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 object_count;
 220 static unsigned long marks_set_count;
 221 static unsigned long object_count_by_type[1 << TYPE_BITS];
 222 static unsigned long duplicate_count_by_type[1 << TYPE_BITS];
 223 static unsigned long delta_count_by_type[1 << TYPE_BITS];
 224
 225 /* Memory pools */
 226 static size_t mem_pool_alloc = 2*1024*1024 - sizeof(struct mem_pool);
 227 static size_t total_allocd;
 228 static struct mem_pool *mem_pool;
 229
 230 /* Atom management */
 231 static unsigned int atom_table_sz = 4451;
 232 static unsigned int atom_cnt;
 233 static struct atom_str **atom_table;
 234
 235 /* The .pack file being generated */
 236 static const char *base_name;
 237 static unsigned int pack_count;
 238 static char *pack_name;
 239 static char *idx_name;
 240 static struct packed_git *pack_data;
 241 static int pack_fd;
 242 static unsigned long pack_size;
 243 static unsigned char pack_sha1[20];
 244
 245 /* Table of objects we've written. */
 246 static unsigned int object_entry_alloc = 5000;
 247 static struct object_entry_pool *blocks;
 248 static struct object_entry *object_table[1 << 16];
 249 static struct mark_set *marks;
 250 static const char* mark_file;
 251
 252 /* Our last blob */
 253 static struct last_object last_blob;
 254
 255 /* Tree management */
 256 static unsigned int tree_entry_alloc = 1000;
 257 static void *avail_tree_entry;
 258 static unsigned int avail_tree_table_sz = 100;
 259 static struct avail_tree_content **avail_tree_table;
 260 static struct dbuf old_tree;
 261 static struct dbuf new_tree;
 262
 263 /* Branch data */
 264 static unsigned long max_active_branches = 5;
 265 static unsigned long cur_active_branches;
 266 static unsigned long branch_table_sz = 1039;
 267 static struct branch **branch_table;
 268 static struct branch *active_branches;
 269
 270 /* Tag data */
 271 static struct tag *first_tag;
 272 static struct tag *last_tag;
 273
 274 /* Input stream parsing */
 275 static struct strbuf command_buf;
 276 static unsigned long next_mark;
 277 static struct dbuf new_data;
 278 static FILE* branch_log;
 279
 280
 281 static void alloc_objects(unsigned int cnt)
 282 {
 283         struct object_entry_pool *b;
 284
 285         b = xmalloc(sizeof(struct object_entry_pool)
 286                 + cnt * sizeof(struct object_entry));
 287         b->next_pool = blocks;
 288         b->next_free = b->entries;
 289         b->end = b->entries + cnt;
 290         blocks = b;
 291         alloc_count += cnt;
 292 }
 293
 294 static struct object_entry* new_object(unsigned char *sha1)
 295 {
 296         struct object_entry *e;
 297
 298         if (blocks->next_free == blocks->end)
 299                 alloc_objects(object_entry_alloc);
 300
 301         e = blocks->next_free++;
 302         hashcpy(e->sha1, sha1);
 303         return e;
 304 }
 305
 306 static struct object_entry* find_object(unsigned char *sha1)
 307 {
 308         unsigned int h = sha1[0] << 8 | sha1[1];
 309         struct object_entry *e;
 310         for (e = object_table[h]; e; e = e->next)
 311                 if (!hashcmp(sha1, e->sha1))
 312                         return e;
 313         return NULL;
 314 }
 315
 316 static struct object_entry* insert_object(unsigned char *sha1)
 317 {
 318         unsigned int h = sha1[0] << 8 | sha1[1];
 319         struct object_entry *e = object_table[h];
 320         struct object_entry *p = NULL;
 321
 322         while (e) {
 323                 if (!hashcmp(sha1, e->sha1))
 324                         return e;
 325                 p = e;
 326                 e = e->next;
 327         }
 328
 329         e = new_object(sha1);
 330         e->next = NULL;
 331         e->offset = 0;
 332         if (p)
 333                 p->next = e;
 334         else
 335                 object_table[h] = e;
 336         return e;
 337 }
 338
 339 static unsigned int hc_str(const char *s, size_t len)
 340 {
 341         unsigned int r = 0;
 342         while (len-- > 0)
 343                 r = r * 31 + *s++;
 344         return r;
 345 }
 346
 347 static void* pool_alloc(size_t len)
 348 {
 349         struct mem_pool *p;
 350         void *r;
 351
 352         for (p = mem_pool; p; p = p->next_pool)
 353                 if ((p->end - p->next_free >= len))
 354                         break;
 355
 356         if (!p) {
 357                 if (len >= (mem_pool_alloc/2)) {
 358                         total_allocd += len;
 359                         return xmalloc(len);
 360                 }
 361                 total_allocd += sizeof(struct mem_pool) + mem_pool_alloc;
 362                 p = xmalloc(sizeof(struct mem_pool) + mem_pool_alloc);
 363                 p->next_pool = mem_pool;
 364                 p->next_free = p->space;
 365                 p->end = p->next_free + mem_pool_alloc;
 366                 mem_pool = p;
 367         }
 368
 369         r = p->next_free;
 370         /* round out to a pointer alignment */
 371         if (len & (sizeof(void*) - 1))
 372                 len += sizeof(void*) - (len & (sizeof(void*) - 1));
 373         p->next_free += len;
 374         return r;
 375 }
 376
 377 static void* pool_calloc(size_t count, size_t size)
 378 {
 379         size_t len = count * size;
 380         void *r = pool_alloc(len);
 381         memset(r, 0, len);
 382         return r;
 383 }
 384
 385 static char* pool_strdup(const char *s)
 386 {
 387         char *r = pool_alloc(strlen(s) + 1);
 388         strcpy(r, s);
 389         return r;
 390 }
 391
 392 static void size_dbuf(struct dbuf *b, size_t maxlen)
 393 {
 394         if (b->buffer) {
 395                 if (b->capacity >= maxlen)
 396                         return;
 397                 free(b->buffer);
 398         }
 399         b->capacity = ((maxlen / 1024) + 1) * 1024;
 400         b->buffer = xmalloc(b->capacity);
 401 }
 402
 403 static void insert_mark(unsigned long idnum, struct object_entry *oe)
 404 {
 405         struct mark_set *s = marks;
 406         while ((idnum >> s->shift) >= 1024) {
 407                 s = pool_calloc(1, sizeof(struct mark_set));
 408                 s->shift = marks->shift + 10;
 409                 s->data.sets[0] = marks;
 410                 marks = s;
 411         }
 412         while (s->shift) {
 413                 unsigned long i = idnum >> s->shift;
 414                 idnum -= i << s->shift;
 415                 if (!s->data.sets[i]) {
 416                         s->data.sets[i] = pool_calloc(1, sizeof(struct mark_set));
 417                         s->data.sets[i]->shift = s->shift - 10;
 418                 }
 419                 s = s->data.sets[i];
 420         }
 421         if (!s->data.marked[idnum])
 422                 marks_set_count++;
 423         s->data.marked[idnum] = oe;
 424 }
 425
 426 static struct object_entry* find_mark(unsigned long idnum)
 427 {
 428         unsigned long orig_idnum = idnum;
 429         struct mark_set *s = marks;
 430         struct object_entry *oe = NULL;
 431         if ((idnum >> s->shift) < 1024) {
 432                 while (s && s->shift) {
 433                         unsigned long i = idnum >> s->shift;
 434                         idnum -= i << s->shift;
 435                         s = s->data.sets[i];
 436                 }
 437                 if (s)
 438                         oe = s->data.marked[idnum];
