fast-import.c

   1 /*
   2 Format of STDIN stream:
   3
   4   stream ::= cmd*;
   5
   6   cmd ::= new_blob
   7         | new_commit
   8         | new_tag
   9         | reset_branch
  10         ;
  11
  12   new_blob ::= 'blob' lf
  13         mark?
  14     file_content;
  15   file_content ::= data;
  16
  17   new_commit ::= 'commit' sp ref_str lf
  18     mark?
  19     ('author' sp name '<' email '>' ts tz lf)?
  20     'committer' sp name '<' email '>' ts tz lf
  21     commit_msg
  22     ('from' sp (ref_str | hexsha1 | sha1exp_str | idnum) lf)?
  23     ('merge' sp (ref_str | hexsha1 | sha1exp_str | idnum) lf)*
  24     file_change*
  25     lf;
  26   commit_msg ::= data;
  27
  28   file_change ::= 'M' sp mode sp (hexsha1 | idnum) sp path_str lf
  29                 | 'D' sp path_str lf
  30                 ;
  31   mode ::= '644' | '755';
  32
  33   new_tag ::= 'tag' sp tag_str lf
  34     'from' sp (ref_str | hexsha1 | sha1exp_str | idnum) lf
  35         'tagger' sp name '<' email '>' ts tz lf
  36     tag_msg;
  37   tag_msg ::= data;
  38
  39   reset_branch ::= 'reset' sp ref_str lf
  40     ('from' sp (ref_str | hexsha1 | sha1exp_str | idnum) lf)?
  41     lf;
  42
  43   checkpoint ::= 'checkpoint' lf
  44     lf;
  45
  46      # note: the first idnum in a stream should be 1 and subsequent
  47      # idnums should not have gaps between values as this will cause
  48      # the stream parser to reserve space for the gapped values.  An
  49          # idnum can be updated in the future to a new object by issuing
  50      # a new mark directive with the old idnum.
  51          #
  52   mark ::= 'mark' sp idnum lf;
  53
  54      # note: declen indicates the length of binary_data in bytes.
  55      # declen does not include the lf preceeding or trailing the
  56      # binary data.
  57      #
  58   data ::= 'data' sp declen lf
  59     binary_data
  60         lf;
  61
  62      # note: quoted strings are C-style quoting supporting \c for
  63      # common escapes of 'c' (e..g \n, \t, \\, \") or \nnn where nnn
  64          # is the signed byte value in octal.  Note that the only
  65      # characters which must actually be escaped to protect the
  66      # stream formatting is: \, " and LF.  Otherwise these values
  67          # are UTF8.
  68      #
  69   ref_str     ::= ref     | '"' quoted(ref)     '"' ;
  70   sha1exp_str ::= sha1exp | '"' quoted(sha1exp) '"' ;
  71   tag_str     ::= tag     | '"' quoted(tag)     '"' ;
  72   path_str    ::= path    | '"' quoted(path)    '"' ;
  73
  74   declen ::= # unsigned 32 bit value, ascii base10 notation;
  75   bigint ::= # unsigned integer value, ascii base10 notation;
  76   binary_data ::= # file content, not interpreted;
  77
  78   sp ::= # ASCII space character;
  79   lf ::= # ASCII newline (LF) character;
  80
  81      # note: a colon (':') must precede the numerical value assigned to
  82          # an idnum.  This is to distinguish it from a ref or tag name as
  83      # GIT does not permit ':' in ref or tag strings.
  84          #
  85   idnum   ::= ':' bigint;
  86   path    ::= # GIT style file path, e.g. "a/b/c";
  87   ref     ::= # GIT ref name, e.g. "refs/heads/MOZ_GECKO_EXPERIMENT";
  88   tag     ::= # GIT tag name, e.g. "FIREFOX_1_5";
  89   sha1exp ::= # Any valid GIT SHA1 expression;
  90   hexsha1 ::= # SHA1 in hexadecimal format;
  91
  92      # note: name and email are UTF8 strings, however name must not
  93          # contain '<' or lf and email must not contain any of the
  94      # following: '<', '>', lf.
  95          #
  96   name  ::= # valid GIT author/committer name;
  97   email ::= # valid GIT author/committer email;
  98   ts    ::= # time since the epoch in seconds, ascii base10 notation;
  99   tz    ::= # GIT style timezone;
 100 */
 101
 102 #include "builtin.h"
 103 #include "cache.h"
 104 #include "object.h"
 105 #include "blob.h"
 106 #include "tree.h"
 107 #include "delta.h"
 108 #include "pack.h"
 109 #include "refs.h"
 110 #include "csum-file.h"
 111 #include "strbuf.h"
 112 #include "quote.h"
 113
 114 struct object_entry
 115 {
 116         struct object_entry *next;
 117         unsigned long offset;
 118         unsigned type : TYPE_BITS;
 119         unsigned pack_id : 16;
 120         unsigned char sha1[20];
 121 };
 122
 123 struct object_entry_pool
 124 {
 125         struct object_entry_pool *next_pool;
 126         struct object_entry *next_free;
 127         struct object_entry *end;
 128         struct object_entry entries[FLEX_ARRAY]; /* more */
 129 };
 130
 131 struct mark_set
 132 {
 133         union {
 134                 struct object_entry *marked[1024];
 135                 struct mark_set *sets[1024];
 136         } data;
 137         unsigned int shift;
 138 };
 139
 140 struct last_object
 141 {
 142         void *data;
 143         unsigned long len;
 144         unsigned long offset;
 145         unsigned int depth;
 146         unsigned no_free:1;
 147 };
 148
 149 struct mem_pool
 150 {
 151         struct mem_pool *next_pool;
 152         char *next_free;
 153         char *end;
 154         char space[FLEX_ARRAY]; /* more */
 155 };
 156
 157 struct atom_str
 158 {
 159         struct atom_str *next_atom;
 160         unsigned int str_len;
 161         char str_dat[FLEX_ARRAY]; /* more */
 162 };
 163
 164 struct tree_content;
 165 struct tree_entry
 166 {
 167         struct tree_content *tree;
 168         struct atom_str* name;
 169         struct tree_entry_ms
 170         {
 171                 unsigned int mode;
 172                 unsigned char sha1[20];
 173         } versions[2];
 174 };
 175
 176 struct tree_content
 177 {
 178         unsigned int entry_capacity; /* must match avail_tree_content */
 179         unsigned int entry_count;
 180         unsigned int delta_depth;
 181         struct tree_entry *entries[FLEX_ARRAY]; /* more */
 182 };
 183
 184 struct avail_tree_content
 185 {
 186         unsigned int entry_capacity; /* must match tree_content */
 187         struct avail_tree_content *next_avail;
 188 };
 189
 190 struct branch
 191 {
 192         struct branch *table_next_branch;
 193         struct branch *active_next_branch;
 194         const char *name;
 195         struct tree_entry branch_tree;
 196         unsigned long last_commit;
 197         unsigned int pack_id;
 198         unsigned char sha1[20];
 199 };
 200
 201 struct tag
 202 {
 203         struct tag *next_tag;
 204         const char *name;
 205         unsigned int pack_id;
 206         unsigned char sha1[20];
 207 };
 208
 209 struct dbuf
 210 {
 211         void *buffer;
 212         size_t capacity;
 213 };
 214
 215 struct hash_list
 216 {
 217         struct hash_list *next;
 218         unsigned char sha1[20];
 219 };
 220
 221 /* Configured limits on output */
 222 static unsigned long max_depth = 10;
 223 static unsigned long max_packsize = (1LL << 32) - 1;
 224 static uintmax_t max_objects = -1;
 225
 226 /* Stats and misc. counters */
 227 static uintmax_t alloc_count;
 228 static uintmax_t marks_set_count;
 229 static uintmax_t object_count_by_type[1 << TYPE_BITS];
 230 static uintmax_t duplicate_count_by_type[1 << TYPE_BITS];
 231 static uintmax_t delta_count_by_type[1 << TYPE_BITS];
 232 static unsigned long object_count;
 233 static unsigned long branch_count;
 234 static unsigned long branch_load_count;
 235
 236 /* Memory pools */
 237 static size_t mem_pool_alloc = 2*1024*1024 - sizeof(struct mem_pool);
 238 static size_t total_allocd;
 239 static struct mem_pool *mem_pool;
 240
 241 /* Atom management */
 242 static unsigned int atom_table_sz = 4451;
 243 static unsigned int atom_cnt;
 244 static struct atom_str **atom_table;
 245
 246 /* The .pack file being generated */
 247 static unsigned int pack_id;
 248 static struct packed_git *pack_data;
 249 static struct packed_git **all_packs;
 250 static unsigned long pack_size;
 251
 252 /* Table of objects we've written. */
 253 static unsigned int object_entry_alloc = 5000;
 254 static struct object_entry_pool *blocks;
 255 static struct object_entry *object_table[1 << 16];
 256 static struct mark_set *marks;
 257 static const char* mark_file;
 258
 259 /* Our last blob */
 260 static struct last_object last_blob;
 261
 262 /* Tree management */
 263 static unsigned int tree_entry_alloc = 1000;
 264 static void *avail_tree_entry;
 265 static unsigned int avail_tree_table_sz = 100;
 266 static struct avail_tree_content **avail_tree_table;
 267 static struct dbuf old_tree;
 268 static struct dbuf new_tree;
 269
 270 /* Branch data */
 271 static unsigned long max_active_branches = 5;
 272 static unsigned long cur_active_branches;
 273 static unsigned long branch_table_sz = 1039;
 274 static struct branch **branch_table;
 275 static struct branch *active_branches;
 276
 277 /* Tag data */
 278 static struct tag *first_tag;
 279 static struct tag *last_tag;
 280
 281 /* Input stream parsing */
 282 static struct strbuf command_buf;
 283 static uintmax_t next_mark;
 284 static struct dbuf new_data;
 285 static FILE* branch_log;
