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      # note: the first idnum in a stream should be 1 and subsequent
  44      # idnums should not have gaps between values as this will cause
  45      # the stream parser to reserve space for the gapped values.  An
  46          # idnum can be updated in the future to a new object by issuing
  47      # a new mark directive with the old idnum.
  48          #
  49   mark ::= 'mark' sp idnum lf;
  50
  51      # note: declen indicates the length of binary_data in bytes.
  52      # declen does not include the lf preceeding or trailing the
  53      # binary data.
  54      #
  55   data ::= 'data' sp declen lf
  56     binary_data
  57         lf;
  58
  59      # note: quoted strings are C-style quoting supporting \c for
  60      # common escapes of 'c' (e..g \n, \t, \\, \") or \nnn where nnn
  61          # is the signed byte value in octal.  Note that the only
  62      # characters which must actually be escaped to protect the
  63      # stream formatting is: \, " and LF.  Otherwise these values
  64          # are UTF8.
  65      #
  66   ref_str     ::= ref     | '"' quoted(ref)     '"' ;
  67   sha1exp_str ::= sha1exp | '"' quoted(sha1exp) '"' ;
  68   tag_str     ::= tag     | '"' quoted(tag)     '"' ;
  69   path_str    ::= path    | '"' quoted(path)    '"' ;
  70
  71   declen ::= # unsigned 32 bit value, ascii base10 notation;
  72   binary_data ::= # file content, not interpreted;
  73
  74   sp ::= # ASCII space character;
  75   lf ::= # ASCII newline (LF) character;
  76
  77      # note: a colon (':') must precede the numerical value assigned to
  78          # an idnum.  This is to distinguish it from a ref or tag name as
  79      # GIT does not permit ':' in ref or tag strings.
  80          #
  81   idnum   ::= ':' declen;
  82   path    ::= # GIT style file path, e.g. "a/b/c";
  83   ref     ::= # GIT ref name, e.g. "refs/heads/MOZ_GECKO_EXPERIMENT";
  84   tag     ::= # GIT tag name, e.g. "FIREFOX_1_5";
  85   sha1exp ::= # Any valid GIT SHA1 expression;
  86   hexsha1 ::= # SHA1 in hexadecimal format;
  87
  88      # note: name and email are UTF8 strings, however name must not
  89          # contain '<' or lf and email must not contain any of the
  90      # following: '<', '>', lf.
  91          #
  92   name  ::= # valid GIT author/committer name;
  93   email ::= # valid GIT author/committer email;
  94   ts    ::= # time since the epoch in seconds, ascii base10 notation;
  95   tz    ::= # GIT style timezone;
  96 */
  97
  98 #include "builtin.h"
  99 #include "cache.h"
 100 #include "object.h"
 101 #include "blob.h"
 102 #include "tree.h"
 103 #include "delta.h"
 104 #include "pack.h"
 105 #include "refs.h"
 106 #include "csum-file.h"
 107 #include "strbuf.h"
 108 #include "quote.h"
 109
 110 struct object_entry
 111 {
 112         struct object_entry *next;
 113         unsigned long offset;
 114         unsigned type : TYPE_BITS;
 115         unsigned char sha1[20];
 116 };
 117
 118 struct object_entry_pool
 119 {
 120         struct object_entry_pool *next_pool;
 121         struct object_entry *next_free;
 122         struct object_entry *end;
 123         struct object_entry entries[FLEX_ARRAY]; /* more */
 124 };
 125
 126 struct mark_set
 127 {
 128         int shift;
 129         union {
 130                 struct object_entry *marked[1024];
 131                 struct mark_set *sets[1024];
 132         } data;
 133 };
 134
 135 struct last_object
 136 {
 137         void *data;
 138         unsigned long len;
 139         unsigned long offset;
 140         unsigned int depth;
 141         unsigned no_free:1;
 142 };
 143
 144 struct mem_pool
 145 {
 146         struct mem_pool *next_pool;
 147         char *next_free;
 148         char *end;
 149         char space[FLEX_ARRAY]; /* more */
 150 };
 151
 152 struct atom_str
 153 {
 154         struct atom_str *next_atom;
 155         int str_len;
 156         char str_dat[FLEX_ARRAY]; /* more */
 157 };
 158
 159 struct tree_content;
 160 struct tree_entry
 161 {
 162         struct tree_content *tree;
 163         struct atom_str* name;
 164         struct tree_entry_ms
 165         {
 166                 unsigned int mode;
 167                 unsigned char sha1[20];
 168         } versions[2];
 169 };
 170
 171 struct tree_content
 172 {
 173         unsigned int entry_capacity; /* must match avail_tree_content */
 174         unsigned int entry_count;
 175         unsigned int delta_depth;
 176         struct tree_entry *entries[FLEX_ARRAY]; /* more */
 177 };
 178
 179 struct avail_tree_content
 180 {
 181         unsigned int entry_capacity; /* must match tree_content */
 182         struct avail_tree_content *next_avail;
 183 };
 184
 185 struct branch
 186 {
 187         struct branch *table_next_branch;
 188         struct branch *active_next_branch;
 189         const char *name;
 190         unsigned long last_commit;
 191         struct tree_entry branch_tree;
 192         unsigned char sha1[20];
 193 };
 194
 195 struct tag
 196 {
 197         struct tag *next_tag;
 198         const char *name;
 199         unsigned char sha1[20];
 200 };
 201
 202 struct dbuf
 203 {
 204         void *buffer;
 205         size_t capacity;
 206 };
 207
 208 struct hash_list
 209 {
 210         struct hash_list *next;
 211         unsigned char sha1[20];
 212 };
 213
 214 /* Stats and misc. counters */
 215 static unsigned long max_depth = 10;
 216 static unsigned long alloc_count;
 217 static unsigned long branch_count;
 218 static unsigned long branch_load_count;
 219 static unsigned long remap_count;
 220 static unsigned long object_count;
 221 static unsigned long duplicate_count;
 222 static unsigned long marks_set_count;
 223 static unsigned long object_count_by_type[1 << TYPE_BITS];
 224 static unsigned long duplicate_count_by_type[1 << TYPE_BITS];
 225 static unsigned long delta_count_by_type[1 << TYPE_BITS];
 226
 227 /* Memory pools */
 228 static size_t mem_pool_alloc = 2*1024*1024 - sizeof(struct mem_pool);
 229 static size_t total_allocd;
 230 static struct mem_pool *mem_pool;
 231
 232 /* Atom management */
 233 static unsigned int atom_table_sz = 4451;
 234 static unsigned int atom_cnt;
 235 static struct atom_str **atom_table;
 236
 237 /* The .pack file being generated */
 238 static struct packed_git *pack_data;
 239 static int pack_fd;
 240 static unsigned long pack_size;
 241 static unsigned char pack_sha1[20];
 242
 243 /* Table of objects we've written. */
 244 static unsigned int object_entry_alloc = 5000;
 245 static struct object_entry_pool *blocks;
 246 static struct object_entry *object_table[1 << 16];
 247 static struct mark_set *marks;
 248 static const char* mark_file;
 249
 250 /* Our last blob */
 251 static struct last_object last_blob;
 252
 253 /* Tree management */
 254 static unsigned int tree_entry_alloc = 1000;
 255 static void *avail_tree_entry;
 256 static unsigned int avail_tree_table_sz = 100;
 257 static struct avail_tree_content **avail_tree_table;
 258 static struct dbuf old_tree;
 259 static struct dbuf new_tree;
 260
 261 /* Branch data */
 262 static unsigned long max_active_branches = 5;
 263 static unsigned long cur_active_branches;
 264 static unsigned long branch_table_sz = 1039;
 265 static struct branch **branch_table;
 266 static struct branch *active_branches;
 267
 268 /* Tag data */
 269 static struct tag *first_tag;
 270 static struct tag *last_tag;
 271
 272 /* Input stream parsing */
 273 static struct strbuf command_buf;
 274 static unsigned long next_mark;
 275 static struct dbuf new_data;
 276 static FILE* branch_log;
 277
 278
 279 static void alloc_objects(unsigned int cnt)
 280 {
 281         struct object_entry_pool *b;
 282
 283         b = xmalloc(sizeof(struct object_entry_pool)
 284                 + cnt * sizeof(struct object_entry));
 285         b->next_pool = blocks;
 286         b->next_free = b->entries;
 287         b->end = b->entries + cnt;
 288         blocks = b;
 289         alloc_count += cnt;
 290 }
 291
 292 static struct object_entry* new_object(unsigned char *sha1)
 293 {
 294         struct object_entry *e;
