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 '>' when lf)?
  21     'committer' sp name '<' email '>' when 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_clr
  30     | file_del
  31     | file_rnm
  32     | file_cpy
  33     | file_obm
  34     | file_inm;
  35   file_clr ::= 'deleteall' lf;
  36   file_del ::= 'D' sp path_str lf;
  37   file_rnm ::= 'R' sp path_str sp path_str lf;
  38   file_cpy ::= 'C' sp path_str sp path_str lf;
  39   file_obm ::= 'M' sp mode sp (hexsha1 | idnum) sp path_str lf;
  40   file_inm ::= 'M' sp mode sp 'inline' sp path_str lf
  41     data;
  42
  43   new_tag ::= 'tag' sp tag_str lf
  44     'from' sp (ref_str | hexsha1 | sha1exp_str | idnum) lf
  45     'tagger' sp name '<' email '>' when lf
  46     tag_msg;
  47   tag_msg ::= data;
  48
  49   reset_branch ::= 'reset' sp ref_str lf
  50     ('from' sp (ref_str | hexsha1 | sha1exp_str | idnum) lf)?
  51     lf?;
  52
  53   checkpoint ::= 'checkpoint' lf
  54     lf?;
  55
  56      # note: the first idnum in a stream should be 1 and subsequent
  57      # idnums should not have gaps between values as this will cause
  58      # the stream parser to reserve space for the gapped values.  An
  59      # idnum can be updated in the future to a new object by issuing
  60      # a new mark directive with the old idnum.
  61      #
  62   mark ::= 'mark' sp idnum lf;
  63   data ::= (delimited_data | exact_data)
  64     lf?;
  65
  66     # note: delim may be any string but must not contain lf.
  67     # data_line may contain any data but must not be exactly
  68     # delim.
  69   delimited_data ::= 'data' sp '<<' delim lf
  70     (data_line lf)*
  71     delim lf;
  72
  73      # note: declen indicates the length of binary_data in bytes.
  74      # declen does not include the lf preceeding the binary data.
  75      #
  76   exact_data ::= 'data' sp declen lf
  77     binary_data;
  78
  79      # note: quoted strings are C-style quoting supporting \c for
  80      # common escapes of 'c' (e..g \n, \t, \\, \") or \nnn where nnn
  81      # is the signed byte value in octal.  Note that the only
  82      # characters which must actually be escaped to protect the
  83      # stream formatting is: \, " and LF.  Otherwise these values
  84      # are UTF8.
  85      #
  86   ref_str     ::= ref;
  87   sha1exp_str ::= sha1exp;
  88   tag_str     ::= tag;
  89   path_str    ::= path    | '"' quoted(path)    '"' ;
  90   mode        ::= '100644' | '644'
  91                 | '100755' | '755'
  92                 | '120000'
  93                 ;
  94
  95   declen ::= # unsigned 32 bit value, ascii base10 notation;
  96   bigint ::= # unsigned integer value, ascii base10 notation;
  97   binary_data ::= # file content, not interpreted;
  98
  99   when         ::= raw_when | rfc2822_when;
 100   raw_when     ::= ts sp tz;
 101   rfc2822_when ::= # Valid RFC 2822 date and time;
 102
 103   sp ::= # ASCII space character;
 104   lf ::= # ASCII newline (LF) character;
 105
 106      # note: a colon (':') must precede the numerical value assigned to
 107      # an idnum.  This is to distinguish it from a ref or tag name as
 108      # GIT does not permit ':' in ref or tag strings.
 109      #
 110   idnum   ::= ':' bigint;
 111   path    ::= # GIT style file path, e.g. "a/b/c";
 112   ref     ::= # GIT ref name, e.g. "refs/heads/MOZ_GECKO_EXPERIMENT";
 113   tag     ::= # GIT tag name, e.g. "FIREFOX_1_5";
 114   sha1exp ::= # Any valid GIT SHA1 expression;
 115   hexsha1 ::= # SHA1 in hexadecimal format;
 116
 117      # note: name and email are UTF8 strings, however name must not
 118      # contain '<' or lf and email must not contain any of the
 119      # following: '<', '>', lf.
 120      #
 121   name  ::= # valid GIT author/committer name;
 122   email ::= # valid GIT author/committer email;
 123   ts    ::= # time since the epoch in seconds, ascii base10 notation;
 124   tz    ::= # GIT style timezone;
 125
 126      # note: comments may appear anywhere in the input, except
 127      # within a data command.  Any form of the data command
 128      # always escapes the related input from comment processing.
 129      #
 130      # In case it is not clear, the '#' that starts the comment
 131      # must be the first character on that the line (an lf have
 132      # preceeded it).
 133      #
 134   comment ::= '#' not_lf* lf;
 135   not_lf  ::= # Any byte that is not ASCII newline (LF);
 136 */
 137
 138 #include "builtin.h"
 139 #include "cache.h"
 140 #include "object.h"
 141 #include "blob.h"
 142 #include "tree.h"
 143 #include "commit.h"
 144 #include "delta.h"
 145 #include "pack.h"
 146 #include "refs.h"
 147 #include "csum-file.h"
 148 #include "strbuf.h"
 149 #include "quote.h"
 150
 151 #define PACK_ID_BITS 16
 152 #define MAX_PACK_ID ((1<<PACK_ID_BITS)-1)
 153
 154 struct object_entry
 155 {
 156         struct object_entry *next;
 157         uint32_t offset;
 158         unsigned type : TYPE_BITS;
 159         unsigned pack_id : PACK_ID_BITS;
 160         unsigned char sha1[20];
 161 };
 162
 163 struct object_entry_pool
 164 {
 165         struct object_entry_pool *next_pool;
 166         struct object_entry *next_free;
 167         struct object_entry *end;
 168         struct object_entry entries[FLEX_ARRAY]; /* more */
 169 };
 170
 171 struct mark_set
 172 {
 173         union {
 174                 struct object_entry *marked[1024];
 175                 struct mark_set *sets[1024];
 176         } data;
 177         unsigned int shift;
 178 };
 179
 180 struct last_object
 181 {
 182         void *data;
 183         unsigned long len;
 184         uint32_t offset;
 185         unsigned int depth;
 186         unsigned no_free:1;
 187 };
 188
 189 struct mem_pool
 190 {
 191         struct mem_pool *next_pool;
 192         char *next_free;
 193         char *end;
 194         char space[FLEX_ARRAY]; /* more */
 195 };
 196
 197 struct atom_str
 198 {
 199         struct atom_str *next_atom;
 200         unsigned short str_len;
 201         char str_dat[FLEX_ARRAY]; /* more */
 202 };
 203
 204 struct tree_content;
 205 struct tree_entry
 206 {
 207         struct tree_content *tree;
 208         struct atom_str* name;
 209         struct tree_entry_ms
 210         {
 211                 uint16_t mode;
 212                 unsigned char sha1[20];
 213         } versions[2];
 214 };
 215
 216 struct tree_content
 217 {
 218         unsigned int entry_capacity; /* must match avail_tree_content */
 219         unsigned int entry_count;
 220         unsigned int delta_depth;
 221         struct tree_entry *entries[FLEX_ARRAY]; /* more */
 222 };
 223
 224 struct avail_tree_content
 225 {
 226         unsigned int entry_capacity; /* must match tree_content */
 227         struct avail_tree_content *next_avail;
 228 };
 229
 230 struct branch
 231 {
 232         struct branch *table_next_branch;
 233         struct branch *active_next_branch;
 234         const char *name;
 235         struct tree_entry branch_tree;
 236         uintmax_t last_commit;
 237         unsigned active : 1;
 238         unsigned pack_id : PACK_ID_BITS;
 239         unsigned char sha1[20];
 240 };
 241
 242 struct tag
 243 {
 244         struct tag *next_tag;
 245         const char *name;
 246         unsigned int pack_id;
 247         unsigned char sha1[20];
 248 };
 249
 250 struct dbuf
 251 {
 252         void *buffer;
 253         size_t capacity;
 254 };
 255
 256 struct hash_list
 257 {
 258         struct hash_list *next;
 259         unsigned char sha1[20];
 260 };
 261
 262 typedef enum {
 263         WHENSPEC_RAW = 1,
 264         WHENSPEC_RFC2822,
 265         WHENSPEC_NOW,
 266 } whenspec_type;
 267
 268 /* Configured limits on output */
 269 static unsigned long max_depth = 10;
 270 static off_t max_packsize = (1LL << 32) - 1;
 271 static int force_update;
 272
 273 /* Stats and misc. counters */
 274 static uintmax_t alloc_count;
 275 static uintmax_t marks_set_count;
 276 static uintmax_t object_count_by_type[1 << TYPE_BITS];
 277 static uintmax_t duplicate_count_by_type[1 << TYPE_BITS];
 278 static uintmax_t delta_count_by_type[1 << TYPE_BITS];
 279 static unsigned long object_count;
 280 static unsigned long branch_count;
 281 static unsigned long branch_load_count;
 282 static int failure;
 283 static FILE *pack_edges;
 284
 285 /* Memory pools */
 286 static size_t mem_pool_alloc = 2*1024*1024 - sizeof(struct mem_pool);
 287 static size_t total_allocd;
 288 static struct mem_pool *mem_pool;
 289
 290 /* Atom management */
 291 static unsigned int atom_table_sz = 4451;
 292 static unsigned int atom_cnt;
 293 static struct atom_str **atom_table;
 294
 295 /* The .pack file being generated */
 296 static unsigned int pack_id;
 297 static struct packed_git *pack_data;
 298 static struct packed_git **all_packs;
 299 static unsigned long pack_size;
 300
 301 /* Table of objects we've written. */
 302 static unsigned int object_entry_alloc = 5000;
 303 static struct object_entry_pool *blocks;
 304 static struct object_entry *object_table[1 << 16];
 305 static struct mark_set *marks;
 306 static const char* mark_file;
 307
 308 /* Our last blob */
 309 static struct last_object last_blob;
 310
 311 /* Tree management */
 312 static unsigned int tree_entry_alloc = 1000;
 313 static void *avail_tree_entry;
 314 static unsigned int avail_tree_table_sz = 100;
 315 static struct avail_tree_content **avail_tree_table;
 316 static struct dbuf old_tree;
 317 static struct dbuf new_tree;
 318
 319 /* Branch data */
 320 static unsigned long max_active_branches = 5;
 321 static unsigned long cur_active_branches;
 322 static unsigned long branch_table_sz = 1039;
 323 static struct branch **branch_table;
 324 static struct branch *active_branches;
 325
 326 /* Tag data */
 327 static struct tag *first_tag;
 328 static struct tag *last_tag;
 329
 330 /* Input stream parsing */
 331 static whenspec_type whenspec = WHENSPEC_RAW;
 332 static struct strbuf command_buf;
 333 static int unread_command_buf;
 334 static uintmax_t next_mark;
 335 static struct dbuf new_data;
 336
 337
 338 static void alloc_objects(unsigned int cnt)
 339 {
 340         struct object_entry_pool *b;
 341
 342         b = xmalloc(sizeof(struct object_entry_pool)
 343                 + cnt * sizeof(struct object_entry));
 344         b->next_pool = blocks;
 345         b->next_free = b->entries;
 346         b->end = b->entries + cnt;
 347         blocks = b;
 348         alloc_count += cnt;
 349 }
 350
 351 static struct object_entry *new_object(unsigned char *sha1)
 352 {
 353         struct object_entry *e;
 354
 355         if (blocks->next_free == blocks->end)