 439         }
 440         if (!oe)
 441                 die("mark :%lu not declared", orig_idnum);
 442         return oe;
 443 }
 444
 445 static struct atom_str* to_atom(const char *s, size_t len)
 446 {
 447         unsigned int hc = hc_str(s, len) % atom_table_sz;
 448         struct atom_str *c;
 449
 450         for (c = atom_table[hc]; c; c = c->next_atom)
 451                 if (c->str_len == len && !strncmp(s, c->str_dat, len))
 452                         return c;
 453
 454         c = pool_alloc(sizeof(struct atom_str) + len + 1);
 455         c->str_len = len;
 456         strncpy(c->str_dat, s, len);
 457         c->str_dat[len] = 0;
 458         c->next_atom = atom_table[hc];
 459         atom_table[hc] = c;
 460         atom_cnt++;
 461         return c;
 462 }
 463
 464 static struct branch* lookup_branch(const char *name)
 465 {
 466         unsigned int hc = hc_str(name, strlen(name)) % branch_table_sz;
 467         struct branch *b;
 468
 469         for (b = branch_table[hc]; b; b = b->table_next_branch)
 470                 if (!strcmp(name, b->name))
 471                         return b;
 472         return NULL;
 473 }
 474
 475 static struct branch* new_branch(const char *name)
 476 {
 477         unsigned int hc = hc_str(name, strlen(name)) % branch_table_sz;
 478         struct branch* b = lookup_branch(name);
 479
 480         if (b)
 481                 die("Invalid attempt to create duplicate branch: %s", name);
 482         if (check_ref_format(name))
 483                 die("Branch name doesn't conform to GIT standards: %s", name);
 484
 485         b = pool_calloc(1, sizeof(struct branch));
 486         b->name = pool_strdup(name);
 487         b->table_next_branch = branch_table[hc];
 488         b->branch_tree.versions[0].mode = S_IFDIR;
 489         b->branch_tree.versions[1].mode = S_IFDIR;
 490         branch_table[hc] = b;
 491         branch_count++;
 492         return b;
 493 }
 494
 495 static unsigned int hc_entries(unsigned int cnt)
 496 {
 497         cnt = cnt & 7 ? (cnt / 8) + 1 : cnt / 8;
 498         return cnt < avail_tree_table_sz ? cnt : avail_tree_table_sz - 1;
 499 }
 500
 501 static struct tree_content* new_tree_content(unsigned int cnt)
 502 {
 503         struct avail_tree_content *f, *l = NULL;
 504         struct tree_content *t;
 505         unsigned int hc = hc_entries(cnt);
 506
 507         for (f = avail_tree_table[hc]; f; l = f, f = f->next_avail)
 508                 if (f->entry_capacity >= cnt)
 509                         break;
 510
 511         if (f) {
 512                 if (l)
 513                         l->next_avail = f->next_avail;
 514                 else
 515                         avail_tree_table[hc] = f->next_avail;
 516         } else {
 517                 cnt = cnt & 7 ? ((cnt / 8) + 1) * 8 : cnt;
 518                 f = pool_alloc(sizeof(*t) + sizeof(t->entries[0]) * cnt);
 519                 f->entry_capacity = cnt;
 520         }
 521
 522         t = (struct tree_content*)f;
 523         t->entry_count = 0;
 524         t->delta_depth = 0;
 525         return t;
 526 }
 527
 528 static void release_tree_entry(struct tree_entry *e);
 529 static void release_tree_content(struct tree_content *t)
 530 {
 531         struct avail_tree_content *f = (struct avail_tree_content*)t;
 532         unsigned int hc = hc_entries(f->entry_capacity);
 533         f->next_avail = avail_tree_table[hc];
 534         avail_tree_table[hc] = f;
 535 }
 536
 537 static void release_tree_content_recursive(struct tree_content *t)
 538 {
 539         unsigned int i;
 540         for (i = 0; i < t->entry_count; i++)
 541                 release_tree_entry(t->entries[i]);
 542         release_tree_content(t);
 543 }
 544
 545 static struct tree_content* grow_tree_content(
 546         struct tree_content *t,
 547         int amt)
 548 {
 549         struct tree_content *r = new_tree_content(t->entry_count + amt);
 550         r->entry_count = t->entry_count;
 551         r->delta_depth = t->delta_depth;
 552         memcpy(r->entries,t->entries,t->entry_count*sizeof(t->entries[0]));
 553         release_tree_content(t);
 554         return r;
 555 }
 556
 557 static struct tree_entry* new_tree_entry()
 558 {
 559         struct tree_entry *e;
 560
 561         if (!avail_tree_entry) {
 562                 unsigned int n = tree_entry_alloc;
 563                 total_allocd += n * sizeof(struct tree_entry);
 564                 avail_tree_entry = e = xmalloc(n * sizeof(struct tree_entry));
 565                 while (n-- > 1) {
 566                         *((void**)e) = e + 1;
 567                         e++;
 568                 }
 569                 *((void**)e) = NULL;
 570         }
 571
 572         e = avail_tree_entry;
 573         avail_tree_entry = *((void**)e);
 574         return e;
 575 }
 576
 577 static void release_tree_entry(struct tree_entry *e)
 578 {
 579         if (e->tree)
 580                 release_tree_content_recursive(e->tree);
 581         *((void**)e) = avail_tree_entry;
 582         avail_tree_entry = e;
 583 }
 584
 585 static void yread(int fd, void *buffer, size_t length)
 586 {
 587         ssize_t ret = 0;
 588         while (ret < length) {
 589                 ssize_t size = xread(fd, (char *) buffer + ret, length - ret);
 590                 if (!size)
 591                         die("Read from descriptor %i: end of stream", fd);
 592                 if (size < 0)
 593                         die("Read from descriptor %i: %s", fd, strerror(errno));
 594                 ret += size;
 595         }
 596 }
 597
 598 static void start_packfile()
 599 {
 600         struct pack_header hdr;
 601
 602         pack_count++;
 603         pack_name = xmalloc(strlen(base_name) + 11);
 604         idx_name = xmalloc(strlen(base_name) + 11);
 605         sprintf(pack_name, "%s%5.5i.pack", base_name, pack_count);
 606         sprintf(idx_name, "%s%5.5i.idx", base_name, pack_count);
 607
 608         pack_fd = open(pack_name, O_RDWR|O_CREAT|O_EXCL, 0666);
 609         if (pack_fd < 0)
 610                 die("Can't create %s: %s", pack_name, strerror(errno));
 611
 612         pack_data = xcalloc(1, sizeof(*pack_data) + strlen(pack_name) + 2);
 613         strcpy(pack_data->pack_name, pack_name);
 614         pack_data->pack_fd = pack_fd;
 615
 616         hdr.hdr_signature = htonl(PACK_SIGNATURE);
 617         hdr.hdr_version = htonl(2);
 618         hdr.hdr_entries = 0;
 619
 620         write_or_die(pack_fd, &hdr, sizeof(hdr));
 621         pack_size = sizeof(hdr);
 622         object_count = 0;
 623 }
 624
 625 static void fixup_header_footer()
 626 {
 627         SHA_CTX c;
 628         char hdr[8];
 629         unsigned long cnt;
 630         char *buf;
 631
 632         if (lseek(pack_fd, 0, SEEK_SET) != 0)
 633                 die("Failed seeking to start: %s", strerror(errno));
 634
 635         SHA1_Init(&c);
 636         yread(pack_fd, hdr, 8);
 637         SHA1_Update(&c, hdr, 8);
 638
 639         cnt = htonl(object_count);
 640         SHA1_Update(&c, &cnt, 4);
 641         write_or_die(pack_fd, &cnt, 4);
 642
 643         buf = xmalloc(128 * 1024);
 644         for (;;) {
 645                 size_t n = xread(pack_fd, buf, 128 * 1024);
 646                 if (n <= 0)
 647                         break;
 648                 SHA1_Update(&c, buf, n);
 649         }
 650         free(buf);