 295
 296         if (blocks->next_free == blocks->end)
 297                 alloc_objects(object_entry_alloc);
 298
 299         e = blocks->next_free++;
 300         hashcpy(e->sha1, sha1);
 301         return e;
 302 }
 303
 304 static struct object_entry* find_object(unsigned char *sha1)
 305 {
 306         unsigned int h = sha1[0] << 8 | sha1[1];
 307         struct object_entry *e;
 308         for (e = object_table[h]; e; e = e->next)
 309                 if (!hashcmp(sha1, e->sha1))
 310                         return e;
 311         return NULL;
 312 }
 313
 314 static struct object_entry* insert_object(unsigned char *sha1)
 315 {
 316         unsigned int h = sha1[0] << 8 | sha1[1];
 317         struct object_entry *e = object_table[h];
 318         struct object_entry *p = NULL;
 319
 320         while (e) {
 321                 if (!hashcmp(sha1, e->sha1))
 322                         return e;
 323                 p = e;
 324                 e = e->next;
 325         }
 326
 327         e = new_object(sha1);
 328         e->next = NULL;
 329         e->offset = 0;
 330         if (p)
 331                 p->next = e;
 332         else
 333                 object_table[h] = e;
 334         return e;
 335 }
 336
 337 static unsigned int hc_str(const char *s, size_t len)
 338 {
 339         unsigned int r = 0;
 340         while (len-- > 0)
 341                 r = r * 31 + *s++;
 342         return r;
 343 }
 344
 345 static void* pool_alloc(size_t len)
 346 {
 347         struct mem_pool *p;
 348         void *r;
 349
 350         for (p = mem_pool; p; p = p->next_pool)
 351                 if ((p->end - p->next_free >= len))
 352                         break;
 353
 354         if (!p) {
 355                 if (len >= (mem_pool_alloc/2)) {
 356                         total_allocd += len;
 357                         return xmalloc(len);
 358                 }
 359                 total_allocd += sizeof(struct mem_pool) + mem_pool_alloc;
 360                 p = xmalloc(sizeof(struct mem_pool) + mem_pool_alloc);
 361                 p->next_pool = mem_pool;
 362                 p->next_free = p->space;
 363                 p->end = p->next_free + mem_pool_alloc;
 364                 mem_pool = p;
 365         }
 366
 367         r = p->next_free;
 368         /* round out to a pointer alignment */
 369         if (len & (sizeof(void*) - 1))
 370                 len += sizeof(void*) - (len & (sizeof(void*) - 1));
 371         p->next_free += len;
 372         return r;
 373 }
 374
 375 static void* pool_calloc(size_t count, size_t size)
 376 {
 377         size_t len = count * size;
 378         void *r = pool_alloc(len);
 379         memset(r, 0, len);
 380         return r;
 381 }
 382
 383 static char* pool_strdup(const char *s)
 384 {
 385         char *r = pool_alloc(strlen(s) + 1);
 386         strcpy(r, s);
 387         return r;
 388 }
 389
 390 static void size_dbuf(struct dbuf *b, size_t maxlen)
 391 {
 392         if (b->buffer) {
 393                 if (b->capacity >= maxlen)
 394                         return;
 395                 free(b->buffer);
 396         }
 397         b->capacity = ((maxlen / 1024) + 1) * 1024;
 398         b->buffer = xmalloc(b->capacity);
 399 }
 400
 401 static void insert_mark(unsigned long idnum, struct object_entry *oe)
 402 {
 403         struct mark_set *s = marks;
 404         while ((idnum >> s->shift) >= 1024) {
 405                 s = pool_calloc(1, sizeof(struct mark_set));
 406                 s->shift = marks->shift + 10;
 407                 s->data.sets[0] = marks;
 408                 marks = s;
 409         }
 410         while (s->shift) {
 411                 unsigned long i = idnum >> s->shift;
 412                 idnum -= i << s->shift;
 413                 if (!s->data.sets[i]) {
 414                         s->data.sets[i] = pool_calloc(1, sizeof(struct mark_set));
 415                         s->data.sets[i]->shift = s->shift - 10;
 416                 }
 417                 s = s->data.sets[i];
 418         }
 419         if (!s->data.marked[idnum])
 420                 marks_set_count++;
 421         s->data.marked[idnum] = oe;
 422 }
 423
 424 static struct object_entry* find_mark(unsigned long idnum)
 425 {
 426         unsigned long orig_idnum = idnum;
 427         struct mark_set *s = marks;
 428         struct object_entry *oe = NULL;
 429         if ((idnum >> s->shift) < 1024) {
 430                 while (s && s->shift) {
 431                         unsigned long i = idnum >> s->shift;
 432                         idnum -= i << s->shift;
 433                         s = s->data.sets[i];
 434                 }
 435                 if (s)
 436                         oe = s->data.marked[idnum];
 437         }
 438         if (!oe)
 439                 die("mark :%lu not declared", orig_idnum);
 440         return oe;
 441 }
 442
 443 static struct atom_str* to_atom(const char *s, size_t len)
 444 {
 445         unsigned int hc = hc_str(s, len) % atom_table_sz;
 446         struct atom_str *c;
 447
 448         for (c = atom_table[hc]; c; c = c->next_atom)
 449                 if (c->str_len == len && !strncmp(s, c->str_dat, len))
 450                         return c;
 451
 452         c = pool_alloc(sizeof(struct atom_str) + len + 1);
 453         c->str_len = len;
 454         strncpy(c->str_dat, s, len);
 455         c->str_dat[len] = 0;
 456         c->next_atom = atom_table[hc];
 457         atom_table[hc] = c;
 458         atom_cnt++;
 459         return c;
 460 }
 461
 462 static struct branch* lookup_branch(const char *name)
 463 {
 464         unsigned int hc = hc_str(name, strlen(name)) % branch_table_sz;
 465         struct branch *b;
 466
 467         for (b = branch_table[hc]; b; b = b->table_next_branch)
 468                 if (!strcmp(name, b->name))
 469                         return b;
 470         return NULL;
 471 }
 472
 473 static struct branch* new_branch(const char *name)
 474 {
 475         unsigned int hc = hc_str(name, strlen(name)) % branch_table_sz;
 476         struct branch* b = lookup_branch(name);
 477
 478         if (b)
 479                 die("Invalid attempt to create duplicate branch: %s", name);
 480         if (check_ref_format(name))
 481                 die("Branch name doesn't conform to GIT standards: %s", name);
 482
 483         b = pool_calloc(1, sizeof(struct branch));
 484         b->name = pool_strdup(name);
 485         b->table_next_branch = branch_table[hc];
 486         b->branch_tree.versions[0].mode = S_IFDIR;
 487         b->branch_tree.versions[1].mode = S_IFDIR;
 488         branch_table[hc] = b;
 489         branch_count++;
 490         return b;
 491 }
 492
 493 static unsigned int hc_entries(unsigned int cnt)
 494 {
 495         cnt = cnt & 7 ? (cnt / 8) + 1 : cnt / 8;
 496         return cnt < avail_tree_table_sz ? cnt : avail_tree_table_sz - 1;
 497 }
 498
 499 static struct tree_content* new_tree_content(unsigned int cnt)
 500 {
 501         struct avail_tree_content *f, *l = NULL;
 502         struct tree_content *t;
 503         unsigned int hc = hc_entries(cnt);
 504
 505         for (f = avail_tree_table[hc]; f; l = f, f = f->next_avail)
 506                 if (f->entry_capacity >= cnt)
 507                         break;
 508
 509         if (f) {
 510                 if (l)
 511                         l->next_avail = f->next_avail;
 512                 else
 513                         avail_tree_table[hc] = f->next_avail;
 514         } else {
 515                 cnt = cnt & 7 ? ((cnt / 8) + 1) * 8 : cnt;
 516                 f = pool_alloc(sizeof(*t) + sizeof(t->entries[0]) * cnt);
 517                 f->entry_capacity = cnt;
 518         }
 519
 520         t = (struct tree_content*)f;
 521         t->entry_count = 0;
 522         t->delta_depth = 0;
 523         return t;
 524 }
 525
 526 static void release_tree_entry(struct tree_entry *e);
 527 static void release_tree_content(struct tree_content *t)
 528 {
 529         struct avail_tree_content *f = (struct avail_tree_content*)t;
 530         unsigned int hc = hc_entries(f->entry_capacity);
 531         f->next_avail = avail_tree_table[hc];