 356                 alloc_objects(object_entry_alloc);
 357
 358         e = blocks->next_free++;
 359         hashcpy(e->sha1, sha1);
 360         return e;
 361 }
 362
 363 static struct object_entry *find_object(unsigned char *sha1)
 364 {
 365         unsigned int h = sha1[0] << 8 | sha1[1];
 366         struct object_entry *e;
 367         for (e = object_table[h]; e; e = e->next)
 368                 if (!hashcmp(sha1, e->sha1))
 369                         return e;
 370         return NULL;
 371 }
 372
 373 static struct object_entry *insert_object(unsigned char *sha1)
 374 {
 375         unsigned int h = sha1[0] << 8 | sha1[1];
 376         struct object_entry *e = object_table[h];
 377         struct object_entry *p = NULL;
 378
 379         while (e) {
 380                 if (!hashcmp(sha1, e->sha1))
 381                         return e;
 382                 p = e;
 383                 e = e->next;
 384         }
 385
 386         e = new_object(sha1);
 387         e->next = NULL;
 388         e->offset = 0;
 389         if (p)
 390                 p->next = e;
 391         else
 392                 object_table[h] = e;
 393         return e;
 394 }
 395
 396 static unsigned int hc_str(const char *s, size_t len)
 397 {
 398         unsigned int r = 0;
 399         while (len-- > 0)
 400                 r = r * 31 + *s++;
 401         return r;
 402 }
 403
 404 static void *pool_alloc(size_t len)
 405 {
 406         struct mem_pool *p;
 407         void *r;
 408
 409         for (p = mem_pool; p; p = p->next_pool)
 410                 if ((p->end - p->next_free >= len))
 411                         break;
 412
 413         if (!p) {
 414                 if (len >= (mem_pool_alloc/2)) {
 415                         total_allocd += len;
 416                         return xmalloc(len);
 417                 }
 418                 total_allocd += sizeof(struct mem_pool) + mem_pool_alloc;
 419                 p = xmalloc(sizeof(struct mem_pool) + mem_pool_alloc);
 420                 p->next_pool = mem_pool;
 421                 p->next_free = p->space;
 422                 p->end = p->next_free + mem_pool_alloc;
 423                 mem_pool = p;
 424         }
 425
 426         r = p->next_free;
 427         /* round out to a pointer alignment */
 428         if (len & (sizeof(void*) - 1))
 429                 len += sizeof(void*) - (len & (sizeof(void*) - 1));
 430         p->next_free += len;
 431         return r;
 432 }
 433
 434 static void *pool_calloc(size_t count, size_t size)
 435 {
 436         size_t len = count * size;
 437         void *r = pool_alloc(len);
 438         memset(r, 0, len);
 439         return r;
 440 }
 441
 442 static char *pool_strdup(const char *s)
 443 {
 444         char *r = pool_alloc(strlen(s) + 1);
 445         strcpy(r, s);
 446         return r;
 447 }
 448
 449 static void size_dbuf(struct dbuf *b, size_t maxlen)
 450 {
 451         if (b->buffer) {
 452                 if (b->capacity >= maxlen)
 453                         return;
 454                 free(b->buffer);
 455         }
 456         b->capacity = ((maxlen / 1024) + 1) * 1024;
 457         b->buffer = xmalloc(b->capacity);
 458 }
 459
 460 static void insert_mark(uintmax_t idnum, struct object_entry *oe)
 461 {
 462         struct mark_set *s = marks;
 463         while ((idnum >> s->shift) >= 1024) {
 464                 s = pool_calloc(1, sizeof(struct mark_set));
 465                 s->shift = marks->shift + 10;
 466                 s->data.sets[0] = marks;
 467                 marks = s;
 468         }
 469         while (s->shift) {
 470                 uintmax_t i = idnum >> s->shift;
 471                 idnum -= i << s->shift;
 472                 if (!s->data.sets[i]) {
 473                         s->data.sets[i] = pool_calloc(1, sizeof(struct mark_set));
 474                         s->data.sets[i]->shift = s->shift - 10;
 475                 }
 476                 s = s->data.sets[i];
 477         }
 478         if (!s->data.marked[idnum])
 479                 marks_set_count++;
 480         s->data.marked[idnum] = oe;
 481 }
 482
 483 static struct object_entry *find_mark(uintmax_t idnum)
 484 {
 485         uintmax_t orig_idnum = idnum;
 486         struct mark_set *s = marks;
 487         struct object_entry *oe = NULL;
 488         if ((idnum >> s->shift) < 1024) {
 489                 while (s && s->shift) {
 490                         uintmax_t i = idnum >> s->shift;
 491                         idnum -= i << s->shift;
 492                         s = s->data.sets[i];
 493                 }
 494                 if (s)
 495                         oe = s->data.marked[idnum];
 496         }
 497         if (!oe)
 498                 die("mark :%" PRIuMAX " not declared", orig_idnum);
 499         return oe;
 500 }
 501
 502 static struct atom_str *to_atom(const char *s, unsigned short len)
 503 {
 504         unsigned int hc = hc_str(s, len) % atom_table_sz;
 505         struct atom_str *c;
 506
 507         for (c = atom_table[hc]; c; c = c->next_atom)
 508                 if (c->str_len == len && !strncmp(s, c->str_dat, len))
 509                         return c;
 510
 511         c = pool_alloc(sizeof(struct atom_str) + len + 1);
 512         c->str_len = len;
 513         strncpy(c->str_dat, s, len);
 514         c->str_dat[len] = 0;
 515         c->next_atom = atom_table[hc];
 516         atom_table[hc] = c;
 517         atom_cnt++;
 518         return c;
 519 }
 520
 521 static struct branch *lookup_branch(const char *name)
 522 {
 523         unsigned int hc = hc_str(name, strlen(name)) % branch_table_sz;
 524         struct branch *b;
 525
 526         for (b = branch_table[hc]; b; b = b->table_next_branch)
 527                 if (!strcmp(name, b->name))
 528                         return b;
 529         return NULL;
 530 }
 531
 532 static struct branch *new_branch(const char *name)
 533 {
 534         unsigned int hc = hc_str(name, strlen(name)) % branch_table_sz;
 535         struct branch* b = lookup_branch(name);
 536
 537         if (b)
 538                 die("Invalid attempt to create duplicate branch: %s", name);
 539         switch (check_ref_format(name)) {
 540         case  0: break; /* its valid */
 541         case -2: break; /* valid, but too few '/', allow anyway */
 542         default:
 543                 die("Branch name doesn't conform to GIT standards: %s", name);
 544         }
 545
 546         b = pool_calloc(1, sizeof(struct branch));
 547         b->name = pool_strdup(name);
 548         b->table_next_branch = branch_table[hc];
 549         b->branch_tree.versions[0].mode = S_IFDIR;
 550         b->branch_tree.versions[1].mode = S_IFDIR;
 551         b->active = 0;
 552         b->pack_id = MAX_PACK_ID;
 553         branch_table[hc] = b;
 554         branch_count++;
 555         return b;
 556 }
 557
 558 static unsigned int hc_entries(unsigned int cnt)
 559 {
 560         cnt = cnt & 7 ? (cnt / 8) + 1 : cnt / 8;
 561         return cnt < avail_tree_table_sz ? cnt : avail_tree_table_sz - 1;
 562 }
 563
 564 static struct tree_content *new_tree_content(unsigned int cnt)
 565 {
 566         struct avail_tree_content *f, *l = NULL;
 567         struct tree_content *t;
 568         unsigned int hc = hc_entries(cnt);
 569
 570         for (f = avail_tree_table[hc]; f; l = f, f = f->next_avail)
 571                 if (f->entry_capacity >= cnt)
 572                         break;
 573
 574         if (f) {
 575                 if (l)
 576                         l->next_avail = f->next_avail;
 577                 else
 578                         avail_tree_table[hc] = f->next_avail;
 579         } else {
 580                 cnt = cnt & 7 ? ((cnt / 8) + 1) * 8 : cnt;
 581                 f = pool_alloc(sizeof(*t) + sizeof(t->entries[0]) * cnt);
 582                 f->entry_capacity = cnt;
 583         }
 584
 585         t = (struct tree_content*)f;
 586         t->entry_count = 0;
 587         t->delta_depth = 0;
 588         return t;
 589 }
 590
 591 static void release_tree_entry(struct tree_entry *e);
 592 static void release_tree_content(struct tree_content *t)
 593 {
 594         struct avail_tree_content *f = (struct avail_tree_content*)t;
 595         unsigned int hc = hc_entries(f->entry_capacity);
 596         f->next_avail = avail_tree_table[hc];
 597         avail_tree_table[hc] = f;
 598 }
 599
 600 static void release_tree_content_recursive(struct tree_content *t)
 601 {
 602         unsigned int i;
 603         for (i = 0; i < t->entry_count; i++)
 604                 release_tree_entry(t->entries[i]);
 605         release_tree_content(t);
 606 }
 607
 608 static struct tree_content *grow_tree_content(
 609         struct tree_content *t,
 610         int amt)
 611 {
 612         struct tree_content *r = new_tree_content(t->entry_count + amt);
 613         r->entry_count = t->entry_count;
 614         r->delta_depth = t->delta_depth;
 615         memcpy(r->entries,t->entries,t->entry_count*sizeof(t->entries[0]));
 616         release_tree_content(t);
 617         return r;
 618 }
 619
 620 static struct tree_entry *new_tree_entry(void)
 621 {
 622         struct tree_entry *e;
 623
 624         if (!avail_tree_entry) {
 625                 unsigned int n = tree_entry_alloc;
 626                 total_allocd += n * sizeof(struct tree_entry);
 627                 avail_tree_entry = e = xmalloc(n * sizeof(struct tree_entry));
 628                 while (n-- > 1) {
 629                         *((void**)e) = e + 1;
 630                         e++;
 631                 }
 632                 *((void**)e) = NULL;
 633         }
 634
 635         e = avail_tree_entry;
 636         avail_tree_entry = *((void**)e);
 637         return e;
 638 }
 639
 640 static void release_tree_entry(struct tree_entry *e)
 641 {
 642         if (e->tree)
 643                 release_tree_content_recursive(e->tree);
 644         *((void**)e) = avail_tree_entry;
 645         avail_tree_entry = e;
 646 }
 647
 648 static struct tree_content *dup_tree_content(struct tree_content *s)
 649 {
 650         struct tree_content *d;
 651         struct tree_entry *a, *b;
 652         unsigned int i;
 653
 654         if (!s)
 655                 return NULL;
 656         d = new_tree_content(s->entry_count);
 657         for (i = 0; i < s->entry_count; i++) {
 658                 a = s->entries[i];
 659                 b = new_tree_entry();
 660                 memcpy(b, a, sizeof(*a));
 661                 if (a->tree && is_null_sha1(b->versions[1].sha1))
 662                         b->tree = dup_tree_content(a->tree);
 663                 else
