src/odb_loose.c

   1 /*
   2  * Copyright (C) the libgit2 contributors. All rights reserved.
   3  *
   4  * This file is part of libgit2, distributed under the GNU GPL v2 with
   5  * a Linking Exception. For full terms see the included COPYING file.
   6  */
   7
   8 #include "common.h"
   9
  10 #include <zlib.h>
  11 #include "git2/object.h"
  12 #include "git2/sys/odb_backend.h"
  13 #include "fileops.h"
  14 #include "hash.h"
  15 #include "odb.h"
  16 #include "delta.h"
  17 #include "filebuf.h"
  18 #include "object.h"
  19 #include "zstream.h"
  20
  21 #include "git2/odb_backend.h"
  22 #include "git2/types.h"
  23
  24 /* maximum possible header length */
  25 #define MAX_HEADER_LEN 64
  26
  27 typedef struct { /* object header data */
  28         git_object_t type; /* object type */
  29         size_t  size; /* object size */
  30 } obj_hdr;
  31
  32 typedef struct {
  33         git_odb_stream stream;
  34         git_filebuf fbuf;
  35 } loose_writestream;
  36
  37 typedef struct {
  38         git_odb_stream stream;
  39         git_map map;
  40         char start[MAX_HEADER_LEN];
  41         size_t start_len;
  42         size_t start_read;
  43         git_zstream zstream;
  44 } loose_readstream;
  45
  46 typedef struct loose_backend {
  47         git_odb_backend parent;
  48
  49         int object_zlib_level; /** loose object zlib compression level. */
  50         int fsync_object_files; /** loose object file fsync flag. */
  51         mode_t object_file_mode;
  52         mode_t object_dir_mode;
  53
  54         size_t objects_dirlen;
  55         char objects_dir[GIT_FLEX_ARRAY];
  56 } loose_backend;
  57
  58 /* State structure for exploring directories,
  59  * in order to locate objects matching a short oid.
  60  */
  61 typedef struct {
  62         size_t dir_len;
  63         unsigned char short_oid[GIT_OID_HEXSZ]; /* hex formatted oid to match */
  64         size_t short_oid_len;
  65         int found;                              /* number of matching
  66                                                  * objects already found */
  67         unsigned char res_oid[GIT_OID_HEXSZ];   /* hex formatted oid of
  68                                                  * the object found */
  69 } loose_locate_object_state;
  70
  71
  72 /***********************************************************
  73  *
  74  * MISCELLANEOUS HELPER FUNCTIONS
  75  *
  76  ***********************************************************/
  77
  78 static int object_file_name(
  79         git_buf *name, const loose_backend *be, const git_oid *id)
  80 {
  81         size_t alloclen;
  82
  83         /* expand length for object root + 40 hex sha1 chars + 2 * '/' + '\0' */
  84         GIT_ERROR_CHECK_ALLOC_ADD(&alloclen, be->objects_dirlen, GIT_OID_HEXSZ);
  85         GIT_ERROR_CHECK_ALLOC_ADD(&alloclen, alloclen, 3);
  86         if (git_buf_grow(name, alloclen) < 0)
  87                 return -1;
  88
  89         git_buf_set(name, be->objects_dir, be->objects_dirlen);
  90         git_path_to_dir(name);
  91
  92         /* loose object filename: aa/aaa... (41 bytes) */
  93         git_oid_pathfmt(name->ptr + name->size, id);
  94         name->size += GIT_OID_HEXSZ + 1;
  95         name->ptr[name->size] = '\0';
  96
  97         return 0;
  98 }
  99
 100 static int object_mkdir(const git_buf *name, const loose_backend *be)
 101 {
 102         return git_futils_mkdir_relative(
 103                 name->ptr + be->objects_dirlen, be->objects_dir, be->object_dir_mode,
 104                 GIT_MKDIR_PATH | GIT_MKDIR_SKIP_LAST | GIT_MKDIR_VERIFY_DIR, NULL);
 105 }
 106
 107 static int parse_header_packlike(
 108         obj_hdr *out, size_t *out_len, const unsigned char *data, size_t len)
 109 {
 110         unsigned long c;
 111         size_t shift, size, used = 0;
 112
 113         if (len == 0)
 114                 goto on_error;
 115
 116         c = data[used++];
 117         out->type = (c >> 4) & 7;
 118
 119         size = c & 15;
 120         shift = 4;
 121         while (c & 0x80) {
 122                 if (len <= used)
 123                         goto on_error;
 124
 125                 if (sizeof(size_t) * 8 <= shift)
 126                         goto on_error;
 127
 128                 c = data[used++];
 129                 size += (c & 0x7f) << shift;
 130                 shift += 7;
 131         }
 132
 133         out->size = size;
 134
 135         if (out_len)
 136                 *out_len = used;
 137
 138         return 0;
 139
 140 on_error:
 141         git_error_set(GIT_ERROR_OBJECT, "failed to parse loose object: invalid header");
 142         return -1;
 143 }
 144
 145 static int parse_header(
 146         obj_hdr *out,
 147         size_t *out_len,
 148         const unsigned char *_data,
 149         size_t data_len)
 150 {
 151         const char *data = (char *)_data;
 152         size_t i, typename_len, size_idx, size_len;
 153         int64_t size;
 154
 155         *out_len = 0;
 156
 157         /* find the object type name */
 158         for (i = 0, typename_len = 0; i < data_len; i++, typename_len++) {