 532         avail_tree_table[hc] = f;
 533 }
 534
 535 static void release_tree_content_recursive(struct tree_content *t)
 536 {
 537         unsigned int i;
 538         for (i = 0; i < t->entry_count; i++)
 539                 release_tree_entry(t->entries[i]);
 540         release_tree_content(t);
 541 }
 542
 543 static struct tree_content* grow_tree_content(
 544         struct tree_content *t,
 545         int amt)
 546 {
 547         struct tree_content *r = new_tree_content(t->entry_count + amt);
 548         r->entry_count = t->entry_count;
 549         r->delta_depth = t->delta_depth;
 550         memcpy(r->entries,t->entries,t->entry_count*sizeof(t->entries[0]));
 551         release_tree_content(t);
 552         return r;
 553 }
 554
 555 static struct tree_entry* new_tree_entry()
 556 {
 557         struct tree_entry *e;
 558
 559         if (!avail_tree_entry) {
 560                 unsigned int n = tree_entry_alloc;
 561                 total_allocd += n * sizeof(struct tree_entry);
 562                 avail_tree_entry = e = xmalloc(n * sizeof(struct tree_entry));
 563                 while (n-- > 1) {
 564                         *((void**)e) = e + 1;
 565                         e++;
 566                 }
 567                 *((void**)e) = NULL;
 568         }
 569
 570         e = avail_tree_entry;
 571         avail_tree_entry = *((void**)e);
 572         return e;
 573 }
 574
 575 static void release_tree_entry(struct tree_entry *e)
 576 {
 577         if (e->tree)
 578                 release_tree_content_recursive(e->tree);
 579         *((void**)e) = avail_tree_entry;
 580         avail_tree_entry = e;
 581 }
 582
 583 static void yread(int fd, void *buffer, size_t length)
 584 {
 585         ssize_t ret = 0;
 586         while (ret < length) {
 587                 ssize_t size = xread(fd, (char *) buffer + ret, length - ret);
 588                 if (!size)
 589                         die("Read from descriptor %i: end of stream", fd);
 590                 if (size < 0)
 591                         die("Read from descriptor %i: %s", fd, strerror(errno));
 592                 ret += size;
 593         }
 594 }
 595
 596 static size_t encode_header(
 597         enum object_type type,
 598         size_t size,
 599         unsigned char *hdr)
 600 {
 601         int n = 1;
 602         unsigned char c;
 603
 604         if (type < OBJ_COMMIT || type > OBJ_REF_DELTA)
 605                 die("bad type %d", type);
 606
 607         c = (type << 4) | (size & 15);
 608         size >>= 4;
 609         while (size) {
 610                 *hdr++ = c | 0x80;
 611                 c = size & 0x7f;
 612                 size >>= 7;
 613                 n++;
 614         }
 615         *hdr = c;
 616         return n;
 617 }
 618
 619 static int store_object(
 620         enum object_type type,
 621         void *dat,
 622         size_t datlen,
 623         struct last_object *last,
 624         unsigned char *sha1out,
 625         unsigned long mark)
 626 {
 627         void *out, *delta;
 628         struct object_entry *e;
 629         unsigned char hdr[96];
 630         unsigned char sha1[20];
 631         unsigned long hdrlen, deltalen;
 632         SHA_CTX c;
 633         z_stream s;
 634
 635         hdrlen = sprintf((char*)hdr,"%s %lu",type_names[type],datlen) + 1;
 636         SHA1_Init(&c);
 637         SHA1_Update(&c, hdr, hdrlen);
 638         SHA1_Update(&c, dat, datlen);
 639         SHA1_Final(sha1, &c);
 640         if (sha1out)
 641                 hashcpy(sha1out, sha1);
 642
 643         e = insert_object(sha1);
 644         if (mark)
 645                 insert_mark(mark, e);
 646         if (e->offset) {
 647                 duplicate_count++;
 648                 duplicate_count_by_type[type]++;
 649                 return 1;
 650         }
 651         e->type = type;
 652         e->offset = pack_size;
 653         object_count++;
 654         object_count_by_type[type]++;
 655
 656         if (last && last->data && last->depth < max_depth)
 657                 delta = diff_delta(last->data, last->len,
 658                         dat, datlen,
 659                         &deltalen, 0);
 660         else
 661                 delta = 0;
 662
 663         memset(&s, 0, sizeof(s));
 664         deflateInit(&s, zlib_compression_level);
 665
 666         if (delta) {
 667                 unsigned long ofs = e->offset - last->offset;
 668                 unsigned pos = sizeof(hdr) - 1;
 669
 670                 delta_count_by_type[type]++;
 671                 last->depth++;
 672                 s.next_in = delta;
 673                 s.avail_in = deltalen;
 674
 675                 hdrlen = encode_header(OBJ_OFS_DELTA, deltalen, hdr);
 676                 write_or_die(pack_fd, hdr, hdrlen);
 677                 pack_size += hdrlen;
 678
 679                 hdr[pos] = ofs & 127;
 680                 while (ofs >>= 7)
 681                         hdr[--pos] = 128 | (--ofs & 127);
 682                 write_or_die(pack_fd, hdr + pos, sizeof(hdr) - pos);
 683                 pack_size += sizeof(hdr) - pos;
 684         } else {
 685                 if (last)
 686                         last->depth = 0;
 687                 s.next_in = dat;
 688                 s.avail_in = datlen;
 689                 hdrlen = encode_header(type, datlen, hdr);
 690                 write_or_die(pack_fd, hdr, hdrlen);
 691                 pack_size += hdrlen;
 692         }
 693
 694         s.avail_out = deflateBound(&s, s.avail_in);
 695         s.next_out = out = xmalloc(s.avail_out);
 696         while (deflate(&s, Z_FINISH) == Z_OK)
 697                 /* nothing */;
 698         deflateEnd(&s);
 699
 700         write_or_die(pack_fd, out, s.total_out);
 701         pack_size += s.total_out;
 702
 703         free(out);
 704         if (delta)
 705                 free(delta);
 706         if (last) {
 707                 if (last->data && !last->no_free)
 708                         free(last->data);
 709                 last->data = dat;
 710                 last->offset = e->offset;
 711                 last->len = datlen;
 712         }
 713         return 0;
 714 }
 715
 716 static void *gfi_unpack_entry(unsigned long ofs, unsigned long *sizep)
 717 {
 718         char type[20];
 719         pack_data->pack_size = pack_size + 20;
 720         return unpack_entry(pack_data, ofs, type, sizep);
 721 }
 722
 723 static const char *get_mode(const char *str, unsigned int *modep)
 724 {
 725         unsigned char c;
 726         unsigned int mode = 0;
 727
 728         while ((c = *str++) != ' ') {
 729                 if (c < '0' || c > '7')
 730                         return NULL;
 731                 mode = (mode << 3) + (c - '0');
 732         }
 733         *modep = mode;
 734         return str;
 735 }
 736
 737 static void load_tree(struct tree_entry *root)
 738 {
 739         unsigned char* sha1 = root->versions[1].sha1;
 740         struct object_entry *myoe;
 741         struct tree_content *t;
 742         unsigned long size;
 743         char *buf;
 744         const char *c;
 745
 746         root->tree = t = new_tree_content(8);
 747         if (is_null_sha1(sha1))
 748                 return;
 749
 750         myoe = find_object(sha1);
 751         if (myoe) {
 752                 if (myoe->type != OBJ_TREE)
 753                         die("Not a tree: %s", sha1_to_hex(sha1));
 754                 t->delta_depth = 0;
 755                 buf = gfi_unpack_entry(myoe->offset, &size);
 756         } else {
 757                 char type[20];
 758                 buf = read_sha1_file(sha1, type, &size);
 759                 if (!buf || strcmp(type, tree_type))
 760                         die("Can't load tree %s", sha1_to_hex(sha1));
 761         }
 762
 763         c = buf;
 764         while (c != (buf + size)) {
 765                 struct tree_entry *e = new_tree_entry();
 766
 767                 if (t->entry_count == t->entry_capacity)
 768                         root->tree = t = grow_tree_content(t, 8);
 769                 t->entries[t->entry_count++] = e;
 770