 664                         b->tree = NULL;
 665                 d->entries[i] = b;
 666         }
 667         d->entry_count = s->entry_count;
 668         d->delta_depth = s->delta_depth;
 669
 670         return d;
 671 }
 672
 673 static void start_packfile(void)
 674 {
 675         static char tmpfile[PATH_MAX];
 676         struct packed_git *p;
 677         struct pack_header hdr;
 678         int pack_fd;
 679
 680         snprintf(tmpfile, sizeof(tmpfile),
 681                 "%s/tmp_pack_XXXXXX", get_object_directory());
 682         pack_fd = xmkstemp(tmpfile);
 683         p = xcalloc(1, sizeof(*p) + strlen(tmpfile) + 2);
 684         strcpy(p->pack_name, tmpfile);
 685         p->pack_fd = pack_fd;
 686
 687         hdr.hdr_signature = htonl(PACK_SIGNATURE);
 688         hdr.hdr_version = htonl(2);
 689         hdr.hdr_entries = 0;
 690         write_or_die(p->pack_fd, &hdr, sizeof(hdr));
 691
 692         pack_data = p;
 693         pack_size = sizeof(hdr);
 694         object_count = 0;
 695
 696         all_packs = xrealloc(all_packs, sizeof(*all_packs) * (pack_id + 1));
 697         all_packs[pack_id] = p;
 698 }
 699
 700 static int oecmp (const void *a_, const void *b_)
 701 {
 702         struct object_entry *a = *((struct object_entry**)a_);
 703         struct object_entry *b = *((struct object_entry**)b_);
 704         return hashcmp(a->sha1, b->sha1);
 705 }
 706
 707 static char *create_index(void)
 708 {
 709         static char tmpfile[PATH_MAX];
 710         SHA_CTX ctx;
 711         struct sha1file *f;
 712         struct object_entry **idx, **c, **last, *e;
 713         struct object_entry_pool *o;
 714         uint32_t array[256];
 715         int i, idx_fd;
 716
 717         /* Build the sorted table of object IDs. */
 718         idx = xmalloc(object_count * sizeof(struct object_entry*));
 719         c = idx;
 720         for (o = blocks; o; o = o->next_pool)
 721                 for (e = o->next_free; e-- != o->entries;)
 722                         if (pack_id == e->pack_id)
 723                                 *c++ = e;
 724         last = idx + object_count;
 725         if (c != last)
 726                 die("internal consistency error creating the index");
 727         qsort(idx, object_count, sizeof(struct object_entry*), oecmp);
 728
 729         /* Generate the fan-out array. */
 730         c = idx;
 731         for (i = 0; i < 256; i++) {
 732                 struct object_entry **next = c;;
 733                 while (next < last) {
 734                         if ((*next)->sha1[0] != i)
 735                                 break;
 736                         next++;
 737                 }
 738                 array[i] = htonl(next - idx);
 739                 c = next;
 740         }
 741
 742         snprintf(tmpfile, sizeof(tmpfile),
 743                 "%s/tmp_idx_XXXXXX", get_object_directory());
 744         idx_fd = xmkstemp(tmpfile);
 745         f = sha1fd(idx_fd, tmpfile);
 746         sha1write(f, array, 256 * sizeof(int));
 747         SHA1_Init(&ctx);
 748         for (c = idx; c != last; c++) {
 749                 uint32_t offset = htonl((*c)->offset);
 750                 sha1write(f, &offset, 4);
 751                 sha1write(f, (*c)->sha1, sizeof((*c)->sha1));
 752                 SHA1_Update(&ctx, (*c)->sha1, 20);
 753         }
 754         sha1write(f, pack_data->sha1, sizeof(pack_data->sha1));
 755         sha1close(f, NULL, 1);
 756         free(idx);
 757         SHA1_Final(pack_data->sha1, &ctx);
 758         return tmpfile;
 759 }
 760
 761 static char *keep_pack(char *curr_index_name)
 762 {
 763         static char name[PATH_MAX];
 764         static const char *keep_msg = "fast-import";
 765         int keep_fd;
 766
 767         chmod(pack_data->pack_name, 0444);
 768         chmod(curr_index_name, 0444);
 769
 770         snprintf(name, sizeof(name), "%s/pack/pack-%s.keep",
 771                  get_object_directory(), sha1_to_hex(pack_data->sha1));
 772         keep_fd = open(name, O_RDWR|O_CREAT|O_EXCL, 0600);
 773         if (keep_fd < 0)
 774                 die("cannot create keep file");
 775         write(keep_fd, keep_msg, strlen(keep_msg));
 776         close(keep_fd);
 777
 778         snprintf(name, sizeof(name), "%s/pack/pack-%s.pack",
 779                  get_object_directory(), sha1_to_hex(pack_data->sha1));
 780         if (move_temp_to_file(pack_data->pack_name, name))
 781                 die("cannot store pack file");
 782
 783         snprintf(name, sizeof(name), "%s/pack/pack-%s.idx",
 784                  get_object_directory(), sha1_to_hex(pack_data->sha1));
 785         if (move_temp_to_file(curr_index_name, name))
 786                 die("cannot store index file");
 787         return name;
 788 }
 789
 790 static void unkeep_all_packs(void)
 791 {
 792         static char name[PATH_MAX];
 793         int k;
 794
 795         for (k = 0; k < pack_id; k++) {
 796                 struct packed_git *p = all_packs[k];
 797                 snprintf(name, sizeof(name), "%s/pack/pack-%s.keep",
 798                          get_object_directory(), sha1_to_hex(p->sha1));
 799                 unlink(name);
 800         }
 801 }
 802
 803 static void end_packfile(void)
 804 {
 805         struct packed_git *old_p = pack_data, *new_p;
 806
 807         if (object_count) {
 808                 char *idx_name;
 809                 int i;
 810                 struct branch *b;
 811                 struct tag *t;
 812
 813                 fixup_pack_header_footer(pack_data->pack_fd, pack_data->sha1,
 814                                     pack_data->pack_name, object_count);
 815                 close(pack_data->pack_fd);
 816                 idx_name = keep_pack(create_index());
 817
 818                 /* Register the packfile with core git's machinary. */
 819                 new_p = add_packed_git(idx_name, strlen(idx_name), 1);
 820                 if (!new_p)
 821                         die("core git rejected index %s", idx_name);
 822                 new_p->windows = old_p->windows;
 823                 all_packs[pack_id] = new_p;
 824                 install_packed_git(new_p);
 825
 826                 /* Print the boundary */
 827                 if (pack_edges) {
 828                         fprintf(pack_edges, "%s:", new_p->pack_name);
 829                         for (i = 0; i < branch_table_sz; i++) {
 830                                 for (b = branch_table[i]; b; b = b->table_next_branch) {
 831                                         if (b->pack_id == pack_id)
 832                                                 fprintf(pack_edges, " %s", sha1_to_hex(b->sha1));
 833                                 }
 834                         }
 835                         for (t = first_tag; t; t = t->next_tag) {
 836                                 if (t->pack_id == pack_id)
 837                                         fprintf(pack_edges, " %s", sha1_to_hex(t->sha1));
 838                         }
 839                         fputc('\n', pack_edges);
 840                         fflush(pack_edges);
 841                 }
 842
 843                 pack_id++;
 844         }
 845         else
 846                 unlink(old_p->pack_name);
 847         free(old_p);
 848
 849         /* We can't carry a delta across packfiles. */
 850         free(last_blob.data);
 851         last_blob.data = NULL;
 852         last_blob.len = 0;
 853         last_blob.offset = 0;
 854         last_blob.depth = 0;
 855 }
 856
 857 static void cycle_packfile(void)
 858 {
 859         end_packfile();
 860         start_packfile();
 861 }
 862
 863 static size_t encode_header(
 864         enum object_type type,
 865         size_t size,
 866         unsigned char *hdr)
 867 {
 868         int n = 1;
 869         unsigned char c;
 870
 871         if (type < OBJ_COMMIT || type > OBJ_REF_DELTA)
 872                 die("bad type %d", type);
 873
 874         c = (type << 4) | (size & 15);
 875         size >>= 4;
 876         while (size) {
 877                 *hdr++ = c | 0x80;
 878                 c = size & 0x7f;
 879                 size >>= 7;
 880                 n++;
 881         }
 882         *hdr = c;
 883         return n;
 884 }
 885
 886 static int store_object(
 887         enum object_type type,
 888         void *dat,
 889         size_t datlen,
 890         struct last_object *last,
 891         unsigned char *sha1out,
 892         uintmax_t mark)
 893 {
 894         void *out, *delta;
 895         struct object_entry *e;
 896         unsigned char hdr[96];
 897         unsigned char sha1[20];
 898         unsigned long hdrlen, deltalen;
 899         SHA_CTX c;
 900         z_stream s;
 901
 902         hdrlen = sprintf((char*)hdr,"%s %lu", typename(type),
 903                 (unsigned long)datlen) + 1;
 904         SHA1_Init(&c);
 905         SHA1_Update(&c, hdr, hdrlen);
 906         SHA1_Update(&c, dat, datlen);
 907         SHA1_Final(sha1, &c);
 908         if (sha1out)
 909                 hashcpy(sha1out, sha1);
 910
 911         e = insert_object(sha1);
 912         if (mark)
 913                 insert_mark(mark, e);
 914         if (e->offset) {
 915                 duplicate_count_by_type[type]++;
 916                 return 1;
 917         } else if (find_sha1_pack(sha1, packed_git)) {
 918                 e->type = type;
 919                 e->pack_id = MAX_PACK_ID;
 920                 e->offset = 1; /* just not zero! */
 921                 duplicate_count_by_type[type]++;
 922                 return 1;
 923         }
 924
 925         if (last && last->data && last->depth < max_depth) {
 926                 delta = diff_delta(last->data, last->len,
 927                         dat, datlen,
 928                         &deltalen, 0);
 929                 if (delta && deltalen >= datlen) {
 930                         free(delta);
 931                         delta = NULL;
 932                 }
 933         } else
 934                 delta = NULL;
 935
 936         memset(&s, 0, sizeof(s));
 937         deflateInit(&s, zlib_compression_level);
 938         if (delta) {
 939                 s.next_in = delta;
 940                 s.avail_in = deltalen;
 941         } else {
 942                 s.next_in = dat;
 943                 s.avail_in = datlen;
 944         }
 945         s.avail_out = deflateBound(&s, s.avail_in);
 946         s.next_out = out = xmalloc(s.avail_out);
 947         while (deflate(&s, Z_FINISH) == Z_OK)
 948                 /* nothing */;
 949         deflateEnd(&s);
 950
 951         /* Determine if we should auto-checkpoint. */
 952         if ((pack_size + 60 + s.total_out) > max_packsize
