_tree, p);
1542         if (p_uq)
1543                 free(p_uq);
1544 }
1545
1546 static void cmd_from(struct branch *b)
1547 {
1548         const char *from;
1549         struct branch *s;
1550
1551         if (strncmp("from ", command_buf.buf, 5))
1552                 return;
1553
1554         if (b->last_commit)
1555                 die("Can't reinitailize branch %s", b->name);
1556
1557         from = strchr(command_buf.buf, ' ') + 1;
1558         s = lookup_branch(from);
1559         if (b == s)
1560                 die("Can't create a branch from itself: %s", b->name);
1561         else if (s) {
1562                 unsigned char *t = s->branch_tree.versions[1].sha1;
1563                 hashcpy(b->sha1, s->sha1);
1564                 hashcpy(b->branch_tree.versions[0].sha1, t);
1565                 hashcpy(b->branch_tree.versions[1].sha1, t);
1566         } else if (*from == ':') {
1567                 uintmax_t idnum = strtoumax(from + 1, NULL, 10);
1568                 struct object_entry *oe = find_mark(idnum);
1569                 unsigned long size;
1570                 char *buf;
1571                 if (oe->type != OBJ_COMMIT)
1572                         die("Mark :%ju not a commit", idnum);
1573                 hashcpy(b->sha1, oe->sha1);
1574                 buf = gfi_unpack_entry(oe, &size);
1575                 if (!buf || size < 46)
1576                         die("Not a valid commit: %s", from);
1577                 if (memcmp("tree ", buf, 5)
1578                         || get_sha1_hex(buf + 5, b->branch_tree.versions[1].sha1))
1579                         die("The commit %s is corrupt", sha1_to_hex(b->sha1));
1580                 free(buf);
1581                 hashcpy(b->branch_tree.versions[0].sha1,
1582                         b->branch_tree.versions[1].sha1);
1583         } else if (!get_sha1(from, b->sha1)) {
1584                 if (is_null_sha1(b->sha1)) {
1585                         hashclr(b->branch_tree.versions[0].sha1);
1586                         hashclr(b->branch_tree.versions[1].sha1);
1587                 } else {
1588                         unsigned long size;
1589                         char *buf;
1590
1591                         buf = read_object_with_reference(b->sha1,
1592                                 type_names[OBJ_COMMIT], &size, b->sha1);
1593                         if (!buf || size < 46)
1594                                 die("Not a valid commit: %s", from);
1595                         if (memcmp("tree ", buf, 5)
1596                                 || get_sha1_hex(buf + 5, b->branch_tree.versions[1].sha1))
1597                                 die("The commit %s is corrupt", sha1_to_hex(b->sha1));
1598                         free(buf);
1599                         hashcpy(b->branch_tree.versions[0].sha1,
1600                                 b->branch_tree.versions[1].sha1);
1601                 }
1602         } else
1603                 die("Invalid ref name or SHA1 expression: %s", from);
1604
1605         read_next_command();
1606 }
1607
1608 static struct hash_list* cmd_merge(unsigned int *count)
1609 {
1610         struct hash_list *list = NULL, *n, *e = e;
1611         const char *from;
1612         struct branch *s;
1613
1614         *count = 0;
1615         while (!strncmp("merge ", command_buf.buf, 6)) {
1616                 from = strchr(command_buf.buf, ' ') + 1;
1617                 n = xmalloc(sizeof(*n));
1618                 s = lookup_branch(from);
1619                 if (s)
1620                         hashcpy(n->sha1, s->sha1);
1621                 else if (*from == ':') {
1622                         uintmax_t idnum = strtoumax(from + 1, NULL, 10);
1623                         struct object_entry *oe = find_mark(idnum);
1624                         if (oe->type != OBJ_COMMIT)
1625                                 die("Mark :%ju not a commit", idnum);
1626                         hashcpy(n->sha1, oe->sha1);
1627                 } else if (get_sha1(from, n->sha1))
1628                         die("Invalid ref name or SHA1 expression: %s", from);
1629
1630                 n->next = NULL;