 286
 287
 288 static void alloc_objects(unsigned int cnt)
 289 {
 290         struct object_entry_pool *b;
 291
 292         b = xmalloc(sizeof(struct object_entry_pool)
 293                 + cnt * sizeof(struct object_entry));
 294         b->next_pool = blocks;
 295         b->next_free = b->entries;
 296         b->end = b->entries + cnt;
 297         blocks = b;
 298         alloc_count += cnt;
 299 }
 300
 301 static struct object_entry* new_object(unsigned char *sha1)
 302 {
 303         struct object_entry *e;
 304
 305         if (blocks->next_free == blocks->end)
 306                 alloc_objects(object_entry_alloc);
 307
 308         e = blocks->next_free++;
 309         hashcpy(e->sha1, sha1);
 310         return e;
 311 }
 312
 313 static struct object_entry* find_object(unsigned char *sha1)
 314 {
 315         unsigned int h = sha1[0] << 8 | sha1[1];
 316         struct object_entry *e;
 317         for (e = object_table[h]; e; e = e->next)
 318                 if (!hashcmp(sha1, e->sha1))
 319                         return e;
 320         return NULL;
 321 }
 322
 323 static struct object_entry* insert_object(unsigned char *sha1)
 324 {
 325         unsigned int h = sha1[0] << 8 | sha1[1];
 326         struct object_entry *e = object_table[h];
 327         struct object_entry *p = NULL;
 328
 329         while (e) {
 330                 if (!hashcmp(sha1, e->sha1))
 331                         return e;
 332                 p = e;
 333                 e = e->next;
 334         }
 335
 336         e = new_object(sha1);
 337         e->next = NULL;
 338         e->offset = 0;
 339         if (p)
 340                 p->next = e;
 341         else
 342                 object_table[h] = e;
 343         return e;
 344 }
 345
 346 static unsigned int hc_str(const char *s, size_t len)
 347 {
 348         unsigned int r = 0;
 349         while (len-- > 0)
 350                 r = r * 31 + *s++;
 351         return r;
 352 }
 353
 354 static void* pool_alloc(size_t len)
 355 {
 356         struct mem_pool *p;
 357         void *r;
 358
 359         for (p = mem_pool; p; p = p->next_pool)
 360                 if ((p->end - p->next_free >= len))
 361                         break;
 362
 363         if (!p) {
 364                 if (len >= (mem_pool_alloc/2)) {
 365                         total_allocd += len;
 366                         return xmalloc(len);
 367                 }
 368                 total_allocd += sizeof(struct mem_pool) + mem_pool_alloc;
 369                 p = xmalloc(sizeof(struct mem_pool) + mem_pool_alloc);
 370                 p->next_pool = mem_pool;
 371                 p->next_free = p->space;
 372                 p->end = p->next_free + mem_pool_alloc;
 373                 mem_pool = p;
 374         }
 375
 376         r = p->next_free;
 377         /* round out to a pointer alignment */
 378         if (len & (sizeof(void*) - 1))
 379                 len += sizeof(void*) - (len & (sizeof(void*) - 1));
 380         p->next_free += len;
 381         return r;
 382 }
 383
 384 static void* pool_calloc(size_t count, size_t size)
 385 {
 386         size_t len = count * size;
 387         void *r = pool_alloc(len);
 388         memset(r, 0, len);
 389         return r;
 390 }
 391
 392 static char* pool_strdup(const char *s)
 393 {
 394         char *r = pool_alloc(strlen(s) + 1);
 395         strcpy(r, s);
 396         return r;
 397 }
 398
 399 static void size_dbuf(struct dbuf *b, size_t maxlen)
 400 {
 401         if (b->buffer) {
 402                 if (b->capacity >= maxlen)
 403                         return;
 404                 free(b->buffer);
 405         }
 406         b->capacity = ((maxlen / 1024) + 1) * 1024;
 407         b->buffer = xmalloc(b->capacity);
 408 }
 409
 410 static void insert_mark(uintmax_t idnum, struct object_entry *oe)
 411 {
 412         struct mark_set *s = marks;
 413         while ((idnum >> s->shift) >= 1024) {
 414                 s = pool_calloc(1, sizeof(struct mark_set));
 415                 s->shift = marks->shift + 10;
 416                 s->data.sets[0] = marks;
 417                 marks = s;
 418         }
 419         while (s->shift) {
 420                 uintmax_t i = idnum >> s->shift;
 421                 idnum -= i << s->shift;
 422                 if (!s->data.sets[i]) {
 423                         s->data.sets[i] = pool_calloc(1, sizeof(struct mark_set));
 424                         s->data.sets[i]->shift = s->shift - 10;
 425                 }
 426                 s = s->data.sets[i];
 427         }
 428         if (!s->data.marked[idnum])
 429                 marks_set_count++;
 430         s->data.marked[idnum] = oe;
 431 }
 432
 433 static struct object_entry* find_mark(uintmax_t idnum)
 434 {
 435         uintmax_t orig_idnum = idnum;
 436         struct mark_set *s = marks;
 437         struct object_entry *oe = NULL;
 438         if ((idnum >> s->shift) < 1024) {
 439                 while (s && s->shift) {
 440                         uintmax_t i = idnum >> s->shift;
 441                         idnum -= i << s->shift;
 442                         s = s->data.sets[i];
 443                 }
 444                 if (s)
 445                         oe = s->data.marked[idnum];
 446         }
 447         if (!oe)
 448                 die("mark :%ju not declared", orig_idnum);
 449         return oe;
 450 }
 451
 452 static struct atom_str* to_atom(const char *s, size_t len)
 453 {
 454         unsigned int hc = hc_str(s, len) % atom_table_sz;
 455         struct atom_str *c;
 456
 457         for (c = atom_table[hc]; c; c = c->next_atom)
 458                 if (c->str_len == len && !strncmp(s, c->str_dat, len))
 459                         return c;
 460
 461         c = pool_alloc(sizeof(struct atom_str) + len + 1);
 462         c->str_len = len;
 463         strncpy(c->str_dat, s, len);
 464         c->str_dat[len] = 0;
 465         c->next_atom = atom_table[hc];
 466         atom_table[hc] = c;
 467         atom_cnt++;
 468         return c;
 469 }
 470
 471 static struct branch* lookup_branch(const char *name)
 472 {
 473         unsigned int hc = hc_str(name, strlen(name)) % branch_table_sz;
 474         struct branch *b;
 475
 476         for (b = branch_table[hc]; b; b = b->table_next_branch)
 477                 if (!strcmp(name, b->name))
 478                         return b;
 479         return NULL;
 480 }
 481
 482 static struct branch* new_branch(const char *name)
 483 {
 484         unsigned int hc = hc_str(name, strlen(name)) % branch_table_sz;
 485         struct branch* b = lookup_branch(name);
 486
 487         if (b)
 488                 die("Invalid attempt to create duplicate branch: %s", name);
 489         if (check_ref_format(name))
 490                 die("Branch name doesn't conform to GIT standards: %s", name);
 491
 492         b = pool_calloc(1, sizeof(struct branch));
 493         b->name = pool_strdup(name);
 494         b->table_next_branch = branch_table[hc];
 495         b->branch_tree.versions[0].mode = S_IFDIR;
 496         b->branch_tree.versions[1].mode = S_IFDIR;
 497         branch_table[hc] = b;
 498         branch_count++;
 499         return b;
 500 }
 501
 502 static unsigned int hc_entries(unsigned int cnt)
 503 {
 504         cnt = cnt & 7 ? (cnt / 8) + 1 : cnt / 8;
 505         return cnt < avail_tree_table_sz ? cnt : avail_tree_table_sz - 1;
 506 }
 507
 508 static struct tree_content* new_tree_content(unsigned int cnt)
 509 {
 510         struct avail_tree_content *f, *l = NULL;
 511         struct tree_content *t;
 512         unsigned int hc = hc_entries(cnt);
 513
 514         for (f = avail_tree_table[hc]; f; l = f, f = f->next_avail)
 515                 if (f->entry_capacity >= cnt)
 516                         break;
 517
 518         if (f) {
 519                 if (l)
 520                         l->next_avail = f->next_avail;
 521                 else
 522                         avail_tree_table[hc] = f->next_avail;
 523         } else {
 524                 cnt = cnt & 7 ? ((cnt / 8) + 1) * 8 : cnt;
 525                 f = pool_alloc(sizeof(*t) + sizeof(t->entries[0]) * cnt);
 526                 f->entry_capacity = cnt;
 527         }
 528
 529         t = (struct tree_content*)f;
 530         t->entry_count = 0;
 531         t->delta_depth = 0;
 532         return t;
 533 }
 534
 535 static void release_tree_entry(struct tree_entry *e);
 536 static void release_tree_content(struct tree_content *t)
 537 {
 538         struct avail_tree_content *f = (struct avail_tree_content*)t;
 539         unsigned int hc = hc_entries(f->entry_capacity);
 540         f->next_avail = avail_tree_table[hc];
 541         avail_tree_table[hc] = f;
 542 }
 543
 544 static void release_tree_content_recursive(struct tree_content *t)
 545 {
 546         unsigned int i;
 547         for (i = 0; i < t->entry_count; i++)
 548                 release_tree_entry(t->entries[i]);
 549         release_tree_content(t);
 550 }
 551
 552 static struct tree_content* grow_tree_content(