 953                 || (pack_size + 60 + s.total_out) < pack_size) {
 954
 955                 /* This new object needs to *not* have the current pack_id. */
 956                 e->pack_id = pack_id + 1;
 957                 cycle_packfile();
 958
 959                 /* We cannot carry a delta into the new pack. */
 960                 if (delta) {
 961                         free(delta);
 962                         delta = NULL;
 963
 964                         memset(&s, 0, sizeof(s));
 965                         deflateInit(&s, zlib_compression_level);
 966                         s.next_in = dat;
 967                         s.avail_in = datlen;
 968                         s.avail_out = deflateBound(&s, s.avail_in);
 969                         s.next_out = out = xrealloc(out, s.avail_out);
 970                         while (deflate(&s, Z_FINISH) == Z_OK)
 971                                 /* nothing */;
 972                         deflateEnd(&s);
 973                 }
 974         }
 975
 976         e->type = type;
 977         e->pack_id = pack_id;
 978         e->offset = pack_size;
 979         object_count++;
 980         object_count_by_type[type]++;
 981
 982         if (delta) {
 983                 unsigned long ofs = e->offset - last->offset;
 984                 unsigned pos = sizeof(hdr) - 1;
 985
 986                 delta_count_by_type[type]++;
 987                 last->depth++;
 988
 989                 hdrlen = encode_header(OBJ_OFS_DELTA, deltalen, hdr);
 990                 write_or_die(pack_data->pack_fd, hdr, hdrlen);
 991                 pack_size += hdrlen;
 992
 993                 hdr[pos] = ofs & 127;
 994                 while (ofs >>= 7)
 995                         hdr[--pos] = 128 | (--ofs & 127);
 996                 write_or_die(pack_data->pack_fd, hdr + pos, sizeof(hdr) - pos);
 997                 pack_size += sizeof(hdr) - pos;
 998         } else {
 999                 if (last)
1000                         last->depth = 0;
1001                 hdrlen = encode_header(type, datlen, hdr);
1002                 write_or_die(pack_data->pack_fd, hdr, hdrlen);
1003                 pack_size += hdrlen;
1004         }
1005
1006         write_or_die(pack_data->pack_fd, out, s.total_out);
1007         pack_size += s.total_out;
1008
1009         free(out);
1010         free(delta);
1011         if (last) {
1012                 if (!last->no_free)
1013                         free(last->data);
1014                 last->data = dat;
1015                 last->offset = e->offset;
1016                 last->len = datlen;
1017         }
1018         return 0;
1019 }
1020
1021 static void *gfi_unpack_entry(
1022         struct object_entry *oe,
1023         unsigned long *sizep)
1024 {
1025         enum object_type type;
1026         struct packed_git *p = all_packs[oe->pack_id];
1027         if (p == pack_data)
1028                 p->pack_size = pack_size + 20;
1029         return unpack_entry(p, oe->offset, &type, sizep);
1030 }
1031
1032 static const char *get_mode(const char *str, uint16_t *modep)
1033 {
1034         unsigned char c;
1035         uint16_t mode = 0;
1036
1037         while ((c = *str++) != ' ') {
1038                 if (c < '0' || c > '7')
1039                         return NULL;
1040                 mode = (mode << 3) + (c - '0');
1041         }
1042         *modep = mode;
1043         return str;
1044 }
1045
1046 static void load_tree(struct tree_entry *root)
1047 {
1048         unsigned char* sha1 = root->versions[1].sha1;
1049         struct object_entry *myoe;
1050         struct tree_content *t;
1051         unsigned long size;
1052         char *buf;
1053         const char *c;
1054
1055         root->tree = t = new_tree_content(8);
1056         if (is_null_sha1(sha1))
1057                 return;
1058
1059         myoe = find_object(sha1);
1060         if (myoe && myoe->pack_id != MAX_PACK_ID) {
1061                 if (myoe->type != OBJ_TREE)
1062                         die("Not a tree: %s", sha1_to_hex(sha1));
1063                 t->delta_depth = 0;
1064                 buf = gfi_unpack_entry(myoe, &size);
1065         } else {
1066                 enum object_type type;
1067                 buf = read_sha1_file(sha1, &type, &size);
1068                 if (!buf || type != OBJ_TREE)
1069                         die("Can't load tree %s", sha1_to_hex(sha1));
1070         }
1071
1072         c = buf;
1073         while (c != (buf + size)) {
1074                 struct tree_entry *e = new_tree_entry();
1075
1076                 if (t->entry_count == t->entry_capacity)
1077                         root->tree = t = grow_tree_content(t, t->entry_count);
1078                 t->entries[t->entry_count++] = e;
1079
1080                 e->tree = NULL;
1081                 c = get_mode(c, &e->versions[1].mode);
1082                 if (!c)
1083                         die("Corrupt mode in %s", sha1_to_hex(sha1));
1084                 e->versions[0].mode = e->versions[1].mode;
1085                 e->name = to_atom(c, strlen(c));
1086                 c += e->name->str_len + 1;
1087                 hashcpy(e->versions[0].sha1, (unsigned char*)c);
1088                 hashcpy(e->versions[1].sha1, (unsigned char*)c);
1089                 c += 20;
1090         }
1091         free(buf);
1092 }
1093
1094 static int tecmp0 (const void *_a, const void *_b)
1095 {
1096         struct tree_entry *a = *((struct tree_entry**)_a);
1097         struct tree_entry *b = *((struct tree_entry**)_b);
1098         return base_name_compare(
1099                 a->name->str_dat, a->name->str_len, a->versions[0].mode,
1100                 b->name->str_dat, b->name->str_len, b->versions[0].mode);
1101 }
1102
1103 static int tecmp1 (const void *_a, const void *_b)
1104 {
1105         struct tree_entry *a = *((struct tree_entry**)_a);
1106         struct tree_entry *b = *((struct tree_entry**)_b);
1107         return base_name_compare(
1108                 a->name->str_dat, a->name->str_len, a->versions[1].mode,
1109                 b->name->str_dat, b->name->str_len, b->versions[1].mode);
1110 }
1111
1112 static void mktree(struct tree_content *t,
1113         int v,
1114         unsigned long *szp,
1115         struct dbuf *b)
1116 {
1117         size_t maxlen = 0;
1118         unsigned int i;
1119         char *c;
1120
1121         if (!v)
1122                 qsort(t->entries,t->entry_count,sizeof(t->entries[0]),tecmp0);
1123         else
1124                 qsort(t->entries,t->entry_count,sizeof(t->entries[0]),tecmp1);
1125
1126         for (i = 0; i < t->entry_count; i++) {
1127                 if (t->entries[i]->versions[v].mode)
1128                         maxlen += t->entries[i]->name->str_len + 34;
1129         }
1130
1131         size_dbuf(b, maxlen);
1132         c = b->buffer;
1133         for (i = 0; i < t->entry_count; i++) {
1134                 struct tree_entry *e = t->entries[i];
1135                 if (!e->versions[v].mode)
1136                         continue;
1137                 c += sprintf(c, "%o", (unsigned int)e->versions[v].mode);
1138                 *c++ = ' ';
1139                 strcpy(c, e->name->str_dat);
1140                 c += e->name->str_len + 1;
1141                 hashcpy((unsigned char*)c, e->versions[v].sha1);
1142                 c += 20;
1143         }
1144         *szp = c - (char*)b->buffer;
1145 }
1146
1147 static void store_tree(struct tree_entry *root)
1148 {
1149         struct tree_content *t = root->tree;
1150         unsigned int i, j, del;
1151         unsigned long new_len;
1152         struct last_object lo;
1153         struct object_entry *le;
1154
1155         if (!is_null_sha1(root->versions[1].sha1))
1156                 return;
1157
1158         for (i = 0; i < t->entry_count; i++) {
1159                 if (t->entries[i]->tree)
1160                         store_tree(t->entries[i]);
1161         }
1162
1163         le = find_object(root->versions[0].sha1);
1164         if (!S_ISDIR(root->versions[0].mode)
1165                 || !le
1166                 || le->pack_id != pack_id) {
1167                 lo.data = NULL;
1168                 lo.depth = 0;
1169                 lo.no_free = 0;
1170         } else {
1171                 mktree(t, 0, &lo.len, &old_tree);
1172                 lo.data = old_tree.buffer;
1173                 lo.offset = le->offset;
1174                 lo.depth = t->delta_depth;
1175                 lo.no_free = 1;
1176         }
1177
1178         mktree(t, 1, &new_len, &new_tree);
1179         store_object(OBJ_TREE, new_tree.buffer, new_len,
1180                 &lo, root->versions[1].sha1, 0);
1181
1182         t->delta_depth = lo.depth;
1183         for (i = 0, j = 0, del = 0; i < t->entry_count; i++) {
1184                 struct tree_entry *e = t->entries[i];
1185                 if (e->versions[1].mode) {
1186                         e->versions[0].mode = e->versions[1].mode;
1187                         hashcpy(e->versions[0].sha1, e->versions[1].sha1);
1188                         t->entries[j++] = e;
1189                 } else {
1190                         release_tree_entry(e);
1191                         del++;
1192                 }
1193         }
1194         t->entry_count -= del;
1195 }
1196
1197 static int tree_content_set(
1198         struct tree_entry *root,
1199         const char *p,
1200         const unsigned char *sha1,
1201         const uint16_t mode,
1202         struct tree_content *subtree)
1203 {
1204         struct tree_content *t = root->tree;
1205         const char *slash1;
1206         unsigned int i, n;
1207         struct tree_entry *e;
1208
1209         slash1 = strchr(p, '/');
1210         if (slash1)
1211                 n = slash1 - p;
1212         else
1213                 n = strlen(p);
1214         if (!n)
1215                 die("Empty path component found in input");
1216         if (!slash1 && !S_ISDIR(mode) && subtree)
1217                 die("Non-directories cannot have subtrees");
1218
1219         for (i = 0; i < t->entry_count; i++) {
1220                 e = t->entries[i];
1221                 if (e->name->str_len == n && !strncmp(p, e->name->str_dat, n)) {
1222                         if (!slash1) {
1223                                 if (!S_ISDIR(mode)
1224                                                 && e->versions[1].mode == mode
1225                                                 && !hashcmp(e->versions[1].sha1, sha1))
1226                                         return 0;
1227                                 e->versions[1].mode = mode;
1228                                 hashcpy(e->versions[1].sha1, sha1);