 771                 e->tree = NULL;
 772                 c = get_mode(c, &e->versions[1].mode);
 773                 if (!c)
 774                         die("Corrupt mode in %s", sha1_to_hex(sha1));
 775                 e->versions[0].mode = e->versions[1].mode;
 776                 e->name = to_atom(c, strlen(c));
 777                 c += e->name->str_len + 1;
 778                 hashcpy(e->versions[0].sha1, (unsigned char*)c);
 779                 hashcpy(e->versions[1].sha1, (unsigned char*)c);
 780                 c += 20;
 781         }
 782         free(buf);
 783 }
 784
 785 static int tecmp0 (const void *_a, const void *_b)
 786 {
 787         struct tree_entry *a = *((struct tree_entry**)_a);
 788         struct tree_entry *b = *((struct tree_entry**)_b);
 789         return base_name_compare(
 790                 a->name->str_dat, a->name->str_len, a->versions[0].mode,
 791                 b->name->str_dat, b->name->str_len, b->versions[0].mode);
 792 }
 793
 794 static int tecmp1 (const void *_a, const void *_b)
 795 {
 796         struct tree_entry *a = *((struct tree_entry**)_a);
 797         struct tree_entry *b = *((struct tree_entry**)_b);
 798         return base_name_compare(
 799                 a->name->str_dat, a->name->str_len, a->versions[1].mode,
 800                 b->name->str_dat, b->name->str_len, b->versions[1].mode);
 801 }
 802
 803 static void mktree(struct tree_content *t,
 804         int v,
 805         unsigned long *szp,
 806         struct dbuf *b)
 807 {
 808         size_t maxlen = 0;
 809         unsigned int i;
 810         char *c;
 811
 812         if (!v)
 813                 qsort(t->entries,t->entry_count,sizeof(t->entries[0]),tecmp0);
 814         else
 815                 qsort(t->entries,t->entry_count,sizeof(t->entries[0]),tecmp1);
 816
 817         for (i = 0; i < t->entry_count; i++) {
 818                 if (t->entries[i]->versions[v].mode)
 819                         maxlen += t->entries[i]->name->str_len + 34;
 820         }
 821
 822         size_dbuf(b, maxlen);
 823         c = b->buffer;
 824         for (i = 0; i < t->entry_count; i++) {
 825                 struct tree_entry *e = t->entries[i];
 826                 if (!e->versions[v].mode)
 827                         continue;
 828                 c += sprintf(c, "%o", e->versions[v].mode);
 829                 *c++ = ' ';
 830                 strcpy(c, e->name->str_dat);
 831                 c += e->name->str_len + 1;
 832                 hashcpy((unsigned char*)c, e->versions[v].sha1);
 833                 c += 20;
 834         }
 835         *szp = c - (char*)b->buffer;
 836 }
 837
 838 static void store_tree(struct tree_entry *root)
 839 {
 840         struct tree_content *t = root->tree;
 841         unsigned int i, j, del;
 842         unsigned long new_len;
 843         struct last_object lo;
 844         struct object_entry *le;
 845
 846         if (!is_null_sha1(root->versions[1].sha1))
 847                 return;
 848
 849         for (i = 0; i < t->entry_count; i++) {
 850                 if (t->entries[i]->tree)
 851                         store_tree(t->entries[i]);
 852         }
 853
 854         le = find_object(root->versions[0].sha1);
 855         if (!S_ISDIR(root->versions[0].mode) || !le) {
 856                 lo.data = NULL;
 857                 lo.depth = 0;
 858         } else {
 859                 mktree(t, 0, &lo.len, &old_tree);
 860                 lo.data = old_tree.buffer;
 861                 lo.offset = le->offset;
 862                 lo.depth = t->delta_depth;
 863                 lo.no_free = 1;
 864         }
 865
 866         mktree(t, 1, &new_len, &new_tree);
 867         store_object(OBJ_TREE, new_tree.buffer, new_len,
 868                 &lo, root->versions[1].sha1, 0);
 869
 870         t->delta_depth = lo.depth;
 871         for (i = 0, j = 0, del = 0; i < t->entry_count; i++) {
 872                 struct tree_entry *e = t->entries[i];
 873                 if (e->versions[1].mode) {
 874                         e->versions[0].mode = e->versions[1].mode;
 875                         hashcpy(e->versions[0].sha1, e->versions[1].sha1);
 876                         t->entries[j++] = e;
 877                 } else {
 878                         release_tree_entry(e);
 879                         del++;
 880                 }
 881         }
 882         t->entry_count -= del;
 883 }
 884
 885 static int tree_content_set(
 886         struct tree_entry *root,
 887         const char *p,
 888         const unsigned char *sha1,
 889         const unsigned int mode)
 890 {
 891         struct tree_content *t = root->tree;
 892         const char *slash1;
 893         unsigned int i, n;
 894         struct tree_entry *e;
 895
 896         slash1 = strchr(p, '/');
 897         if (slash1)
 898                 n = slash1 - p;
 899         else
 900                 n = strlen(p);
 901
 902         for (i = 0; i < t->entry_count; i++) {
 903                 e = t->entries[i];
 904                 if (e->name->str_len == n && !strncmp(p, e->name->str_dat, n)) {
 905                         if (!slash1) {
 906                                 if (e->versions[1].mode == mode
 907                                                 && !hashcmp(e->versions[1].sha1, sha1))
 908                                         return 0;
 909                                 e->versions[1].mode = mode;
 910                                 hashcpy(e->versions[1].sha1, sha1);
 911                                 if (e->tree) {
 912                                         release_tree_content_recursive(e->tree);
 913                                         e->tree = NULL;
 914                                 }
 915                                 hashclr(root->versions[1].sha1);
 916                                 return 1;
 917                         }
 918                         if (!S_ISDIR(e->versions[1].mode)) {
 919                                 e->tree = new_tree_content(8);
 920                                 e->versions[1].mode = S_IFDIR;
 921                         }
 922                         if (!e->tree)
 923                                 load_tree(e);
 924                         if (tree_content_set(e, slash1 + 1, sha1, mode)) {
 925                                 hashclr(root->versions[1].sha1);
 926                                 return 1;
 927                         }
 928                         return 0;
 929                 }
 930         }
 931
 932         if (t->entry_count == t->entry_capacity)
 933                 root->tree = t = grow_tree_content(t, 8);
 934         e = new_tree_entry();
 935         e->name = to_atom(p, n);
 936         e->versions[0].mode = 0;
 937         hashclr(e->versions[0].sha1);
 938         t->entries[t->entry_count++] = e;
 939         if (slash1) {
 940                 e->tree = new_tree_content(8);
 941                 e->versions[1].mode = S_IFDIR;
 942                 tree_content_set(e, slash1 + 1, sha1, mode);
 943         } else {
 944                 e->tree = NULL;
 945                 e->versions[1].mode = mode;
 946                 hashcpy(e->versions[1].sha1, sha1);
 947         }
 948         hashclr(root->versions[1].sha1);
 949         return 1;
 950 }
 951
 952 static int tree_content_remove(struct tree_entry *root, const char *p)
 953 {
 954         struct tree_content *t = root->tree;
 955         const char *slash1;
 956         unsigned int i, n;
 957         struct tree_entry *e;
 958
 959         slash1 = strchr(p, '/');
 960         if (slash1)
 961                 n = slash1 - p;
 962         else
 963                 n = strlen(p);
 964
 965         for (i = 0; i < t->entry_count; i++) {
 966                 e = t->entries[i];
 967                 if (e->name->str_len == n && !strncmp(p, e->name->str_dat, n)) {
 968                         if (!slash1 || !S_ISDIR(e->versions[1].mode))
 969                                 goto del_entry;
 970                         if (!e->tree)
 971                                 load_tree(e);
 972                         if (tree_content_remove(e, slash1 + 1)) {