 553         struct tree_content *t,
 554         int amt)
 555 {
 556         struct tree_content *r = new_tree_content(t->entry_count + amt);
 557         r->entry_count = t->entry_count;
 558         r->delta_depth = t->delta_depth;
 559         memcpy(r->entries,t->entries,t->entry_count*sizeof(t->entries[0]));
 560         release_tree_content(t);
 561         return r;
 562 }
 563
 564 static struct tree_entry* new_tree_entry()
 565 {
 566         struct tree_entry *e;
 567
 568         if (!avail_tree_entry) {
 569                 unsigned int n = tree_entry_alloc;
 570                 total_allocd += n * sizeof(struct tree_entry);
 571                 avail_tree_entry = e = xmalloc(n * sizeof(struct tree_entry));
 572                 while (n-- > 1) {
 573                         *((void**)e) = e + 1;
 574                         e++;
 575                 }
 576                 *((void**)e) = NULL;
 577         }
 578
 579         e = avail_tree_entry;
 580         avail_tree_entry = *((void**)e);
 581         return e;
 582 }
 583
 584 static void release_tree_entry(struct tree_entry *e)
 585 {
 586         if (e->tree)
 587                 release_tree_content_recursive(e->tree);
 588         *((void**)e) = avail_tree_entry;
 589         avail_tree_entry = e;
 590 }
 591
 592 static void start_packfile()
 593 {
 594         static char tmpfile[PATH_MAX];
 595         struct packed_git *p;
 596         struct pack_header hdr;
 597         int pack_fd;
 598
 599         snprintf(tmpfile, sizeof(tmpfile),
 600                 "%s/pack_XXXXXX", get_object_directory());
 601         pack_fd = mkstemp(tmpfile);
 602         if (pack_fd < 0)
 603                 die("Can't create %s: %s", tmpfile, strerror(errno));
 604         p = xcalloc(1, sizeof(*p) + strlen(tmpfile) + 2);
 605         strcpy(p->pack_name, tmpfile);
 606         p->pack_fd = pack_fd;
 607
 608         hdr.hdr_signature = htonl(PACK_SIGNATURE);
 609         hdr.hdr_version = htonl(2);
 610         hdr.hdr_entries = 0;
 611         write_or_die(p->pack_fd, &hdr, sizeof(hdr));
 612
 613         pack_data = p;
 614         pack_size = sizeof(hdr);
 615         object_count = 0;
 616
 617         all_packs = xrealloc(all_packs, sizeof(*all_packs) * (pack_id + 1));
 618         all_packs[pack_id] = p;
 619 }
 620
 621 static void fixup_header_footer()
 622 {
 623         int pack_fd = pack_data->pack_fd;
 624         SHA_CTX c;
 625         char hdr[8];
 626         unsigned long cnt;
 627         char *buf;
 628
 629         if (lseek(pack_fd, 0, SEEK_SET) != 0)
 630                 die("Failed seeking to start: %s", strerror(errno));
 631
 632         SHA1_Init(&c);
 633         if (read_in_full(pack_fd, hdr, 8) != 8)
 634                 die("Unable to reread header of %s", pack_data->pack_name);
 635         SHA1_Update(&c, hdr, 8);
 636
 637         cnt = htonl(object_count);
 638         SHA1_Update(&c, &cnt, 4);
 639         write_or_die(pack_fd, &cnt, 4);
 640
 641         buf = xmalloc(128 * 1024);
 642         for (;;) {
 643                 size_t n = xread(pack_fd, buf, 128 * 1024);
 644                 if (n <= 0)
 645                         break;
 646                 SHA1_Update(&c, buf, n);
 647         }
 648         free(buf);
 649
 650         SHA1_Final(pack_data->sha1, &c);
 651         write_or_die(pack_fd, pack_data->sha1, sizeof(pack_data->sha1));
 652         close(pack_fd);
 653 }
 654
 655 static int oecmp (const void *a_, const void *b_)
 656 {
 657         struct object_entry *a = *((struct object_entry**)a_);
 658         struct object_entry *b = *((struct object_entry**)b_);
 659         return hashcmp(a->sha1, b->sha1);
 660 }
 661
 662 static char* create_index()
 663 {
 664         static char tmpfile[PATH_MAX];
 665         SHA_CTX ctx;
 666         struct sha1file *f;
 667         struct object_entry **idx, **c, **last, *e;
 668         struct object_entry_pool *o;
 669         unsigned int array[256];
 670         int i, idx_fd;
 671
 672         /* Build the sorted table of object IDs. */
 673         idx = xmalloc(object_count * sizeof(struct object_entry*));
 674         c = idx;
 675         for (o = blocks; o; o = o->next_pool)
 676                 for (e = o->next_free; e-- != o->entries;)
 677                         if (pack_id == e->pack_id)
 678                                 *c++ = e;
 679         last = idx + object_count;
 680         if (c != last)
 681                 die("internal consistency error creating the index");
 682         qsort(idx, object_count, sizeof(struct object_entry*), oecmp);
 683
 684         /* Generate the fan-out array. */
 685         c = idx;
 686         for (i = 0; i < 256; i++) {
 687                 struct object_entry **next = c;;
 688                 while (next < last) {
 689                         if ((*next)->sha1[0] != i)
 690                                 break;
 691                         next++;
 692                 }
 693                 array[i] = htonl(next - idx);
 694                 c = next;
 695         }
 696
 697         snprintf(tmpfile, sizeof(tmpfile),
 698                 "%s/index_XXXXXX", get_object_directory());
 699         idx_fd = mkstemp(tmpfile);
 700         if (idx_fd < 0)
 701                 die("Can't create %s: %s", tmpfile, strerror(errno));
 702         f = sha1fd(idx_fd, tmpfile);
 703         sha1write(f, array, 256 * sizeof(int));
 704         SHA1_Init(&ctx);
 705         for (c = idx; c != last; c++) {
 706                 unsigned int offset = htonl((*c)->offset);
 707                 sha1write(f, &offset, 4);
 708                 sha1write(f, (*c)->sha1, sizeof((*c)->sha1));
 709                 SHA1_Update(&ctx, (*c)->sha1, 20);
 710         }
 711         sha1write(f, pack_data->sha1, sizeof(pack_data->sha1));
 712         sha1close(f, NULL, 1);
 713         free(idx);
 714         SHA1_Final(pack_data->sha1, &ctx);
 715         return tmpfile;
 716 }
 717
 718 static char* keep_pack(char *curr_index_name)
 719 {
 720         static char name[PATH_MAX];
 721         static char *keep_msg = "fast-import";
 722         int keep_fd;
 723
 724         chmod(pack_data->pack_name, 0444);
 725         chmod(curr_index_name, 0444);
 726
 727         snprintf(name, sizeof(name), "%s/pack/pack-%s.keep",
 728                  get_object_directory(), sha1_to_hex(pack_data->sha1));
 729         keep_fd = open(name, O_RDWR|O_CREAT|O_EXCL, 0600);
 730         if (keep_fd < 0)
 731                 die("cannot create keep file");
 732         write(keep_fd, keep_msg, strlen(keep_msg));
 733         close(keep_fd);
 734
 735         snprintf(name, sizeof(name), "%s/pack/pack-%s.pack",
 736                  get_object_directory(), sha1_to_hex(pack_data->sha1));
 737         if (move_temp_to_file(pack_data->pack_name, name))
 738                 die("cannot store pack file");
 739
 740         snprintf(name, sizeof(name), "%s/pack/pack-%s.idx",
 741                  get_object_directory(), sha1_to_hex(pack_data->sha1));
 742         if (move_temp_to_file(curr_index_name, name))
 743                 die("cannot store index file");
 744         return name;
 745 }
 746
 747 static void unkeep_all_packs()
 748 {
 749         static char name[PATH_MAX];
 750         int k;
 751
 752         for (k = 0; k < pack_id; k++) {
 753                 struct packed_git *p = all_packs[k];
 754                 snprintf(name, sizeof(name), "%s/pack/pack-%s.keep",
 755                          get_object_directory(), sha1_to_hex(p->sha1));
 756                 unlink(name);
 757         }
 758 }
 759
 760 static void end_packfile()
 761 {
 762         struct packed_git *old_p = pack_data, *new_p;
 763
 764         if (object_count) {
 765                 char *idx_name;
 766                 int i;
 767                 struct branch *b;
 768                 struct tag *t;
 769
 770                 fixup_header_footer();
 771                 idx_name = keep_pack(create_index());
 772
 773                 /* Register the packfile with core git's machinary. */
 774                 new_p = add_packed_git(idx_name, strlen(idx_name), 1);
 775                 if (!new_p)
 776                         die("core git rejected index %s", idx_name);
 777                 new_p->windows = old_p->windows;
 778                 all_packs[pack_id] = new_p;
 779                 install_packed_git(new_p);
 780
 781                 /* Print the boundary */
 782                 fprintf(stdout, "%s:", new_p->pack_name);
 783                 for (i = 0; i < branch_table_sz; i++) {
 784                         for (b = branch_table[i]; b; b = b->table_next_branch) {
 785                                 if (b->pack_id == pack_id)
 786                                         fprintf(stdout, " %s", sha1_to_hex(b->sha1));
 787                         }
 788                 }
 789                 for (t = first_tag; t; t = t->next_tag) {
 790                         if (t->pack_id == pack_id)
 791                                 fprintf(stdout, " %s", sha1_to_hex(t->sha1));
 792                 }
 793                 fputc('\n', stdout);