1229                                 if (e->tree)
1230                                         release_tree_content_recursive(e->tree);
1231                                 e->tree = subtree;
1232                                 hashclr(root->versions[1].sha1);
1233                                 return 1;
1234                         }
1235                         if (!S_ISDIR(e->versions[1].mode)) {
1236                                 e->tree = new_tree_content(8);
1237                                 e->versions[1].mode = S_IFDIR;
1238                         }
1239                         if (!e->tree)
1240                                 load_tree(e);
1241                         if (tree_content_set(e, slash1 + 1, sha1, mode, subtree)) {
1242                                 hashclr(root->versions[1].sha1);
1243                                 return 1;
1244                         }
1245                         return 0;
1246                 }
1247         }
1248
1249         if (t->entry_count == t->entry_capacity)
1250                 root->tree = t = grow_tree_content(t, t->entry_count);
1251         e = new_tree_entry();
1252         e->name = to_atom(p, n);
1253         e->versions[0].mode = 0;
1254         hashclr(e->versions[0].sha1);
1255         t->entries[t->entry_count++] = e;
1256         if (slash1) {
1257                 e->tree = new_tree_content(8);
1258                 e->versions[1].mode = S_IFDIR;
1259                 tree_content_set(e, slash1 + 1, sha1, mode, subtree);
1260         } else {
1261                 e->tree = subtree;
1262                 e->versions[1].mode = mode;
1263                 hashcpy(e->versions[1].sha1, sha1);
1264         }
1265         hashclr(root->versions[1].sha1);
1266         return 1;
1267 }
1268
1269 static int tree_content_remove(
1270         struct tree_entry *root,
1271         const char *p,
1272         struct tree_entry *backup_leaf)
1273 {
1274         struct tree_content *t = root->tree;
1275         const char *slash1;
1276         unsigned int i, n;
1277         struct tree_entry *e;
1278
1279         slash1 = strchr(p, '/');
1280         if (slash1)
1281                 n = slash1 - p;
1282         else
1283                 n = strlen(p);
1284
1285         for (i = 0; i < t->entry_count; i++) {
1286                 e = t->entries[i];
1287                 if (e->name->str_len == n && !strncmp(p, e->name->str_dat, n)) {
1288                         if (!slash1 || !S_ISDIR(e->versions[1].mode))
1289                                 goto del_entry;
1290                         if (!e->tree)
1291                                 load_tree(e);
1292                         if (tree_content_remove(e, slash1 + 1, backup_leaf)) {
1293                                 for (n = 0; n < e->tree->entry_count; n++) {
1294                                         if (e->tree->entries[n]->versions[1].mode) {
1295                                                 hashclr(root->versions[1].sha1);
1296                                                 return 1;
1297                                         }
1298                                 }
1299                                 backup_leaf = NULL;
1300                                 goto del_entry;
1301                         }
1302                         return 0;
1303                 }
1304         }
1305         return 0;
1306
1307 del_entry:
1308         if (backup_leaf)
1309                 memcpy(backup_leaf, e, sizeof(*backup_leaf));
1310         else if (e->tree)
1311                 release_tree_content_recursive(e->tree);
1312         e->tree = NULL;
1313         e->versions[1].mode = 0;
1314         hashclr(e->versions[1].sha1);
1315         hashclr(root->versions[1].sha1);
1316         return 1;
1317 }
1318
1319 static int tree_content_get(
1320         struct tree_entry *root,
1321         const char *p,
1322         struct tree_entry *leaf)
1323 {
1324         struct tree_content *t = root->tree;
1325         const char *slash1;
1326         unsigned int i, n;
1327         struct tree_entry *e;
1328
1329         slash1 = strchr(p, '/');
1330         if (slash1)
1331                 n = slash1 - p;
1332         else
1333                 n = strlen(p);
1334
1335         for (i = 0; i < t->entry_count; i++) {
1336                 e = t->entries[i];
1337                 if (e->name->str_len == n && !strncmp(p, e->name->str_dat, n)) {
1338                         if (!slash1) {
1339                                 memcpy(leaf, e, sizeof(*leaf));
1340                                 if (e->tree && is_null_sha1(e->versions[1].sha1))
1341                                         leaf->tree = dup_tree_content(e->tree);
1342                                 else
1343                                         leaf->tree = NULL;
1344                                 return 1;
1345                         }
1346                         if (!S_ISDIR(e->versions[1].mode))
1347                                 return 0;
1348                         if (!e->tree)
1349                                 load_tree(e);
1350                         return tree_content_get(e, slash1 + 1, leaf);
1351                 }
1352         }
1353         return 0;
1354 }
1355
1356 static int update_branch(struct branch *b)
1357 {
1358         static const char *msg = "fast-import";
1359         struct ref_lock *lock;
1360         unsigned char old_sha1[20];
1361
1362         if (read_ref(b->name, old_sha1))
1363                 hashclr(old_sha1);
1364         lock = lock_any_ref_for_update(b->name, old_sha1, 0);
1365         if (!lock)
1366                 return error("Unable to lock %s", b->name);
1367         if (!force_update && !is_null_sha1(old_sha1)) {
1368                 struct commit *old_cmit, *new_cmit;
1369
1370                 old_cmit = lookup_commit_reference_gently(old_sha1, 0);
1371                 new_cmit = lookup_commit_reference_gently(b->sha1, 0);
1372                 if (!old_cmit || !new_cmit) {
1373                         unlock_ref(lock);
1374                         return error("Branch %s is missing commits.", b->name);
1375                 }
1376
1377                 if (!in_merge_bases(old_cmit, &new_cmit, 1)) {
1378                         unlock_ref(lock);
1379                         warning("Not updating %s"
1380                                 " (new tip %s does not contain %s)",
1381                                 b->name, sha1_to_hex(b->sha1), sha1_to_hex(old_sha1));
1382                         return -1;
1383                 }
1384         }
1385         if (write_ref_sha1(lock, b->sha1, msg) < 0)
1386                 return error("Unable to update %s", b->name);
1387         return 0;
1388 }
1389
1390 static void dump_branches(void)
1391 {
1392         unsigned int i;
1393         struct branch *b;
1394
1395         for (i = 0; i < branch_table_sz; i++) {
1396                 for (b = branch_table[i]; b; b = b->table_next_branch)
1397                         failure |= update_branch(b);
1398         }
1399 }
1400
1401 static void dump_tags(void)
1402 {
1403         static const char *msg = "fast-import";
1404         struct tag *t;
1405         struct ref_lock *lock;
1406         char ref_name[PATH_MAX];
1407
1408         for (t = first_tag; t; t = t->next_tag) {
1409                 sprintf(ref_name, "tags/%s", t->name);
1410                 lock = lock_ref_sha1(ref_name, NULL);
1411                 if (!lock || write_ref_sha1(lock, t->sha1, msg) < 0)
1412                         failure |= error("Unable to update %s", ref_name);
1413         }
1414 }
1415
1416 static void dump_marks_helper(FILE *f,
1417         uintmax_t base,
1418         struct mark_set *m)
1419 {
1420         uintmax_t k;
1421         if (m->shift) {
1422                 for (k = 0; k < 1024; k++) {
1423                         if (m->data.sets[k])
1424                                 dump_marks_helper(f, (base + k) << m->shift,
1425                                         m->data.sets[k]);
1426                 }
1427         } else {
1428                 for (k = 0; k < 1024; k++) {
1429                         if (m->data.marked[k])
1430                                 fprintf(f, ":%" PRIuMAX " %s\n", base + k,
1431                                         sha1_to_hex(m->data.marked[k]->sha1));
1432                 }
1433         }
1434 }
1435
1436 static void dump_marks(void)
1437 {
1438         static struct lock_file mark_lock;
1439         int mark_fd;
1440         FILE *f;
1441
1442         if (!mark_file)
1443                 return;
1444
1445         mark_fd = hold_lock_file_for_update(&mark_lock, mark_file, 0);
1446         if (mark_fd < 0) {
1447                 failure |= error("Unable to write marks file %s: %s",
1448                         mark_file, strerror(errno));
1449                 return;
1450         }
1451
1452         f = fdopen(mark_fd, "w");
1453         if (!f) {
1454                 rollback_lock_file(&mark_lock);
1455                 failure |= error("Unable to write marks file %s: %s",
1456                         mark_file, strerror(errno));
1457                 return;
1458         }
1459
1460         dump_marks_helper(f, 0, marks);
1461         fclose(f);
1462         if (commit_lock_file(&mark_lock))
1463                 failure |= error("Unable to write marks file %s: %s",
1464                         mark_file, strerror(errno));
1465 }
1466
1467 static void read_next_command(void)
1468 {
1469         do {
1470                 if (unread_command_buf)
1471                         unread_command_buf = 0;
1472                 else
1473                         read_line(&command_buf, stdin, '\n');
1474         } while (!command_buf.eof && command_buf.buf[0] == '#');
1475 }
1476
1477 static void skip_optional_lf()
1478 {
1479         int term_char = fgetc(stdin);
1480         if (term_char != '\n' && term_char != EOF)
1481                 ungetc(term_char, stdin);
1482 }
1483
1484 static void cmd_mark(void)
1485 {
1486         if (!prefixcmp(command_buf.buf, "mark :")) {
1487                 next_mark = strtoumax(command_buf.buf + 6, NULL, 10);
1488                 read_next_command();
1489         }
1490         else
1491                 next_mark = 0;
1492 }
1493
1494 static void *cmd_data (size_t *size)
1495 {
1496         size_t length;
1497         char *buffer;
1498
1499         if (prefixcmp(command_buf.buf, "data "))
1500                 die("Expected 'data n' command, found: %s", command_buf.buf);
1501
1502         if (!prefixcmp(command_buf.buf + 5, "<<")) {
1503                 char *term = xstrdup(command_buf.buf + 5 + 2);
1504                 size_t sz = 8192, term_len = command_buf.len - 5 - 2;