 651
 652         SHA1_Final(pack_sha1, &c);
 653         write_or_die(pack_fd, pack_sha1, sizeof(pack_sha1));
 654 }
 655
 656 static int oecmp (const void *a_, const void *b_)
 657 {
 658         struct object_entry *a = *((struct object_entry**)a_);
 659         struct object_entry *b = *((struct object_entry**)b_);
 660         return hashcmp(a->sha1, b->sha1);
 661 }
 662
 663 static void write_index(const char *idx_name)
 664 {
 665         struct sha1file *f;
 666         struct object_entry **idx, **c, **last, *e;
 667         struct object_entry_pool *o;
 668         unsigned int array[256];
 669         int i;
 670
 671         /* Build the sorted table of object IDs. */
 672         idx = xmalloc(object_count * sizeof(struct object_entry*));
 673         c = idx;
 674         for (o = blocks; o; o = o->next_pool)
 675                 for (e = o->entries; e != o->next_free; e++)
 676                         *c++ = e;
 677         last = idx + object_count;
 678         qsort(idx, object_count, sizeof(struct object_entry*), oecmp);
 679
 680         /* Generate the fan-out array. */
 681         c = idx;
 682         for (i = 0; i < 256; i++) {
 683                 struct object_entry **next = c;;
 684                 while (next < last) {
 685                         if ((*next)->sha1[0] != i)
 686                                 break;
 687                         next++;
 688                 }
 689                 array[i] = htonl(next - idx);
 690                 c = next;
 691         }
 692
 693         f = sha1create("%s", idx_name);
 694         sha1write(f, array, 256 * sizeof(int));
 695         for (c = idx; c != last; c++) {
 696                 unsigned int offset = htonl((*c)->offset);
 697                 sha1write(f, &offset, 4);
 698                 sha1write(f, (*c)->sha1, sizeof((*c)->sha1));
 699         }
 700         sha1write(f, pack_sha1, sizeof(pack_sha1));
 701         sha1close(f, NULL, 1);
 702         free(idx);
 703 }
 704
 705 static void end_packfile()
 706 {
 707         fixup_header_footer();
 708         close(pack_fd);
 709         write_index(idx_name);
 710
 711         free(pack_name);
 712         free(idx_name);
 713         free(pack_data);
 714 }
 715
 716 static size_t encode_header(
 717         enum object_type type,
 718         size_t size,
 719         unsigned char *hdr)
 720 {
 721         int n = 1;
 722         unsigned char c;
 723
 724         if (type < OBJ_COMMIT || type > OBJ_REF_DELTA)
 725                 die("bad type %d", type);
 726
 727         c = (type << 4) | (size & 15);
 728         size >>= 4;
 729         while (size) {
 730                 *hdr++ = c | 0x80;
 731                 c = size & 0x7f;
 732                 size >>= 7;
 733                 n++;
 734         }
 735         *hdr = c;
 736         return n;
 737 }
 738
 739 static int store_object(
 740         enum object_type type,
 741         void *dat,
 742         size_t datlen,
 743         struct last_object *last,
 744         unsigned char *sha1out,
 745         unsigned long mark)
 746 {
 747         void *out, *delta;
 748         struct object_entry *e;
 749         unsigned char hdr[96];
 750         unsigned char sha1[20];
 751         unsigned long hdrlen, deltalen;
 752         SHA_CTX c;
 753         z_stream s;
 754
 755         hdrlen = sprintf((char*)hdr,"%s %lu",type_names[type],datlen) + 1;
 756         SHA1_Init(&c);
 757         SHA1_Update(&c, hdr, hdrlen);
 758         SHA1_Update(&c, dat, datlen);
 759         SHA1_Final(sha1, &c);
 760         if (sha1out)
 761                 hashcpy(sha1out, sha1);
 762
 763         e = insert_object(sha1);
 764         if (mark)
 765                 insert_mark(mark, e);
 766         if (e->offset) {
 767                 duplicate_count_by_type[type]++;
 768                 return 1;
 769         }
 770         e->type = type;
 771         e->offset = pack_size;
 772         object_count++;
 773         object_count_by_type[type]++;
 774
 775         if (last && last->data && last->depth < max_depth)
 776                 delta = diff_delta(last->data, last->len,
 777                         dat, datlen,
 778                         &deltalen, 0);
 779         else
 780                 delta = 0;
 781
 782         memset(&s, 0, sizeof(s));
 783         deflateInit(&s, zlib_compression_level);
 784
 785         if (delta) {
 786                 unsigned long ofs = e->offset - last->offset;
 787                 unsigned pos = sizeof(hdr) - 1;
 788
 789                 delta_count_by_type[type]++;
 790                 last->depth++;
 791                 s.next_in = delta;
 792                 s.avail_in = deltalen;
 793
 794                 hdrlen = encode_header(OBJ_OFS_DELTA, deltalen, hdr);
 795                 write_or_die(pack_fd, hdr, hdrlen);
 796                 pack_size += hdrlen;
 797
 798                 hdr[pos] = ofs & 127;
 799                 while (ofs >>= 7)
 800                         hdr[--pos] = 128 | (--ofs & 127);
 801                 write_or_die(pack_fd, hdr + pos, sizeof(hdr) - pos);
 802                 pack_size += sizeof(hdr) - pos;
 803         } else {
 804                 if (last)
 805                         last->depth = 0;
 806                 s.next_in = dat;
 807                 s.avail_in = datlen;
 808                 hdrlen = encode_header(type, datlen, hdr);
 809                 write_or_die(pack_fd, hdr, hdrlen);
 810                 pack_size += hdrlen;
 811         }
 812
 813         s.avail_out = deflateBound(&s, s.avail_in);
 814         s.next_out = out = xmalloc(s.avail_out);
 815         while (deflate(&s, Z_FINISH) == Z_OK)
 816                 /* nothing */;
 817         deflateEnd(&s);
 818
 819         write_or_die(pack_fd, out, s.total_out);
 820         pack_size += s.total_out;
 821
 822         free(out);
 823         if (delta)
 824                 free(delta);
 825         if (last) {
 826                 if (last->data && !last->no_free)
 827                         free(last->data);
 828                 last->data = dat;
 829                 last->offset = e->offset;
 830                 last->len = datlen;
 831         }
 832         return 0;
 833 }
 834
 835 static void *gfi_unpack_entry(unsigned long ofs, unsigned long *sizep)
 836 {
 837         char type[20];
 838         pack_data->pack_size = pack_size + 20;
 839         return unpack_entry(pack_data, ofs, type, sizep);
 840 }
 841
 842 static const char *get_mode(const char *str, unsigned int *modep)
 843 {
 844         unsigned char c;
 845         unsigned int mode = 0;
 846
 847         while ((c = *str++) != ' ') {
 848                 if (c < '0' || c > '7')
 849                         return NULL;
 850                 mode = (mode << 3) + (c - '0');
 851         }
 852         *modep = mode;
 853         return str;
 854 }
 855
 856 static void load_tree(struct tree_entry *root)
 857 {
 858         unsigned char* sha1 = root->versions[1].sha1;
 859         struct object_entry *myoe;
 860         struct tree_content *t;
 861         unsigned long size;
 862         char *buf;
 863         const char *c;
 864
 865         root->tree = t = new_tree_content(8);
 866         if (is_null_sha1(sha1))
 867                 return;
 868
 869         myoe = find_object(sha1);
 870         if (myoe) {
 871                 if (myoe->type != OBJ_TREE)