 794
 795                 pack_id++;
 796         }
 797         else
 798                 unlink(old_p->pack_name);
 799         free(old_p);
 800
 801         /* We can't carry a delta across packfiles. */
 802         free(last_blob.data);
 803         last_blob.data = NULL;
 804         last_blob.len = 0;
 805         last_blob.offset = 0;
 806         last_blob.depth = 0;
 807 }
 808
 809 static void checkpoint()
 810 {
 811         end_packfile();
 812         start_packfile();
 813 }
 814
 815 static size_t encode_header(
 816         enum object_type type,
 817         size_t size,
 818         unsigned char *hdr)
 819 {
 820         int n = 1;
 821         unsigned char c;
 822
 823         if (type < OBJ_COMMIT || type > OBJ_REF_DELTA)
 824                 die("bad type %d", type);
 825
 826         c = (type << 4) | (size & 15);
 827         size >>= 4;
 828         while (size) {
 829                 *hdr++ = c | 0x80;
 830                 c = size & 0x7f;
 831                 size >>= 7;
 832                 n++;
 833         }
 834         *hdr = c;
 835         return n;
 836 }
 837
 838 static int store_object(
 839         enum object_type type,
 840         void *dat,
 841         size_t datlen,
 842         struct last_object *last,
 843         unsigned char *sha1out,
 844         uintmax_t mark)
 845 {
 846         void *out, *delta;
 847         struct object_entry *e;
 848         unsigned char hdr[96];
 849         unsigned char sha1[20];
 850         unsigned long hdrlen, deltalen;
 851         SHA_CTX c;
 852         z_stream s;
 853
 854         hdrlen = sprintf((char*)hdr,"%s %lu",type_names[type],datlen) + 1;
 855         SHA1_Init(&c);
 856         SHA1_Update(&c, hdr, hdrlen);
 857         SHA1_Update(&c, dat, datlen);
 858         SHA1_Final(sha1, &c);
 859         if (sha1out)
 860                 hashcpy(sha1out, sha1);
 861
 862         e = insert_object(sha1);
 863         if (mark)
 864                 insert_mark(mark, e);
 865         if (e->offset) {
 866                 duplicate_count_by_type[type]++;
 867                 return 1;
 868         }
 869
 870         if (last && last->data && last->depth < max_depth) {
 871                 delta = diff_delta(last->data, last->len,
 872                         dat, datlen,
 873                         &deltalen, 0);
 874                 if (delta && deltalen >= datlen) {
 875                         free(delta);
 876                         delta = NULL;
 877                 }
 878         } else
 879                 delta = NULL;
 880
 881         memset(&s, 0, sizeof(s));
 882         deflateInit(&s, zlib_compression_level);
 883         if (delta) {
 884                 s.next_in = delta;
 885                 s.avail_in = deltalen;
 886         } else {
 887                 s.next_in = dat;
 888                 s.avail_in = datlen;
 889         }
 890         s.avail_out = deflateBound(&s, s.avail_in);
 891         s.next_out = out = xmalloc(s.avail_out);
 892         while (deflate(&s, Z_FINISH) == Z_OK)
 893                 /* nothing */;
 894         deflateEnd(&s);
 895
 896         /* Determine if we should auto-checkpoint. */
 897         if ((object_count + 1) > max_objects
 898                 || (object_count + 1) < object_count
 899                 || (pack_size + 60 + s.total_out) > max_packsize
 900                 || (pack_size + 60 + s.total_out) < pack_size) {
 901
 902                 /* This new object needs to *not* have the current pack_id. */
 903                 e->pack_id = pack_id + 1;
 904                 checkpoint();
 905
 906                 /* We cannot carry a delta into the new pack. */
 907                 if (delta) {
 908                         free(delta);
 909                         delta = NULL;
 910
 911                         memset(&s, 0, sizeof(s));
 912                         deflateInit(&s, zlib_compression_level);
 913                         s.next_in = dat;
 914                         s.avail_in = datlen;
 915                         s.avail_out = deflateBound(&s, s.avail_in);
 916                         s.next_out = out = xrealloc(out, s.avail_out);
 917                         while (deflate(&s, Z_FINISH) == Z_OK)
 918                                 /* nothing */;
 919                         deflateEnd(&s);
 920                 }
 921         }
 922
 923         e->type = type;
 924         e->pack_id = pack_id;
 925         e->offset = pack_size;
 926         object_count++;
 927         object_count_by_type[type]++;
 928
 929         if (delta) {
 930                 unsigned long ofs = e->offset - last->offset;
 931                 unsigned pos = sizeof(hdr) - 1;
 932
 933                 delta_count_by_type[type]++;
 934                 last->depth++;
 935
 936                 hdrlen = encode_header(OBJ_OFS_DELTA, deltalen, hdr);
 937                 write_or_die(pack_data->pack_fd, hdr, hdrlen);
 938                 pack_size += hdrlen;
 939
 940                 hdr[pos] = ofs & 127;
 941                 while (ofs >>= 7)
 942                         hdr[--pos] = 128 | (--ofs & 127);
 943                 write_or_die(pack_data->pack_fd, hdr + pos, sizeof(hdr) - pos);
 944                 pack_size += sizeof(hdr) - pos;
 945         } else {
 946                 if (last)
 947                         last->depth = 0;
 948                 hdrlen = encode_header(type, datlen, hdr);
 949                 write_or_die(pack_data->pack_fd, hdr, hdrlen);
 950                 pack_size += hdrlen;
 951         }
 952
 953         write_or_die(pack_data->pack_fd, out, s.total_out);
 954         pack_size += s.total_out;
 955
 956         free(out);
 957         if (delta)
 958                 free(delta);
 959         if (last) {
 960                 if (last->data && !last->no_free)
 961                         free(last->data);
 962                 last->data = dat;
 963                 last->offset = e->offset;
 964                 last->len = datlen;
 965         }
 966         return 0;
 967 }
 968
 969 static void *gfi_unpack_entry(
 970         struct object_entry *oe,
 971         unsigned long *sizep)
 972 {
 973         static char type[20];
 974         struct packed_git *p = all_packs[oe->pack_id];
 975         if (p == pack_data)
 976                 p->pack_size = pack_size + 20;
 977         return unpack_entry(p, oe->offset, type, sizep);
 978 }
 979
 980 static const char *get_mode(const char *str, unsigned int *modep)
 981 {
 982         unsigned char c;
 983         unsigned int mode = 0;
 984
 985         while ((c = *str++) != ' ') {
 986                 if (c < '0' || c > '7')
 987                         return NULL;
 988                 mode = (mode << 3) + (c - '0');
 989         }
 990         *modep = mode;
 991         return str;
 992 }
 993
 994 static void load_tree(struct tree_entry *root)
 995 {
 996         unsigned char* sha1 = root->versions[1].sha1;
 997         struct object_entry *myoe;
 998         struct tree_content *t;
 999         unsigned long size;
1000         char *buf;
1001         const char *c;
1002
1003         root->tree = t = new_tree_content(8);
1004         if (is_null_sha1(sha1))
1005                 return;
1006
1007         myoe = find_object(sha1);
1008         if (myoe) {
1009                 if (myoe->type != OBJ_TREE)
1010                         die("Not a tree: %s", sha1_to_hex(sha1));
1011                 t->delta_depth = 0;
1012                 buf = gfi_unpack_entry(myoe, &size);
1013         } else {
1014                 char type[20];
1015                 buf = read_sha1_file(sha1, type, &size);
1016                 if (!buf || strcmp(type, tree_type))
1017                         die("Can't load tree %s", sha1_to_hex(sha1));
1018         }
1019
1020         c = buf;
1021         while (c != (buf + size)) {
1022                 struct tree_entry *e = new_tree_entry();
1023
1024                 if (t->entry_count == t->entry_capacity)
1025                         root->tree = t = grow_tree_content(t, 8);
1026                 t->entries[t->entry_count++] = e;
1027
1028                 e->tree = NULL;
1029                 c = get_mode(c, &e->versions[1].mode);
1030                 if (!c)
1031                         die("Corrupt mode in %s", sha1_to_hex(sha1));
1032                 e->versions[0].mode = e->versions[1].mode;
1033                 e->name = to_atom(c, strlen(c));
1034                 c += e->name->str_len + 1;
1035                 hashcpy(e->versions[0].sha1, (unsigned char*)c);
1036                 hashcpy(e->versions[1].sha1, (unsigned char*)c);
1037                 c += 20;
1038         }
1039         free(buf);
1040 }
1041
1042 static int tecmp0 (const void *_a, const void *_b)
1043 {
1044         struct tree_entry *a = *((struct tree_entry**)_a);
1045         struct tree_entry *b = *((struct tree_entry**)_b);
1046         return base_name_compare(
1047                 a->name->str_dat, a->name->str_len, a->versions[0].mode,
1048                 b->name->str_dat, b->name->str_len, b->versions[0].mode);