1505                 length = 0;
1506                 buffer = xmalloc(sz);
1507                 for (;;) {
1508                         read_line(&command_buf, stdin, '\n');
1509                         if (command_buf.eof)
1510                                 die("EOF in data (terminator '%s' not found)", term);
1511                         if (term_len == command_buf.len
1512                                 && !strcmp(term, command_buf.buf))
1513                                 break;
1514                         ALLOC_GROW(buffer, length + command_buf.len, sz);
1515                         memcpy(buffer + length,
1516                                 command_buf.buf,
1517                                 command_buf.len - 1);
1518                         length += command_buf.len - 1;
1519                         buffer[length++] = '\n';
1520                 }
1521                 free(term);
1522         }
1523         else {
1524                 size_t n = 0;
1525                 length = strtoul(command_buf.buf + 5, NULL, 10);
1526                 buffer = xmalloc(length);
1527                 while (n < length) {
1528                         size_t s = fread(buffer + n, 1, length - n, stdin);
1529                         if (!s && feof(stdin))
1530                                 die("EOF in data (%lu bytes remaining)",
1531                                         (unsigned long)(length - n));
1532                         n += s;
1533                 }
1534         }
1535
1536         skip_optional_lf();
1537         *size = length;
1538         return buffer;
1539 }
1540
1541 static int validate_raw_date(const char *src, char *result, int maxlen)
1542 {
1543         const char *orig_src = src;
1544         char *endp, sign;
1545
1546         strtoul(src, &endp, 10);
1547         if (endp == src || *endp != ' ')
1548                 return -1;
1549
1550         src = endp + 1;
1551         if (*src != '-' && *src != '+')
1552                 return -1;
1553         sign = *src;
1554
1555         strtoul(src + 1, &endp, 10);
1556         if (endp == src || *endp || (endp - orig_src) >= maxlen)
1557                 return -1;
1558
1559         strcpy(result, orig_src);
1560         return 0;
1561 }
1562
1563 static char *parse_ident(const char *buf)
1564 {
1565         const char *gt;
1566         size_t name_len;
1567         char *ident;
1568
1569         gt = strrchr(buf, '>');
1570         if (!gt)
1571                 die("Missing > in ident string: %s", buf);
1572         gt++;
1573         if (*gt != ' ')
1574                 die("Missing space after > in ident string: %s", buf);
1575         gt++;
1576         name_len = gt - buf;
1577         ident = xmalloc(name_len + 24);
1578         strncpy(ident, buf, name_len);
1579
1580         switch (whenspec) {
1581         case WHENSPEC_RAW:
1582                 if (validate_raw_date(gt, ident + name_len, 24) < 0)
1583                         die("Invalid raw date \"%s\" in ident: %s", gt, buf);
1584                 break;
1585         case WHENSPEC_RFC2822:
1586                 if (parse_date(gt, ident + name_len, 24) < 0)
1587                         die("Invalid rfc2822 date \"%s\" in ident: %s", gt, buf);
1588                 break;
1589         case WHENSPEC_NOW:
1590                 if (strcmp("now", gt))
1591                         die("Date in ident must be 'now': %s", buf);
1592                 datestamp(ident + name_len, 24);
1593                 break;
1594         }
1595
1596         return ident;
1597 }
1598
1599 static void cmd_new_blob(void)
1600 {
1601         size_t l;
1602         void *d;
1603
1604         read_next_command();
1605         cmd_mark();
1606         d = cmd_data(&l);
1607
1608         if (store_object(OBJ_BLOB, d, l, &last_blob, NULL, next_mark))
1609                 free(d);
1610 }
1611
1612 static void unload_one_branch(void)
1613 {
1614         while (cur_active_branches
1615                 && cur_active_branches >= max_active_branches) {
1616                 uintmax_t min_commit = ULONG_MAX;
1617                 struct branch *e, *l = NULL, *p = NULL;
1618
1619                 for (e = active_branches; e; e = e->active_next_branch) {
1620                         if (e->last_commit < min_commit) {
1621                                 p = l;
1622                                 min_commit = e->last_commit;
1623                         }
1624                         l = e;
1625                 }
1626
1627                 if (p) {
1628                         e = p->active_next_branch;
1629                         p->active_next_branch = e->active_next_branch;
1630                 } else {
1631                         e = active_branches;
1632                         active_branches = e->active_next_branch;
1633                 }
1634                 e->active = 0;
1635                 e->active_next_branch = NULL;
1636                 if (e->branch_tree.tree) {
1637                         release_tree_content_recursive(e->branch_tree.tree);
1638                         e->branch_tree.tree = NULL;
1639                 }
1640                 cur_active_branches--;
1641         }
1642 }
1643
1644 static void load_branch(struct branch *b)
1645 {
1646         load_tree(&b->branch_tree);
1647         if (!b->active) {
1648                 b->active = 1;
1649                 b->active_next_branch = active_branches;
1650                 active_branches = b;
1651                 cur_active_branches++;
1652                 branch_load_count++;
1653         }
1654 }
1655
1656 static void file_change_m(struct branch *b)
1657 {
1658         const char *p = command_buf.buf + 2;
1659         char *p_uq;
1660         const char *endp;
1661         struct object_entry *oe = oe;
1662         unsigned char sha1[20];
1663         uint16_t mode, inline_data = 0;
1664
1665         p = get_mode(p, &mode);
1666         if (!p)
1667                 die("Corrupt mode: %s", command_buf.buf);
1668         switch (mode) {
1669         case S_IFREG | 0644:
1670         case S_IFREG | 0755:
1671         case S_IFLNK:
1672         case 0644:
1673         case 0755:
1674                 /* ok */
1675                 break;
1676         default:
1677                 die("Corrupt mode: %s", command_buf.buf);
1678         }
1679
1680         if (*p == ':') {
1681                 char *x;
1682                 oe = find_mark(strtoumax(p + 1, &x, 10));
1683                 hashcpy(sha1, oe->sha1);
1684                 p = x;
1685         } else if (!prefixcmp(p, "inline")) {
1686                 inline_data = 1;
1687                 p += 6;
1688         } else {
1689                 if (get_sha1_hex(p, sha1))
1690                         die("Invalid SHA1: %s", command_buf.buf);
1691                 oe = find_object(sha1);
1692                 p += 40;
1693         }
1694         if (*p++ != ' ')
1695                 die("Missing space after SHA1: %s", command_buf.buf);
1696
1697         p_uq = unquote_c_style(p, &endp);
1698         if (p_uq) {
1699                 if (*endp)
1700                         die("Garbage after path in: %s", command_buf.buf);
1701                 p = p_uq;
1702         }
1703
1704         if (inline_data) {
1705                 size_t l;
1706                 void *d;
1707                 if (!p_uq)
1708                         p = p_uq = xstrdup(p);
1709                 read_next_command();
1710                 d = cmd_data(&l);
1711                 if (store_object(OBJ_BLOB, d, l, &last_blob, sha1, 0))
1712                         free(d);
1713         } else if (oe) {
1714                 if (oe->type != OBJ_BLOB)
1715                         die("Not a blob (actually a %s): %s",
1716                                 command_buf.buf, typename(oe->type));
1717         } else {
1718                 enum object_type type = sha1_object_info(sha1, NULL);
1719                 if (type < 0)
1720                         die("Blob not found: %s", command_buf.buf);
1721                 if (type != OBJ_BLOB)
1722                         die("Not a blob (actually a %s): %s",
1723                             typename(type), command_buf.buf);
1724         }
1725
1726         tree_content_set(&b->branch_tree, p, sha1, S_IFREG | mode, NULL);
1727         free(p_uq);
1728 }
1729
1730 static void file_change_d(struct branch *b)
1731 {
1732         const char *p = command_buf.buf + 2;
1733         char *p_uq;
1734         const char *endp;
1735
1736         p_uq = unquote_c_style(p, &endp);
1737         if (p_uq) {
1738                 if (*endp)
1739                         die("Garbage after path in: %s", command_buf.buf);
1740                 p = p_uq;
1741         }
1742         tree_content_remove(&b->branch_tree, p, NULL);
1743         free(p_uq);
1744 }
1745
1746 static void file_change_cr(struct branch *b, int rename)
1747 {
1748         const char *s, *d;
1749         char *s_uq, *d_uq;
1750         const char *endp;
1751         struct tree_entry leaf;
1752
1753         s = command_buf.buf + 2;
1754         s_uq = unquote_c_style(s, &endp);
1755         if (s_uq) {
1756                 if (*endp != ' ')
1757                         die("Missing space after source: %s", command_buf.buf);
1758         }
1759         else {
1760                 endp = strchr(s, ' ');
1761                 if (!endp)
1762                         die("Missing space after source: %s", command_buf.buf);
1763                 s_uq = xmalloc(endp - s + 1);
1764                 memcpy(s_uq, s, endp - s);
1765                 s_uq[endp - s] = 0;
1766         }
1767         s = s_uq;
1768
1769         endp++;
1770         if (!*endp)
1771                 die("Missing dest: %s", command_buf.buf);
1772
1773         d = endp;
1774         d_uq = unquote_c_style(d, &endp);
1775         if (d_uq) {
1776                 if (*endp)
1777                         die("Garbage after dest in: %s", command_buf.buf);
1778                 d = d_uq;
1779         }
1780
1781         memset(&leaf, 0, sizeof(leaf));
1782         if (rename)
1783                 tree_content_remove(&b->branch_tree, s, &leaf);
1784         else
1785                 tree_content_get(&b->branch_tree, s, &leaf);
1786         if (!leaf.versions[1].mode)
1787                 die("Path %s not in branch", s);
1788         tree_content_set(&b->branch_tree, d,
1789                 leaf.versions[1].sha1,
1790                 leaf.versions[1].mode,
1791                 leaf.tree);
1792
1793         free(s_uq);
1794         free(d_uq);
1795 }
1796
1797 static void file_change_deleteall(struct branch *b)
1798 {
1799         release_tree_content_recursive(b->branch_tree.tree);
1800         hashclr(b->branch_tree.versions[0].sha1);
