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

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