1049 }
1050
1051 static int tecmp1 (const void *_a, const void *_b)
1052 {
1053         struct tree_entry *a = *((struct tree_entry**)_a);
1054         struct tree_entry *b = *((struct tree_entry**)_b);
1055         return base_name_compare(
1056                 a->name->str_dat, a->name->str_len, a->versions[1].mode,
1057                 b->name->str_dat, b->name->str_len, b->versions[1].mode);
1058 }
1059
1060 static void mktree(struct tree_content *t,
1061         int v,
1062         unsigned long *szp,
1063         struct dbuf *b)
1064 {
1065         size_t maxlen = 0;
1066         unsigned int i;
1067         char *c;
1068
1069         if (!v)
1070                 qsort(t->entries,t->entry_count,sizeof(t->entries[0]),tecmp0);
1071         else
1072                 qsort(t->entries,t->entry_count,sizeof(t->entries[0]),tecmp1);
1073
1074         for (i = 0; i < t->entry_count; i++) {
1075                 if (t->entries[i]->versions[v].mode)
1076                         maxlen += t->entries[i]->name->str_len + 34;
1077         }
1078
1079         size_dbuf(b, maxlen);
1080         c = b->buffer;
1081         for (i = 0; i < t->entry_count; i++) {
1082                 struct tree_entry *e = t->entries[i];
1083                 if (!e->versions[v].mode)
1084                         continue;
1085                 c += sprintf(c, "%o", e->versions[v].mode);
1086                 *c++ = ' ';
1087                 strcpy(c, e->name->str_dat);
1088                 c += e->name->str_len + 1;
1089                 hashcpy((unsigned char*)c, e->versions[v].sha1);
1090                 c += 20;
1091         }
1092         *szp = c - (char*)b->buffer;
1093 }
1094
1095 static void store_tree(struct tree_entry *root)
1096 {
1097         struct tree_content *t = root->tree;
1098         unsigned int i, j, del;
1099         unsigned long new_len;
1100         struct last_object lo;
1101         struct object_entry *le;
1102
1103         if (!is_null_sha1(root->versions[1].sha1))
1104                 return;
1105
1106         for (i = 0; i < t->entry_count; i++) {
1107                 if (t->entries[i]->tree)
1108                         store_tree(t->entries[i]);
1109         }
1110
1111         le = find_object(root->versions[0].sha1);
1112         if (!S_ISDIR(root->versions[0].mode)
1113                 || !le
1114                 || le->pack_id != pack_id) {
1115                 lo.data = NULL;
1116                 lo.depth = 0;
1117         } else {
1118                 mktree(t, 0, &lo.len, &old_tree);
1119                 lo.data = old_tree.buffer;
1120                 lo.offset = le->offset;
1121                 lo.depth = t->delta_depth;
1122                 lo.no_free = 1;
1123         }
1124
1125         mktree(t, 1, &new_len, &new_tree);
1126         store_object(OBJ_TREE, new_tree.buffer, new_len,
1127                 &lo, root->versions[1].sha1, 0);
1128
1129         t->delta_depth = lo.depth;
1130         for (i = 0, j = 0, del = 0; i < t->entry_count; i++) {
1131                 struct tree_entry *e = t->entries[i];
1132                 if (e->versions[1].mode) {
1133                         e->versions[0].mode = e->versions[1].mode;
1134                         hashcpy(e->versions[0].sha1, e->versions[1].sha1);
1135                         t->entries[j++] = e;
1136                 } else {
1137                         release_tree_entry(e);
1138                         del++;
1139                 }
1140         }
1141         t->entry_count -= del;
1142 }
1143
1144 static int tree_content_set(
1145         struct tree_entry *root,
1146         const char *p,
1147         const unsigned char *sha1,
1148         const unsigned int mode)
1149 {
1150         struct tree_content *t = root->tree;
1151         const char *slash1;
1152         unsigned int i, n;
1153         struct tree_entry *e;
1154
1155         slash1 = strchr(p, '/');
1156         if (slash1)
1157                 n = slash1 - p;
1158         else
1159                 n = strlen(p);
1160
1161         for (i = 0; i < t->entry_count; i++) {
1162                 e = t->entries[i];
1163                 if (e->name->str_len == n && !strncmp(p, e->name->str_dat, n)) {
1164                         if (!slash1) {
1165                                 if (e->versions[1].mode == mode
1166                                                 && !hashcmp(e->versions[1].sha1, sha1))
1167                                         return 0;
1168                                 e->versions[1].mode = mode;
1169                                 hashcpy(e->versions[1].sha1, sha1);
1170                                 if (e->tree) {
1171                                         release_tree_content_recursive(e->tree);
1172                                         e->tree = NULL;
1173                                 }
1174                                 hashclr(root->versions[1].sha1);
1175                                 return 1;
1176                         }
1177                         if (!S_ISDIR(e->versions[1].mode)) {
1178                                 e->tree = new_tree_content(8);
1179                                 e->versions[1].mode = S_IFDIR;
1180                         }
1181                         if (!e->tree)
1182                                 load_tree(e);
1183                         if (tree_content_set(e, slash1 + 1, sha1, mode)) {
1184                                 hashclr(root->versions[1].sha1);
1185                                 return 1;
1186                         }
1187                         return 0;
1188                 }
1189         }
1190
1191         if (t->entry_count == t->entry_capacity)
1192                 root->tree = t = grow_tree_content(t, 8);
1193         e = new_tree_entry();
1194         e->name = to_atom(p, n);
1195         e->versions[0].mode = 0;
1196         hashclr(e->versions[0].sha1);
1197         t->entries[t->entry_count++] = e;
1198         if (slash1) {
1199                 e->tree = new_tree_content(8);
1200                 e->versions[1].mode = S_IFDIR;
1201                 tree_content_set(e, slash1 + 1, sha1, mode);
1202         } else {
1203                 e->tree = NULL;
1204                 e->versions[1].mode = mode;
1205                 hashcpy(e->versions[1].sha1, sha1);
1206         }
1207         hashclr(root->versions[1].sha1);
1208         return 1;
1209 }
1210
1211 static int tree_content_remove(struct tree_entry *root, const char *p)
1212 {
1213         struct tree_content *t = root->tree;
1214         const char *slash1;
1215         unsigned int i, n;
1216         struct tree_entry *e;
1217
1218         slash1 = strchr(p, '/');
1219         if (slash1)
1220                 n = slash1 - p;
1221         else
1222                 n = strlen(p);
1223
1224         for (i = 0; i < t->entry_count; i++) {
1225                 e = t->entries[i];
1226                 if (e->name->str_len == n && !strncmp(p, e->name->str_dat, n)) {
1227                         if (!slash1 || !S_ISDIR(e->versions[1].mode))
1228                                 goto del_entry;
1229                         if (!e->tree)
1230                                 load_tree(e);
1231                         if (tree_content_remove(e, slash1 + 1)) {
1232                                 for (n = 0; n < e->tree->entry_count; n++) {
1233                                         if (e->tree->entries[n]->versions[1].mode) {
1234                                                 hashclr(root->versions[1].sha1);
1235                                                 return 1;
1236                                         }
1237                                 }
1238                                 goto del_entry;
1239                         }
1240                         return 0;
1241                 }
1242         }
1243         return 0;
1244
1245 del_entry:
1246         if (e->tree) {
1247                 release_tree_content_recursive(e->tree);
1248                 e->tree = NULL;
1249         }
1250         e->versions[1].mode = 0;
1251         hashclr(e->versions[1].sha1);
1252         hashclr(root->versions[1].sha1);
1253         return 1;
1254 }
1255
1256 static void dump_branches()
1257 {
1258         static const char *msg = "fast-import";
1259         unsigned int i;
1260         struct branch *b;
1261         struct ref_lock *lock;
1262
1263         for (i = 0; i < branch_table_sz; i++) {
1264                 for (b = branch_table[i]; b; b = b->table_next_branch) {
1265                         lock = lock_any_ref_for_update(b->name, NULL);
1266                         if (!lock || write_ref_sha1(lock, b->sha1, msg) < 0)
1267                                 die("Can't write %s", b->name);
1268                 }
1269         }
1270 }
1271
1272 static void dump_tags()
1273 {
1274         static const char *msg = "fast-import";
1275         struct tag *t;
1276         struct ref_lock *lock;
1277         char path[PATH_MAX];
1278
1279         for (t = first_tag; t; t = t->next_tag) {
1280                 sprintf(path, "refs/tags/%s", t->name);