1801         hashclr(b->branch_tree.versions[1].sha1);
1802         load_tree(&b->branch_tree);
1803 }
1804
1805 static void cmd_from_commit(struct branch *b, char *buf, unsigned long size)
1806 {
1807         if (!buf || size < 46)
1808                 die("Not a valid commit: %s", sha1_to_hex(b->sha1));
1809         if (memcmp("tree ", buf, 5)
1810                 || get_sha1_hex(buf + 5, b->branch_tree.versions[1].sha1))
1811                 die("The commit %s is corrupt", sha1_to_hex(b->sha1));
1812         hashcpy(b->branch_tree.versions[0].sha1,
1813                 b->branch_tree.versions[1].sha1);
1814 }
1815
1816 static void cmd_from_existing(struct branch *b)
1817 {
1818         if (is_null_sha1(b->sha1)) {
1819                 hashclr(b->branch_tree.versions[0].sha1);
1820                 hashclr(b->branch_tree.versions[1].sha1);
1821         } else {
1822                 unsigned long size;
1823                 char *buf;
1824
1825                 buf = read_object_with_reference(b->sha1,
1826                         commit_type, &size, b->sha1);
1827                 cmd_from_commit(b, buf, size);
1828                 free(buf);
1829         }
1830 }
1831
1832 static int cmd_from(struct branch *b)
1833 {
1834         const char *from;
1835         struct branch *s;
1836
1837         if (prefixcmp(command_buf.buf, "from "))
1838                 return 0;
1839
1840         if (b->branch_tree.tree) {
1841                 release_tree_content_recursive(b->branch_tree.tree);
1842                 b->branch_tree.tree = NULL;
1843         }
1844
1845         from = strchr(command_buf.buf, ' ') + 1;
1846         s = lookup_branch(from);
1847         if (b == s)
1848                 die("Can't create a branch from itself: %s", b->name);
1849         else if (s) {
1850                 unsigned char *t = s->branch_tree.versions[1].sha1;
1851                 hashcpy(b->sha1, s->sha1);
1852                 hashcpy(b->branch_tree.versions[0].sha1, t);
1853                 hashcpy(b->branch_tree.versions[1].sha1, t);
1854         } else if (*from == ':') {
1855                 uintmax_t idnum = strtoumax(from + 1, NULL, 10);
1856                 struct object_entry *oe = find_mark(idnum);
1857                 if (oe->type != OBJ_COMMIT)
1858                         die("Mark :%" PRIuMAX " not a commit", idnum);
1859                 hashcpy(b->sha1, oe->sha1);
1860                 if (oe->pack_id != MAX_PACK_ID) {
1861                         unsigned long size;
1862                         char *buf = gfi_unpack_entry(oe, &size);
1863                         cmd_from_commit(b, buf, size);
1864                         free(buf);
1865                 } else
1866                         cmd_from_existing(b);
1867         } else if (!get_sha1(from, b->sha1))
1868                 cmd_from_existing(b);
1869         else
1870                 die("Invalid ref name or SHA1 expression: %s", from);
1871
1872         read_next_command();
1873         return 1;
1874 }
1875
1876 static struct hash_list *cmd_merge(unsigned int *count)
1877 {
1878         struct hash_list *list = NULL, *n, *e = e;
1879         const char *from;
1880         struct branch *s;
1881
1882         *count = 0;
1883         while (!prefixcmp(command_buf.buf, "merge ")) {
1884                 from = strchr(command_buf.buf, ' ') + 1;
1885                 n = xmalloc(sizeof(*n));
1886                 s = lookup_branch(from);
1887                 if (s)
1888                         hashcpy(n->sha1, s->sha1);
1889                 else if (*from == ':') {
1890                         uintmax_t idnum = strtoumax(from + 1, NULL, 10);
1891                         struct object_entry *oe = find_mark(idnum);
1892                         if (oe->type != OBJ_COMMIT)
1893                                 die("Mark :%" PRIuMAX " not a commit", idnum);
1894                         hashcpy(n->sha1, oe->sha1);
1895                 } else if (!get_sha1(from, n->sha1)) {
1896                         unsigned long size;
1897                         char *buf = read_object_with_reference(n->sha1,
1898                                 commit_type, &size, n->sha1);
1899                         if (!buf || size < 46)
1900                                 die("Not a valid commit: %s", from);
1901                         free(buf);
1902                 } else
1903                         die("Invalid ref name or SHA1 expression: %s", from);
1904
1905                 n->next = NULL;
1906                 if (list)
1907                         e->next = n;
1908                 else
1909                         list = n;
1910                 e = n;
1911                 (*count)++;
1912                 read_next_command();
1913         }
1914         return list;
1915 }
1916
1917 static void cmd_new_commit(void)
1918 {
1919         struct branch *b;
1920         void *msg;
1921         size_t msglen;
1922         char *sp;
1923         char *author = NULL;
1924         char *committer = NULL;
1925         struct hash_list *merge_list = NULL;
1926         unsigned int merge_count;
1927
1928         /* Obtain the branch name from the rest of our command */
1929         sp = strchr(command_buf.buf, ' ') + 1;
1930         b = lookup_branch(sp);
1931         if (!b)
1932                 b = new_branch(sp);
1933
1934         read_next_command();
1935         cmd_mark();
1936         if (!prefixcmp(command_buf.buf, "author ")) {
1937                 author = parse_ident(command_buf.buf + 7);
1938                 read_next_command();
1939         }
1940         if (!prefixcmp(command_buf.buf, "committer ")) {
1941                 committer = parse_ident(command_buf.buf + 10);
1942                 read_next_command();
1943         }
1944         if (!committer)
1945                 die("Expected committer but didn't get one");
1946         msg = cmd_data(&msglen);
1947         read_next_command();
1948         cmd_from(b);
1949         merge_list = cmd_merge(&merge_count);
1950
1951         /* ensure the branch is active/loaded */
1952         if (!b->branch_tree.tree || !max_active_branches) {
1953                 unload_one_branch();
1954                 load_branch(b);
1955         }
1956
1957         /* file_change* */
1958         while (!command_buf.eof && command_buf.len > 1) {
1959                 if (!prefixcmp(command_buf.buf, "M "))
1960                         file_change_m(b);
1961                 else if (!prefixcmp(command_buf.buf, "D "))
1962                         file_change_d(b);
1963                 else if (!prefixcmp(command_buf.buf, "R "))
1964                         file_change_cr(b, 1);
1965                 else if (!prefixcmp(command_buf.buf, "C "))
1966                         file_change_cr(b, 0);
1967                 else if (!strcmp("deleteall", command_buf.buf))
1968                         file_change_deleteall(b);
1969                 else {
1970                         unread_command_buf = 1;
1971                         break;
1972                 }
1973                 read_next_command();
1974         }
1975
1976         /* build the tree and the commit */
1977         store_tree(&b->branch_tree);
1978         hashcpy(b->branch_tree.versions[0].sha1,
1979                 b->branch_tree.versions[1].sha1);
1980         size_dbuf(&new_data, 114 + msglen
1981                 + merge_count * 49
1982                 + (author
1983                         ? strlen(author) + strlen(committer)
1984                         : 2 * strlen(committer)));
1985         sp = new_data.buffer;
1986         sp += sprintf(sp, "tree %s\n",
1987                 sha1_to_hex(b->branch_tree.versions[1].sha1));
1988         if (!is_null_sha1(b->sha1))
1989                 sp += sprintf(sp, "parent %s\n", sha1_to_hex(b->sha1));
1990         while (merge_list) {
1991                 struct hash_list *next = merge_list->next;
1992                 sp += sprintf(sp, "parent %s\n", sha1_to_hex(merge_list->sha1));
1993                 free(merge_list);
1994                 merge_list = next;
1995         }
1996         sp += sprintf(sp, "author %s\n", author ? author : committer);
1997         sp += sprintf(sp, "committer %s\n", committer);
1998         *sp++ = '\n';
1999         memcpy(sp, msg, msglen);
2000         sp += msglen;
2001         free(author);
2002         free(committer);
2003         free(msg);
2004
2005         if (!store_object(OBJ_COMMIT,
2006                 new_data.buffer, sp - (char*)new_data.buffer,
2007                 NULL, b->sha1, next_mark))
2008                 b->pack_id = pack_id;
2009         b->last_commit = object_count_by_type[OBJ_COMMIT];
2010 }
2011
2012 static void cmd_new_tag(void)
2013 {
2014         char *sp;
2015         const char *from;
2016         char *tagger;
2017         struct branch *s;
2018         void *msg;
2019         size_t msglen;
2020         struct tag *t;
2021         uintmax_t from_mark = 0;
2022         unsigned char sha1[20];
2023
2024         /* Obtain the new tag name from the rest of our command */
2025         sp = strchr(command_buf.buf, ' ') + 1;
2026         t = pool_alloc(sizeof(struct tag));
2027         t->next_tag = NULL;
2028         t->name = pool_strdup(sp);
2029         if (last_tag)
2030                 last_tag->next_tag = t;
2031         else
2032                 first_tag = t;
2033         last_tag = t;
2034         read_next_command();
2035
2036         /* from ... */
2037         if (prefixcmp(command_buf.buf, "from "))
2038                 die("Expected from command, got %s", command_buf.buf);
2039         from = strchr(command_buf.buf, ' ') + 1;
2040         s = lookup_branch(from);
2041         if (s) {
2042                 hashcpy(sha1, s->sha1);
2043         } else if (*from == ':') {
2044                 struct object_entry *oe;
2045                 from_mark = strtoumax(from + 1, NULL, 10);
2046                 oe = find_mark(from_mark);
2047                 if (oe->type != OBJ_COMMIT)
2048                         die("Mark :%" PRIuMAX " not a commit", from_mark);
2049                 hashcpy(sha1, oe->sha1);
2050         } else if (!get_sha1(from, sha1)) {
2051                 unsigned long size;
2052                 char *buf;
2053
2054                 buf = read_object_with_reference(sha1,
2055                         commit_type, &size, sha1);
2056                 if (!buf || size < 46)
2057                         die("Not a valid commit: %s", from);
2058                 free(buf);
2059         } else
2060                 die("Invalid ref name or SHA1 expression: %s", from);
2061         read_next_command();
2062
2063         /* tagger ... */
2064         if (prefixcmp(command_buf.buf, "tagger "))