 159                 if (data[i] == ' ')
 160                         break;
 161         }
 162
 163         if (typename_len == data_len)
 164                 goto on_error;
 165
 166         out->type = git_object_stringn2type(data, typename_len);
 167
 168         size_idx = typename_len + 1;
 169         for (i = size_idx, size_len = 0; i < data_len; i++, size_len++) {
 170                 if (data[i] == '\0')
 171                         break;
 172         }
 173
 174         if (i == data_len)
 175                 goto on_error;
 176
 177         if (git__strntol64(&size, &data[size_idx], size_len, NULL, 10) < 0 ||
 178                 size < 0)
 179                 goto on_error;
 180
 181         if ((uint64_t)size > SIZE_MAX) {
 182                 git_error_set(GIT_ERROR_OBJECT, "object is larger than available memory");
 183                 return -1;
 184         }
 185
 186         out->size = (size_t)size;
 187
 188         if (GIT_ADD_SIZET_OVERFLOW(out_len, i, 1))
 189                 goto on_error;
 190
 191         return 0;
 192
 193 on_error:
 194         git_error_set(GIT_ERROR_OBJECT, "failed to parse loose object: invalid header");
 195         return -1;
 196 }
 197
 198 static int is_zlib_compressed_data(unsigned char *data, size_t data_len)
 199 {
 200         unsigned int w;
 201
 202         if (data_len < 2)
 203                 return 0;
 204
 205         w = ((unsigned int)(data[0]) << 8) + data[1];
 206         return (data[0] & 0x8F) == 0x08 && !(w % 31);
 207 }
 208
 209 /***********************************************************
 210  *
 211  * ODB OBJECT READING & WRITING
 212  *
 213  * Backend for the public API; read headers and full objects
 214  * from the ODB. Write raw data to the ODB.
 215  *
 216  ***********************************************************/
 217
 218
 219 /*
 220  * At one point, there was a loose object format that was intended to
 221  * mimic the format used in pack-files. This was to allow easy copying
 222  * of loose object data into packs. This format is no longer used, but
 223  * we must still read it.
 224  */
 225 static int read_loose_packlike(git_rawobj *out, git_buf *obj)
 226 {
 227         git_buf body = GIT_BUF_INIT;
 228         const unsigned char *obj_data;
 229         obj_hdr hdr;
 230         size_t obj_len, head_len, alloc_size;
 231         int error;
 232
 233         obj_data = (unsigned char *)obj->ptr;
 234         obj_len = obj->size;
 235
 236         /*
 237          * read the object header, which is an (uncompressed)
 238          * binary encoding of the object type and size.
 239          */
 240         if ((error = parse_header_packlike(&hdr, &head_len, obj_data, obj_len)) < 0)
 241                 goto done;
 242
 243         if (!git_object_typeisloose(hdr.type) || head_len > obj_len) {
 244                 git_error_set(GIT_ERROR_ODB, "failed to inflate loose object");
 245                 error = -1;
 246                 goto done;
 247         }
 248
 249         obj_data += head_len;
 250         obj_len -= head_len;
 251
 252         /*
 253          * allocate a buffer and inflate the data into it
 254          */
 255         if (GIT_ADD_SIZET_OVERFLOW(&alloc_size, hdr.size, 1) ||
 256                 git_buf_init(&body, alloc_size) < 0) {
 257                 error = -1;
 258                 goto done;
 259         }
 260
 261         if ((error = git_zstream_inflatebuf(&body, obj_data, obj_len)) < 0)
 262                 goto done;
 263
 264         out->len = hdr.size;
 265         out->type = hdr.type;
 266         out->data = git_buf_detach(&body);
 267
 268 done:
 269         git_buf_dispose(&body);
 270         return error;
 271 }
 272
 273 static int read_loose_standard(git_rawobj *out, git_buf *obj)
 274 {
 275         git_zstream zstream = GIT_ZSTREAM_INIT;
 276         unsigned char head[MAX_HEADER_LEN], *body = NULL;
 277         size_t decompressed, head_len, body_len, alloc_size;
 278         obj_hdr hdr;
 279         int error;
 280
 281         if ((error = git_zstream_init(&zstream, GIT_ZSTREAM_INFLATE)) < 0 ||
 282                 (error = git_zstream_set_input(&zstream, git_buf_cstr(obj), git_buf_len(obj))) < 0)
 283                 goto done;
 284
 285         decompressed = sizeof(head);
 286
 287         /*
 288          * inflate the initial part of the compressed buffer in order to
 289          * parse the header; read the largest header possible, then push the
 290          * remainder into the body buffer.
 291          */
 292         if ((error = git_zstream_get_output(head, &decompressed, &zstream)) < 0 ||
 293                 (error = parse_header(&hdr, &head_len, head, decompressed)) < 0)
 294                 goto done;
 295
 296         if (!git_object_typeisloose(hdr.type)) {
 297                 git_error_set(GIT_ERROR_ODB, "failed to inflate disk object");
 298                 error = -1;
 299                 goto done;
 300         }
 301
 302         /*
 303          * allocate a buffer and inflate the object data into it
 304          * (including the initial sequence in the head buffer).