 973                                 for (n = 0; n < e->tree->entry_count; n++) {
 974                                         if (e->tree->entries[n]->versions[1].mode) {
 975                                                 hashclr(root->versions[1].sha1);
 976                                                 return 1;
 977                                         }
 978                                 }
 979                                 goto del_entry;
 980                         }
 981                         return 0;
 982                 }
 983         }
 984         return 0;
 985
 986 del_entry:
 987         if (e->tree) {
 988                 release_tree_content_recursive(e->tree);
 989                 e->tree = NULL;
 990         }
 991         e->versions[1].mode = 0;
 992         hashclr(e->versions[1].sha1);
 993         hashclr(root->versions[1].sha1);
 994         return 1;
 995 }
 996
 997 static void init_pack_header()
 998 {
 999         struct pack_header hdr;
1000
1001         hdr.hdr_signature = htonl(PACK_SIGNATURE);
1002         hdr.hdr_version = htonl(2);
1003         hdr.hdr_entries = 0;
1004
1005         write_or_die(pack_fd, &hdr, sizeof(hdr));
1006         pack_size = sizeof(hdr);
1007 }
1008
1009 static void fixup_header_footer()
1010 {
1011         SHA_CTX c;
1012         char hdr[8];
1013         unsigned long cnt;
1014         char *buf;
1015         size_t n;
1016
1017         if (lseek(pack_fd, 0, SEEK_SET) != 0)
1018                 die("Failed seeking to start: %s", strerror(errno));
1019
1020         SHA1_Init(&c);
1021         yread(pack_fd, hdr, 8);
1022         SHA1_Update(&c, hdr, 8);
1023
1024         cnt = htonl(object_count);
1025         SHA1_Update(&c, &cnt, 4);
1026         write_or_die(pack_fd, &cnt, 4);
1027
1028         buf = xmalloc(128 * 1024);
1029         for (;;) {
1030                 n = xread(pack_fd, buf, 128 * 1024);
1031                 if (n <= 0)
1032                         break;
1033                 SHA1_Update(&c, buf, n);
1034         }
1035         free(buf);
1036
1037         SHA1_Final(pack_sha1, &c);
1038         write_or_die(pack_fd, pack_sha1, sizeof(pack_sha1));
1039 }
1040
1041 static int oecmp (const void *_a, const void *_b)
1042 {
1043         struct object_entry *a = *((struct object_entry**)_a);
1044         struct object_entry *b = *((struct object_entry**)_b);
1045         return hashcmp(a->sha1, b->sha1);
1046 }
1047
1048 static void write_index(const char *idx_name)
1049 {
1050         struct sha1file *f;
1051         struct object_entry **idx, **c, **last;
1052         struct object_entry *e;
1053         struct object_entry_pool *o;
1054         unsigned int array[256];
1055         int i;
1056
1057         /* Build the sorted table of object IDs. */
1058         idx = xmalloc(object_count * sizeof(struct object_entry*));
1059         c = idx;
1060         for (o = blocks; o; o = o->next_pool)
1061                 for (e = o->entries; e != o->next_free; e++)
1062                         *c++ = e;
1063         last = idx + object_count;
1064         qsort(idx, object_count, sizeof(struct object_entry*), oecmp);
1065
1066         /* Generate the fan-out array. */
1067         c = idx;
1068         for (i = 0; i < 256; i++) {
1069                 struct object_entry **next = c;;
1070                 while (next < last) {
1071                         if ((*next)->sha1[0] != i)
1072                                 break;
1073                         next++;
1074                 }
1075                 array[i] = htonl(next - idx);
1076                 c = next;
1077         }
1078
1079         f = sha1create("%s", idx_name);
1080         sha1write(f, array, 256 * sizeof(int));
1081         for (c = idx; c != last; c++) {
1082                 unsigned int offset = htonl((*c)->offset);
1083                 sha1write(f, &offset, 4);
1084                 sha1write(f, (*c)->sha1, sizeof((*c)->sha1));
1085         }
1086         sha1write(f, pack_sha1, sizeof(pack_sha1));
1087         sha1close(f, NULL, 1);
1088         free(idx);
1089 }
1090
1091 static void dump_branches()
1092 {
1093         static const char *msg = "fast-import";
1094         unsigned int i;
1095         struct branch *b;
1096         struct ref_lock *lock;
1097
1098         for (i = 0; i < branch_table_sz; i++) {
1099                 for (b = branch_table[i]; b; b = b->table_next_branch) {
1100                         lock = lock_any_ref_for_update(b->name, NULL);
1101                         if (!lock || write_ref_sha1(lock, b->sha1, msg) < 0)
1102                                 die("Can't write %s", b->name);
1103                 }
1104         }
1105 }
1106
1107 static void dump_tags()
1108 {
1109         static const char *msg = "fast-import";
1110         struct tag *t;
1111         struct ref_lock *lock;
1112         char path[PATH_MAX];
1113
1114         for (t = first_tag; t; t = t->next_tag) {
1115                 sprintf(path, "refs/tags/%s", t->name);
1116                 lock = lock_any_ref_for_update(path, NULL);
1117                 if (!lock || write_ref_sha1(lock, t->sha1, msg) < 0)
1118                         die("Can't write %s", path);
1119         }
1120 }
1121
1122 static void dump_marks_helper(FILE *f,
1123         unsigned long base,
1124         struct mark_set *m)
1125 {
1126         int k;
1127         if (m->shift) {
1128                 for (k = 0; k < 1024; k++) {
1129                         if (m->data.sets[k])
1130                                 dump_marks_helper(f, (base + k) << m->shift,
1131                                         m->data.sets[k]);
1132                 }
1133         } else {
1134                 for (k = 0; k < 1024; k++) {
1135                         if (m->data.marked[k])
1136                                 fprintf(f, ":%lu %s\n", base + k,
1137                                         sha1_to_hex(m->data.marked[k]->sha1));
1138                 }
1139         }
1140 }
1141
1142 static void dump_marks()
1143 {
1144         if (mark_file)
1145         {
1146                 FILE *f = fopen(mark_file, "w");
1147                 dump_marks_helper(f, 0, marks);
1148                 fclose(f);
1149         }
1150 }
1151
1152 static void read_next_command()
1153 {
1154         read_line(&command_buf, stdin, '\n');
1155 }
1156
1157 static void cmd_mark()
1158 {
1159         if (!strncmp("mark :", command_buf.buf, 6)) {
1160                 next_mark = strtoul(command_buf.buf + 6, NULL, 10);
1161                 read_next_command();
1162         }
1163         else
1164                 next_mark = 0;
1165 }
1166
1167 static void* cmd_data (size_t *size)
1168 {
1169         size_t n = 0;
1170         void *buffer;
1171         size_t length;
1172
1173         if (strncmp("data ", command_buf.buf, 5))
1174                 die("Expected 'data n' command, found: %s", command_buf.buf);
1175
1176         length = strtoul(command_buf.buf + 5, NULL, 10);
1177         buffer = xmalloc(length);
1178
1179         while (n < length) {
1180                 size_t s = fread((char*)buffer + n, 1, length - n, stdin);
1181                 if (!s && feof(stdin))
1182                         die("EOF in data (%lu bytes remaining)", length - n);
1183                 n += s;
1184         }
1185
1186         if (fgetc(stdin) != '\n')
1187                 die("An lf did not trail the binary data as expected.");
1188
1189         *size = length;
1190         return buffer;
1191 }
1192
1193 static void cmd_new_blob()
1194 {
1195         size_t l;
1196         void *d;
1197
1198         read_next_command();
1199         cmd_mark();
1200         d = cmd_data(&l);
1201
1202         if (store_object(OBJ_BLOB, d, l, &last_blob, NULL, next_mark))
1203                 free(d);
1204 }
1205
1206 static void unload_one_branch()
1207 {
1208         while (cur_active_branches
1209                 && cur_active_branches >= max_active_branches) {
1210                 unsigned long min_commit = ULONG_MAX;
1211                 struct branch *e, *l = NULL, *p = NULL;
1212
1213                 for (e = active_branches; e; e = e->active_next_branch) {
1214                         if (e->last_commit < min_commit) {