 872                         die("Not a tree: %s", sha1_to_hex(sha1));
 873                 t->delta_depth = 0;
 874                 buf = gfi_unpack_entry(myoe->offset, &size);
 875         } else {
 876                 char type[20];
 877                 buf = read_sha1_file(sha1, type, &size);
 878                 if (!buf || strcmp(type, tree_type))
 879                         die("Can't load tree %s", sha1_to_hex(sha1));
 880         }
 881
 882         c = buf;
 883         while (c != (buf + size)) {
 884                 struct tree_entry *e = new_tree_entry();
 885
 886                 if (t->entry_count == t->entry_capacity)
 887                         root->tree = t = grow_tree_content(t, 8);
 888                 t->entries[t->entry_count++] = e;
 889
 890                 e->tree = NULL;
 891                 c = get_mode(c, &e->versions[1].mode);
 892                 if (!c)
 893                         die("Corrupt mode in %s", sha1_to_hex(sha1));
 894                 e->versions[0].mode = e->versions[1].mode;
 895                 e->name = to_atom(c, strlen(c));
 896                 c += e->name->str_len + 1;
 897                 hashcpy(e->versions[0].sha1, (unsigned char*)c);
 898                 hashcpy(e->versions[1].sha1, (unsigned char*)c);
 899                 c += 20;
 900         }
 901         free(buf);
 902 }
 903
 904 static int tecmp0 (const void *_a, const void *_b)
 905 {
 906         struct tree_entry *a = *((struct tree_entry**)_a);
 907         struct tree_entry *b = *((struct tree_entry**)_b);
 908         return base_name_compare(
 909                 a->name->str_dat, a->name->str_len, a->versions[0].mode,
 910                 b->name->str_dat, b->name->str_len, b->versions[0].mode);
 911 }
 912
 913 static int tecmp1 (const void *_a, const void *_b)
 914 {
 915         struct tree_entry *a = *((struct tree_entry**)_a);
 916         struct tree_entry *b = *((struct tree_entry**)_b);
 917         return base_name_compare(
 918                 a->name->str_dat, a->name->str_len, a->versions[1].mode,
 919                 b->name->str_dat, b->name->str_len, b->versions[1].mode);
 920 }
 921
 922 static void mktree(struct tree_content *t,
 923         int v,
 924         unsigned long *szp,
 925         struct dbuf *b)
 926 {
 927         size_t maxlen = 0;
 928         unsigned int i;
 929         char *c;
 930
 931         if (!v)
 932                 qsort(t->entries,t->entry_count,sizeof(t->entries[0]),tecmp0);
 933         else
 934                 qsort(t->entries,t->entry_count,sizeof(t->entries[0]),tecmp1);
 935
 936         for (i = 0; i < t->entry_count; i++) {
 937                 if (t->entries[i]->versions[v].mode)
 938                         maxlen += t->entries[i]->name->str_len + 34;
 939         }
 940
 941         size_dbuf(b, maxlen);
 942         c = b->buffer;
 943         for (i = 0; i < t->entry_count; i++) {
 944                 struct tree_entry *e = t->entries[i];
 945                 if (!e->versions[v].mode)
 946                         continue;
 947                 c += sprintf(c, "%o", e->versions[v].mode);
 948                 *c++ = ' ';
 949                 strcpy(c, e->name->str_dat);
 950                 c += e->name->str_len + 1;
 951                 hashcpy((unsigned char*)c, e->versions[v].sha1);
 952                 c += 20;
 953         }
 954         *szp = c - (char*)b->buffer;
 955 }
 956
 957 static void store_tree(struct tree_entry *root)
 958 {
 959         struct tree_content *t = root->tree;
 960         unsigned int i, j, del;
 961         unsigned long new_len;
 962         struct last_object lo;
 963         struct object_entry *le;
 964
 965         if (!is_null_sha1(root->versions[1].sha1))
 966                 return;
 967
 968         for (i = 0; i < t->entry_count; i++) {
 969                 if (t->entries[i]->tree)
 970                         store_tree(t->entries[i]);
 971         }
 972
 973         le = find_object(root->versions[0].sha1);
 974         if (!S_ISDIR(root->versions[0].mode) || !le) {
 975                 lo.data = NULL;
 976                 lo.depth = 0;
 977         } else {
 978                 mktree(t, 0, &lo.len, &old_tree);
 979                 lo.data = old_tree.buffer;
 980                 lo.offset = le->offset;
 981                 lo.depth = t->delta_depth;
 982                 lo.no_free = 1;
 983         }
 984
 985         mktree(t, 1, &new_len, &new_tree);
 986         store_object(OBJ_TREE, new_tree.buffer, new_len,
 987                 &lo, root->versions[1].sha1, 0);
 988
 989         t->delta_depth = lo.depth;
 990         for (i = 0, j = 0, del = 0; i < t->entry_count; i++) {
 991                 struct tree_entry *e = t->entries[i];
 992                 if (e->versions[1].mode) {
 993                         e->versions[0].mode = e->versions[1].mode;
 994                         hashcpy(e->versions[0].sha1, e->versions[1].sha1);
 995                         t->entries[j++] = e;
 996                 } else {
 997                         release_tree_entry(e);
 998                         del++;
 999                 }
1000         }
1001         t->entry_count -= del;
1002 }
1003
1004 static int tree_content_set(
1005         struct tree_entry *root,
1006         const char *p,
1007         const unsigned char *sha1,
1008         const unsigned int mode)
1009 {
1010         struct tree_content *t = root->tree;
1011         const char *slash1;
1012         unsigned int i, n;
1013         struct tree_entry *e;
1014
1015         slash1 = strchr(p, '/');
1016         if (slash1)
1017                 n = slash1 - p;
1018         else
1019                 n = strlen(p);
1020
1021         for (i = 0; i < t->entry_count; i++) {
1022                 e = t->entries[i];
1023                 if (e->name->str_len == n && !strncmp(p, e->name->str_dat, n)) {
1024                         if (!slash1) {
1025                                 if (e->versions[1].mode == mode
1026                                                 && !hashcmp(e->versions[1].sha1, sha1))
1027                                         return 0;
1028                                 e->versions[1].mode = mode;
1029                                 hashcpy(e->versions[1].sha1, sha1);
1030                                 if (e->tree) {
1031                                         release_tree_content_recursive(e->tree);
1032                                         e->tree = NULL;
1033                                 }
1034                                 hashclr(root->versions[1].sha1);
1035                                 return 1;
1036                         }
1037                         if (!S_ISDIR(e->versions[1].mode)) {
1038                                 e->tree = new_tree_content(8);
1039                                 e->versions[1].mode = S_IFDIR;
1040                         }
1041                         if (!e->tree)
1042                                 load_tree(e);
1043                         if (tree_content_set(e, slash1 + 1, sha1, mode)) {
1044                                 hashclr(root->versions[1].sha1);
1045                                 return 1;
1046                         }
1047                         return 0;
1048                 }
1049         }
1050
1051         if (t->entry_count == t->entry_capacity)
1052                 root->tree = t = grow_tree_content(t, 8);
1053         e = new_tree_entry();