 305          */
 306         if (GIT_ADD_SIZET_OVERFLOW(&alloc_size, hdr.size, 1) ||
 307                 (body = git__malloc(alloc_size)) == NULL) {
 308                 error = -1;
 309                 goto done;
 310         }
 311
 312         assert(decompressed >= head_len);
 313         body_len = decompressed - head_len;
 314
 315         if (body_len)
 316                 memcpy(body, head + head_len, body_len);
 317
 318         decompressed = hdr.size - body_len;
 319         if ((error = git_zstream_get_output(body + body_len, &decompressed, &zstream)) < 0)
 320                 goto done;
 321
 322         if (!git_zstream_done(&zstream)) {
 323                 git_error_set(GIT_ERROR_ZLIB, "failed to finish zlib inflation: stream aborted prematurely");
 324                 error = -1;
 325                 goto done;
 326         }
 327
 328         body[hdr.size] = '\0';
 329
 330         out->data = body;
 331         out->len = hdr.size;
 332         out->type = hdr.type;
 333
 334 done:
 335         if (error < 0)
 336                 git__free(body);
 337
 338         git_zstream_free(&zstream);
 339         return error;
 340 }
 341
 342 static int read_loose(git_rawobj *out, git_buf *loc)
 343 {
 344         int error;
 345         git_buf obj = GIT_BUF_INIT;
 346
 347         assert(out && loc);
 348
 349         if (git_buf_oom(loc))
 350                 return -1;
 351
 352         out->data = NULL;
 353         out->len = 0;
 354         out->type = GIT_OBJECT_INVALID;
 355
 356         if ((error = git_futils_readbuffer(&obj, loc->ptr)) < 0)
 357                 goto done;
 358
 359         if (!is_zlib_compressed_data((unsigned char *)obj.ptr, obj.size))
 360                 error = read_loose_packlike(out, &obj);
 361         else
 362                 error = read_loose_standard(out, &obj);
 363
 364 done:
 365         git_buf_dispose(&obj);
 366         return error;
 367 }
 368
 369 static int read_header_loose_packlike(
 370         git_rawobj *out, const unsigned char *data, size_t len)
 371 {
 372         obj_hdr hdr;
 373         size_t header_len;
 374         int error;
 375
 376         if ((error = parse_header_packlike(&hdr, &header_len, data, len)) < 0)
 377                 return error;
 378
 379         out->len = hdr.size;
 380         out->type = hdr.type;
 381
 382         return error;
 383 }
 384
 385 static int read_header_loose_standard(
 386         git_rawobj *out, const unsigned char *data, size_t len)
 387 {
 388         git_zstream zs = GIT_ZSTREAM_INIT;
 389         obj_hdr hdr;
 390         unsigned char inflated[MAX_HEADER_LEN];
 391         size_t header_len, inflated_len = sizeof(inflated);
 392         int error;
 393
 394         if ((error = git_zstream_init(&zs, GIT_ZSTREAM_INFLATE)) < 0 ||
 395                 (error = git_zstream_set_input(&zs, data, len)) < 0 ||
 396                 (error = git_zstream_get_output_chunk(inflated, &inflated_len, &zs)) < 0 ||
 397                 (error = parse_header(&hdr, &header_len, inflated, inflated_len)) < 0)
 398                 goto done;
 399
 400         out->len = hdr.size;
 401         out->type = hdr.type;
 402
 403 done:
 404         git_zstream_free(&zs);
 405         return error;
 406 }
 407
 408 static int read_header_loose(git_rawobj *out, git_buf *loc)
 409 {
 410         unsigned char obj[1024];
 411         int fd, obj_len, error;
 412
 413         assert(out && loc);
 414
 415         if (git_buf_oom(loc))
 416                 return -1;
 417
 418         out->data = NULL;
 419
 420         if ((error = fd = git_futils_open_ro(loc->ptr)) < 0 ||
 421                 (error = obj_len = p_read(fd, obj, sizeof(obj))) < 0)
 422                 goto done;
 423
 424         if (!is_zlib_compressed_data(obj, (size_t)obj_len))
 425                 error = read_header_loose_packlike(out, obj, (size_t)obj_len);
 426         else
 427                 error = read_header_loose_standard(out, obj, (size_t)obj_len);
 428
 429         if (!error && !git_object_typeisloose(out->type)) {
 430                 git_error_set(GIT_ERROR_ZLIB, "failed to read loose object header");
 431                 error = -1;
 432                 goto done;
 433         }
 434
 435 done:
 436         if (fd >= 0)
 437                 p_close(fd);
 438         return error;
 439 }
 440
 441 static int locate_object(
 442         git_buf *object_location,
 443         loose_backend *backend,
 444         const git_oid *oid)
 445 {
 446         int error = object_file_name(object_location, backend, oid);
 447
 448         if (!error && !git_path_exists(object_location->ptr))
 449                 return GIT_ENOTFOUND;
 450
 451         return error;
 452 }
 453
 454 /* Explore an entry of a directory and see if it matches a short oid */
 455 static int fn_locate_object_short_oid(void *state, git_buf *pathbuf) {
 456         loose_locate_object_state *sstate = (loose_locate_object_state *)state;
 457
 458         if (git_buf_len(pathbuf) - sstate->dir_len != GIT_OID_HEXSZ - 2) {