1215                                 p = l;
1216                                 min_commit = e->last_commit;
1217                         }
1218                         l = e;
1219                 }
1220
1221                 if (p) {
1222                         e = p->active_next_branch;
1223                         p->active_next_branch = e->active_next_branch;
1224                 } else {
1225                         e = active_branches;
1226                         active_branches = e->active_next_branch;
1227                 }
1228                 e->active_next_branch = NULL;
1229                 if (e->branch_tree.tree) {
1230                         release_tree_content_recursive(e->branch_tree.tree);
1231                         e->branch_tree.tree = NULL;
1232                 }
1233                 cur_active_branches--;
1234         }
1235 }
1236
1237 static void load_branch(struct branch *b)
1238 {
1239         load_tree(&b->branch_tree);
1240         b->active_next_branch = active_branches;
1241         active_branches = b;
1242         cur_active_branches++;
1243         branch_load_count++;
1244 }
1245
1246 static void file_change_m(struct branch *b)
1247 {
1248         const char *p = command_buf.buf + 2;
1249         char *p_uq;
1250         const char *endp;
1251         struct object_entry *oe;
1252         unsigned char sha1[20];
1253         unsigned int mode;
1254         char type[20];
1255
1256         p = get_mode(p, &mode);
1257         if (!p)
1258                 die("Corrupt mode: %s", command_buf.buf);
1259         switch (mode) {
1260         case S_IFREG | 0644:
1261         case S_IFREG | 0755:
1262         case S_IFLNK:
1263         case 0644:
1264         case 0755:
1265                 /* ok */
1266                 break;
1267         default:
1268                 die("Corrupt mode: %s", command_buf.buf);
1269         }
1270
1271         if (*p == ':') {
1272                 char *x;
1273                 oe = find_mark(strtoul(p + 1, &x, 10));
1274                 hashcpy(sha1, oe->sha1);
1275                 p = x;
1276         } else {
1277                 if (get_sha1_hex(p, sha1))
1278                         die("Invalid SHA1: %s", command_buf.buf);
1279                 oe = find_object(sha1);
1280                 p += 40;
1281         }
1282         if (*p++ != ' ')
1283                 die("Missing space after SHA1: %s", command_buf.buf);
1284
1285         p_uq = unquote_c_style(p, &endp);
1286         if (p_uq) {
1287                 if (*endp)
1288                         die("Garbage after path in: %s", command_buf.buf);
1289                 p = p_uq;
1290         }
1291
1292         if (oe) {
1293                 if (oe->type != OBJ_BLOB)
1294                         die("Not a blob (actually a %s): %s",
1295                                 command_buf.buf, type_names[oe->type]);
1296         } else {
1297                 if (sha1_object_info(sha1, type, NULL))
1298                         die("Blob not found: %s", command_buf.buf);
1299                 if (strcmp(blob_type, type))
1300                         die("Not a blob (actually a %s): %s",
1301                                 command_buf.buf, type);
1302         }
1303
1304         tree_content_set(&b->branch_tree, p, sha1, S_IFREG | mode);
1305
1306         if (p_uq)
1307                 free(p_uq);
1308 }
1309
1310 static void file_change_d(struct branch *b)
1311 {
1312         const char *p = command_buf.buf + 2;
1313         char *p_uq;
1314         const char *endp;
1315
1316         p_uq = unquote_c_style(p, &endp);
1317         if (p_uq) {
1318                 if (*endp)
1319                         die("Garbage after path in: %s", command_buf.buf);
1320                 p = p_uq;
1321         }
1322         tree_content_remove(&b->branch_tree, p);
1323         if (p_uq)
1324                 free(p_uq);
1325 }
1326
1327 static void cmd_from(struct branch *b)
1328 {
1329         const char *from, *endp;
1330         char *str_uq;
1331         struct branch *s;
1332
1333         if (strncmp("from ", command_buf.buf, 5))
1334                 return;
1335
1336         if (b->last_commit)
1337                 die("Can't reinitailize branch %s", b->name);
1338
1339         from = strchr(command_buf.buf, ' ') + 1;
1340         str_uq = unquote_c_style(from, &endp);
1341         if (str_uq) {
1342                 if (*endp)
1343                         die("Garbage after string in: %s", command_buf.buf);
1344                 from = str_uq;
1345         }
1346
1347         s = lookup_branch(from);
1348         if (b == s)
1349                 die("Can't create a branch from itself: %s", b->name);
1350         else if (s) {
1351                 unsigned char *t = s->branch_tree.versions[1].sha1;
1352                 hashcpy(b->sha1, s->sha1);
1353                 hashcpy(b->branch_tree.versions[0].sha1, t);
1354                 hashcpy(b->branch_tree.versions[1].sha1, t);
1355         } else if (*from == ':') {
1356                 unsigned long idnum = strtoul(from + 1, NULL, 10);
1357                 struct object_entry *oe = find_mark(idnum);
1358                 unsigned long size;
1359                 char *buf;
1360                 if (oe->type != OBJ_COMMIT)
1361                         die("Mark :%lu not a commit", idnum);
1362                 hashcpy(b->sha1, oe->sha1);
1363                 buf = gfi_unpack_entry(oe->offset, &size);
1364                 if (!buf || size < 46)
1365                         die("Not a valid commit: %s", from);
1366                 if (memcmp("tree ", buf, 5)
1367                         || get_sha1_hex(buf + 5, b->branch_tree.versions[1].sha1))
1368                         die("The commit %s is corrupt", sha1_to_hex(b->sha1));
1369                 free(buf);
1370                 hashcpy(b->branch_tree.versions[0].sha1,
1371                         b->branch_tree.versions[1].sha1);
1372         } else if (!get_sha1(from, b->sha1)) {
1373                 if (is_null_sha1(b->sha1)) {
1374                         hashclr(b->branch_tree.versions[0].sha1);
1375                         hashclr(b->branch_tree.versions[1].sha1);
1376                 } else {
1377                         unsigned long size;
1378                         char *buf;
1379
1380                         buf = read_object_with_reference(b->sha1,
1381                                 type_names[OBJ_COMMIT], &size, b->sha1);
1382                         if (!buf || size < 46)
1383                                 die("Not a valid commit: %s", from);
1384                         if (memcmp("tree ", buf, 5)
1385                                 || get_sha1_hex(buf + 5, b->branch_tree.versions[1].sha1))
1386                                 die("The commit %s is corrupt", sha1_to_hex(b->sha1));
1387                         free(buf);
1388                         hashcpy(b->branch_tree.versions[0].sha1,
1389                                 b->branch_tree.versions[1].sha1);
1390                 }
1391         } else
1392                 die("Invalid ref name or SHA1 expression: %s", from);
1393
1394         read_next_command();
1395 }
1396
1397 static struct hash_list* cmd_merge(unsigned int *count)
1398 {
1399         struct hash_list *list = NULL, *n, *e;
1400         const char *from, *endp;
1401         char *str_uq;
1402         struct branch *s;
1403
1404         *count = 0;
1405         while (!strncmp("merge ", command_buf.buf, 6)) {
1406                 from = strchr(command_buf.buf, ' ') + 1;
1407                 str_uq = unquote_c_style(from, &endp);
1408                 if (str_uq) {
1409                         if (*endp)
1410                                 die("Garbage after string in: %s", command_buf.buf);
1411                         from = str_uq;
1412                 }
1413
1414                 n = xmalloc(sizeof(*n));
1415                 s = lookup_branch(from);
1416                 if (s)
1417                         hashcpy(n->sha1, s->sha1);
1418                 else if (*from == ':') {
1419                         unsigned long idnum = strtoul(from + 1, NULL, 10);
1420                         struct object_entry *oe = find_mark(idnum);