1281                 lock = lock_any_ref_for_update(path, NULL);
1282                 if (!lock || write_ref_sha1(lock, t->sha1, msg) < 0)
1283                         die("Can't write %s", path);
1284         }
1285 }
1286
1287 static void dump_marks_helper(FILE *f,
1288         uintmax_t base,
1289         struct mark_set *m)
1290 {
1291         uintmax_t k;
1292         if (m->shift) {
1293                 for (k = 0; k < 1024; k++) {
1294                         if (m->data.sets[k])
1295                                 dump_marks_helper(f, (base + k) << m->shift,
1296                                         m->data.sets[k]);
1297                 }
1298         } else {
1299                 for (k = 0; k < 1024; k++) {
1300                         if (m->data.marked[k])
1301                                 fprintf(f, ":%ju %s\n", base + k,
1302                                         sha1_to_hex(m->data.marked[k]->sha1));
1303                 }
1304         }
1305 }
1306
1307 static void dump_marks()
1308 {
1309         if (mark_file)
1310         {
1311                 FILE *f = fopen(mark_file, "w");
1312                 dump_marks_helper(f, 0, marks);
1313                 fclose(f);
1314         }
1315 }
1316
1317 static void read_next_command()
1318 {
1319         read_line(&command_buf, stdin, '\n');
1320 }
1321
1322 static void cmd_mark()
1323 {
1324         if (!strncmp("mark :", command_buf.buf, 6)) {
1325                 next_mark = strtoumax(command_buf.buf + 6, NULL, 10);
1326                 read_next_command();
1327         }
1328         else
1329                 next_mark = 0;
1330 }
1331
1332 static void* cmd_data (size_t *size)
1333 {
1334         size_t n = 0;
1335         void *buffer;
1336         size_t length;
1337
1338         if (strncmp("data ", command_buf.buf, 5))
1339                 die("Expected 'data n' command, found: %s", command_buf.buf);
1340
1341         length = strtoul(command_buf.buf + 5, NULL, 10);
1342         buffer = xmalloc(length);
1343
1344         while (n < length) {
1345                 size_t s = fread((char*)buffer + n, 1, length - n, stdin);
1346                 if (!s && feof(stdin))
1347                         die("EOF in data (%lu bytes remaining)", length - n);
1348                 n += s;
1349         }
1350
1351         if (fgetc(stdin) != '\n')
1352                 die("An lf did not trail the binary data as expected.");
1353
1354         *size = length;
1355         return buffer;
1356 }
1357
1358 static void cmd_new_blob()
1359 {
1360         size_t l;
1361         void *d;
1362
1363         read_next_command();
1364         cmd_mark();
1365         d = cmd_data(&l);
1366
1367         if (store_object(OBJ_BLOB, d, l, &last_blob, NULL, next_mark))
1368                 free(d);
1369 }
1370
1371 static void unload_one_branch()
1372 {
1373         while (cur_active_branches
1374                 && cur_active_branches >= max_active_branches) {
1375                 unsigned long min_commit = ULONG_MAX;
1376                 struct branch *e, *l = NULL, *p = NULL;
1377
1378                 for (e = active_branches; e; e = e->active_next_branch) {
1379                         if (e->last_commit < min_commit) {
1380                                 p = l;
1381                                 min_commit = e->last_commit;
1382                         }
1383                         l = e;
1384                 }
1385
1386                 if (p) {
1387                         e = p->active_next_branch;
1388                         p->active_next_branch = e->active_next_branch;
1389                 } else {
1390                         e = active_branches;
1391                         active_branches = e->active_next_branch;
1392                 }
1393                 e->active_next_branch = NULL;
1394                 if (e->branch_tree.tree) {
1395                         release_tree_content_recursive(e->branch_tree.tree);
1396                         e->branch_tree.tree = NULL;
1397                 }
1398                 cur_active_branches--;
1399         }
1400 }
1401
1402 static void load_branch(struct branch *b)
1403 {
1404         load_tree(&b->branch_tree);
1405         b->active_next_branch = active_branches;
1406         active_branches = b;
1407         cur_active_branches++;
1408         branch_load_count++;
1409 }
1410
1411 static void file_change_m(struct branch *b)
1412 {
1413         const char *p = command_buf.buf + 2;
1414         char *p_uq;
1415         const char *endp;
1416         struct object_entry *oe;
1417         unsigned char sha1[20];
1418         unsigned int mode;
1419         char type[20];
1420
1421         p = get_mode(p, &mode);
1422         if (!p)
1423                 die("Corrupt mode: %s", command_buf.buf);
1424         switch (mode) {
1425         case S_IFREG | 0644:
1426         case S_IFREG | 0755:
1427         case S_IFLNK:
1428         case 0644:
1429         case 0755:
1430                 /* ok */
1431                 break;
1432         default:
1433                 die("Corrupt mode: %s", command_buf.buf);
1434         }
1435
1436         if (*p == ':') {
1437                 char *x;
1438                 oe = find_mark(strtoumax(p + 1, &x, 10));
1439                 hashcpy(sha1, oe->sha1);
1440                 p = x;
1441         } else {
1442                 if (get_sha1_hex(p, sha1))
1443                         die("Invalid SHA1: %s", command_buf.buf);
1444                 oe = find_object(sha1);
1445                 p += 40;
1446         }
1447         if (*p++ != ' ')
1448                 die("Missing space after SHA1: %s", command_buf.buf);
1449
1450         p_uq = unquote_c_style(p, &endp);
1451         if (p_uq) {
1452                 if (*endp)
1453                         die("Garbage after path in: %s", command_buf.buf);
1454                 p = p_uq;
1455         }
1456
1457         if (oe) {
1458                 if (oe->type != OBJ_BLOB)
1459                         die("Not a blob (actually a %s): %s",
1460                                 command_buf.buf, type_names[oe->type]);
1461         } else {
1462                 if (sha1_object_info(sha1, type, NULL))
1463                         die("Blob not found: %s", command_buf.buf);
1464                 if (strcmp(blob_type, type))
1465                         die("Not a blob (actually a %s): %s",
1466                                 command_buf.buf, type);
1467         }
1468
1469         tree_content_set(&b->branch_tree, p, sha1, S_IFREG | mode);
1470
1471         if (p_uq)
1472                 free(p_uq);
1473 }
1474
1475 static void file_change_d(struct branch *b)
1476 {
1477         const char *p = command_buf.buf + 2;
1478         char *p_uq;
1479         const char *endp;
1480
1481         p_uq = unquote_c_style(p, &endp);
1482         if (p_uq) {
1483                 if (*endp)
1484                         die("Garbage after path in: %s", command_buf.buf);
1485                 p = p_uq;
1486         }
1487         tree_content_remove(&b->branch_tree, p);
1488         if (p_uq)
1489                 free(p_uq);
1490 }
1491
1492 static void cmd_from(struct branch *b)
1493 {
1494         const char *from, *endp;
1495         char *str_uq;
1496         struct branch *s;
1497
1498         if (strncmp("from ", command_buf.buf, 5))
1499                 return;
1500
1501         if (b->last_commit)
1502                 die("Can't reinitailize branch %s", b->name);
1503
1504         from = strchr(command_buf.buf, ' ') + 1;
1505         str_uq = unquote_c_style(from, &endp);
1506         if (str_uq) {
1507                 if (*endp)
1508                         die("Garbage after string in: %s", command_buf.buf);
1509                 from = str_uq;
1510         }
1511
1512         s = lookup_branch(from);
1513         if (b == s)
1514                 die("Can't create a branch from itself: %s", b->name);
1515         else if (s) {
1516                 unsigned char *t = s->branch_tree.versions[1].sha1;
1517                 hashcpy(b->sha1, s->sha1);
1518                 hashcpy(b->branch_tree.versions[0].sha1, t);
1519                 hashcpy(b->branch_tree.versions[1].sha1, t);
1520         } else if (*from == ':') {
1521                 uintmax_t idnum = strtoumax(from + 1, NULL, 10);
1522                 struct object_entry *oe = find_mark(idnum);
1523                 unsigned long size;
1524                 char *buf;
1525                 if (oe->type != OBJ_COMMIT)
1526                         die("Mark :%ju not a commit", idnum);
1527                 hashcpy(b->sha1, oe->sha1);
1528                 buf = gfi_unpack_entry(oe, &size);
1529                 if (!buf || size < 46)
1530                         die("Not a valid commit: %s", from);
1531                 if (memcmp("tree ", buf, 5)
1532                         || get_sha1_hex(buf + 5, b->branch_tree.versions[1].sha1))
1533                         die("The commit %s is corrupt", sha1_to_hex(b->sha1));
1534                 free(buf);
1535                 hashcpy(b->branch_tree.versions[0].sha1,
1536                         b->branch_tree.versions[1].sha1);