 459                 /* Entry cannot be an object. Continue to next entry */
 460                 return 0;
 461         }
 462
 463         if (git_path_isdir(pathbuf->ptr) == false) {
 464                 /* We are already in the directory matching the 2 first hex characters,
 465                  * compare the first ncmp characters of the oids */
 466                 if (!memcmp(sstate->short_oid + 2,
 467                         (unsigned char *)pathbuf->ptr + sstate->dir_len,
 468                         sstate->short_oid_len - 2)) {
 469
 470                         if (!sstate->found) {
 471                                 sstate->res_oid[0] = sstate->short_oid[0];
 472                                 sstate->res_oid[1] = sstate->short_oid[1];
 473                                 memcpy(sstate->res_oid+2, pathbuf->ptr+sstate->dir_len, GIT_OID_HEXSZ-2);
 474                         }
 475                         sstate->found++;
 476                 }
 477         }
 478
 479         if (sstate->found > 1)
 480                 return GIT_EAMBIGUOUS;
 481
 482         return 0;
 483 }
 484
 485 /* Locate an object matching a given short oid */
 486 static int locate_object_short_oid(
 487         git_buf *object_location,
 488         git_oid *res_oid,
 489         loose_backend *backend,
 490         const git_oid *short_oid,
 491         size_t len)
 492 {
 493         char *objects_dir = backend->objects_dir;
 494         size_t dir_len = strlen(objects_dir), alloc_len;
 495         loose_locate_object_state state;
 496         int error;
 497
 498         /* prealloc memory for OBJ_DIR/xx/xx..38x..xx */
 499         GIT_ERROR_CHECK_ALLOC_ADD(&alloc_len, dir_len, GIT_OID_HEXSZ);
 500         GIT_ERROR_CHECK_ALLOC_ADD(&alloc_len, alloc_len, 3);
 501         if (git_buf_grow(object_location, alloc_len) < 0)
 502                 return -1;
 503
 504         git_buf_set(object_location, objects_dir, dir_len);
 505         git_path_to_dir(object_location);
 506
 507         /* save adjusted position at end of dir so it can be restored later */
 508         dir_len = git_buf_len(object_location);
 509
 510         /* Convert raw oid to hex formatted oid */
 511         git_oid_fmt((char *)state.short_oid, short_oid);
 512
 513         /* Explore OBJ_DIR/xx/ where xx is the beginning of hex formatted short oid */
 514         if (git_buf_put(object_location, (char *)state.short_oid, 3) < 0)
 515                 return -1;
 516         object_location->ptr[object_location->size - 1] = '/';
 517
 518         /* Check that directory exists */
 519         if (git_path_isdir(object_location->ptr) == false)
 520                 return git_odb__error_notfound("no matching loose object for prefix",
 521                         short_oid, len);
 522
 523         state.dir_len = git_buf_len(object_location);
 524         state.short_oid_len = len;
 525         state.found = 0;
 526
 527         /* Explore directory to find a unique object matching short_oid */
 528         error = git_path_direach(
 529                 object_location, 0, fn_locate_object_short_oid, &state);
 530         if (error < 0 && error != GIT_EAMBIGUOUS)
 531                 return error;
 532
 533         if (!state.found)
 534                 return git_odb__error_notfound("no matching loose object for prefix",
 535                         short_oid, len);
 536
 537         if (state.found > 1)
 538                 return git_odb__error_ambiguous("multiple matches in loose objects");
 539
 540         /* Convert obtained hex formatted oid to raw */
 541         error = git_oid_fromstr(res_oid, (char *)state.res_oid);
 542         if (error)
 543                 return error;
 544
 545         /* Update the location according to the oid obtained */
 546         GIT_ERROR_CHECK_ALLOC_ADD(&alloc_len, dir_len, GIT_OID_HEXSZ);
 547         GIT_ERROR_CHECK_ALLOC_ADD(&alloc_len, alloc_len, 2);
 548
 549         git_buf_truncate(object_location, dir_len);
 550         if (git_buf_grow(object_location, alloc_len) < 0)
 551                 return -1;
 552
 553         git_oid_pathfmt(object_location->ptr + dir_len, res_oid);
 554
 555         object_location->size += GIT_OID_HEXSZ + 1;
 556         object_location->ptr[object_location->size] = '\0';
 557
 558         return 0;
 559 }
 560
 561
 562
 563
 564
 565
 566
 567
 568
 569 /***********************************************************
 570  *
 571  * LOOSE BACKEND PUBLIC API
 572  *
 573  * Implement the git_odb_backend API calls
 574  *
 575  ***********************************************************/
 576
 577 static int loose_backend__read_header(size_t *len_p, git_object_t *type_p, git_odb_backend *backend, const git_oid *oid)
 578 {
 579         git_buf object_path = GIT_BUF_INIT;
 580         git_rawobj raw;
 581         int error;
 582
 583         assert(backend && oid);
 584
 585         raw.len = 0;
 586         raw.type = GIT_OBJECT_INVALID;
 587
 588         if (locate_object(&object_path, (loose_backend *)backend, oid) < 0) {