1421                         if (oe->type != OBJ_COMMIT)
1422                                 die("Mark :%lu not a commit", idnum);
1423                         hashcpy(n->sha1, oe->sha1);
1424                 } else if (get_sha1(from, n->sha1))
1425                         die("Invalid ref name or SHA1 expression: %s", from);
1426
1427                 n->next = NULL;
1428                 if (list)
1429                         e->next = n;
1430                 else
1431                         list = n;
1432                 e = n;
1433                 *count++;
1434                 read_next_command();
1435         }
1436         return list;
1437 }
1438
1439 static void cmd_new_commit()
1440 {
1441         struct branch *b;
1442         void *msg;
1443         size_t msglen;
1444         char *str_uq;
1445         const char *endp;
1446         char *sp;
1447         char *author = NULL;
1448         char *committer = NULL;
1449         struct hash_list *merge_list = NULL;
1450         unsigned int merge_count;
1451
1452         /* Obtain the branch name from the rest of our command */
1453         sp = strchr(command_buf.buf, ' ') + 1;
1454         str_uq = unquote_c_style(sp, &endp);
1455         if (str_uq) {
1456                 if (*endp)
1457                         die("Garbage after ref in: %s", command_buf.buf);
1458                 sp = str_uq;
1459         }
1460         b = lookup_branch(sp);
1461         if (!b)
1462                 b = new_branch(sp);
1463         if (str_uq)
1464                 free(str_uq);
1465
1466         read_next_command();
1467         cmd_mark();
1468         if (!strncmp("author ", command_buf.buf, 7)) {
1469                 author = strdup(command_buf.buf);
1470                 read_next_command();
1471         }
1472         if (!strncmp("committer ", command_buf.buf, 10)) {
1473                 committer = strdup(command_buf.buf);
1474                 read_next_command();
1475         }
1476         if (!committer)
1477                 die("Expected committer but didn't get one");
1478         msg = cmd_data(&msglen);
1479         read_next_command();
1480         cmd_from(b);
1481         merge_list = cmd_merge(&merge_count);
1482
1483         /* ensure the branch is active/loaded */
1484         if (!b->branch_tree.tree || !max_active_branches) {
1485                 unload_one_branch();
1486                 load_branch(b);
1487         }
1488
1489         /* file_change* */
1490         for (;;) {
1491                 if (1 == command_buf.len)
1492                         break;
1493                 else if (!strncmp("M ", command_buf.buf, 2))
1494                         file_change_m(b);
1495                 else if (!strncmp("D ", command_buf.buf, 2))
1496                         file_change_d(b);
1497                 else
1498                         die("Unsupported file_change: %s", command_buf.buf);
1499                 read_next_command();
1500         }
1501
1502         /* build the tree and the commit */
1503         store_tree(&b->branch_tree);
1504         hashcpy(b->branch_tree.versions[0].sha1,
1505                 b->branch_tree.versions[1].sha1);
1506         size_dbuf(&new_data, 97 + msglen
1507                 + merge_count * 49
1508                 + (author
1509                         ? strlen(author) + strlen(committer)
1510                         : 2 * strlen(committer)));
1511         sp = new_data.buffer;
1512         sp += sprintf(sp, "tree %s\n",
1513                 sha1_to_hex(b->branch_tree.versions[1].sha1));
1514         if (!is_null_sha1(b->sha1))
1515                 sp += sprintf(sp, "parent %s\n", sha1_to_hex(b->sha1));
1516         while (merge_list) {
1517                 struct hash_list *next = merge_list->next;
1518                 sp += sprintf(sp, "parent %s\n", sha1_to_hex(merge_list->sha1));
1519                 free(merge_list);
1520                 merge_list = next;
1521         }
1522         if (author)
1523                 sp += sprintf(sp, "%s\n", author);
1524         else
1525                 sp += sprintf(sp, "author %s\n", committer + 10);
1526         sp += sprintf(sp, "%s\n\n", committer);
1527         memcpy(sp, msg, msglen);
1528         sp += msglen;
1529         if (author)
1530                 free(author);
1531         free(committer);
1532         free(msg);
1533
1534         store_object(OBJ_COMMIT,
1535                 new_data.buffer, sp - (char*)new_data.buffer,
1536                 NULL, b->sha1, next_mark);
1537         b->last_commit = object_count_by_type[OBJ_COMMIT];
1538
1539         if (branch_log) {
1540                 int need_dq = quote_c_style(b->name, NULL, NULL, 0);
1541                 fprintf(branch_log, "commit ");
1542                 if (need_dq) {
1543                         fputc('"', branch_log);
1544                         quote_c_style(b->name, NULL, branch_log, 0);
1545                         fputc('"', branch_log);
1546                 } else
1547                         fprintf(branch_log, "%s", b->name);
1548                 fprintf(branch_log," :%lu %s\n",next_mark,sha1_to_hex(b->sha1));
1549         }
1550 }
1551
1552 static void cmd_new_tag()
1553 {
1554         char *str_uq;
1555         const char *endp;
1556         char *sp;
1557         const char *from;
1558         char *tagger;
1559         struct branch *s;
1560         void *msg;
1561         size_t msglen;
1562         struct tag *t;
1563         unsigned long from_mark = 0;
1564         unsigned char sha1[20];
1565
1566         /* Obtain the new tag name from the rest of our command */
1567         sp = strchr(command_buf.buf, ' ') + 1;
1568         str_uq = unquote_c_style(sp, &endp);
1569         if (str_uq) {
1570                 if (*endp)
1571                         die("Garbage after tag name in: %s", command_buf.buf);
1572                 sp = str_uq;
1573         }
1574         t = pool_alloc(sizeof(struct tag));
1575         t->next_tag = NULL;
1576         t->name = pool_strdup(sp);
1577         if (last_tag)
1578                 last_tag->next_tag = t;
1579         else
1580                 first_tag = t;
1581         last_tag = t;
1582         if (str_uq)
1583                 free(str_uq);
1584         read_next_command();
1585
1586         /* from ... */
1587         if (strncmp("from ", command_buf.buf, 5))
1588                 die("Expected from command, got %s", command_buf.buf);
1589
1590         from = strchr(command_buf.buf, ' ') + 1;
1591         str_uq = unquote_c_style(from, &endp);
1592         if (str_uq) {
1593                 if (*endp)
1594                         die("Garbage after string in: %s", command_buf.buf);
1595                 from = str_uq;
1596         }
1597
1598         s = lookup_branch(from);
1599         if (s) {
1600                 hashcpy(sha1, s->sha1);
1601         } else if (*from == ':') {
1602                 from_mark = strtoul(from + 1, NULL, 10);
1603                 struct object_entry *oe = find_mark(from_mark);
1604                 if (oe->type != OBJ_COMMIT)
1605                         die("Mark :%lu not a commit", from_mark);
1606                 hashcpy(sha1, oe->sha1);
1607         } else if (!get_sha1(from, sha1)) {
1608                 unsigned long size;
1609                 char *buf;
1610
1611                 buf = read_object_with_reference(sha1,
1612                         type_names[OBJ_COMMIT], &size, sha1);
1613                 if (!buf || size < 46)
1614                         die("Not a valid commit: %s", from);
1615                 free(buf);
1616         } else
1617                 die("Invalid ref name or SHA1 expression: %s", from);
1618
1619         if (str_uq)
1620                 free(str_uq);
1621         read_next_command();
1622
1623         /* tagger ... */
1624         if (strncmp("tagger ", command_buf.buf, 7))
1625                 die("Expected tagger command, got %s", command_buf.buf);
1626         tagger = strdup(command_buf.buf);
1627
1628         /* tag payload/message */
1629         read_next_command();
1630         msg = cmd_data(&msglen);
1631
1632         /* build the tag object */