1054         e->name = to_atom(p, n);
1055         e->versions[0].mode = 0;
1056         hashclr(e->versions[0].sha1);
1057         t->entries[t->entry_count++] = e;
1058         if (slash1) {
1059                 e->tree = new_tree_content(8);
1060                 e->versions[1].mode = S_IFDIR;
1061                 tree_content_set(e, slash1 + 1, sha1, mode);
1062         } else {
1063                 e->tree = NULL;
1064                 e->versions[1].mode = mode;
1065                 hashcpy(e->versions[1].sha1, sha1);
1066         }
1067         hashclr(root->versions[1].sha1);
1068         return 1;
1069 }
1070
1071 static int tree_content_remove(struct tree_entry *root, const char *p)
1072 {
1073         struct tree_content *t = root->tree;
1074         const char *slash1;
1075         unsigned int i, n;
1076         struct tree_entry *e;
1077
1078         slash1 = strchr(p, '/');
1079         if (slash1)
1080                 n = slash1 - p;
1081         else
1082                 n = strlen(p);
1083
1084         for (i = 0; i < t->entry_count; i++) {
1085                 e = t->entries[i];
1086                 if (e->name->str_len == n && !strncmp(p, e->name->str_dat, n)) {
1087                         if (!slash1 || !S_ISDIR(e->versions[1].mode))
1088                                 goto del_entry;
1089                         if (!e->tree)
1090                                 load_tree(e);
1091                         if (tree_content_remove(e, slash1 + 1)) {
1092                                 for (n = 0; n < e->tree->entry_count; n++) {
1093                                         if (e->tree->entries[n]->versions[1].mode) {
1094                                                 hashclr(root->versions[1].sha1);
1095                                                 return 1;
1096                                         }
1097                                 }
1098                                 goto del_entry;
1099                         }
1100                         return 0;
1101                 }
1102         }
1103         return 0;
1104
1105 del_entry:
1106         if (e->tree) {
1107                 release_tree_content_recursive(e->tree);
1108                 e->tree = NULL;
1109         }
1110         e->versions[1].mode = 0;
1111         hashclr(e->versions[1].sha1);
1112         hashclr(root->versions[1].sha1);
1113         return 1;
1114 }
1115
1116 static void dump_branches()
1117 {
1118         static const char *msg = "fast-import";
1119         unsigned int i;
1120         struct branch *b;
1121         struct ref_lock *lock;
1122
1123         for (i = 0; i < branch_table_sz; i++) {
1124                 for (b = branch_table[i]; b; b = b->table_next_branch) {
1125                         lock = lock_any_ref_for_update(b->name, NULL);
1126                         if (!lock || write_ref_sha1(lock, b->sha1, msg) < 0)
1127                                 die("Can't write %s", b->name);
1128                 }
1129         }
1130 }
1131
1132 static void dump_tags()
1133 {
1134         static const char *msg = "fast-import";
1135         struct tag *t;
1136         struct ref_lock *lock;
1137         char path[PATH_MAX];
1138
1139         for (t = first_tag; t; t = t->next_tag) {
1140                 sprintf(path, "refs/tags/%s", t->name);
1141                 lock = lock_any_ref_for_update(path, NULL);
1142                 if (!lock || write_ref_sha1(lock, t->sha1, msg) < 0)
1143                         die("Can't write %s", path);
1144         }
1145 }
1146
1147 static void dump_marks_helper(FILE *f,
1148         unsigned long base,
1149         struct mark_set *m)
1150 {
1151         int k;
1152         if (m->shift) {
1153                 for (k = 0; k < 1024; k++) {
1154                         if (m->data.sets[k])
1155                                 dump_marks_helper(f, (base + k) << m->shift,
1156                                         m->data.sets[k]);
1157                 }
1158         } else {
1159                 for (k = 0; k < 1024; k++) {
1160                         if (m->data.marked[k])
1161                                 fprintf(f, ":%lu %s\n", base + k,
1162                                         sha1_to_hex(m->data.marked[k]->sha1));
1163                 }
1164         }
1165 }
1166
1167 static void dump_marks()
1168 {
1169         if (mark_file)
1170         {
1171                 FILE *f = fopen(mark_file, "w");
1172                 dump_marks_helper(f, 0, marks);
1173                 fclose(f);
1174         }
1175 }
1176
1177 static void read_next_command()
1178 {
1179         read_line(&command_buf, stdin, '\n');
1180 }
1181
1182 static void cmd_mark()
1183 {
1184         if (!strncmp("mark :", command_buf.buf, 6)) {
1185                 next_mark = strtoul(command_buf.buf + 6, NULL, 10);
1186                 read_next_command();
1187         }
1188         else
1189                 next_mark = 0;
1190 }
1191
1192 static void* cmd_data (size_t *size)
1193 {
1194         size_t n = 0;
1195         void *buffer;
1196         size_t length;
1197
1198         if (strncmp("data ", command_buf.buf, 5))
1199                 die("Expected 'data n' command, found: %s", command_buf.buf);
1200
1201         length = strtoul(command_buf.buf + 5, NULL, 10);
1202         buffer = xmalloc(length);
1203
1204         while (n < length) {
1205                 size_t s = fread((char*)buffer + n, 1, length - n, stdin);
1206                 if (!s && feof(stdin))
1207                         die("EOF in data (%lu bytes remaining)", length - n);
1208                 n += s;
1209         }
1210
1211         if (fgetc(stdin) != '\n')
1212                 die("An lf did not trail the binary data as expected.");
1213
1214         *size = length;
1215         return buffer;
1216 }
1217
1218 static void cmd_new_blob()
1219 {
1220         size_t l;
1221         void *d;
1222
1223         read_next_command();
1224         cmd_mark();
1225         d = cmd_data(&l);
1226
1227         if (store_object(OBJ_BLOB, d, l, &last_blob, NULL, next_mark))
1228                 free(d);
1229 }
1230
1231 static void unload_one_branch()
1232 {
1233         while (cur_active_branches
1234                 && cur_active_branches >= max_active_branches) {
1235                 unsigned long min_commit = ULONG_MAX;
1236                 struct branch *e, *l = NULL, *p = NULL;
1237
1238                 for (e = active_branches; e; e = e->active_next_branch) {
1239                         if (e->last_commit < min_commit) {
1240                                 p = l;
1241                                 min_commit = e->last_commit;
1242                         }
1243                         l = e;
1244                 }
1245
1246                 if (p) {
1247                         e = p->active_next_branch;
1248                         p->active_next_branch = e->active_next_branch;
1249                 } else {
1250                         e = active_branches;
1251                         active_branches = e->active_next_branch;
1252                 }
1253                 e->active_next_branch = NULL;
1254                 if (e->branch_tree.tree) {
1255                         release_tree_content_recursive(e->branch_tree.tree);
1256                         e->branch_tree.tree = NULL;
1257                 }
1258                 cur_active_branches--;
1259         }
1260 }
1261