 589                 error = git_odb__error_notfound("no matching loose object",
 590                         oid, GIT_OID_HEXSZ);
 591         } else if ((error = read_header_loose(&raw, &object_path)) == 0) {
 592                 *len_p = raw.len;
 593                 *type_p = raw.type;
 594         }
 595
 596         git_buf_dispose(&object_path);
 597
 598         return error;
 599 }
 600
 601 static int loose_backend__read(void **buffer_p, size_t *len_p, git_object_t *type_p, git_odb_backend *backend, const git_oid *oid)
 602 {
 603         git_buf object_path = GIT_BUF_INIT;
 604         git_rawobj raw;
 605         int error = 0;
 606
 607         assert(backend && oid);
 608
 609         if (locate_object(&object_path, (loose_backend *)backend, oid) < 0) {
 610                 error = git_odb__error_notfound("no matching loose object",
 611                         oid, GIT_OID_HEXSZ);
 612         } else if ((error = read_loose(&raw, &object_path)) == 0) {
 613                 *buffer_p = raw.data;
 614                 *len_p = raw.len;
 615                 *type_p = raw.type;
 616         }
 617
 618         git_buf_dispose(&object_path);
 619
 620         return error;
 621 }
 622
 623 static int loose_backend__read_prefix(
 624         git_oid *out_oid,
 625         void **buffer_p,
 626         size_t *len_p,
 627         git_object_t *type_p,
 628         git_odb_backend *backend,
 629         const git_oid *short_oid,
 630         size_t len)
 631 {
 632         int error = 0;
 633
 634         assert(len >= GIT_OID_MINPREFIXLEN && len <= GIT_OID_HEXSZ);
 635
 636         if (len == GIT_OID_HEXSZ) {
 637                 /* We can fall back to regular read method */
 638                 error = loose_backend__read(buffer_p, len_p, type_p, backend, short_oid);
 639                 if (!error)
 640                         git_oid_cpy(out_oid, short_oid);
 641         } else {
 642                 git_buf object_path = GIT_BUF_INIT;
 643                 git_rawobj raw;
 644
 645                 assert(backend && short_oid);
 646
 647                 if ((error = locate_object_short_oid(&object_path, out_oid,
 648                                 (loose_backend *)backend, short_oid, len)) == 0 &&
 649                         (error = read_loose(&raw, &object_path)) == 0)
 650                 {
 651                         *buffer_p = raw.data;
 652                         *len_p = raw.len;
 653                         *type_p = raw.type;
 654                 }
 655
 656                 git_buf_dispose(&object_path);
 657         }
 658
 659         return error;
 660 }
 661
 662 static int loose_backend__exists(git_odb_backend *backend, const git_oid *oid)
 663 {
 664         git_buf object_path = GIT_BUF_INIT;
 665         int error;
 666
 667         assert(backend && oid);
 668
 669         error = locate_object(&object_path, (loose_backend *)backend, oid);
 670
 671         git_buf_dispose(&object_path);
 672
 673         return !error;
 674 }
 675
 676 static int loose_backend__exists_prefix(
 677         git_oid *out, git_odb_backend *backend, const git_oid *short_id, size_t len)
 678 {
 679         git_buf object_path = GIT_BUF_INIT;
 680         int error;
 681
 682         assert(backend && out && short_id && len >= GIT_OID_MINPREFIXLEN);
 683
 684         error = locate_object_short_oid(
 685                 &object_path, out, (loose_backend *)backend, short_id, len);
 686
 687         git_buf_dispose(&object_path);
 688
 689         return error;
 690 }
 691
 692 struct foreach_state {
 693         size_t dir_len;
 694         git_odb_foreach_cb cb;
 695         void *data;
 696 };
 697
 698 GIT_INLINE(int) filename_to_oid(git_oid *oid, const char *ptr)
 699 {
 700         int v, i = 0;
 701         if (strlen(ptr) != GIT_OID_HEXSZ+1)
 702                 return -1;
 703
 704         if (ptr[2] != '/') {
 705                 return -1;
 706         }
 707
 708         v = (git__fromhex(ptr[i]) << 4) | git__fromhex(ptr[i+1]);
 709         if (v < 0)
 710                 return -1;
 711
 712         oid->id[0] = (unsigned char) v;
 713
 714         ptr += 3;
 715         for (i = 0; i < 38; i += 2) {
 716                 v = (git__fromhex(ptr[i]) << 4) | git__fromhex(ptr[i + 1]);
 717                 if (v < 0)
 718                         return -1;
 719
 720                 oid->id[1 + i/2] = (unsigned char) v;
 721         }
 722
 723         return 0;
 724 }
 725
 726 static int foreach_object_dir_cb(void *_state, git_buf *path)
 727 {
 728         git_oid oid;
 729         struct foreach_state *state = (struct foreach_state *) _state;
 730
 731         if (filename_to_oid(&oid, path->ptr + state->dir_len) < 0)
 732                 return 0;
 733
 734         return git_error_set_after_callback_function(
 735                 state->cb(&oid, state->data), "git_odb_foreach");
 736 }
 737
 738 static int foreach_cb(void *_state, git_buf *path)
 739 {