2065                 die("Expected tagger command, got %s", command_buf.buf);
2066         tagger = parse_ident(command_buf.buf + 7);
2067
2068         /* tag payload/message */
2069         read_next_command();
2070         msg = cmd_data(&msglen);
2071
2072         /* build the tag object */
2073         size_dbuf(&new_data, 67+strlen(t->name)+strlen(tagger)+msglen);
2074         sp = new_data.buffer;
2075         sp += sprintf(sp, "object %s\n", sha1_to_hex(sha1));
2076         sp += sprintf(sp, "type %s\n", commit_type);
2077         sp += sprintf(sp, "tag %s\n", t->name);
2078         sp += sprintf(sp, "tagger %s\n", tagger);
2079         *sp++ = '\n';
2080         memcpy(sp, msg, msglen);
2081         sp += msglen;
2082         free(tagger);
2083         free(msg);
2084
2085         if (store_object(OBJ_TAG, new_data.buffer,
2086                 sp - (char*)new_data.buffer,
2087                 NULL, t->sha1, 0))
2088                 t->pack_id = MAX_PACK_ID;
2089         else
2090                 t->pack_id = pack_id;
2091 }
2092
2093 static void cmd_reset_branch(void)
2094 {
2095         struct branch *b;
2096         char *sp;
2097
2098         /* Obtain the branch name from the rest of our command */
2099         sp = strchr(command_buf.buf, ' ') + 1;
2100         b = lookup_branch(sp);
2101         if (b) {
2102                 hashclr(b->sha1);
2103                 hashclr(b->branch_tree.versions[0].sha1);
2104                 hashclr(b->branch_tree.versions[1].sha1);
2105                 if (b->branch_tree.tree) {
2106                         release_tree_content_recursive(b->branch_tree.tree);
2107                         b->branch_tree.tree = NULL;
2108                 }
2109         }
2110         else
2111                 b = new_branch(sp);
2112         read_next_command();
2113         if (!cmd_from(b) && command_buf.len > 1)
2114                 unread_command_buf = 1;
2115 }
2116
2117 static void cmd_checkpoint(void)
2118 {
2119         if (object_count) {
2120                 cycle_packfile();
2121                 dump_branches();
2122                 dump_tags();
2123                 dump_marks();
2124         }
2125         skip_optional_lf();
2126 }
2127
2128 static void import_marks(const char *input_file)
2129 {
2130         char line[512];
2131         FILE *f = fopen(input_file, "r");
2132         if (!f)
2133                 die("cannot read %s: %s", input_file, strerror(errno));
2134         while (fgets(line, sizeof(line), f)) {
2135                 uintmax_t mark;
2136                 char *end;
2137                 unsigned char sha1[20];
2138                 struct object_entry *e;
2139
2140                 end = strchr(line, '\n');
2141                 if (line[0] != ':' || !end)
2142                         die("corrupt mark line: %s", line);
2143                 *end = 0;
2144                 mark = strtoumax(line + 1, &end, 10);
2145                 if (!mark || end == line + 1
2146                         || *end != ' ' || get_sha1(end + 1, sha1))
2147                         die("corrupt mark line: %s", line);
2148                 e = find_object(sha1);
2149                 if (!e) {
2150                         enum object_type type = sha1_object_info(sha1, NULL);
2151                         if (type < 0)
2152                                 die("object not found: %s", sha1_to_hex(sha1));
2153                         e = insert_object(sha1);
2154                         e->type = type;
2155                         e->pack_id = MAX_PACK_ID;
2156                         e->offset = 1; /* just not zero! */
2157                 }
2158                 insert_mark(mark, e);
2159         }
2160         fclose(f);
2161 }
2162
2163 static const char fast_import_usage[] =
2164 "git-fast-import [--date-format=f] [--max-pack-size=n] [--depth=n] [--active-branches=n] [--export-marks=marks.file]";
2165
2166 int main(int argc, const char **argv)
2167 {
2168         int i, show_stats = 1;
2169
2170         git_config(git_default_config);
2171         alloc_objects(object_entry_alloc);
2172         strbuf_init(&command_buf);
2173         atom_table = xcalloc(atom_table_sz, sizeof(struct atom_str*));
2174         branch_table = xcalloc(branch_table_sz, sizeof(struct branch*));
2175         avail_tree_table = xcalloc(avail_tree_table_sz, sizeof(struct avail_tree_content*));
2176         marks = pool_calloc(1, sizeof(struct mark_set));
2177
2178         for (i = 1; i < argc; i++) {
2179                 const char *a = argv[i];
2180
2181                 if (*a != '-' || !strcmp(a, "--"))
2182                         break;
2183                 else if (!prefixcmp(a, "--date-format=")) {
2184                         const char *fmt = a + 14;
2185                         if (!strcmp(fmt, "raw"))
2186                                 whenspec = WHENSPEC_RAW;
2187                         else if (!strcmp(fmt, "rfc2822"))
2188                                 whenspec = WHENSPEC_RFC2822;
2189                         else if (!strcmp(fmt, "now"))
2190                                 whenspec = WHENSPEC_NOW;
2191                         else
2192                                 die("unknown --date-format argument %s", fmt);
2193                 }
2194                 else if (!prefixcmp(a, "--max-pack-size="))
2195                         max_packsize = strtoumax(a + 16, NULL, 0) * 1024 * 1024;
2196                 else if (!prefixcmp(a, "--depth="))
2197                         max_depth = strtoul(a + 8, NULL, 0);
2198                 else if (!prefixcmp(a, "--active-branches="))
2199                         max_active_branches = strtoul(a + 18, NULL, 0);
2200                 else if (!prefixcmp(a, "--import-marks="))
2201                         import_marks(a + 15);
2202                 else if (!prefixcmp(a, "--export-marks="))
2203                         mark_file = a + 15;
2204                 else if (!prefixcmp(a, "--export-pack-edges=")) {
2205                         if (pack_edges)
2206                                 fclose(pack_edges);
2207                         pack_edges = fopen(a + 20, "a");
2208                         if (!pack_edges)
2209                                 die("Cannot open %s: %s", a + 20, strerror(errno));
2210                 } else if (!strcmp(a, "--force"))
2211                         force_update = 1;
2212                 else if (!strcmp(a, "--quiet"))
2213                         show_stats = 0;
2214                 else if (!strcmp(a, "--stats"))
2215                         show_stats = 1;
2216                 else
2217                         die("unknown option %s", a);
2218         }
2219         if (i != argc)
2220                 usage(fast_import_usage);
2221
2222         prepare_packed_git();
2223         start_packfile();
2224         for (;;) {
2225                 read_next_command();
2226                 if (command_buf.eof)
2227                         break;
2228                 else if (!strcmp("blob", command_buf.buf))
2229                         cmd_new_blob();
2230                 else if (!prefixcmp(command_buf.buf, "commit "))
2231                         cmd_new_commit();
2232                 else if (!prefixcmp(command_buf.buf, "tag "))
2233                         cmd_new_tag();
2234                 else if (!prefixcmp(command_buf.buf, "reset "))
2235                         cmd_reset_branch();
2236                 else if (!strcmp("checkpoint", command_buf.buf))
2237                         cmd_checkpoint();
2238                 else
2239                         die("Unsupported command: %s", command_buf.buf);
2240         }
2241         end_packfile();
2242
2243         dump_branches();
2244         dump_tags();
2245         unkeep_all_packs();
2246         dump_marks();
2247
2248         if (pack_edges)
2249                 fclose(pack_edges);
2250
2251         if (show_stats) {
2252                 uintmax_t total_count = 0, duplicate_count = 0;
2253                 for (i = 0; i < ARRAY_SIZE(object_count_by_type); i++)
2254                         total_count += object_count_by_type[i];
2255                 for (i = 0; i < ARRAY_SIZE(duplicate_count_by_type); i++)
2256                         duplicate_count += duplicate_count_by_type[i];
2257
2258                 fprintf(stderr, "%s statistics:\n", argv[0]);
2259                 fprintf(stderr, "---------------------------------------------------------------------\n");
2260                 fprintf(stderr, "Alloc'd objects: %10" PRIuMAX "\n", alloc_count);
2261                 fprintf(stderr, "Total objects:   %10" PRIuMAX " (%10" PRIuMAX " duplicates                  )\n", total_count, duplicate_count);
2262                 fprintf(stderr, "      blobs  :   %10" PRIuMAX " (%10" PRIuMAX " duplicates %10" PRIuMAX " deltas)\n", object_count_by_type[OBJ_BLOB], duplicate_count_by_type[OBJ_BLOB], delta_count_by_type[OBJ_BLOB]);
2263                 fprintf(stderr, "      trees  :   %10" PRIuMAX " (%10" PRIuMAX " duplicates %10" PRIuMAX " deltas)\n", object_count_by_type[OBJ_TREE], duplicate_count_by_type[OBJ_TREE], delta_count_by_type[OBJ_TREE]);
2264                 fprintf(stderr, "      commits:   %10" PRIuMAX " (%10" PRIuMAX " duplicates %10" PRIuMAX " deltas)\n", object_count_by_type[OBJ_COMMIT], duplicate_count_by_type[OBJ_COMMIT], delta_count_by_type[OBJ_COMMIT]);
2265                 fprintf(stderr, "      tags   :   %10" PRIuMAX " (%10" PRIuMAX " duplicates %10" PRIuMAX " deltas)\n", object_count_by_type[OBJ_TAG], duplicate_count_by_type[OBJ_TAG], delta_count_by_type[OBJ_TAG]);
2266                 fprintf(stderr, "Total branches:  %10lu (%10lu loads     )\n", branch_count, branch_load_count);
2267                 fprintf(stderr, "      marks:     %10" PRIuMAX " (%10" PRIuMAX " unique    )\n", (((uintmax_t)1) << marks->shift) * 1024, marks_set_count);
2268                 fprintf(stderr, "      atoms:     %10u\n", atom_cnt);
2269                 fprintf(stderr, "Memory total:    %10" PRIuMAX " KiB\n", (total_allocd + alloc_count*sizeof(struct object_entry))/1024);
2270                 fprintf(stderr, "       pools:    %10lu KiB\n", (unsigned long)(total_allocd/1024));
2271                 fprintf(stderr, "     objects:    %10" PRIuMAX " KiB\n", (alloc_count*sizeof(struct object_entry))/1024);
2272                 fprintf(stderr, "---------------------------------------------------------------------\n");
2273                 pack_report();
2274                 fprintf(stderr, "---------------------------------------------------------------------\n");
2275                 fprintf(stderr, "\n");
2276         }
2277
2278         return failure ? 1 : 0;
2279 }