1633         size_dbuf(&new_data, 67+strlen(t->name)+strlen(tagger)+msglen);
1634         sp = new_data.buffer;
1635         sp += sprintf(sp, "object %s\n", sha1_to_hex(sha1));
1636         sp += sprintf(sp, "type %s\n", type_names[OBJ_COMMIT]);
1637         sp += sprintf(sp, "tag %s\n", t->name);
1638         sp += sprintf(sp, "%s\n\n", tagger);
1639         memcpy(sp, msg, msglen);
1640         sp += msglen;
1641         free(tagger);
1642         free(msg);
1643
1644         store_object(OBJ_TAG, new_data.buffer, sp - (char*)new_data.buffer,
1645                 NULL, t->sha1, 0);
1646
1647         if (branch_log) {
1648                 int need_dq = quote_c_style(t->name, NULL, NULL, 0);
1649                 fprintf(branch_log, "tag ");
1650                 if (need_dq) {
1651                         fputc('"', branch_log);
1652                         quote_c_style(t->name, NULL, branch_log, 0);
1653                         fputc('"', branch_log);
1654                 } else
1655                         fprintf(branch_log, "%s", t->name);
1656                 fprintf(branch_log," :%lu %s\n",from_mark,sha1_to_hex(t->sha1));
1657         }
1658 }
1659
1660 static void cmd_reset_branch()
1661 {
1662         struct branch *b;
1663         char *str_uq;
1664         const char *endp;
1665         char *sp;
1666
1667         /* Obtain the branch name from the rest of our command */
1668         sp = strchr(command_buf.buf, ' ') + 1;
1669         str_uq = unquote_c_style(sp, &endp);
1670         if (str_uq) {
1671                 if (*endp)
1672                         die("Garbage after ref in: %s", command_buf.buf);
1673                 sp = str_uq;
1674         }
1675         b = lookup_branch(sp);
1676         if (b) {
1677                 b->last_commit = 0;
1678                 if (b->branch_tree.tree) {
1679                         release_tree_content_recursive(b->branch_tree.tree);
1680                         b->branch_tree.tree = NULL;
1681                 }
1682         }
1683         else
1684                 b = new_branch(sp);
1685         if (str_uq)
1686                 free(str_uq);
1687         read_next_command();
1688         cmd_from(b);
1689 }
1690
1691 static const char fast_import_usage[] =
1692 "git-fast-import [--objects=n] [--depth=n] [--active-branches=n] [--export-marks=marks.file] [--branch-log=log] temp.pack";
1693
1694 int main(int argc, const char **argv)
1695 {
1696         const char *base_name;
1697         int i;
1698         unsigned long est_obj_cnt = object_entry_alloc;
1699         char *pack_name;
1700         char *idx_name;
1701         struct stat sb;
1702
1703         setup_ident();
1704         git_config(git_default_config);
1705
1706         for (i = 1; i < argc; i++) {
1707                 const char *a = argv[i];
1708
1709                 if (*a != '-' || !strcmp(a, "--"))
1710                         break;
1711                 else if (!strncmp(a, "--objects=", 10))
1712                         est_obj_cnt = strtoul(a + 10, NULL, 0);
1713                 else if (!strncmp(a, "--depth=", 8))
1714                         max_depth = strtoul(a + 8, NULL, 0);
1715                 else if (!strncmp(a, "--active-branches=", 18))
1716                         max_active_branches = strtoul(a + 18, NULL, 0);
1717                 else if (!strncmp(a, "--export-marks=", 15))
1718                         mark_file = a + 15;
1719                 else if (!strncmp(a, "--branch-log=", 13)) {
1720                         branch_log = fopen(a + 13, "w");
1721                         if (!branch_log)
1722                                 die("Can't create %s: %s", a + 13, strerror(errno));
1723                 }
1724                 else
1725                         die("unknown option %s", a);
1726         }
1727         if ((i+1) != argc)
1728                 usage(fast_import_usage);
1729         base_name = argv[i];
1730
1731         pack_name = xmalloc(strlen(base_name) + 6);
1732         sprintf(pack_name, "%s.pack", base_name);
1733         idx_name = xmalloc(strlen(base_name) + 5);
1734         sprintf(idx_name, "%s.idx", base_name);
1735
1736         pack_fd = open(pack_name, O_RDWR|O_CREAT|O_EXCL, 0666);
1737         if (pack_fd < 0)
1738                 die("Can't create %s: %s", pack_name, strerror(errno));
1739
1740         pack_data = xcalloc(1, sizeof(*pack_data) + strlen(pack_name) + 2);
1741         strcpy(pack_data->pack_name, pack_name);
1742         pack_data->pack_fd = pack_fd;
1743
1744         init_pack_header();
1745         alloc_objects(est_obj_cnt);
1746         strbuf_init(&command_buf);
1747
1748         atom_table = xcalloc(atom_table_sz, sizeof(struct atom_str*));
1749         branch_table = xcalloc(branch_table_sz, sizeof(struct branch*));
1750         avail_tree_table = xcalloc(avail_tree_table_sz, sizeof(struct avail_tree_content*));
1751         marks = pool_calloc(1, sizeof(struct mark_set));
1752
1753         for (;;) {
1754                 read_next_command();
1755                 if (command_buf.eof)
1756                         break;
1757                 else if (!strcmp("blob", command_buf.buf))
1758                         cmd_new_blob();
1759                 else if (!strncmp("commit ", command_buf.buf, 7))
1760                         cmd_new_commit();
1761                 else if (!strncmp("tag ", command_buf.buf, 4))
1762                         cmd_new_tag();
1763                 else if (!strncmp("reset ", command_buf.buf, 6))
1764                         cmd_reset_branch();
1765                 else
1766                         die("Unsupported command: %s", command_buf.buf);
1767         }
1768
1769         fixup_header_footer();
1770         close(pack_fd);
1771         write_index(idx_name);
1772         dump_branches();
1773         dump_tags();
1774         dump_marks();
1775         if (branch_log)
1776                 fclose(branch_log);
1777
1778         fprintf(stderr, "%s statistics:\n", argv[0]);
1779         fprintf(stderr, "---------------------------------------------------------------------\n");
1780         fprintf(stderr, "Alloc'd objects: %10lu (%10lu overflow  )\n", alloc_count, alloc_count - est_obj_cnt);
1781         fprintf(stderr, "Total objects:   %10lu (%10lu duplicates                  )\n", object_count, duplicate_count);
1782         fprintf(stderr, "      blobs  :   %10lu (%10lu duplicates %10lu deltas)\n", object_count_by_type[OBJ_BLOB], duplicate_count_by_type[OBJ_BLOB], delta_count_by_type[OBJ_BLOB]);
1783         fprintf(stderr, "      trees  :   %10lu (%10lu duplicates %10lu deltas)\n", object_count_by_type[OBJ_TREE], duplicate_count_by_type[OBJ_TREE], delta_count_by_type[OBJ_TREE]);
1784         fprintf(stderr, "      commits:   %10lu (%10lu duplicates %10lu deltas)\n", object_count_by_type[OBJ_COMMIT], duplicate_count_by_type[OBJ_COMMIT], delta_count_by_type[OBJ_COMMIT]);
1785         fprintf(stderr, "      tags   :   %10lu (%10lu duplicates %10lu deltas)\n", object_count_by_type[OBJ_TAG], duplicate_count_by_type[OBJ_TAG], delta_count_by_type[OBJ_TAG]);
1786         fprintf(stderr, "Total branches:  %10lu (%10lu loads     )\n", branch_count, branch_load_count);
1787         fprintf(stderr, "      marks:     %10u (%10lu unique    )\n", (1 << marks->shift) * 1024, marks_set_count);
1788         fprintf(stderr, "      atoms:     %10u\n", atom_cnt);
1789         fprintf(stderr, "Memory total:    %10lu KiB\n", (total_allocd + alloc_count*sizeof(struct object_entry))/1024);
1790         fprintf(stderr, "       pools:    %10lu KiB\n", total_allocd/1024);
1791         fprintf(stderr, "     objects:    %10lu KiB\n", (alloc_count*sizeof(struct object_entry))/1024);
1792         fprintf(stderr, "Pack remaps:     %10lu\n", remap_count);
1793         stat(pack_name, &sb);
1794         fprintf(stderr, "Pack size:       %10lu KiB\n", (unsigned long)(sb.st_size/1024));
1795         stat(idx_name, &sb);
1796         fprintf(stderr, "Index size:      %10lu KiB\n", (unsigned long)(sb.st_size/1024));
1797         fprintf(stderr, "---------------------------------------------------------------------\n");
1798
1799         fprintf(stderr, "\n");
1800
1801         return 0;
1802 }