1631                 if (list)
1632                         e->next = n;
1633                 else
1634                         list = n;
1635                 e = n;
1636                 (*count)++;
1637                 read_next_command();
1638         }
1639         return list;
1640 }
1641
1642 static void cmd_new_commit(void)
1643 {
1644         struct branch *b;
1645         void *msg;
1646         size_t msglen;
1647         char *sp;
1648         char *author = NULL;
1649         char *committer = NULL;
1650         struct hash_list *merge_list = NULL;
1651         unsigned int merge_count;
1652
1653         /* Obtain the branch name from the rest of our command */
1654         sp = strchr(command_buf.buf, ' ') + 1;
1655         b = lookup_branch(sp);
1656         if (!b)
1657                 b = new_branch(sp);
1658
1659         read_next_command();
1660         cmd_mark();
1661         if (!strncmp("author ", command_buf.buf, 7)) {
1662                 author = strdup(command_buf.buf);
1663                 read_next_command();
1664         }
1665         if (!strncmp("committer ", command_buf.buf, 10)) {
1666                 committer = strdup(command_buf.buf);
1667                 read_next_command();
1668         }
1669         if (!committer)
1670                 die("Expected committer but didn't get one");
1671         msg = cmd_data(&msglen);
1672         read_next_command();
1673         cmd_from(b);
1674         merge_list = cmd_merge(&merge_count);
1675
1676         /* ensure the branch is active/loaded */
1677         if (!b->branch_tree.tree || !max_active_branches) {
1678                 unload_one_branch();
1679                 load_branch(b);
1680         }
1681
1682         /* file_change* */
1683         for (;;) {
1684                 if (1 == command_buf.len)
1685                         break;
1686                 else if (!strncmp("M ", command_buf.buf, 2))
1687                         file_change_m(b);
1688                 else if (!strncmp("D ", command_buf.buf, 2))
1689                         file_change_d(b);
1690                 else
1691                         die("Unsupported file_change: %s", command_buf.buf);
1692                 read_next_command();
1693         }
1694
1695         /* build the tree and the commit */
1696         store_tree(&b->branch_tree);
1697         hashcpy(b->branch_tree.versions[0].sha1,
1698                 b->branch_tree.versions[1].sha1);
1699         size_dbuf(&new_data, 97 + msglen
1700                 + merge_count * 49
1701                 + (author
1702                         ? strlen(author) + strlen(committer)
1703                         : 2 * strlen(committer)));
1704         sp = new_data.buffer;
1705         sp += sprintf(sp, "tree %s\n",
1706                 sha1_to_hex(b->branch_tree.versions[1].sha1));
1707         if (!is_null_sha1(b->sha1))
1708                 sp += sprintf(sp, "parent %s\n", sha1_to_hex(b->sha1));
1709         while (merge_list) {
1710                 struct hash_list *next = merge_list->next;
1711                 sp += sprintf(sp, "parent %s\n", sha1_to_hex(merge_list->sha1));
1712                 free(merge_list);
1713                 merge_list = next;
1714         }
1715         if (author)
1716                 sp += sprintf(sp, "%s\n", author);
1717         else
1718                 sp += sprintf(sp, "author %s\n", committer + 10);
1719         sp += sprintf(sp, "%s\n\n", committer);
1720         memcpy(sp, msg, msglen);
1721         sp += msglen;
1722         if (author)
1723                 free(author);
1724         free(committer);
1725         free(msg);
1726
1727         if (!store_object(OBJ_COMMIT,
1728                 new_data.buffer, sp - (char*)new_data.buffer,
1729                 NULL, b->sha1, next_mark))
1730                 b->pack_id = pack_id;
1731         b->last_commit = object_count_by_type[OBJ_COMMIT];
1732 }
1733
1734 static void cmd_new_tag(void)
1735 {
1736         char *sp;
1737         const char *from;
1738         char *tagger;
1739         struct branch *s;
1740         void *msg;
1741         size_t msglen;
1742         struct tag *t;