1262 static void load_branch(struct branch *b)
1263 {
1264         load_tree(&b->branch_tree);
1265         b->active_next_branch = active_branches;
1266         active_branches = b;
1267         cur_active_branches++;
1268         branch_load_count++;
1269 }
1270
1271 static void file_change_m(struct branch *b)
1272 {
1273         const char *p = command_buf.buf + 2;
1274         char *p_uq;
1275         const char *endp;
1276         struct object_entry *oe;
1277         unsigned char sha1[20];
1278         unsigned int mode;
1279         char type[20];
1280
1281         p = get_mode(p, &mode);
1282         if (!p)
1283                 die("Corrupt mode: %s", command_buf.buf);
1284         switch (mode) {
1285         case S_IFREG | 0644:
1286         case S_IFREG | 0755:
1287         case S_IFLNK:
1288         case 0644:
1289         case 0755:
1290                 /* ok */
1291                 break;
1292         default:
1293                 die("Corrupt mode: %s", command_buf.buf);
1294         }
1295
1296         if (*p == ':') {
1297                 char *x;
1298                 oe = find_mark(strtoul(p + 1, &x, 10));
1299                 hashcpy(sha1, oe->sha1);
1300                 p = x;
1301         } else {
1302                 if (get_sha1_hex(p, sha1))
1303                         die("Invalid SHA1: %s", command_buf.buf);
1304                 oe = find_object(sha1);
1305                 p += 40;
1306         }
1307         if (*p++ != ' ')
1308                 die("Missing space after SHA1: %s", command_buf.buf);
1309
1310         p_uq = unquote_c_style(p, &endp);
1311         if (p_uq) {
1312                 if (*endp)
1313                         die("Garbage after path in: %s", command_buf.buf);
1314                 p = p_uq;
1315         }
1316
1317         if (oe) {
1318                 if (oe->type != OBJ_BLOB)
1319                         die("Not a blob (actually a %s): %s",
1320                                 command_buf.buf, type_names[oe->type]);
1321         } else {
1322                 if (sha1_object_info(sha1, type, NULL))
1323                         die("Blob not found: %s", command_buf.buf);
1324                 if (strcmp(blob_type, type))
1325                         die("Not a blob (actually a %s): %s",
1326                                 command_buf.buf, type);
1327         }
1328
1329         tree_content_set(&b->branch_tree, p, sha1, S_IFREG | mode);
1330
1331         if (p_uq)
1332                 free(p_uq);
1333 }
1334
1335 static void file_change_d(struct branch *b)
1336 {
1337         const char *p = command_buf.buf + 2;
1338         char *p_uq;
1339         const char *endp;
1340
1341         p_uq = unquote_c_style(p, &endp);
1342         if (p_uq) {
1343                 if (*endp)
1344                         die("Garbage after path in: %s", command_buf.buf);
1345                 p = p_uq;
1346         }
1347         tree_content_remove(&b->branch_tree, p);
1348         if (p_uq)
1349                 free(p_uq);
1350 }
1351
1352 static void cmd_from(struct branch *b)
1353 {
1354         const char *from, *endp;
1355         char *str_uq;
1356         struct branch *s;
1357
1358         if (strncmp("from ", command_buf.buf, 5))
1359                 return;
1360
1361         if (b->last_commit)
1362                 die("Can't reinitailize branch %s", b->name);
1363
1364         from = strchr(command_buf.buf, ' ') + 1;
1365         str_uq = unquote_c_style(from, &endp);
1366         if (str_uq) {
1367                 if (*endp)
1368                         die("Garbage after string in: %s", command_buf.buf);
1369                 from = str_uq;
1370         }
1371
1372         s = lookup_branch(from);
1373         if (b == s)
1374                 die("Can't create a branch from itself: %s", b->name);
1375         else if (s) {
1376                 unsigned char *t = s->branch_tree.versions[1].sha1;
1377                 hashcpy(b->sha1, s->sha1);
1378                 hashcpy(b->branch_tree.versions[0].sha1, t);
1379                 hashcpy(b->branch_tree.versions[1].sha1, t);
1380         } else if (*from == ':') {
1381                 unsigned long idnum = strtoul(from + 1, NULL, 10);
1382                 struct object_entry *oe = find_mark(idnum);
1383                 unsigned long size;
1384                 char *buf;
1385                 if (oe->type != OBJ_COMMIT)
1386                         die("Mark :%lu not a commit", idnum);
1387                 hashcpy(b->sha1, oe->sha1);
1388                 buf = gfi_unpack_entry(oe->offset, &size);
1389                 if (!buf || size < 46)
1390                         die("Not a valid commit: %s", from);
1391                 if (memcmp("tree ", buf, 5)
1392                         || get_sha1_hex(buf + 5, b->branch_tree.versions[1].sha1))
1393                         die("The commit %s is corrupt", sha1_to_hex(b->sha1));
1394                 free(buf);
1395                 hashcpy(b->branch_tree.versions[0].sha1,
1396                         b->branch_tree.versions[1].sha1);
1397         } else if (!get_sha1(from, b->sha1)) {
1398                 if (is_null_sha1(b->sha1)) {
1399                         hashclr(b->branch_tree.versions[0].sha1);
1400                         hashclr(b->branch_tree.versions[1].sha1);
1401                 } else {
1402                         unsigned long size;
1403                         char *buf;
1404
1405                         buf = read_object_with_reference(b->sha1,
1406                                 type_names[OBJ_COMMIT], &size, b->sha1);
1407                         if (!buf || size < 46)
1408                                 die("Not a valid commit: %s", from);
1409                         if (memcmp("tree ", buf, 5)
1410                                 || get_sha1_hex(buf + 5, b->branch_tree.versions[1].sha1))
1411                                 die("The commit %s is corrupt", sha1_to_hex(b->sha1));
1412                         free(buf);
1413                         hashcpy(b->branch_tree.versions[0].sha1,
1414                                 b->branch_tree.versions[1].sha1);
1415                 }
1416         } else
1417                 die("Invalid ref name or SHA1 expression: %s", from);
1418
1419         read_next_command();
1420 }
1421
1422 static struct hash_list* cmd_merge(unsigned int *count)
1423 {
1424         struct hash_list *list = NULL, *n, *e;
1425         const char *from, *endp;
1426         char *str_uq;
1427         struct branch *s;
1428
1429         *count = 0;
1430         while (!strncmp("merge ", command_buf.buf, 6)) {
1431                 from = strchr(command_buf.buf, ' ') + 1;
1432                 str_uq = unquote_c_style(from, &endp);
1433                 if (str_uq) {
1434                         if (*endp)
1435                                 die("Garbage after string in: %s", command_buf.buf);
1436                         from = str_uq;
1437                 }
1438
1439                 n = xmalloc(sizeof(*n));
1440                 s = lookup_branch(from);
1441                 if (s)
1442                         hashcpy(n->sha1, s->sha1);
1443                 else if (*from == ':') {
1444                         unsigned long idnum = strtoul(from + 1, NULL, 10);
1445                         struct object_entry *oe = find_mark(idnum);
1446                         if (oe->type != OBJ_COMMIT)
1447                                 die("Mark :%lu not a commit", idnum);