 740         struct foreach_state *state = (struct foreach_state *) _state;
 741
 742         /* non-dir is some stray file, ignore it */
 743         if (!git_path_isdir(git_buf_cstr(path)))
 744                 return 0;
 745
 746         return git_path_direach(path, 0, foreach_object_dir_cb, state);
 747 }
 748
 749 static int loose_backend__foreach(git_odb_backend *_backend, git_odb_foreach_cb cb, void *data)
 750 {
 751         char *objects_dir;
 752         int error;
 753         git_buf buf = GIT_BUF_INIT;
 754         struct foreach_state state;
 755         loose_backend *backend = (loose_backend *) _backend;
 756
 757         assert(backend && cb);
 758
 759         objects_dir = backend->objects_dir;
 760
 761         git_buf_sets(&buf, objects_dir);
 762         git_path_to_dir(&buf);
 763         if (git_buf_oom(&buf))
 764                 return -1;
 765
 766         memset(&state, 0, sizeof(state));
 767         state.cb = cb;
 768         state.data = data;
 769         state.dir_len = git_buf_len(&buf);
 770
 771         error = git_path_direach(&buf, 0, foreach_cb, &state);
 772
 773         git_buf_dispose(&buf);
 774
 775         return error;
 776 }
 777
 778 static int loose_backend__writestream_finalize(git_odb_stream *_stream, const git_oid *oid)
 779 {
 780         loose_writestream *stream = (loose_writestream *)_stream;
 781         loose_backend *backend = (loose_backend *)_stream->backend;
 782         git_buf final_path = GIT_BUF_INIT;
 783         int error = 0;
 784
 785         if (object_file_name(&final_path, backend, oid) < 0 ||
 786                 object_mkdir(&final_path, backend) < 0)
 787                 error = -1;
 788         else
 789                 error = git_filebuf_commit_at(
 790                         &stream->fbuf, final_path.ptr);
 791
 792         git_buf_dispose(&final_path);
 793
 794         return error;
 795 }
 796
 797 static int loose_backend__writestream_write(git_odb_stream *_stream, const char *data, size_t len)
 798 {
 799         loose_writestream *stream = (loose_writestream *)_stream;
 800         return git_filebuf_write(&stream->fbuf, data, len);
 801 }
 802
 803 static void loose_backend__writestream_free(git_odb_stream *_stream)
 804 {
 805         loose_writestream *stream = (loose_writestream *)_stream;
 806
 807         git_filebuf_cleanup(&stream->fbuf);
 808         git__free(stream);
 809 }
 810
 811 static int filebuf_flags(loose_backend *backend)
 812 {
 813         int flags = GIT_FILEBUF_TEMPORARY |
 814                 (backend->object_zlib_level << GIT_FILEBUF_DEFLATE_SHIFT);
 815
 816         if (backend->fsync_object_files || git_repository__fsync_gitdir)
 817                 flags |= GIT_FILEBUF_FSYNC;
 818
 819         return flags;
 820 }
 821
 822 static int loose_backend__writestream(git_odb_stream **stream_out, git_odb_backend *_backend, git_off_t length, git_object_t type)
 823 {
 824         loose_backend *backend;
 825         loose_writestream *stream = NULL;
 826         char hdr[MAX_HEADER_LEN];
 827         git_buf tmp_path = GIT_BUF_INIT;
 828         size_t hdrlen;
 829         int error;
 830
 831         assert(_backend && length >= 0);
 832
 833         backend = (loose_backend *)_backend;
 834         *stream_out = NULL;
 835
 836         if ((error = git_odb__format_object_header(&hdrlen,
 837                 hdr, sizeof(hdr), length, type)) < 0)
 838                 return error;
 839
 840         stream = git__calloc(1, sizeof(loose_writestream));
 841         GIT_ERROR_CHECK_ALLOC(stream);
 842
 843         stream->stream.backend = _backend;
 844         stream->stream.read = NULL; /* read only */
 845         stream->stream.write = &loose_backend__writestream_write;
 846         stream->stream.finalize_write = &loose_backend__writestream_finalize;
 847         stream->stream.free = &loose_backend__writestream_free;
 848         stream->stream.mode = GIT_STREAM_WRONLY;
 849
 850         if (git_buf_joinpath(&tmp_path, backend->objects_dir, "tmp_object") < 0 ||
 851                 git_filebuf_open(&stream->fbuf, tmp_path.ptr, filebuf_flags(backend),
 852                         backend->object_file_mode) < 0 ||
 853                 stream->stream.write((git_odb_stream *)stream, hdr, hdrlen) < 0)
 854         {
 855                 git_filebuf_cleanup(&stream->fbuf);
 856                 git__free(stream);
 857                 stream = NULL;
 858         }
 859         git_buf_dispose(&tmp_path);
 860         *stream_out = (git_odb_stream *)stream;
 861
 862         return !stream ? -1 : 0;
 863 }
 864
 865 static int loose_backend__readstream_read(
 866         git_odb_stream *_stream,
 867         char *buffer,
 868         size_t buffer_len)
 869 {
 870         loose_readstream *stream = (loose_readstream *)_stream;
 871         size_t start_remain = stream->start_len - stream->start_read;
 872         int total = 0, error;
 873
 874         /*
 875          * if we read more than just the header in the initial read, play
 876          * that back for the caller.