1743         uintmax_t from_mark = 0;
1744         unsigned char sha1[20];
1745
1746         /* Obtain the new tag name from the rest of our command */
1747         sp = strchr(command_buf.buf, ' ') + 1;
1748         t = pool_alloc(sizeof(struct tag));
1749         t->next_tag = NULL;
1750         t->name = pool_strdup(sp);
1751         if (last_tag)
1752                 last_tag->next_tag = t;
1753         else
1754                 first_tag = t;
1755         last_tag = t;
1756         read_next_command();
1757
1758         /* from ... */
1759         if (strncmp("from ", command_buf.buf, 5))
1760                 die("Expected from command, got %s", command_buf.buf);
1761         from = strchr(command_buf.buf, ' ') + 1;
1762         s = lookup_branch(from);
1763         if (s) {
1764                 hashcpy(sha1, s->sha1);
1765         } else if (*from == ':') {
1766                 struct object_entry *oe;
1767                 from_mark = strtoumax(from + 1, NULL, 10);
1768                 oe = find_mark(from_mark);
1769                 if (oe->type != OBJ_COMMIT)
1770                         die("Mark :%ju not a commit", from_mark);
1771                 hashcpy(sha1, oe->sha1);
1772         } else if (!get_sha1(from, sha1)) {
1773                 unsigned long size;
1774                 char *buf;
1775
1776                 buf = read_object_with_reference(sha1,
1777                         type_names[OBJ_COMMIT], &size, sha1);
1778                 if (!buf || size < 46)
1779                         die("Not a valid commit: %s", from);
1780                 free(buf);
1781         } else
1782                 die("Invalid ref name or SHA1 expression: %s", from);
1783         read_next_command();
1784
1785         /* tagger ... */
1786         if (strncmp("tagger ", command_buf.buf, 7))
1787                 die("Expected tagger command, got %s", command_buf.buf);
1788         tagger = strdup(command_buf.buf);
1789
1790         /* tag payload/message */
1791         read_next_command();
1792         msg = cmd_data(&msglen);
1793
1794         /* build the tag object */
1795         size_dbuf(&new_data, 67+strlen(t->name)+strlen(tagger)+msglen);
1796         sp = new_data.buffer;
1797         sp += sprintf(sp, "object %s\n", sha1_to_hex(sha1));
1798         sp += sprintf(sp, "type %s\n", type_names[OBJ_COMMIT]);
1799         sp += sprintf(sp, "tag %s\n", t->name);
1800         sp += sprintf(sp, "%s\n\n", tagger);
1801         memcpy(sp, msg, msglen);
1802         sp += msglen;
1803         free(tagger);
1804         free(msg);
1805
1806         if (store_object(OBJ_TAG, new_data.buffer,
1807                 sp - (char*)new_data.buffer,
1808                 NULL, t->sha1, 0))
1809                 t->pack_id = MAX_PACK_ID;
1810         else
1811                 t->pack_id = pack_id;
1812 }
1813
1814 static void cmd_reset_branch(void)
1815 {
1816         struct branch *b;
1817         char *sp;
1818
1819         /* Obtain the branch name from the rest of our command */
1820         sp = strchr(command_buf.buf, ' ') + 1;
1821         b = lookup_branch(sp);
1822         if (b) {
1823                 b->last_commit = 0;
1824                 if (b->branch_tree.tree) {
1825                         release_tree_content_recursive(b->branch_tree.tree);
1826                         b->branch_tree.tree = NULL;
1827                 }
1828         }
1829         else
1830                 b = new_branch(sp);
1831         read_next_command();
1832         cmd_from(b);
1833 }
1834
1835 static void cmd_checkpoint(void)
1836 {
1837         if (object_count)
1838                 checkpoint();
1839         read_next_command();
1840 }
1841
1842 static const char fast_import_usage[] =
1843 "git-fast-import [--depth=n] [--active-branches=n] [--export-marks=marks.file] [--branch-log=log]";
1844
1845 int main(int argc, const char **argv)
1846 {
1847         int i;
1848         uintmax_t total_count, duplicate_count;
1849
1850         git_config(git_default_config);
1851
1852         for (i = 1; i < argc; i++) {
1853                 const char *a = argv[i];
1854