1448                         hashcpy(n->sha1, oe->sha1);
1449                 } else if (get_sha1(from, n->sha1))
1450                         die("Invalid ref name or SHA1 expression: %s", from);
1451
1452                 n->next = NULL;
1453                 if (list)
1454                         e->next = n;
1455                 else
1456                         list = n;
1457                 e = n;
1458                 *count++;
1459                 read_next_command();
1460         }
1461         return list;
1462 }
1463
1464 static void cmd_new_commit()
1465 {
1466         struct branch *b;
1467         void *msg;
1468         size_t msglen;
1469         char *str_uq;
1470         const char *endp;
1471         char *sp;
1472         char *author = NULL;
1473         char *committer = NULL;
1474         struct hash_list *merge_list = NULL;
1475         unsigned int merge_count;
1476
1477         /* Obtain the branch name from the rest of our command */
1478         sp = strchr(command_buf.buf, ' ') + 1;
1479         str_uq = unquote_c_style(sp, &endp);
1480         if (str_uq) {
1481                 if (*endp)
1482                         die("Garbage after ref in: %s", command_buf.buf);
1483                 sp = str_uq;
1484         }
1485         b = lookup_branch(sp);
1486         if (!b)
1487                 b = new_branch(sp);
1488         if (str_uq)
1489                 free(str_uq);
1490
1491         read_next_command();
1492         cmd_mark();
1493         if (!strncmp("author ", command_buf.buf, 7)) {
1494                 author = strdup(command_buf.buf);
1495                 read_next_command();
1496         }
1497         if (!strncmp("committer ", command_buf.buf, 10)) {
1498                 committer = strdup(command_buf.buf);
1499                 read_next_command();
1500         }
1501         if (!committer)
1502                 die("Expected committer but didn't get one");
1503         msg = cmd_data(&msglen);
1504         read_next_command();
1505         cmd_from(b);
1506         merge_list = cmd_merge(&merge_count);
1507
1508         /* ensure the branch is active/loaded */
1509         if (!b->branch_tree.tree || !max_active_branches) {
1510                 unload_one_branch();
1511                 load_branch(b);
1512         }
1513
1514         /* file_change* */
1515         for (;;) {
1516                 if (1 == command_buf.len)
1517                         break;
1518                 else if (!strncmp("M ", command_buf.buf, 2))
1519                         file_change_m(b);
1520                 else if (!strncmp("D ", command_buf.buf, 2))
1521                         file_change_d(b);
1522                 else
1523                         die("Unsupported file_change: %s", command_buf.buf);
1524                 read_next_command();
1525         }
1526
1527         /* build the tree and the commit */
1528         store_tree(&b->branch_tree);
1529         hashcpy(b->branch_tree.versions[0].sha1,
1530                 b->branch_tree.versions[1].sha1);
1531         size_dbuf(&new_data, 97 + msglen
1532                 + merge_count * 49
1533                 + (author
1534                         ? strlen(author) + strlen(committer)
1535                         : 2 * strlen(committer)));
1536         sp = new_data.buffer;
1537         sp += sprintf(sp, "tree %s\n",
1538                 sha1_to_hex(b->branch_tree.versions[1].sha1));
1539         if (!is_null_sha1(b->sha1))
1540                 sp += sprintf(sp, "parent %s\n", sha1_to_hex(b->sha1));
1541         while (merge_list) {
1542                 struct hash_list *next = merge_list->next;
1543                 sp += sprintf(sp, "parent %s\n", sha1_to_hex(merge_list->sha1));
1544                 free(merge_list);
1545                 merge_list = next;
1546         }
1547         if (author)
1548                 sp += sprintf(sp, "%s\n", author);
1549         else
1550                 sp += sprintf(sp, "author %s\n", committer + 10);
1551         sp += sprintf(sp, "%s\n\n", committer);
1552         memcpy(sp, msg, msglen);
1553         sp += msglen;
1554         if (author)
1555                 free(author);
1556         free(committer);
1557         free(msg);
1558
1559         store_object(OBJ_COMMIT,
1560                 new_data.buffer, sp - (char*)new_data.buffer,
1561                 NULL, b->sha1, next_mark);
1562         b->last_commit = object_count_by_type[OBJ_COMMIT];
1563
1564         if (branch_log) {
1565                 int need_dq = quote_c_style(b->name, NULL, NULL, 0);
1566                 fprintf(branch_log, "commit ");
1567                 if (need_dq) {
1568                         fputc('"', branch_log);
1569                         quote_c_style(b->name, NULL, branch_log, 0);
1570                         fputc('"', branch_log);
1571                 } else
1572                         fprintf(branch_log, "%s", b->name);
1573                 fprintf(branch_log," :%lu %s\n",next_mark,sha1_to_hex(b->sha1));
1574         }
1575 }
1576
1577 static void cmd_new_tag()
1578 {
1579         char *str_uq;
1580         const char *endp;
1581         char *sp;
1582         const char *from;
1583         char *tagger;
1584         struct branch *s;
1585         void *msg;
1586         size_t msglen;
1587         struct tag *t;
1588         unsigned long from_mark = 0;
1589         unsigned char sha1[20];
1590
1591         /* Obtain the new tag name from the rest of our command */
1592         sp = strchr(command_buf.buf, ' ') + 1;
1593         str_uq = unquote_c_style(sp, &endp);
1594         if (str_uq) {
1595                 if (*endp)
1596                         die("Garbage after tag name in: %s", command_buf.buf);
1597                 sp = str_uq;
1598         }
1599         t = pool_alloc(sizeof(struct tag));
1600         t->next_tag = NULL;
1601         t->name = pool_strdup(sp);
1602         if (last_tag)
1603                 last_tag->next_tag = t;
1604         else
1605                 first_tag = t;
1606         last_tag = t;
1607         if (str_uq)
1608                 free(str_uq);
1609         read_next_command();
1610
1611         /* from ... */
1612         if (strncmp("from ", command_buf.buf, 5))
1613                 die("Expected from command, got %s", command_buf.buf);
1614
1615         from = strchr(command_buf.buf, ' ') + 1;
1616         str_uq = unquote_c_style(from, &endp);
1617         if (str_uq) {
1618                 if (*endp)
1619                         die("Garbage after string in: %s", command_buf.buf);
1620                 from = str_uq;
1621         }
1622
1623         s = lookup_branch(from);
1624         if (s) {
1625                 hashcpy(sha1, s->sha1);
1626         } else if (*from == ':') {
1627                 from_mark = strtoul(from + 1, NULL, 10);
1628                 struct object_entry *oe = find_mark(from_mark);
1629                 if (oe->type != OBJ_COMMIT)
1630                         die("Mark :%lu not a commit", from_mark);
1631                 hashcpy(sha1, oe->sha1);
1632         } else if (!get_sha1(from, sha1)) {
1633                 unsigned long size;
1634                 char *buf;
1635
1636                 buf = read_object_with_reference(sha1,
1637                         type_names[OBJ_COMMIT], &size, sha1);
1638                 if (!buf || size < 46)
1639                         die("Not a valid commit: %s", from);