1537         } else if (!get_sha1(from, b->sha1)) {
1538                 if (is_null_sha1(b->sha1)) {
1539                         hashclr(b->branch_tree.versions[0].sha1);
1540                         hashclr(b->branch_tree.versions[1].sha1);
1541                 } else {
1542                         unsigned long size;
1543                         char *buf;
1544
1545                         buf = read_object_with_reference(b->sha1,
1546                                 type_names[OBJ_COMMIT], &size, b->sha1);
1547                         if (!buf || size < 46)
1548                                 die("Not a valid commit: %s", from);
1549                         if (memcmp("tree ", buf, 5)
1550                                 || get_sha1_hex(buf + 5, b->branch_tree.versions[1].sha1))
1551                                 die("The commit %s is corrupt", sha1_to_hex(b->sha1));
1552                         free(buf);
1553                         hashcpy(b->branch_tree.versions[0].sha1,
1554                                 b->branch_tree.versions[1].sha1);
1555                 }
1556         } else
1557                 die("Invalid ref name or SHA1 expression: %s", from);
1558
1559         read_next_command();
1560 }
1561
1562 static struct hash_list* cmd_merge(unsigned int *count)
1563 {
1564         struct hash_list *list = NULL, *n, *e;
1565         const char *from, *endp;
1566         char *str_uq;
1567         struct branch *s;
1568
1569         *count = 0;
1570         while (!strncmp("merge ", command_buf.buf, 6)) {
1571                 from = strchr(command_buf.buf, ' ') + 1;
1572                 str_uq = unquote_c_style(from, &endp);
1573                 if (str_uq) {
1574                         if (*endp)
1575                                 die("Garbage after string in: %s", command_buf.buf);
1576                         from = str_uq;
1577                 }
1578
1579                 n = xmalloc(sizeof(*n));
1580                 s = lookup_branch(from);
1581                 if (s)
1582                         hashcpy(n->sha1, s->sha1);
1583                 else if (*from == ':') {
1584                         uintmax_t idnum = strtoumax(from + 1, NULL, 10);
1585                         struct object_entry *oe = find_mark(idnum);
1586                         if (oe->type != OBJ_COMMIT)
1587                                 die("Mark :%ju not a commit", idnum);
1588                         hashcpy(n->sha1, oe->sha1);
1589                 } else if (get_sha1(from, n->sha1))
1590                         die("Invalid ref name or SHA1 expression: %s", from);
1591
1592                 n->next = NULL;
1593                 if (list)
1594                         e->next = n;
1595                 else
1596                         list = n;
1597                 e = n;
1598                 *count++;
1599                 read_next_command();
1600         }
1601         return list;
1602 }
1603
1604 static void cmd_new_commit()
1605 {
1606         struct branch *b;
1607         void *msg;
1608         size_t msglen;
1609         char *str_uq;
1610         const char *endp;
1611         char *sp;
1612         char *author = NULL;
1613         char *committer = NULL;
1614         struct hash_list *merge_list = NULL;
1615         unsigned int merge_count;
1616
1617         /* Obtain the branch name from the rest of our command */
1618         sp = strchr(command_buf.buf, ' ') + 1;
1619         str_uq = unquote_c_style(sp, &endp);
1620         if (str_uq) {
1621                 if (*endp)
1622                         die("Garbage after ref in: %s", command_buf.buf);
1623                 sp = str_uq;
1624         }
1625         b = lookup_branch(sp);
1626         if (!b)
1627                 b = new_branch(sp);
1628         if (str_uq)
1629                 free(str_uq);
1630
1631         read_next_command();
1632         cmd_mark();
1633         if (!strncmp("author ", command_buf.buf, 7)) {
1634                 author = strdup(command_buf.buf);
1635                 read_next_command();
1636         }
1637         if (!strncmp("committer ", command_buf.buf, 10)) {
1638                 committer = strdup(command_buf.buf);
1639                 read_next_command();
1640         }
1641         if (!committer)
1642                 die("Expected committer but didn't get one");
1643         msg = cmd_data(&msglen);
1644         read_next_command();
1645         cmd_from(b);
1646         merge_list = cmd_merge(&merge_count);
1647
1648         /* ensure the branch is active/loaded */
1649         if (!b->branch_tree.tree || !max_active_branches) {
1650                 unload_one_branch();
1651                 load_branch(b);
1652         }
1653
1654         /* file_change* */
1655         for (;;) {
1656                 if (1 == command_buf.len)
1657                         break;
1658                 else if (!strncmp("M ", command_buf.buf, 2))
1659                         file_change_m(b);
1660                 else if (!strncmp("D ", command_buf.buf, 2))
1661                         file_change_d(b);
1662                 else
1663                         die("Unsupported file_change: %s", command_buf.buf);
1664                 read_next_command();
1665         }
1666
1667         /* build the tree and the commit */
1668         store_tree(&b->branch_tree);
1669         hashcpy(b->branch_tree.versions[0].sha1,
1670                 b->branch_tree.versions[1].sha1);
1671         size_dbuf(&new_data, 97 + msglen
1672                 + merge_count * 49
1673                 + (author
1674                         ? strlen(author) + strlen(committer)
1675                         : 2 * strlen(committer)));
1676         sp = new_data.buffer;
1677         sp += sprintf(sp, "tree %s\n",
1678                 sha1_to_hex(b->branch_tree.versions[1].sha1));
1679         if (!is_null_sha1(b->sha1))
1680                 sp += sprintf(sp, "parent %s\n", sha1_to_hex(b->sha1));
1681         while (merge_list) {
1682                 struct hash_list *next = merge_list->next;
1683                 sp += sprintf(sp, "parent %s\n", sha1_to_hex(merge_list->sha1));
1684                 free(merge_list);
1685                 merge_list = next;
1686         }
1687         if (author)
1688                 sp += sprintf(sp, "%s\n", author);
1689         else
1690                 sp += sprintf(sp, "author %s\n", committer + 10);
1691         sp += sprintf(sp, "%s\n\n", committer);
1692         memcpy(sp, msg, msglen);
1693         sp += msglen;
1694         if (author)
1695                 free(author);
1696         free(committer);
1697         free(msg);
1698
1699         store_object(OBJ_COMMIT,
1700                 new_data.buffer, sp - (char*)new_data.buffer,
1701                 NULL, b->sha1, next_mark);
1702         b->last_commit = object_count_by_type[OBJ_COMMIT];
1703         b->pack_id = pack_id;
1704
1705         if (branch_log) {
1706                 int need_dq = quote_c_style(b->name, NULL, NULL, 0);
1707                 fprintf(branch_log, "commit ");
1708                 if (need_dq) {
1709                         fputc('"', branch_log);
1710                         quote_c_style(b->name, NULL, branch_log, 0);
1711                         fputc('"', branch_log);
1712                 } else
1713                         fprintf(branch_log, "%s", b->name);
1714                 fprintf(branch_log," :%ju %s\n",next_mark,sha1_to_hex(b->sha1));
1715         }
1716 }
1717
1718 static void cmd_new_tag()
1719 {
1720         char *str_uq;
1721         const char *endp;
1722         char *sp;
1723         const char *from;
1724         char *tagger;
1725         struct branch *s;
1726         void *msg;
1727         size_t msglen;
1728         struct tag *t;
1729         uintmax_t from_mark = 0;
1730         unsigned char sha1[20];
1731
1732         /* Obtain the new tag name from the rest of our command */
1733         sp = strchr(command_buf.buf, ' ') + 1;
1734         str_uq = unquote_c_style(sp, &endp);
1735         if (str_uq) {
1736                 if (*endp)
1737                         die("Garbage after tag name in: %s", command_buf.buf);
1738                 sp = str_uq;
1739         }
1740         t = pool_alloc(sizeof(struct tag));
1741         t->next_tag = NULL;
1742         t->name = pool_strdup(sp);
1743         if (last_tag)
1744                 last_tag->next_tag = t;
1745         else
1746                 first_tag = t;
1747         last_tag = t;
1748         if (str_uq)
1749                 free(str_uq);
1750         read_next_command();
1751
1752         /* from ... */
1753         if (strncmp("from ", command_buf.buf, 5))
1754                 die("Expected from command, got %s", command_buf.buf);
1755
1756         from = strchr(command_buf.buf, ' ') + 1;
1757         str_uq = unquote_c_style(from, &endp);
1758         if (str_uq) {
1759                 if (*endp)
1760                         die("Garbage after string in: %s", command_buf.buf);
1761                 from = str_uq;
1762         }
1763
1764         s = lookup_branch(from);
1765         if (s) {