1855                 if (*a != '-' || !strcmp(a, "--"))
1856                         break;
1857                 else if (!strncmp(a, "--max-pack-size=", 16))
1858                         max_packsize = strtoumax(a + 16, NULL, 0) * 1024 * 1024;
1859                 else if (!strncmp(a, "--depth=", 8))
1860                         max_depth = strtoul(a + 8, NULL, 0);
1861                 else if (!strncmp(a, "--active-branches=", 18))
1862                         max_active_branches = strtoul(a + 18, NULL, 0);
1863                 else if (!strncmp(a, "--export-marks=", 15))
1864                         mark_file = a + 15;
1865                 else
1866                         die("unknown option %s", a);
1867         }
1868         if (i != argc)
1869                 usage(fast_import_usage);
1870
1871         alloc_objects(object_entry_alloc);
1872         strbuf_init(&command_buf);
1873
1874         atom_table = xcalloc(atom_table_sz, sizeof(struct atom_str*));
1875         branch_table = xcalloc(branch_table_sz, sizeof(struct branch*));
1876         avail_tree_table = xcalloc(avail_tree_table_sz, sizeof(struct avail_tree_content*));
1877         marks = pool_calloc(1, sizeof(struct mark_set));
1878
1879         start_packfile();
1880         for (;;) {
1881                 read_next_command();
1882                 if (command_buf.eof)
1883                         break;
1884                 else if (!strcmp("blob", command_buf.buf))
1885                         cmd_new_blob();
1886                 else if (!strncmp("commit ", command_buf.buf, 7))
1887                         cmd_new_commit();
1888                 else if (!strncmp("tag ", command_buf.buf, 4))
1889                         cmd_new_tag();
1890                 else if (!strncmp("reset ", command_buf.buf, 6))
1891                         cmd_reset_branch();
1892                 else if (!strcmp("checkpoint", command_buf.buf))
1893                         cmd_checkpoint();
1894                 else
1895                         die("Unsupported command: %s", command_buf.buf);
1896         }
1897         end_packfile();
1898
1899         dump_branches();
1900         dump_tags();
1901         unkeep_all_packs();
1902         dump_marks();
1903
1904         total_count = 0;
1905         for (i = 0; i < ARRAY_SIZE(object_count_by_type); i++)
1906                 total_count += object_count_by_type[i];
1907         duplicate_count = 0;
1908         for (i = 0; i < ARRAY_SIZE(duplicate_count_by_type); i++)
1909                 duplicate_count += duplicate_count_by_type[i];
1910
1911         fprintf(stderr, "%s statistics:\n", argv[0]);
1912         fprintf(stderr, "---------------------------------------------------------------------\n");
1913         fprintf(stderr, "Alloc'd objects: %10ju\n", alloc_count);
1914         fprintf(stderr, "Total objects:   %10ju (%10ju duplicates                  )\n", total_count, duplicate_count);
1915         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]);
1916         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]);
1917         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]);
1918         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]);
1919         fprintf(stderr, "Total branches:  %10lu (%10lu loads     )\n", branch_count, branch_load_count);
1920         fprintf(stderr, "      marks:     %10ju (%10ju unique    )\n", (((uintmax_t)1) << marks->shift) * 1024, marks_set_count);
1921         fprintf(stderr, "      atoms:     %10u\n", atom_cnt);
1922         fprintf(stderr, "Memory total:    %10ju KiB\n", (total_allocd + alloc_count*sizeof(struct object_entry))/1024);
1923         fprintf(stderr, "       pools:    %10lu KiB\n", total_allocd/1024);
1924         fprintf(stderr, "     objects:    %10ju KiB\n", (alloc_count*sizeof(struct object_entry))/1024);
1925         fprintf(stderr, "---------------------------------------------------------------------\n");
1926         pack_report();
1927         fprintf(stderr, "---------------------------------------------------------------------\n");
1928         fprintf(stderr, "\n");
1929
1930         return 0;
1931 }