 877          */
 878         if (start_remain && buffer_len) {
 879                 size_t chunk = min(start_remain, buffer_len);
 880                 memcpy(buffer, stream->start + stream->start_read, chunk);
 881
 882                 buffer += chunk;
 883                 stream->start_read += chunk;
 884
 885                 total += chunk;
 886                 buffer_len -= chunk;
 887         }
 888
 889         if (buffer_len) {
 890                 size_t chunk = min(buffer_len, INT_MAX);
 891
 892                 if ((error = git_zstream_get_output(buffer, &chunk, &stream->zstream)) < 0)
 893                         return error;
 894
 895                 total += chunk;
 896         }
 897
 898         return total;
 899 }
 900
 901 static void loose_backend__readstream_free(git_odb_stream *_stream)
 902 {
 903         loose_readstream *stream = (loose_readstream *)_stream;
 904
 905         git_futils_mmap_free(&stream->map);
 906         git_zstream_free(&stream->zstream);
 907         git__free(stream);
 908 }
 909
 910 static int loose_backend__readstream_packlike(
 911         obj_hdr *hdr,
 912         loose_readstream *stream)
 913 {
 914         const unsigned char *data;
 915         size_t data_len, head_len;
 916         int error;
 917
 918         data = stream->map.data;
 919         data_len = stream->map.len;
 920
 921         /*
 922          * read the object header, which is an (uncompressed)
 923          * binary encoding of the object type and size.
 924          */
 925         if ((error = parse_header_packlike(hdr, &head_len, data, data_len)) < 0)
 926                 return error;
 927
 928         if (!git_object_typeisloose(hdr->type)) {
 929                 git_error_set(GIT_ERROR_ODB, "failed to inflate loose object");
 930                 return -1;
 931         }
 932
 933         return git_zstream_set_input(&stream->zstream,
 934                 data + head_len, data_len - head_len);
 935 }
 936
 937 static int loose_backend__readstream_standard(
 938         obj_hdr *hdr,
 939         loose_readstream *stream)
 940 {
 941         unsigned char head[MAX_HEADER_LEN];
 942         size_t init, head_len;
 943         int error;
 944
 945         if ((error = git_zstream_set_input(&stream->zstream,
 946                         stream->map.data, stream->map.len)) < 0)
 947                 return error;
 948
 949         init = sizeof(head);
 950
 951         /*
 952          * inflate the initial part of the compressed buffer in order to
 953          * parse the header; read the largest header possible, then store
 954          * it in the `start` field of the stream object.
 955          */
 956         if ((error = git_zstream_get_output(head, &init, &stream->zstream)) < 0 ||
 957                 (error = parse_header(hdr, &head_len, head, init)) < 0)
 958                 return error;
 959
 960         if (!git_object_typeisloose(hdr->type)) {
 961                 git_error_set(GIT_ERROR_ODB, "failed to inflate disk object");
 962                 return -1;
 963         }
 964
 965         if (init > head_len) {
 966                 stream->start_len = init - head_len;
 967                 memcpy(stream->start, head + head_len, init - head_len);
 968         }
 969
 970         return 0;
 971 }
 972
 973 static int loose_backend__readstream(
 974         git_odb_stream **stream_out,
 975         size_t *len_out,
 976         git_object_t *type_out,
 977         git_odb_backend *_backend,
 978         const git_oid *oid)
 979 {
 980         loose_backend *backend;
 981         loose_readstream *stream = NULL;
 982         git_hash_ctx *hash_ctx = NULL;
 983         git_buf object_path = GIT_BUF_INIT;
 984         obj_hdr hdr;
 985         int error = 0;
 986
 987         assert(stream_out && len_out && type_out && _backend && oid);
 988
 989         backend = (loose_backend *)_backend;
 990         *stream_out = NULL;
 991         *len_out = 0;
 992         *type_out = GIT_OBJECT_INVALID;
 993
 994         if (locate_object(&object_path, backend, oid) < 0) {
 995                 error = git_odb__error_notfound("no matching loose object",
 996                         oid, GIT_OID_HEXSZ);
 997                 goto done;
 998         }
 999
1000         stream = git__calloc(1, sizeof(loose_readstream));
1001         GIT_ERROR_CHECK_ALLOC(stream);
1002
1003         hash_ctx = git__malloc(sizeof(git_hash_ctx));
1004         GIT_ERROR_CHECK_ALLOC(hash_ctx);
1005
1006         if ((error = git_hash_ctx_init(hash_ctx)) < 0 ||
1007                 (error = git_futils_mmap_ro_file(&stream->map, object_path.ptr)) < 0 ||
1008                 (error = git_zstream_init(&stream->zstream, GIT_ZSTREAM_INFLATE)) < 0)
1009                 goto done;
1010
1011         /* check for a packlike loose object */
1012         if (!is_zlib_compressed_data(stream->map.data, stream->map.len))
1013                 error = loose_backend__readstream_packlike(&hdr, stream);
1014         else
1015                 error = loose_backend__readstream_standard(&hdr, stream);
1016
1017         if (error < 0)
1018                 goto done;
1019
1020         stream->stream.backend = _backend;
1021         stream->stream.hash_ctx = hash_ctx;
1022         stream->stream.read = &loose_backend__readstream_read;
1023         stream->stream.free = &loose_backend__readstream_free;