1766                 hashcpy(sha1, s->sha1);
1767         } else if (*from == ':') {
1768                 from_mark = strtoumax(from + 1, NULL, 10);
1769                 struct object_entry *oe = find_mark(from_mark);
1770                 if (oe->type != OBJ_COMMIT)
1771                         die("Mark :%ju not a commit", from_mark);
1772                 hashcpy(sha1, oe->sha1);
1773         } else if (!get_sha1(from, sha1)) {
1774                 unsigned long size;
1775                 char *buf;
1776
1777                 buf = read_object_with_reference(sha1,
1778                         type_names[OBJ_COMMIT], &size, sha1);
1779                 if (!buf || size < 46)
1780                         die("Not a valid commit: %s", from);
1781                 free(buf);
1782         } else
1783                 die("Invalid ref name or SHA1 expression: %s", from);
1784
1785         if (str_uq)
1786                 free(str_uq);
1787         read_next_command();
1788
1789         /* tagger ... */
1790         if (strncmp("tagger ", command_buf.buf, 7))
1791                 die("Expected tagger command, got %s", command_buf.buf);
1792         tagger = strdup(command_buf.buf);
1793
1794         /* tag payload/message */
1795         read_next_command();
1796         msg = cmd_data(&msglen);
1797
1798         /* build the tag object */
1799         size_dbuf(&new_data, 67+strlen(t->name)+strlen(tagger)+msglen);
1800         sp = new_data.buffer;
1801         sp += sprintf(sp, "object %s\n", sha1_to_hex(sha1));
1802         sp += sprintf(sp, "type %s\n", type_names[OBJ_COMMIT]);
1803         sp += sprintf(sp, "tag %s\n", t->name);
1804         sp += sprintf(sp, "%s\n\n", tagger);
1805         memcpy(sp, msg, msglen);
1806         sp += msglen;
1807         free(tagger);
1808         free(msg);
1809
1810         store_object(OBJ_TAG, new_data.buffer, sp - (char*)new_data.buffer,
1811                 NULL, t->sha1, 0);
1812         t->pack_id = pack_id;
1813
1814         if (branch_log) {
1815                 int need_dq = quote_c_style(t->name, NULL, NULL, 0);
1816                 fprintf(branch_log, "tag ");
1817                 if (need_dq) {
1818                         fputc('"', branch_log);
1819                         quote_c_style(t->name, NULL, branch_log, 0);
1820                         fputc('"', branch_log);
1821                 } else
1822                         fprintf(branch_log, "%s", t->name);
1823                 fprintf(branch_log," :%ju %s\n",from_mark,sha1_to_hex(t->sha1));
1824         }
1825 }
1826
1827 static void cmd_reset_branch()
1828 {
1829         struct branch *b;
1830         char *str_uq;
1831         const char *endp;
1832         char *sp;
1833
1834         /* Obtain the branch name from the rest of our command */
1835         sp = strchr(command_buf.buf, ' ') + 1;
1836         str_uq = unquote_c_style(sp, &endp);
1837         if (str_uq) {
1838                 if (*endp)
1839                         die("Garbage after ref in: %s", command_buf.buf);
1840                 sp = str_uq;
1841         }
1842         b = lookup_branch(sp);
1843         if (b) {
1844                 b->last_commit = 0;
1845                 if (b->branch_tree.tree) {
1846                         release_tree_content_recursive(b->branch_tree.tree);
1847                         b->branch_tree.tree = NULL;
1848                 }
1849         }
1850         else
1851                 b = new_branch(sp);
1852         if (str_uq)
1853                 free(str_uq);
1854         read_next_command();
1855         cmd_from(b);
1856 }
1857
1858 static void cmd_checkpoint()
1859 {
1860         if (object_count)
1861                 checkpoint();
1862         read_next_command();
1863 }
1864
1865 static const char fast_import_usage[] =
1866 "git-fast-import [--objects=n] [--depth=n] [--active-branches=n] [--export-marks=marks.file] [--branch-log=log]";
1867
1868 int main(int argc, const char **argv)
1869 {
1870         int i;
1871         uintmax_t est_obj_cnt = object_entry_alloc;
1872         uintmax_t total_count, duplicate_count;
1873
1874         setup_ident();
1875         git_config(git_default_config);
1876
1877         for (i = 1; i < argc; i++) {
1878                 const char *a = argv[i];
1879
1880                 if (*a != '-' || !strcmp(a, "--"))
1881                         break;
1882                 else if (!strncmp(a, "--objects=", 10))
1883                         est_obj_cnt = strtoumax(a + 10, NULL, 0);
1884                 else if (!strncmp(a, "--max-objects-per-pack=", 23))
1885                         max_objects = strtoumax(a + 23, NULL, 0);
1886                 else if (!strncmp(a, "--max-pack-size=", 16))
1887                         max_packsize = strtoumax(a + 16, NULL, 0) * 1024 * 1024;
1888                 else if (!strncmp(a, "--depth=", 8))
1889                         max_depth = strtoul(a + 8, NULL, 0);
1890                 else if (!strncmp(a, "--active-branches=", 18))
1891                         max_active_branches = strtoul(a + 18, NULL, 0);
1892                 else if (!strncmp(a, "--export-marks=", 15))
1893                         mark_file = a + 15;
1894                 else if (!strncmp(a, "--branch-log=", 13)) {
1895                         branch_log = fopen(a + 13, "w");
1896                         if (!branch_log)
1897                                 die("Can't create %s: %s", a + 13, strerror(errno));
1898                 }
1899                 else
1900                         die("unknown option %s", a);
1901         }
1902         if (i != argc)
1903                 usage(fast_import_usage);
1904
1905         alloc_objects(est_obj_cnt);
1906         strbuf_init(&command_buf);
1907
1908         atom_table = xcalloc(atom_table_sz, sizeof(struct atom_str*));
1909         branch_table = xcalloc(branch_table_sz, sizeof(struct branch*));
1910         avail_tree_table = xcalloc(avail_tree_table_sz, sizeof(struct avail_tree_content*));
1911         marks = pool_calloc(1, sizeof(struct mark_set));
1912
1913         start_packfile();
1914         for (;;) {
1915                 read_next_command();
1916                 if (command_buf.eof)
1917                         break;
1918                 else if (!strcmp("blob", command_buf.buf))
1919                         cmd_new_blob();
1920                 else if (!strncmp("commit ", command_buf.buf, 7))
1921                         cmd_new_commit();
1922                 else if (!strncmp("tag ", command_buf.buf, 4))
1923                         cmd_new_tag();
1924                 else if (!strncmp("reset ", command_buf.buf, 6))
1925                         cmd_reset_branch();
1926                 else if (!strcmp("checkpoint", command_buf.buf))
1927                         cmd_checkpoint();
1928                 else
1929                         die("Unsupported command: %s", command_buf.buf);
1930         }
1931         end_packfile();
1932
1933         dump_branches();
1934         dump_tags();
1935         unkeep_all_packs();
1936         dump_marks();
1937         if (branch_log)
1938                 fclose(branch_log);
1939
1940         total_count = 0;
1941         for (i = 0; i < ARRAY_SIZE(object_count_by_type); i++)
1942                 total_count += object_count_by_type[i];
1943         duplicate_count = 0;
1944         for (i = 0; i < ARRAY_SIZE(duplicate_count_by_type); i++)
1945                 duplicate_count += duplicate_count_by_type[i];
1946
1947         fprintf(stderr, "%s statistics:\n", argv[0]);
1948         fprintf(stderr, "---------------------------------------------------------------------\n");
1949         fprintf(stderr, "Alloc'd objects: %10ju (%10ju overflow  )\n", alloc_count, alloc_count - est_obj_cnt);
1950         fprintf(stderr, "Total objects:   %10ju (%10ju duplicates                  )\n", total_count, duplicate_count);
1951         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]);
1952         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]);
1953         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]);
1954         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]);
1955         fprintf(stderr, "Total branches:  %10lu (%10lu loads     )\n", branch_count, branch_load_count);
1956         fprintf(stderr, "      marks:     %10ju (%10ju unique    )\n", (((uintmax_t)1) << marks->shift) * 1024, marks_set_count);
1957         fprintf(stderr, "      atoms:     %10u\n", atom_cnt);
1958         fprintf(stderr, "Memory total:    %10ju KiB\n", (total_allocd + alloc_count*sizeof(struct object_entry))/1024);
1959         fprintf(stderr, "       pools:    %10lu KiB\n", total_allocd/1024);
1960         fprintf(stderr, "     objects:    %10ju KiB\n", (alloc_count*sizeof(struct object_entry))/1024);
1961         fprintf(stderr, "---------------------------------------------------------------------\n");
1962         pack_report();
1963         fprintf(stderr, "---------------------------------------------------------------------\n");
1964         fprintf(stderr, "\n");
1965
1966         return 0;
1967 }