1640                 free(buf);
1641         } else
1642                 die("Invalid ref name or SHA1 expression: %s", from);
1643
1644         if (str_uq)
1645                 free(str_uq);
1646         read_next_command();
1647
1648         /* tagger ... */
1649         if (strncmp("tagger ", command_buf.buf, 7))
1650                 die("Expected tagger command, got %s", command_buf.buf);
1651         tagger = strdup(command_buf.buf);
1652
1653         /* tag payload/message */
1654         read_next_command();
1655         msg = cmd_data(&msglen);
1656
1657         /* build the tag object */
1658         size_dbuf(&new_data, 67+strlen(t->name)+strlen(tagger)+msglen);
1659         sp = new_data.buffer;
1660         sp += sprintf(sp, "object %s\n", sha1_to_hex(sha1));
1661         sp += sprintf(sp, "type %s\n", type_names[OBJ_COMMIT]);
1662         sp += sprintf(sp, "tag %s\n", t->name);
1663         sp += sprintf(sp, "%s\n\n", tagger);
1664         memcpy(sp, msg, msglen);
1665         sp += msglen;
1666         free(tagger);
1667         free(msg);
1668
1669         store_object(OBJ_TAG, new_data.buffer, sp - (char*)new_data.buffer,
1670                 NULL, t->sha1, 0);
1671
1672         if (branch_log) {
1673                 int need_dq = quote_c_style(t->name, NULL, NULL, 0);
1674                 fprintf(branch_log, "tag ");
1675                 if (need_dq) {
1676                         fputc('"', branch_log);
1677                         quote_c_style(t->name, NULL, branch_log, 0);
1678                         fputc('"', branch_log);
1679                 } else
1680                         fprintf(branch_log, "%s", t->name);
1681                 fprintf(branch_log," :%lu %s\n",from_mark,sha1_to_hex(t->sha1));
1682         }
1683 }
1684
1685 static void cmd_reset_branch()
1686 {
1687         struct branch *b;
1688         char *str_uq;
1689         const char *endp;
1690         char *sp;
1691
1692         /* Obtain the branch name from the rest of our command */
1693         sp = strchr(command_buf.buf, ' ') + 1;
1694         str_uq = unquote_c_style(sp, &endp);
1695         if (str_uq) {
1696                 if (*endp)
1697                         die("Garbage after ref in: %s", command_buf.buf);
1698                 sp = str_uq;
1699         }
1700         b = lookup_branch(sp);
1701         if (b) {
1702                 b->last_commit = 0;
1703                 if (b->branch_tree.tree) {
1704                         release_tree_content_recursive(b->branch_tree.tree);
1705                         b->branch_tree.tree = NULL;
1706                 }
1707         }
1708         else
1709                 b = new_branch(sp);
1710         if (str_uq)
1711                 free(str_uq);
1712         read_next_command();
1713         cmd_from(b);
1714 }
1715
1716 static const char fast_import_usage[] =
1717 "git-fast-import [--objects=n] [--depth=n] [--active-branches=n] [--export-marks=marks.file] [--branch-log=log] temp.pack";
1718
1719 int main(int argc, const char **argv)
1720 {
1721         int i;
1722         unsigned long est_obj_cnt = object_entry_alloc;
1723         unsigned long duplicate_count;
1724
1725         setup_ident();
1726         git_config(git_default_config);
1727
1728         for (i = 1; i < argc; i++) {
1729                 const char *a = argv[i];
1730
1731                 if (*a != '-' || !strcmp(a, "--"))
1732                         break;
1733                 else if (!strncmp(a, "--objects=", 10))
1734                         est_obj_cnt = strtoul(a + 10, NULL, 0);
1735                 else if (!strncmp(a, "--depth=", 8))
1736                         max_depth = strtoul(a + 8, NULL, 0);
1737                 else if (!strncmp(a, "--active-branches=", 18))
1738                         max_active_branches = strtoul(a + 18, NULL, 0);
1739                 else if (!strncmp(a, "--export-marks=", 15))
1740                         mark_file = a + 15;
1741                 else if (!strncmp(a, "--branch-log=", 13)) {
1742                         branch_log = fopen(a + 13, "w");
1743                         if (!branch_log)
1744                                 die("Can't create %s: %s", a + 13, strerror(errno));
1745                 }
1746                 else
1747                         die("unknown option %s", a);
1748         }
1749         if ((i+1) != argc)
1750                 usage(fast_import_usage);
1751         base_name = argv[i];
1752
1753         alloc_objects(est_obj_cnt);
1754         strbuf_init(&command_buf);
1755
1756         atom_table = xcalloc(atom_table_sz, sizeof(struct atom_str*));
1757         branch_table = xcalloc(branch_table_sz, sizeof(struct branch*));
1758         avail_tree_table = xcalloc(avail_tree_table_sz, sizeof(struct avail_tree_content*));
1759         marks = pool_calloc(1, sizeof(struct mark_set));
1760
1761         start_packfile();
1762         for (;;) {
1763                 read_next_command();
1764                 if (command_buf.eof)
1765                         break;
1766                 else if (!strcmp("blob", command_buf.buf))
1767                         cmd_new_blob();
1768                 else if (!strncmp("commit ", command_buf.buf, 7))
1769                         cmd_new_commit();
1770                 else if (!strncmp("tag ", command_buf.buf, 4))
1771                         cmd_new_tag();
1772                 else if (!strncmp("reset ", command_buf.buf, 6))
1773                         cmd_reset_branch();
1774                 else
1775                         die("Unsupported command: %s", command_buf.buf);
1776         }
1777         end_packfile();
1778
1779         dump_branches();
1780         dump_tags();
1781         dump_marks();
1782         if (branch_log)
1783                 fclose(branch_log);
1784
1785         for (i = 0; i < ARRAY_SIZE(duplicate_count_by_type); i++)
1786                 duplicate_count += duplicate_count_by_type[i];
1787
1788         fprintf(stderr, "%s statistics:\n", argv[0]);
1789         fprintf(stderr, "---------------------------------------------------------------------\n");
1790         fprintf(stderr, "Alloc'd objects: %10lu (%10lu overflow  )\n", alloc_count, alloc_count - est_obj_cnt);
1791         fprintf(stderr, "Total objects:   %10lu (%10lu duplicates                  )\n", object_count, duplicate_count);
1792         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]);
1793         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]);
1794         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]);
1795         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]);
1796         fprintf(stderr, "Total branches:  %10lu (%10lu loads     )\n", branch_count, branch_load_count);
1797         fprintf(stderr, "      marks:     %10u (%10lu unique    )\n", (1 << marks->shift) * 1024, marks_set_count);
1798         fprintf(stderr, "      atoms:     %10u\n", atom_cnt);
1799         fprintf(stderr, "Memory total:    %10lu KiB\n", (total_allocd + alloc_count*sizeof(struct object_entry))/1024);
1800         fprintf(stderr, "       pools:    %10lu KiB\n", total_allocd/1024);
1801         fprintf(stderr, "     objects:    %10lu KiB\n", (alloc_count*sizeof(struct object_entry))/1024);
1802         fprintf(stderr, "---------------------------------------------------------------------\n");
1803         fprintf(stderr, "\n");
1804
1805         return 0;
1806 }