1024
1025         *stream_out = (git_odb_stream *)stream;
1026         *len_out = hdr.size;
1027         *type_out = hdr.type;
1028
1029 done:
1030         if (error < 0) {
1031                 if (stream) {
1032                         git_futils_mmap_free(&stream->map);
1033                         git_zstream_free(&stream->zstream);
1034                         git__free(stream);
1035                 }
1036                 if (hash_ctx) {
1037                         git_hash_ctx_cleanup(hash_ctx);
1038                         git__free(hash_ctx);
1039                 }
1040         }
1041
1042         git_buf_dispose(&object_path);
1043         return error;
1044 }
1045
1046 static int loose_backend__write(git_odb_backend *_backend, const git_oid *oid, const void *data, size_t len, git_object_t type)
1047 {
1048         int error = 0;
1049         git_buf final_path = GIT_BUF_INIT;
1050         char header[MAX_HEADER_LEN];
1051         size_t header_len;
1052         git_filebuf fbuf = GIT_FILEBUF_INIT;
1053         loose_backend *backend;
1054
1055         backend = (loose_backend *)_backend;
1056
1057         /* prepare the header for the file */
1058         if ((error = git_odb__format_object_header(&header_len,
1059                 header, sizeof(header), len, type)) < 0)
1060                 goto cleanup;
1061
1062         if (git_buf_joinpath(&final_path, backend->objects_dir, "tmp_object") < 0 ||
1063                 git_filebuf_open(&fbuf, final_path.ptr, filebuf_flags(backend),
1064                         backend->object_file_mode) < 0)
1065         {
1066                 error = -1;
1067                 goto cleanup;
1068         }
1069
1070         git_filebuf_write(&fbuf, header, header_len);
1071         git_filebuf_write(&fbuf, data, len);
1072
1073         if (object_file_name(&final_path, backend, oid) < 0 ||
1074                 object_mkdir(&final_path, backend) < 0 ||
1075                 git_filebuf_commit_at(&fbuf, final_path.ptr) < 0)
1076                 error = -1;
1077
1078 cleanup:
1079         if (error < 0)
1080                 git_filebuf_cleanup(&fbuf);
1081         git_buf_dispose(&final_path);
1082         return error;
1083 }
1084
1085 static int loose_backend__freshen(
1086         git_odb_backend *_backend,
1087         const git_oid *oid)
1088 {
1089         loose_backend *backend = (loose_backend *)_backend;
1090         git_buf path = GIT_BUF_INIT;
1091         int error;
1092
1093         if (object_file_name(&path, backend, oid) < 0)
1094                 return -1;
1095
1096         error = git_futils_touch(path.ptr, NULL);
1097         git_buf_dispose(&path);
1098
1099         return error;
1100 }
1101
1102 static void loose_backend__free(git_odb_backend *_backend)
1103 {
1104         loose_backend *backend;
1105         assert(_backend);
1106         backend = (loose_backend *)_backend;
1107
1108         git__free(backend);
1109 }
1110
1111 int git_odb_backend_loose(
1112         git_odb_backend **backend_out,
1113         const char *objects_dir,
1114         int compression_level,
1115         int do_fsync,
1116         unsigned int dir_mode,
1117         unsigned int file_mode)
1118 {
1119         loose_backend *backend;
1120         size_t objects_dirlen, alloclen;
1121
1122         assert(backend_out && objects_dir);
1123
1124         objects_dirlen = strlen(objects_dir);
1125
1126         GIT_ERROR_CHECK_ALLOC_ADD(&alloclen, sizeof(loose_backend), objects_dirlen);
1127         GIT_ERROR_CHECK_ALLOC_ADD(&alloclen, alloclen, 2);
1128         backend = git__calloc(1, alloclen);
1129         GIT_ERROR_CHECK_ALLOC(backend);
1130
1131         backend->parent.version = GIT_ODB_BACKEND_VERSION;
1132         backend->objects_dirlen = objects_dirlen;
1133         memcpy(backend->objects_dir, objects_dir, objects_dirlen);
1134         if (backend->objects_dir[backend->objects_dirlen - 1] != '/')
1135                 backend->objects_dir[backend->objects_dirlen++] = '/';
1136
1137         if (compression_level < 0)
1138                 compression_level = Z_BEST_SPEED;
1139
1140         if (dir_mode == 0)
1141                 dir_mode = GIT_OBJECT_DIR_MODE;
1142
1143         if (file_mode == 0)
1144                 file_mode = GIT_OBJECT_FILE_MODE;
1145
1146         backend->object_zlib_level = compression_level;
1147         backend->fsync_object_files = do_fsync;
1148         backend->object_dir_mode = dir_mode;
1149         backend->object_file_mode = file_mode;
1150
1151         backend->parent.read = &loose_backend__read;
1152         backend->parent.write = &loose_backend__write;
1153         backend->parent.read_prefix = &loose_backend__read_prefix;
1154         backend->parent.read_header = &loose_backend__read_header;
1155         backend->parent.writestream = &loose_backend__writestream;
1156         backend->parent.readstream = &loose_backend__readstream;
1157         backend->parent.exists = &loose_backend__exists;
1158         backend->parent.exists_prefix = &loose_backend__exists_prefix;
1159         backend->parent.foreach = &loose_backend__foreach;
1160         backend->parent.freshen = &loose_backend__freshen;
1161         backend->parent.free = &loose_backend__free;
1162
1163         *backend_out = (git_odb_backend *)backend;
1164         return 0;
1165 }